No title

Current Topics in Membranes and Transport VOLUME 24

Membrane Protein Biosynthesis and Turnover

Advisory Board

M . P. Blaustein G. Blobel J . S. Cook P. A. Knauf

Sir H . L. Kornberg C. A. Pasternak W . D. Stein W. Stoeckenius K. J . Ullrich Contributors

Beth A. Rasmussen Vytas A. Bankaitis Enzo Bard Graeme A. Reid Philip J . Bassford, Jr. Enrique RodriguezSamuel W . Cushman Boulan J . Patrick Ryan Guy D. Duffaud Pedro J . 1. Salas Masayori Inouye Annelise 0. Jorgensen Yves-Jacques Schneider Sharon S. Krag tan A. Simpson Susan K. Lehnhardt David H . MacLennan Andre' Trouet Dora Vega de Salas Paul E. March Gunnar von Heijne David E. Misek Jean-Noel Octave Martin Wiedmann H . Steven Wiley Tom A. Rapoport Elizabeth Zubrzycka-Gaarn

C u rrent Topics in Membranes and Transport Edited by

Felix Bronner Department of Oral Biology University of Connecticut Health Center Farmington. Connecticut

VOLUME 24

Membrane Protein Biosynthesis and Turnover Guest Editors Philip A. Knauf

John S. Cook

Department of Radiation Biology and Biophysics The Universiry of Rochester Medical Cenier Rochester, New York

Biology Division Oak Ridge National Laboraioiy Oak Ridge, Tennessee

1985

m

ACADEMIC PRESS, INC. (Harcourr Brace Jovunovich. Publishers)

Orlando San Diego New York London Toronto Montreal Sydney Tokyo

COPYRIGHT o 1985, BY ACADEMIC PRESS, INC. ALL RIGHTS RESERVED. NO PART OF THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM OR BY ANY MEANS, ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR ANY INFORMATION STORAGE AND RETRIEVAL SYSTEM, WITHOUT PERMISSION IN WRITING FROM THE PUBLISHER.

ACADEMIC PRESS, INC. Orlando, Florida 32887

United Kingdom Edition published by ACADEMIC PRESS INC. (LONDON) LTD. 24-28 Oval Road, London N W l 7DX

LIBRARY OF CONGRESS CATALOG CARD NUMBER: 70-1 17091

ISBN: 0-12-153324-7 PRINTED IN THE UNITED STATES OF AMERICA 85 86 87 RX

Y X 7 6 5 4 .i 2 1

Contents Contributors, ix Preface, xi Yale Membrane Transport Processes Volumes, xiii

CHAPTER

1.

Application of the Signal Hypothesis to the Incorporation of Integral Membrane Proteins TOM A. RAPOPORT AND MARTIN WIEDMANN

1. 11. 111. IV.

Introduction, 1 Translocation of Secretory Proteins across Membranes, 5 Incorporation of Proteins into Membranes, 29 Perspectives, 46 References, 47

CHAPTER

2.

Structure and Function of the Signal Peptide GUY D. DUFFAUD, SUSAN K. LEHNHARDT, PAUL E. MARCH, AND MASAYORI INOUYE

I. 11. 111. IV. V. VI.

Introduction, 65 Signal Peptide, 66 Models for Protein Secretion, 75 Experimental Approaches, 81 Components Interacting with the Signal Peptide in Bacteria, 95 Conclusion, 97 References, 98

CHAPTER

3.

The Use of Genetic Techniques to Analyze Protein Export in Escherichia coli VYTAS A. BANKAITIS, J. PATRICK RYAN, BETH A. RASMUSSEN, AND PHILIP J. BASSFORD, JR

I. Introduction, 105 The Use of Gene Fusions to Study Protein Export in E. coli. 107

11.

v

vi

CONTENTS

111. Intragenic Information Specifying Protein Export, 124 IV. Components of the E. coli Protein Export Machinery, 135 V. Summary, 145 References, 146

CHAPTER

4.

Structural and Thermodynamic Aspects of the Transfer of Proteins into and across Membranes GUNNAR VON HEIJNE

I. 11. 111. 1V. V. VI.

Introduction, 151 How Hydrophobic Is Hydrophobic?, 152 The Signal Sequence: A Sequence of Signals, 157 The Transmembrane Segment: Making Friends with Lipids, 163 Protein Export: Rules of the Game, 169 Conclusion, 173 References, 174

CHAPTER

5.

Mechanisms and Functional Role of Glycosylation In Membrane Protein Synthesis SHARON S. KRAG

I. 11. 111. IV.

Introduction, 181 Involvement of Oligosaccharide Lipid Intermediates, 185 Role of the Carbohydrate Moiety, 226 Summary, 232 References, 233

CHAPTER

6.

Protein Sorting in the Secretory Pathway ENRIQUE RODRIGUEZ-BOULAN, DAVID E. MISEK, DORA VEGA DE SALAS, PEDRO J. I. SALAS, AND E N 2 0 BARD

I. 11. 111. IV. V.

Introduction, 252 Molecular Sorting: Definitions and Factors Involved, 252 Molecular Sorting in the Secretory Pathway, 259 Model Systems for the Study of Molecular Sorting in Eukaryotic Cells, 268 Summary and Perspectives, 280 References. 280

vii

CONTENTS CHAPTER

7.

Transport of Proteins into Mitochondria GRAEME A. REID

I.

Introduction: Mitochondrial Biogenesis, 295

11. An Overview of Mitochondria1 Protein Import, 297 111. Are Proteins Transported into Mitochondria Cotranslationally or

Posttranslationally?, 31 3 IV. The Molecular Approach, 316 V. Summary, 327 References. 329

CHAPTER

8.

Assembly of the Sarcoplasmic Reticulum during Muscle Development DAVID H. MACLENNAN, ELIZABETH ZUBRZYCKA-GAARN, AND ANNELISE 0. JORGENSEN

1. Introduction, 338 11. The Sarcoplasmic Reticulum, 338 111. Biogenesis of the Sarcoplasmic Reticulum during Muscle Cell

Differentiation, 342 IV. Models of Sarcoplasmic Reticulum Biogenesis, 35 1 V. Synthesis of Specific Sarcoplasmic Reticulum Proteins, 354 VI. Regulation of the Biosynthesis of Sarcoplasmic Reticulum Proteins, 358 References, 362

CHAPTER

9.

Receptors as Models for the Mechanisms of Membrane Protein Turnover and Dynamics H. STEVEN WILEY

I.

Introduction, 369

11. Definition of Terminology, 370 111. Technical Approaches for Analyzing Receptor Behavior and Function, 372

IV. Ligand Binding and Receptor Number, 374 V. Mechanisms and Models of Receptor Internalization, 385 VI. Receptor Biosynthesis and Recycling, 396 VII. Concluding Remarks, 404 References, 405

CONTENTS CHAPTER

10. The Role of Endocytosis and Lysosomes In Cell Physiology YVES-JACQUES SCHNEIDER, JEAN-NOEL OCTAVE, AND ANDRE TROUET

I. Introduction, 413 The Fate of the Plasma Membrane in the Course of Endocytosis, 418 Ill. The Role of Endocytosis and Lysosomes in Transfenin Iron Uptake, 430 IV. The Intracellular Sorting of Ligands Taken up by Receptor-Mediated Endocytosis, 440 V. Summary and Perspectives, 451 References, 452 11.

CHAPTER

1 1 . Regulation of Glucose Transporter and Hormone Receptor Cycllng by Insulin in the Rat Adipose Cell IAN A. SIMPSON AND SAMUEL W. CUSHMAN

I. Introduction, 459

II. Glucose Transport, 460 111. Insulin-Like Growth Factor I1 Binding, 485 IV. Insulin Binding, 488 V. Summary, 495 References, 496

Index, 505 Contents of Recent Volumes, 517

Contributors Numbers in parentheses indicate the pages o n which the authors' contributions begin

Vytas A. Bankaitis,' Department of Microbiology and Immunology, School of Medicine, University of North Carolina, Chapel Hill, North Carolina 27514 (105) Enzo Bard, Department of Pathology, State University of New York, Downstate Medical Center, Brooklyn, New York 11203 (251) Phllip J. Bassford, Jr., Department of Microbiology and Immunology, School of Medicine, University of North Carolina, Chapel Hill, North Carolina 27514 (105) Samuel W. Cushman, Experimental Diabetes, Metabolism and Nutrition Section, Molecular, Cellular and Nutritional Endocrinology Branch, National Institute of Arthritis, Diabetes. and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland 20205 (459) Guy D. Duffaud, Department of Biochemistry, State University of New York at Stony Brook, Stony Brook, New York I1794 (65) Masayori Inouye, Department of Biochemistry, State University of New York at Stony Brook, Stony Brook, New York 11794 (65) Annelise 0. Jorgensen, Department of Anatomy, Medical Sciences Building, University of Toronto, Toronto, Ontario M5S IAS, Canada (337) Sharon S. Krag, Department of Biochemistry, The Johns Hopkins University, School of Hygiene and Public Health, Baltimore, Maryland 21205 (181) Susan K. Lehnhardt, Department of Biochemistry, State University of New York at Stony Brook, Stony Brook, New York 11794 (65) David H. MacLennan, Banting and Best Department of Medical Research, Charles H. Best Institute, University of Toronto, Toronto. Ontario M5G IL6, Canada (337) Paul E. March, Department of Biochemistry, State University of New York at Stony Brook, Stony Brook, New York 11794 (65) David E. Misek, Department of Pathology, State University of New York, Downstate Medical Center, Brooklyn, New York 11203 (251) Jean-Noel Octave, Laboratoire de Chimie Physiologique, International Institute of Cellular and Molecular Pathology, and Universite Catholique de Louvain, B 1200 Brussels, Belgium (413) Tom A. Rapoport, Zentralinstitut fur Molekularbiologie. Akademie der Wissenschaften der DDR, 11 15 Berlin, German Democratic Republic ( 1 ) Beth A. Rasmussen,' Department of Microbiology and Immunology, School of Medicine, University of North Carolina, Chapel Hill, North Carolina 275 14 (105) Graeme A. Reid,3 Department of Biochemistry, Biocenter, University of Basel, CH-4056 Basel, Switzerland (295) 'Present address: Division of Biology, California Institute of Technology, Pasadena, California 91 125.

2Present address: Department of Molecular Biology, Princeton University, Princeton, New Jersey 08544. "resent address: Department of Microbiology, University of Edinburgh, Edinburgh EH9 3JG, Scotland. ix

X

CONTRIBUTORS

Enrlque Rodriguez-Boulan, Department of Cell Biology and Anatomy, Cornell University Medical College, 1300 York Avenue, New York, New York 10021 (251) J. Patrick Ryan, Department of Microbiology and Immunology, School of Medicine, University of North Carolina, Chapel Hill, North Carolina 27514 (105) Pedro J. 1. Salas, Department of Cell Biology and Anatomy, Cornell University Medical College, 1300 York Avenue, New York, New York 10021 (251) VvesJacques Schnelder, Laboratoire de Chimie Physiologique, International Institute of Cellular and Molecular Pathology, and UniversitC Catholique de Louvain, B 1200 Brussels, Belgium (413) Ian A. Slmpson, Experimental Diabetes, Metabolism and Nutrition Section, Molecular, Cellular and Nutritional Endocrinology Branch, National Institute of Arthritis, Diabetes, and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland 20205 (459) Andre Trouet, Laboratoire de Chimie Physiologique, International Institute of Cellular and Molecular Pathology, and UniversitC Catholique de Louvain, B 1200 Brussels, Belgium (413) Dora Vega de Salas, Department of Cell Biology and Anatomy, Cornell University Medical College, 1300 York Avenue, New York, New York 10021 (251) Gunnar von Heljne, Research Group for Theoretical Biophysics, Department of Theoretical Physics, Royal Institute of Technology, S-100 44 Stockholm, Sweden (151) Martln Wledmann, Zentralinstitut fur Molekularbiologie, Akademie der Wissenschaften der DDR, 1115 Berlin, German Democratic Republic (1) H. Steven Wlley, Department of Pathology, University of Utah College of Medicine, Salt Lake City, Utah 84132 (369) Ellmbeth Zubrzycka-Gaarn, Banting and Best Department of Medical Research, Charles H. Best Institute, University of Toronto, Toronto, Ontario M5G 1L6, Canada (337)

Preface The most outstanding recent development in membrane biology has been the realization that not only cell surfaces but also the membranes bounding internal organelles are highly dynamic structures. It has always been self-evident that, as cells grow, so also must new membranes be synthesized, each with the composition appropriate to its function. Only in the last decade, however, has it become widely appreciated that, once synthesized, morphologically or biochemically identifiable membranes are by no means static structures. They are subject to internalization and turnover, with individual components turning over at different rates. They undergo recycling, with one membrane interacting with another of very different properties and yet returning virtually intact. Membrane-bound vesicles from one organelle fuse in very specific ways with other membranebound organelles, yet each organelle retains its identity. All of this intracellular traffic must be regulated by a system of signals which is ultimately encoded in the cell’s genetic material and which is expressed in the amino acid sequences of membrane proteins, in posttranslational modifications of these proteins by specific enzymes, and in specific recognition proteins residing in various organelles. The table of contents of this book is only a partial but representative listing of the vigorous activities in which cell membranes are engaged and of the current approaches toward discovering the nature of these activities and the signals that regulate them. There are, of course, exceptions to this dynamic picture of cell membranes, notably the surface membranes of mature erythrocytes that are incapable of protein synthesis, but even here endocytosis without hemolysis is demonstrable and may play a significant role in such processes as malarial and other parasitic infections. But the inability of the erythrocyte membrane to turn over and in so doing to repair itself may be importantly related to the finite life span of this cell. As with all major conceptual insights in science, the view of membranes as dynamic entities has had two profoundly positive consequences for the field. It has not only deepened our perception of how the real world works, but has also pointed the way to future areas to be explored, for example, the question of how this intracellular traffic is regulated. The chapters in this volume, therefore, serve not as monuments, but as signposts.

PHILIPA. KNAUF JOHN S . COOK xi

This Page Intentionally Left Blank

Yale Membrane Transport Processes Volumes Joseph F. Hoffman (ed). (1978). “Membrane Transport Processes,” Vol. 1. Raven, New York. Daniel C. Tosteson, Yu. A. Ovchinnikov, and Ramon Latorre (eds.). (1978). “Membrane Transport Processes,” Vol. 2. Raven, New York. Charles F. Stevens and Richard W. Tsien (eds.). (1979). “Membrane Transport Processes,” Vol. 3: Ion Permeation through Membrane Channels. Raven, New York. Emile L. Boulpaep (ed.). (1980). “Cellular Mechanisms of Renal Tubular Ion Transport”: Volume 13 of Current Topics in Membranes and Transport (F. Bronner and A. Kleinzeller, eds.). Academic Press, New York. William H. Miller (ed.). (1981). “Molecular Mechanisms of Photoreceptor Transduction”: Volume 15 of Current Topics in Membranes and Transport (F. Bronner and A. Kleinzeller, eds.). Academic Press, New York. Clifford L. Slayman (ed.). (1982). “Electrogenic Ion Pumps”: Volume 16 of Current Topics in Membranes and Transport (A. Kleinzeller and F. Bronner, eds.). Academic Press, New York. Joseph F. Hoffman and Bliss Forbush 111 (eds.). (1983). “Structure, Mechanism, and Function of the Na/K Pump”: Volume 19 of Current Topics in Membranes and Transport (F. Bronner and A. Kleinzeller, eds.). Academic Press, New York. James B. Wade and Simon A. Lewis (eds.). (1984). “Molecular Approaches to Epithelial Transport”: Volume 20 of Current Topics in Membranes and Transport (A. Kleinzeller and F. Bronner, eds.). Academic Press, New York. Edward A. Adelberg and Carolyn W. Slayman (eds.). (1985). “Genes and Membranes: Transport Proteins and Receptors”: Volume 23 of Current Topics in Membranes and Transport (F. Bronner and A. Kleinzeller, eds.). Academic Press, New York.

xiii

This Page Intentionally Left Blank

CURRENT TOPICS IN MEMBRANES AND TRANSPORT. VOLUME 24

Chapter 7

Application of the Signal Hypothesis to the Incorporation of Integral Membrane Proteins TOM A . RAPOPORT /VVD MARTIN WIEDMA” Zenrralinstitut ,fur Molekularbiologie Akademie der Wissenschaften der DDR Berlin, German Democratic Republic

..................

I

........

5

.................

26

D. Uncleaved and Internal Signal Sequences., . . . . . . . . . . . . . . . . . . . IV. Perspectives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.

INTRODUCTION

A. What Are the Problems? Most polypeptides in a eukaryotic cell and all polypeptides in a prokaryotic organism are synthesized in the cytoplasm. Their location, however, is at many 1

Copyright Q 1985 by Acddemic Prerr. Inc All right, of reproducuon in dny form reserved ISBN 0-12-153324-7

2

TOM A. RAPOPORT AND MARTIN WIEDMANN

different sites. Specific sets of proteins are found in cell organelles, and a given polypeptide is generally found in only one cell compartment. How does a protein “know” where it belongs? What are the signals in a polypeptide for the transport from the site of synthesis to its final site? Closely connected with these problems is the question concerning the mechanism of transport. In many cases polypeptides have to traverse at least one membrane which is normally impermeable for such molecules. Thus, there must be mechanisms for the selective and vectorial translocation of polypeptides across membranes. For membrane proteins an additional problem exists. Such polypeptides can have different orientations in the membrane (see Table I). A membrane protein can span the phospholipid bilayer once or several times, it can have its Nterminus on the outside or inside of the cell, or it can just be embedded into the bilayer from either side. A given polypeptide chain has a defined and characteristic orientation in the membrane; and again one may ask how this specific incorporation is coded for and brought about.

B. Signals for Protein Localization Although all membranes contain specific proteins, the faculty of transporting proteins is limited to a few. In an animal cell these include the rough endoplasmic reticulum (RER) membrane and the mitochondria1 membrane (Blobel, 1980). Most likely, the peroxisomal membrane also has such a competence. In a plant cell, the chloroplast membrane exhibits translocation of polypeptides. In bacteria, the cytoplasmic membrane may be the exclusive site of protein translocation (Michaelis and Beckwith, 1982). It follows that all other membranes derive their membrane-spanning polypeptides from those which are translocation competent and may therefore be called receiver membranes. For instance, plasma membrane proteins of a eukaryotic cell are not directly incorporated into their target membrane but are initially found in the RER membrane (see for example, Sabban et al., 1981; Lodish et al., 1981; Palade, 1975). The transport from the membrane which is translocation competent to the receiving one may occur in two ways. The polypeptides may be transported by lateral movement in the plane of the membrane. For example, it is likely that proteins diffuse from the rough ER to the contiguous smooth ER. The second possibility is by budding-fusion processes, a way used by proteins destined for the plasma membrane. Thus, many membrane proteins reach their final localization in two steps: first, incorporation into a translocation competent membrane, and, second, sorting out and transport to the receiver membrane. Presumably, most polypeptides do not change their orientation within the membrane after their initial incorporation into the phospholipid bilayer (see, however, Farquhar et al., 1974; Little and Widnell, 1975).

3

1, APPLICATION OF THE SIGNAL HYPOTHESIS

TABLE I MLMBKANE PROTEINS PHOSPHOLIIW BILAYER

SOME ORIENTATIONS OF INTEGRAL

TO THE

Orientation in the membraneo

WITH

RESPECT

Example

C protein of VSV C

-i-

Neuraminidase of influenza A virus

N

EcJ

Isomaltase-sucrase complex of intestine

('?)h

EN

EC

Cytochrome b5 &

Em N

C

Bacteriorhodopsin

Band 111 of erythrocytes ( ? ) b *

-

EN E

N 1'

EC, Ectoplasmic; EN, endoplasrnic; N , N-terminus; C. C-terminus.

' Structure not completely known.

It follows that a membrane protein needs up to three signals acting in the following order: 1 . A signal directing the polypeptide to a translocation competent membrane 2. Signal(s) for the defined incorporation into the membrane 3. Sorting signal(s) for further transport to the receiver membrane

It is likely that a fourth type of signal is required to keep a polypeptide at a certain site. This can be simply an association with other proteins either in the membrane itself or outside. Of course, two or more of the different signals may coincide.

4

TOM A. RAPOPORT AND MARTIN WIEDMANN

What are these signals? Where are they located? Generally speaking, they must be coded for in the gene for the corresponding polypeptide. It appears that they are only decoded at the protein level. In other words, the polypeptide itself contains the information for its final destination. It should be kept in mind, however, that it is possible that in some cases cell polarity, spatial sequestration of mRNA, or “helper” polypeptides (guiding another protein which does not carry its own signal) may be responsible. In general, however, each polypeptide appears to have its own signal(s). What is the nature of these signals? The simplest possibility is that the signals are segments of the polypeptide chain-so-called topogenic amino acid sequences (Blobel, 1980). In fact, much evidence exists in favor of this idea, at least for the actual process of protein transfer across membranes. However, it is conceivable that some of the signals are only expressed after folding of the polypeptide so that distant amino acid residues constitute a “topogenic site. This may be particularly relevant for sorting signals. One may raise the question whether modifications carried out on the polypeptide chain may provide topogenic information. In fact, for lysosomal proteins the attachment of mannose 6-phosphate residues is one way of sorting (for review, see Hasilik, 1980). Of course, the information for this modification must ultimately be coded for in a specific amino acid sequence. Some membrane proteins are anchored in the membrane also by covalently attached fatty acid chains (Inouye and Halegoua, 1980; Schmidt, 1982). However, for the actual process of translocation of proteins across membranes, modifications are not likely to play a decisive role. Translocation can occur in the absence of modification events, and the latter occur after at least parts of the polypeptide chain have traversed the membrane. This article is mainly concerned with the translocation of polypeptides across the endoplasmic reticulum membrane. The signal hypothesis, which will be central in it, was initially proposed to explain the biosynthesis of secreted polypeptides. It has become increasingly clear that much of the knowledge may be extended to membrane proteins inserted into the RER membrane. Indeed, there exists strong evidence that the basic process of translocation across the membrane is the same for both types of polypeptides (see Section 111,B): the signals for initiation of the process are the same and the constituents of the transport machinery are identical. Even today, most of the details on the mechanism of translocation are studied on secretory proteins. We shall therefore start with a consideration of these polypeptides and shall later turn to membrane proteins. Translocation across the endoplasmic reticulum membrane and that across the cytoplasmic membrane in bacteria have turned out to be very similar. This article concentrates on the eukaryotic system and only deals with the prokaryotic one where comparisons are made or arguments are borrowed. Separate articles in this volume (Chapters 2 and 3) are devoted to details of the prokaryotic system. Several reviews on the same or related subjects have appeared during the last ”

I . APPLICATION OF THE SIGNAL HYPOTHESIS

5

years (Davis and Tai, 1980; E m et a f . , 1980a; Wickner, 1979; Wickner et al., 1980; Lodish et al., 1981; Kreil, 1981; Warren, 1981; Blobel, 1982; Sabatini et a f . , 1982; Michaelis and Beckwith, 1982; Heinrich, 1982).

C. Some Definitions Much confusion is due to insufficient clarity of terms. In order to avoid misunderstandings, we provide some of the definitions which are used in this article. An integral membrane protein is defined operationally as a polypeptide which can only be removed from the membrane by desintegration of the phospholipid bilayer, such as by detergent treatment. If a protein associated with a membrane can be removed by salt washing, pH effects, etc., it is called a peripheral membrane protein. This article deals only with integral membrane proteins. It should be noted that the definition does not imply any mechanistic or structural feature of the polypeptide or of its biosynthesis. Two topological sides of cellular membranes can be distinguished. When one side faces the cytoplasm, it is called endoplasmic and the other side ectoplasmic (Blobel, 1982). Translocation across and insertion into a membrane are used essentially synonymously for membrane proteins, in contrast to others (Blobel, 1980). These terms are distinguished, however, from embedding. The latter expression is limited to situations in which a polypeptide chain or a part of it interacts with the phospholipid bilayer, does not have a folded domain on the other side of a membrane, and does not need a translocation system for its membrane incorporation (see Section III,A and B). This definition does not exclude the possibility that an embedded polypeptide spans the membrane and has some amino acid residues at the ectoplasmic side. fncorporation of a protein into a membrane is used to include both translocated and embedded polypeptides. Proteins may traverse the membrane once or several times (see Table I). In accord with Blobel (1980, 1982) we distinguish between monotopic (those having no part translocated), bitopic (those traversing the membrane once), and polytopic (those spanning the membranes several times) polypeptides.

II. TRANSLOCATION OF SECRETORY PROTEINS ACROSS MEMBRANES

A. The Signal Hypothesis: Short History and Basic Facts It was recognized some 20 years ago that in eukaryotes the actual translocation process for secretory proteins does not involve the plasma membrane but rather

6

TOM A. RAPOPORT AND MARTIN WIEDMANN

an intracellular membrane. Work mainly of the group of Palade (see Palade, 1975) has shown that ribosomes attached to the RER membrane deliver their translation products to the lumen of the ER tubings. By contrast, cytoplasmic proteins are synthesized on free ribosomes. A cotranslational mode of transfer of secretory proteins across the membrane was indicated by the facts that newly synthesized proteins were not found in the cytosol (Redman et al., 1966) and even incomplete polypeptide chains were located within the microsomal membrane (Redman and Sabatini, 1966; Sabatini and Blobel, 1970). Initially, there was the suggestion that polypeptides with different destinations are made by different classes of ribosomes. However, it was shown later that ribosomes circulate between bound and free states (Borgese er al., 1973). Furthermore, both secretory and cytoplasmic proteins are efficiently synthesized in the same cell-free translation systems, and experiments to be described below show that addition of microsomes suffices for translocation of polypeptides (Blobel and Dobberstein, 1975a,b). Whereas the specialization of ribosomes in the synthesis of a certain class of proteins is unlikely, it is possible that factor(s) present in the cell-free translation system are exclusively required for the biosynthesis of secretory proteins. In principle, the signal for ribosome binding to the RER membrane could either lie in the mRNA or in the nascent polypeptide chain. In fact, some papers claimed a direct recognition of the mRNA by the RER membrane (Baglioni et al., 1971; Mechler and Vassalli, 1975). However, no further evidence was presented. Blobel and Sabatini (1971) proposed that the signal is localized in the Nterminus of the growing secretory polypeptide chain. They proposed a model which at that time was based only on circumstantial evidence. The ingenious idea was that all secretory polypeptides should have a common amino acid sequence at their N-terminus which is directing the ribosome to the RER membrane in the manner of a zip code. Binding of the ribosome would open a pore through which the growing peptide chain is translocated to the lumen. Before completion of the polypeptide chain, the signal peptide would be cleaved off. The first experimental support for the existence of a signal peptide was provided by Milstein et al. (1972) for the immunoglobulin light chain. They showed that the product of cell-free translation of mRNA was some 20 amino acid residues longer than the corresponding cellular product. They presumed that the initiation of translation occurred correctly in vitro and showed that the part missing in the mature protein was at the N-terminus. They speculated that this cleaved-off peptide served as a signal to direct the secretory protein across the RER membrane. Swan et qL’(l972) also found that the cell-free synthesized light chain was larger than the cellular product, but an interpretation was not given.

1. APPLICATION OF THE SIGNAL HYPOTHESIS

7

Schechter et a / . (1975) presented the first data on the primary structure of the N-terminal segment. Similar extensions were thereafter found for other secretory proteins synthesized in cell-free systems (see Table 11). By means of radiosequencing, partial or complete amino acid sequences could be deduced for the signal peptides. Even at that time it turned out that, contrary to the original expectation, a common primary structure of the peptides does not exist. The final formulation of the signal hypothesis was given by Blobel and Dobberstein (1975a,b) in two outstanding papers. Their scheme (Fig. 1) was based on the achievement of an in virro reconstitution of the translocation process. Dog pancreatic microsomes were used because of their high content of bound ribosomes and low content of ribonucleases. The following results lead to the scheme in Fig. 1: 1 . In the absence of membranes, the precursor to the immunoglobulin light chain was synthesized. When RER membranes from dog pancreas were present during translation, the mature polypeptide was obtained. Membranes added after translation did not lead to processing of the protein. 2. Translocation of the processed polypeptide across the membrane was indicated by its being protected against added proteases. In the presence of detergent, the polypeptide was degraded. 3. Polysomes isolated from myeloma cells and detached from membranes by detergent treatment were allowed to complete their nascent polypeptides in vitro. After short incubation times the processed light chains were obtained, whereas after longer times the precursor was found. This was consistent with the idea that ribosomes sitting near the 5'-end of the mRNA carry nascent chains not yet processed whereas those closer to the 3'-end carry processed nascent chains. This experiment was the first to suggest that the cleavage of precursors is not an in vitro artifact. It also indicated that cleavage of the signal peptide occurs cotranslationally. Reconstitution of the translation in vitro was also achieved with ascites tumor

FIG.1. Schematic representation of the signal hypothesis. Codons which are adjacent to and after the initiation codon AUG and which code for the signal peptide are indicated by a zigzag region in the rnRNA. The signal sequence is indicated by a dashed line. Proteolytic cleavage is indicated by the presence of signal peptides (short dashed lines) within the intracisternal space. For details see text. ~ . Vol. 67, pp. 835-851, Reproduced from Blobel and Dobberstein, The J o u r n a / o ~ C e / I B i o l o g1975, by copyright permission of The Rockefeller University Press.

TABLE Il SIGNALSEQUENCES OF EUKARYOTIC AND PROKARYOTIC PROTEINS" Protein Eukaryotie secretory proteins Bovine proparathyroid hormone Rat 1 proinsulin Rat 2 proinsulin

0)

carp proinsulin Rat growth honnone Rat prolactin Chicken lysozyme Chicken ovomucoid Chicken conalbumin Dog trypsinogen 1 Dog trypsinogen 2 Leukocyte interferon Fibroblast (FI) interferon Human ul-antitrypsin Honeybee promellitin Bovine proalbumin Ovine casein p Ovine casein a s l Ovine casein K Mouse MOPC-41 myeloma L chain

-30

-20

- 10

-1

1

References

Habener et al. (1978) MALWMRFLPLLALLVLWEKPPW

_F

MALwIRFLFFLALLILWEKPP@

_F

MAVWIQAGALLFLLAVSS-W!J N WSQTPWLLTFSLLCLLWPQEM L MNS~S-ARKGTLLLLMMS~FCQN4IQT T. MRSLLILVLCFLPLAALG K MAMAGV~~FSFVLCGFLPD~

MPSSVSWGILLLAGLCCLVPVSU E M KF L V X V A L V F M W E Y A !J MKWVTnLLLFISGSAFS K MKVLILAXLVALALA R MKLLILTXLVAVALA R MXKXILLWXILALXLPXLIA Q M ~ ~ ~ P A Q I F G F L L L L F P G T RQ ~

Chan et al. (1976, 1979); Villa-Komaroff er al. (1978) Chan et al. (1976, 1979); Villa-Komaroff er al. (1978) Hahn et al. (1983) Seeburg et al. (1977) McKean and Maurer (1978) Thibodeau et al. (1978) Thibodeau er al. (1978) Thibodeau er al. (1978) Devillers-Thiery er al. (1975) Devillers-Thiery et al. (1975) Taniguchi et al. (1980) Taniguchi et al. (1980) Gross et al. (1983) Suchanek et al. (1978) Strauss et al. (1978) Gaye et al. (1977) Gaye et al. (1977) Gaye er al. (1977) Burstein and Schechter (1978)

Mouse MOPC-321 myeloma L chain Mouse Vh L chain

(D

METDTLLLWVLLLWVF'GSTG

Q

Burstein and Schechter (1978)

MAWTSLILSLLALCSGASS

Q

Tonegawa et a/. (1978)

MKCLLYLAFLFIHVNC MKTNSYIFCLVFA -

K

Q

Lingappa et a/. (1978a) Min Jou er al. (1980)

MAIIYLILLFTAVRG --

D

Min Jou et al. (1980)

MNTQILVFALVAVrPTNA

D

Min Jou et al. (1980)

MVLTLLLIICLALEDS

E

Claudio er a/. (1983)

Eukaryotic membrane proteins VSV glycoprotein Influenza A hemagglutinin (AIVictoria/375) Influenza A hemagglutinin (AlJap. 1305157) Influenza A hemagglutinin (AIFPVIRostock/34) Torpedo acetylcholine receptor y subunit Mouse MOPC-104E myeloma L chain H-2 antigen (A H-2Kd)

MAWISLILSXELSSZS --

Q

Burstein and Schechter (1978)

MAFTTLLLLLAAALAF'TQTM

G -

Dobberstein et

Prokaryotic secretory proteins E. coli alkaline phosphatase

MKQSTIAL ALLPLLFTPJTKA

R

MKIKTGARILALSALTTMMFS ASALA -

K

Inouye and Beckwith (1977); Kikuchi e t a / . (1981) Bedouelle et a / . (1980)

D

Oxender et a/.(1980)

E . coli maltose binding pro-

tein E. coli leucine-specific binding protein E. coli leucine-isoleucinevaline binding protein Salmonella ryphimurium histidine binding protein

MKANAKTIIAGMIALAISHT~

(1982)

Michaelis and Beckwith (1982)

MNIKGKALLAGCIA L A F S N W MKKLALSLSLVLAFSS ATAAFA -

a/.

A -

Higgins and Ames (1981)

(continued)

TABLE I1 (continued) Protein

Salmonella typhimurium 1y sine-arginine-omithine binding protein E. coli am@ p-lactamase E. coli TEM p-lactamase E. coli arabinose binding protein

h e r membrane proteins of E . coli Phage fd major coat protein Phage fd minor coat protein Outer membrane proteins of E . cot5 Lipoprotein

-30

-20

- 10

-1

1

MKKTVLALSLLIGLGATAASYA

A

Higgins and Ames (1981)

MFKTMXALLITASCSTFF MS~Q~&VALIPFFAAFCLPVFA

MKXTKLVLGAVILT~LSXWA

A H E

Jaurin and Gmndstrom (1981) Sutcliffe (1978) Wilson and Hogg (1980)

MKKSLVLKASVAVATLVPMFA MKKLLFAIPLVVPWSHS

A A

Sigimoto et al. (1977) Schaller et al. (1978)

h P A

MQTKLVLGAVILGSTE C WRKLPLAVAVAAGVMSAQAMA V MKKTAIAIAVALAGFATVVW A

OmpF OmPC PhoE

MMKRNILAVIVPALLVAGT-&IA MK-KV&3LLVPALLVAGAAN MKKSTLALVVMGIVAS&SI@

LamB

References

A A A

Inouye er al. (1977) Hedgpeth er al. (1980) Beck and Bremmer (1980); Mowa et al. (1980) Mutoh et al. (1982) Mizuno et al. (1983) Overbeeke ef al. (1983)

Selection of hydrophobic amino acid residues (underlined) is based on data by von Heijne (1981). Amino acids are symbolized by the one-letter code, i.e., A = Ala, C = Cys, D = Asp, E = Glu, F = Phe, G = Gly, H = His, I = Ile, K = Lys, L = Leu, M = Met, N - Asn, P = Pro, Q = Gln, R = Arg, S = Ser, T = Thr, V = Val, W = Trp, Y = Tyr, X = unknown.

1. APPLICATION OF THE SIGNAL HYPOTHESIS

11

microsomes by Szeczesna and Boime (1976), a finding which confirmed these results. Based on these main facts and experiments both by Blobel’s group and others (Mach et al., 1973; Schechter, 1973; Kemper et al., 1974; Boime et al., 1975; Suchanek et al., 1975; see also Scheele et ul., 1980), the signal hypothesis was postulated (see Fig. 1): The biosynthesis of a secretory protein starts, like that of a cytoplasmic one, at a free ribosome. When the N-terminus emerges completely from the ribosome [about 40 amino acid residues are required to span the ribosome (Malkin and Rich, 1967; Blobel and Sabatini, 1970)], the signal peptide located at the Nterminus is recognized by a receptor protein in the RER membrane and the ribosome binding is triggered. According to the signal hypothesis, the polypeptide chain would cross the hydrophobic phospholipid bilayer through a protein channel. During or shortly after completion of the protein the signal peptide is cleaved off by an enzyme, called signal peptidase, which is located at the luminal side of the endoplasmic reticulum. When the ribosome has reached the stop codon on the mRNA, the translation complex disintegrates, the pore in the membrane disappears, and the ribosome becomes free again. We shall discuss the evidence for the various steps in detail in the following sections. Some parts of the original scheme have to be changed and others can be detailed (see Fig. 2, Section 11,E). However, the general ideas remain valid.

B. The Recognition of Signal Peptides 1. SIGNAL PEPTIDES INITIATE THE TRANSLOCATION OF P O L Y P E P r l D E CHAINS With only few exceptions (see Section II,B,7), all secretory proteins contain a cleavable signal peptide at their N-terminus. The signal peptides bear no sequence homology and their length is variable from 15 to 35 amino acid residues (Table 11). This variability is observed even if the proteins stem from a single cell. The only common and obvious feature is a stretch of hydrophobic amino acid residues in the middle of the peptides. This stretch is uninterrupted and has a minimum length of 6-7 residues. Things may be different, however, under extreme conditions, e.g., in halophilic bacteria [bacteriohodopsin has a precursor with a N-terminal extension which is not very hydrophobic (Dunn et al., 1981)]. In many, but not all, signal peptides, there are basic residues preceding the usually observed hydrophobic core (see Table 11). Are these peptides really the signals for translocation? Several lines of evidence support this conclusion. ( l ) Only polypeptides destined for translocation across the RER membrane or across the inner membrane in bacteria contain a

12

TOM A. RAPOPORT AND MARTIN WIEDMANN

cleavable N-terminal peptide with a stretch of hydrophobic residues in the middle. (2) When the signal peptide is removed, secretion is prevented (Talmadge et al., 1980a; Gething and Sambrook, 1982). (3) Point mutations and deletions affecting the secretion of individual polypeptides all map in the signal peptide (Bedouelle et al., 1980; E m and Silhavy, 1980, 1982; Emr er al., 1980b; Koshland et al., 1982). (4)Incorporation of P-hydroxyleucine (HO-Leu) instead of leucine into secretory polypeptides containing signal peptides with many Leuresidues abolishes translocation across the RER membrane (Hortin and Boime, 1980). Actually, the most convincing proof for the role of the signal peptide would be if one could show that a polypeptide normally located in the cytoplasm is secreted when a signal peptide is added at its N-terminus. Such an experiment was carried out for P-galactosidase (Silhavy et al., 1977; Bassford et al., 1979; Ito et af., 1981) but with unexpected results: the fusion polypeptide became a membrane protein or severely inhibited the translocation of many other exported proteins. These results were interpreted to indicate that a polypeptide chain must be compatible with secretion; p-galactosidase apparently has parts in its polypeptide chain which are difficult to translocate across a phospholipid bilayer (von Heijne and Blomberg, 1979; von Heijne, 1980). Nevertheless, the experiments indicated that initiation of secretion occurred correctly by the fusion with the signal peptide. However, it appears that the signal peptide is not sufficient since a fusion product of a short piece of the periplasmic lamB protein (including the entire signal sequence) and the P-galactosidase was exclusively found in the cytoplasm (Moreno et al., 1980; Hall et al., 1982). More recently, Lingappa et a f . (1983) have constructed a plasmid which contained a globin gene preceded by nucleotides coding for a signal peptide. Using in vitro transcription and translation of the RNA in a wheat germ cell-free system they could show that the globin polypeptide was completely translocated across microsomal membranes from dog pancreas. Nature itself has provided an example which is most convincing. There are two forms of yeast invertase, a secreted one and a cytoplasmic one. They are coded by two different mRNAs which originate from the same gene (Carlson and Botstein, 1982; Perlman et al., 1982). The only difference is that the signal peptide coding sequence is absent for the cytoplasmic protein, in accordance with its role as a signal for translocation. Although all data are compatible with the idea that signal peptides initiate the translocation process, more experiments on secretory proteins constructed by genetic engineering are desirable. In particular, it is not yet clear whether secretory proteins require features other than the signal peptide. In fact, it has been suggested that a sequence further “downstream” may also be recognized by the translocation machinery in prokaryotes (Benson and Silhavy, 1983).

1. APPLICATION OF THE SIGNAL HYPOTHESIS

13

2 . THERECOGNITION MECHANISM OF THE SIGNAL PEPTIDESIs UBIQUITOUSIN NATURE Before asking what feature of the signal peptide is recognized, it is important to stress that the recognition mechanism appears to be ubiquitous in nature. Not only can cell-free translation systems derived from mammalian or plant cells be functionally supplemented with microsomes from a variety of sources (Blobel and Dobberstein, 1975b; Szczesna and Boime, 1976; Kreibich et al., 1980; Thibodeau and Walsh, 1980), but also Xenopus oocytes when injected with mRNA coding for secretory proteins will secrete the corresponding products efficiently (Lebleu e f al., 1978; Colman and Morser, 1979; Lane et al., 1980; Colman et al., 1981; Lane, 1981; Rapoport, 1981; Zehavi-Willner and Lane, 1977). Even more surprisingly, rat proinsulin is secreted into the periplasmic space of E . coli with either a prokaryotic or a eukaryotic signal peptide (Talmadge et al., 1980a,b). The function of a eukaryotic signal sequence in bacteria has also been shown for ovalbumin (Mercereau-Puijalon et al., 1978; Fraser and Bruce, 1978) and for carp proinsulin (Rapoport et a f . , 1983). However, it is not clear whether this is a general feature. Mature fibroblast interferon could not be detected in the periplasm (Taniguchi e f al., 1980) and rat growth hormone was found associated with membranes rather than secreted (Seeburg et al., 1978). The reverse-the functioning of a prokaryotic signal sequence in eukaryotic cells-has also been demonstrated. Pre-(3-lactamase was correctly processed and translocated when synthesized in a wheat germ cell-free system supplemented with dog pancreatic microsomes (Muller et al., 1982). Plasmid-encoded pre-plactamase was shown to be processed by yeast cells (Roggenkamp et al., 1981). The fact that many different cell types can recognize signal sequences from unrelated sources argues for a common, ubiquitous receptor mechanism. Direct evidence was provided by competition experiments. Nascent secretory proteins competed with each other for in vitro translocation across the RER membrane (Lingappa et al., 1978a). In prokaryotes, accumulation of a precursor containing a signal peptide prevents secretion of many other proteins (Bassford et al., 1979; Ito et al., 1981; E m et al., 1980b), presumably by saturation of a component involved in translocation. Thus, it appears that there is a single recognition mechanism in a cell for all signal peptides. It appears to be universal in nature. 3. Is THEREA SIGNAL PEPTIDERECEPTOR?

Blobel and Dobberstein (1975a,b) proposed the existence of a protein receptor for signal peptides. This was questioned in the light of the greatly differing sequences of the peptides. Their only common property, the hydrophobicity, was taken as evidence for a direct interaction with the phospholipid bilayer (von

14

TOM A. RAPOPORT AND MARTIN WIEDMANN

Heijne and Blomberg, 1979; Engelman and Steitz, 1981). Calculations, mainly by von Heijne, indicated that such an interaction would be energetically possible. The arguments were also based on the findings by Wickner’s group that the procoat protein of f l phage was able to spontaneously traverse the phospholipid bilayer. The only component required in the bilayer appeared to be the signal peptidase (Silver er al., 1981; Ohno-Iwashita and Wickner, 1983). However, these experiments appear to contradict results from the same group (Date et al., 1980a,b) which showed that a membrane potential is required for the incorporation of procoat. A direct interaction of signal peptide and lipid was made unlikely by binding experiments using completed preproteins. Prehn et al. (1980, 1981) demonstrated that completed carp preproinsulin and human placental prelactogen are bound by the RER membrane and compete with each other in vitro, whereas proinsulin or globin are not bound. Binding of the precursor blocked the sites for cotranslational translocation of nascent secretory proteins across the membrane. There was a saturable number of binding sites which were sensitive to trypsin. Binding was observed only for RER membranes, not for smooth membranes or artificial phospholipid bilayers (Bendzko et al., 1982). Similar results were obtained by Habener’s group using preproparathyroid hormone as a model (Majzoub et al, 1980). They also used for competition experiments a synthetic peptide containing the signal peptide. These experiments strongly suggested the existence of a protein receptor at the RER membrane but did not identify the binding entity.

4. THE SIGNALRECOGNITION PARTICLE The recent discovery by Walter and Blobel of the signal recognition particle (SRP) has provided definitive evidence for a receptor in higher eukaryotes (Walter et al., 1981; Walter and Blobel, 1981a,b). The particle was isolated from RER membranes by extraction with high concentrations of KCl. Salt-washed membranes (K-RM) are translocation inactive but can be reactivated by addition of the high-salt extract (Warren and Dobberstein, 1978). SRP was subsequently isolated from the extract by hydrophobic chromatography and sucrose gradient centrifugation and was shown to be the only component required to reactivate K-RMs (Walter and Blobel, 1980). For assay of the reactivation of the K-RMs, the reticulocyte lysate system cannot be used (Walter et al., 1979) since it contains, unlike the wheat germ system, SRP (Meyer et al., 1982b). In initial experiments, Jackson et al. (1980) were unable to inactivate RER membranes by extraction with high concentrations of KC1 and thought that the differing results of Warren and Dobberstein (1978) were due to proteolysis. It is now clear that both KCl and protease treatments abolish the translocation competence of RER membranes.

1. APPLICATION OF THE SIGNAL HYPOTHESIS

15

SRP has a sedimentation constant of 1 I S and consists of six different polypeptide chains of molecular weights 72K, 68K, 54K, 19K, 14K, and 9K (Walter and Blobel, 1980) and a 7 S RNA (Walter and Blobel, 1982). The 7 S RNA was shown to be identical with the 7 SL RNA known for many years as a metabolically stable RNA species present in the nucleus and cytoplasm of the cell. More than 75% of the total cellular 7 SL RNA is contained in the SRP (Walter and Blobel, 1983a). The nucleotide sequence of the RNA was determined by two groups (Ullu et al., 1982; Li et a l . , 1982). The 7 SL RNA contains at both ends segments which correspond to highly repetitive sequences of the genome ( A h sequences). The middle part (S sequence) is repeated less frequently in the genome or is unique. When SRP was incubated with high concentrations of ribonuclease, the 7 S RNA was degraded and the particle lost its receptor activity (Walter and Blobel, 1982). More recently, a more detailed analysis of the organization of the 7 SL RNA in the SRP was carried out by digestion with micrococcal nuclease (Gundelfinger et al., 1983). The four largest polypeptides are associated with the S segment, whereas the two smallest protein constituents are bound to the Alu fragments. The precise role of the RNA or of the polypeptides is not known as yet. However, different approaches are being undertaken to get more information. Walter and Blobel (1983b) have disassembled SRP into native protein and RNA components. When recombined under suitable conditions, they were able to regain a stoichiometric 11 S complex which was fully active. It is hoped that modification of single components and reconstitution into SRP will allow conclusions as to their role. Antibodies prepared against the three largest polypeptides are also being used to probe their function in SRP (Walter and Blobel, 1983a). The evidence for SRP being the signal receptor is strong but circumstantial. In the absence of K-RMs, the addition of SRP to the wheat germ cell-free system led to a severe inhibition of the translation of preprolactin and pre-growth hormone (Walter et al., 1981). Translation of a- and P-globin was not affected by SRP. Using a synchronized translation system, a peptide fragment about 70 amino acid residues in length was found to accumulate when pituitary mRNA was translated in the presence of SRP (Walter and Blobel, 1981b). Addition of K-RMs restored the translation, the peptide fragment disappeared, translocation across the RER membrane occurred, and the signal peptide was removed. It was concluded that SRP affected a site-specific translational arrest which was released by a component present in the K-RMs. The length of the arrested fragment may be interpreted as a direct interaction of SRP with the signal peptide (about 40 amino acid residues would be buried within the ribosome and 30 would emerge from it, including the complete signal sequence). It remains to be established by sequencing that the arrested fragment is related to preprolactin, the major translation product of the pituitary mRNA.

16

TOM A. RAPOPORT AND MARTIN WIEDMANN

Further support for a recognition of the signal peptide by SRP comes from studies on the interaction of SRP with ribosomes. SRP binds to polyribosomes synthesizing secretory proteins but not to those synthesizing globin (Walter et al., 1981). It binds only weakly ( 104-fold less) to monosomes. HO-Leu incorporation, which inhibited translocation of preprolactin across the RER membrane (Hortin and Boime, 1980), abolished the binding of SRP to polyribosomes (Walter et al., 1981). It was therefore concluded that SRP recognizes the Leurich signal peptide. Of course, HO-Leu is incorporated throughout the polypeptide chain, and its effect might not be entirely due to a change in the signal peptide. Even if it were, SRP could interact with some other part of the nascent chain which is influenced somehow by the signal peptide. From this discussion it appears that more definitive proof for the interaction of SRP with the signal sequence is still desirable. Meanwhile, the function of SRP has been shown for a number of other secretory proteins, including apolipoprotein A1 (Stoffel et al., 198I), immunoglobulin light chain (Meyer et af., 1982b), human placental lactogen (Bassiiner et al., 1983), and carp proinsulin (our unpublished results). Further examples for the function of SRP include lysosomal enzyme precursors (Erikson et al., 1983) and precursors of plant storage globulin polypeptides (Bassiiner et af.,1983). As discussed in Section III,C, the incorporation of some membrane proteins into the RER membrane also depends on SRP. It appears therefore that SRP is generally required for translocation across the RER membrane. The two functions of the particle-the exertion of a translational arrest and the restoration of translocation competence of KC1-washed membranes-are thought to be coupled (see Section

11,E). Although SRP has been purified from RER membranes, it is presumed to cycle in the cell among three states: free in the cytoplasm, bound to ribosomes, and bound to RER membranes (Walter and Blobel, 1981b). The assumption is supported by cell fractionation studies employing different concentrations of KCl and determination of the amount of SRP in the various fractions (Walter and Blobel, 1983a). It was found that at low KCl concentrations SRP is primarily located on the RER membranes. With rising salt concentrations, SRP is found on free ribosomes until, at high concentrations, it is entirely in the unbound state. At physiological salt concentrations, SRP is present in all three compartments.

5. EVIDENCE FOR SRP-LIKEFACTORS IN LOWERORGANISMS Definite proof for the existence of SRP in nonmammalian cells is still lacking. However, the fact that signal peptides from prokaryotes (Miiller et al., 1982) and plants (Bassiiner et af., 1983) are recognized by mammalian SRP suggests that similar particles are present in these organisms. Antibodies raised against the larger polypeptides of dog pancreatic SRP de-

1. APPLICATION OF THE SIGNAL HYPOTHESIS

17

tected related proteins in other mammalian tissues but not in amphibians, insects, plants, or Escherichiu coli (Walter and Blobel, 1983a). The middle part of the human 7 SL RNA cross-hybridizes with sequences derived from animals as simple as Xenopus luevis but not with those from Drosophila and yeast (E. Ullu, personal communication). Nevertheless, Xenopus and Drosophila 7 SL RNA form with the SRP polypeptides of dog pancreas particles which have sedimentation constants of about 11 S and which are functionally active (Walter and Blobel, 1983b). These experiments suggest the existence of an SRP in amphibians and insects which is closely related to that of mammals. Recently, evidence for an SRP-like factor in E . coli has been obtained. Several pleiotropic mutants of E . coli exist which affect secretion (for review, see Michaelis and Beckwith, 1982; Duffaud et al., and Bankaitis et a/., both this volume). The secA protein, a 92K polypeptide, appears to be located at the cytoplasmic side of the inner membrane of E . coli (Oliver and Beckwith, 1982). Antibodies were prepared against the secA protein and appear to precipitate a portion of the 6 S RNA from E . coli homogenates (Liebke et al., 1983). The 6 S RNA was sequenced many years ago by Brownlee (1 97 1) and comparison with the sequence of the mammalian 7 SL RNA indicates several regions of homology (P. Walter, personal communication). The secA protein and the 6 S RNA appear to be constituents of a particle with a sedimentation constant of about I 1 S. TheprlA mutation ( E m et al., 1981) mapped on the promoter-distal end of the ribosomal protein gene cluster (Shultz et al., 1982) may also be a component of the E . coli SRP. The presence of this mutation corrects secretion-deficient signal peptide mutants but has no effect on normal signal peptides. Its occurrence in the ribosomal gene cluster suggests that it is regulated in coordination with the synthesis of ribosomes. This is reminiscent of the SRP structure which also has been called a “third ribosomal subunit” (Walter and Blobel, 1982) (only required for the synthesis of proteins to be translocated across the RER membrane or across the cytoplasmic membrane in bacteria). Other identified proteins involved in the secretory process in E. coli, such as the secC class recently found as suppressors of the secA mutants, could correspond to other involved components, e.g., to the SRP receptor (docking protein) of the RER membrane (see Section II,C,l), particularly since not only the translocation but also the synthesis of exported proteins appears to be affected (Ferro-Novick et al., 1984).

6 . WHATFEATURES OF THE SIGNAL SEQUENCES ARE RECOGNIZED? How are signal sequences recognized by a single receptor (SRP) given their large differences in length and amino acid sequence (see Table II)? It has been suggested that signal peptides may self-organize to form identical

18

TOM A. RAPOPORT AND MARTIN WIEDMANN

secondary structures and that this preformed structure would be recognized by the receptor (Steiner et al., 1980a; Garnier el al., 1980; Austen, 1979; von Heijne, 1980). However, on the basis of the prediction method of Chou and Fasman (1978), contradictory results were obtained. By use of a molecular theory (Finkelstein and Ptitsyn, 1977), we have recently confirmed that a common secondary structure cannot be predicted for all signal sequences (Finkelstein et al., 1983). The same result was obtained if the possibility of an interaction between a signal sequence and the hydrophobic surface of a receptor was taken into consideration (Ptitsyn and Finkelstein, 1983). Experiments with a synthetic peptide containing the signal sequence of preproparathyroid hormone also yielded ambiguous results (Rosenblatt et al., 1980). The unequal importance of the residues in the hydrophobic core of a signal peptide for secretion in E . coli ( E m and Silhavy, 1982, 1983) has been taken as evidence for the preformation of an a-helix, but alternative explanations are possible. We have therefore proposed that the continuous stretch of hydrophobic amino acid residues present in all signal sequences is deeply immersed into the hydrophobic pocket of a receptor (SRP). In this case, the requirement of saturation of all H-bond donors and acceptors dictates the structure of the peptide. If the pocket itself does not present any hydrophilic groups, H-bond formation must occur within the signal sequence and a helical (presumably a-helical) conformation is formed. if the pocket contains polar groups, H bonds may be formed with the signal peptide, and the conformation of the buried part is dictated by the interaction with the pocket. In any case, identical backbone conformations are induced for all sequences. Deep immersion into a hydrophobic pocket is only possible for entirely hydrophobic sequences, a finding which explains the effect of mutations in signal sequences which abolish the translocation (Bedouelle et al., 1980; Emr et al., 1980b; Emr and Silhavy, 1980; Koshland et al., 1982). Quantitative considerations based on data by Tanford (1980) indicate that a minimum of about six or seven contiguous hydrophobic residues is required, a prediction in agreement with the actual findings (Table 11). Any strong interaction of the signal peptide with other parts of the protein would lead to significant weakening of the receptor binding. We have, therefore, predicted that a continuous stretch of hydrophobic residues of a polypeptide chain which follows either an unfolded N-terminal part or a folded domain of the protein with a hydrophilic surface will be recognized by SRP and therefore function as a signal for translocation. Taking into account the cotranslational mode of transfer, we propose that it must be the first of such segments in a polypeptide chain (from the Nterminus) that functions as a signal for translocation. The proposed mechanism differs from other known ligand-receptor interactions in which side chains of amino acids play an important role. In the present case the immersion is so deep-and this is only possible for entirely apolar

1. APPLICATION OF THE SIGNAL HYPOTHESIS

19

sequences-that the side chain interactions are unimportant. This explains the lack of any sequence homology among different signal peptides. Though the variation among the signal sequences excludes a complete complementarity between their surfaces and that of the cleft of the receptor, strong binding and hydrophobic interactions are nevertheless possible. A similar case is found for intermediates in the unfolding of proteins (Dolgikh et al., 1981). Yost el al. (1983) have placed the hydrophobic tail of the membrane-bound immunoglobulin M in front of the cytoplasmic globin, but translocation across the RER membrane was not observed. However, the segment was much longer than a normal signal peptide (about 70 amino acid residues) so that this negative result does not directly contradict our hypothesis. Are there any parts in signal sequences of importance for translocation other than the hydrophobic core? Elegant studies by the group of Inouye indicate that the transport is facilitated by the presence of the basic amino acid residues preceding the hydrophobic core of the signal peptide of prelipoprotein, an outer membrane protein of E . coli (Inouye et al., 1982; Vlasuk et al., 1983). The basic amino acid residues, though not absolutely essential, may be helpful by interacting with the negatively charged phospholipid head groups. It is not yet clear whether this result can be generalized to all prokaryotic exported proteins or even to eukaryotes [there are examples of signal peptides with a negative charge close to the N-terminus (see Table II)]. It should be noted that we have only dealt with the initial recognition event. It is conceivable that in a later step of the translocation process the signal peptide by virtue of its hydrophobicity directly interacts with the phospholipid bilayer.

SEQUENCES 7. INTERNAL SIGNAL It came as a great surprise when Palmiter et al. (1978) discovered that ovalbumin, a secreted protein, is synthesized without a cleavable signal peptide. Yet, studies in cell-free protein synthesis systems showed that ovalbumin was sequestered within microsomal vesicles like other secretory proteins (Lingappa et al., 1978b; Palmiter et al., 1980). In addition, proteins with N-terminal signal sequences competed with ovalbumin for sequestration within microsomes. These data suggest that ovalbumin and other signal-containing proteins use a common translocation system. So far ovalbumin has remained an exception for eukaryotes. Although the N-terminal amino acid sequence of ovalbumin is not strikingly hydrophobic, it was believed at first to be the (uncleaved) signal peptide. However, the fact that cloned ovalbumin was secreted from E . coli though the first four amino acid residues were missing (Fraser and Bruce, 1978; MercerauPuijalon et al., 1978) suggested that the signal peptide was internal (there is a

20

TOM A. RAPOPORT AND MARTIN WIEDMANN

charged amino acid residue at position 9). Lingappa et al. (1979) claimed in a much discussed paper that the signal peptide was located on the tryptic peptide 229-276. Their evidence was (1) that the purified tryptic fragment competed with nascent ovalbumin and nascent preprolactin for translocation across the RER membrane in v i m ; (2) that in a synchronized translation system the membranes could be added until about 250 amino acid residues had been polymerized (by contrast, in the case of preprolactin the membranes had to be present before 70 amino acids were linked); and (3) that a sequence resembling the signal peptides of related proteins (lysozyme, ovomucoid) was found on the tryptic peptide. None of these data is fully convincing, and in fact it turned out that the synchronization experiment was erroneous as a result of incomplete inhibition of initiation of translation (Braell and Lodish, 1982a). It was found that membranes actually have to be present before the nascent chain has a length of about 150 residues. The competition experiments are not conclusive since a 1000-fold molar excess of the tryptic peptide was necessary for inhibition of translocation. Finally, a sequence resembling a signal peptide may just reflect a gene duplication rather than a functional peptide. It has been noted by D. F. Steiner that growth hormone contains in the middle of its sequence a peptide resembling its own signal peptide (personal communication). Meek et al. (1982) have provided evidence that the signal peptide in ovalbumin is located within the first 50-60 amino acid residues. By assaying the binding of mRNA to microsomes as a function of time, they concluded that the nascent chains must have a length between 45 and 90 amino acid residues. Similarly, nascent ovalbumin chains as short as 50-60 residues long were already found sequestered within the RER membranes both in vitro and in vivo, whereas those shorter than 50 residues were not retained by the microsomes. Meek et al. (1982) suggest that the signal peptide is located in the ovalbumin chain between positions 26 and 45, a region which is indeed hydrophobic (see Table 111). To retain the hypothesis of a functional N-terminal signal peptide even in ovalbumin, these authors propose a loop structure, with one strand being apolar, the other polar. Such a loop would then be N-terminal in ovalbumin and other secretory proteins; the only difference would be that the amphipathic character of the hairpin is reversed in the former. Baty et al. (1981) found that ovalbumin lacking the first 126 amino acid residues is not secreted from E . coli, a finding consistent with, but no proof of, the assumption that the signal peptide in ovalbumin is located closer to the Nterminus. A definite identification of the internal signal peptide in ovalbumin is probably possible only by genetic means. For instance, the sequence in question should substitute a normal cleavable signal peptide. Mutations in it should abolish the translocation. More recent evidence suggests an internal signal sequence in a prokaryotic

1. APPLICATION OF THE SIGNAL HYPOTHESIS

21

protein. Colicin El is secreted into the periplasm and is not processed (Ebina et al., 1981). The only hydrophobic sequence, according to the nucleotide sequence of the gene, is located near the C-terminus (Yamada et a/., I982b). The protein is secreted posttranslationally at a very slow rate, but its pathway appears to be identical with that of proteins with cleavable N-terminal signal peptides (Yamada et al., 1982a). An internal signal sequence was also constructed artificially using rat preproinsulin as a model (Talmadge et al., 1981). Nucleotides coding for a few residues of the P-galactosidase and some linker residues were put in front of a slightly shortened preproinsulin gene so that 39 amino acid residues are expected from the initiator methionine up to the cleavage site of preproinsulin. Proinsulin was found in the periplasm of E . coli cells. We have extended such a study, using carp preproinsulin as a model (Rapoport et al., 1983). Even when the signal peptide was preceded by 54 amino acid residues (total length of the “signal peptide,” 76 residues), it was operational and cleaved correctly. Coleman et al. (1983) have shown that the signal peptide of the lipoprotein of the E . coli outer membrane was fully active for protein secretion even when it was preceded by up to 145 amino acid residues. In this case it was possible to prove that the internalized signal peptide was functional, rather than a N-terminal one created by proteolytic cleavage of the primary translation product.

C. Translocation of Secretory Polypeptides across Membranes 1. THE “DOCKING PROTEIN”: THESRP RECEPTOR When RER membranes are incubated with low concentrations of trypsin or elastase, a component is cleaved off which can be removed from the membranes by raising the KCI concentration above 0.2 M . When added back to proteaseinactivated membranes, the extract restores the translocation activity (Walter et al., 1979; Jackson er al., 1980; Meyer and Dobberstein, 1980a,b). This component was purified and was identified by Meyer and Dobberstein (1980b) as a 60K polypeptide. Antibodies raised against it precipitated a 72K membrane protein (Meyer et af.,1982a) which yielded the 60K fragment after limited protease digestion. It appears that the 72K protein consists of two domains. Functional reconstitution only requires an ionic interaction between them. The function of the polypeptide was demonstrated using the SRP-induced arrest of translation as an assay. Meyer er al. (1982b) found that the isolated 60K fragment released the translational arrest. Using a different approach, Blobel’s group reached a similar conclusion (Gilmore et a / . , 1982a,b). They isolated a 72K polypeptide from a detergent extract of RER membranes by affinity chromatography using SRP coupled to

22

TOM A. RAPOPORT AND MARTIN WIEDMANN

Sepharose. By immunological means and by peptide mapping, it was shown that this protein is identical with the 72K membrane protein identified by Meyer et al. (1982a). However, in contrast to the findings of the latter authors, it was found that the 72K protein and not the 60K proteolytic fragment was able to release the translational arrest. The deviating results of Meyer et al. (1982b) can be explained by contamination of their 60K preparation with the 72K protein. On the basis of its function, the 72K membrane protein was termed “docking protein” by Meyer et al. (1982b) and “SRP-receptor” by Blobel’s group (Gilmore et al., 1982a,b). The existence of such a SRP-receptor was predicted by Walter and Blobel (1981b).

2. THE ROLEOF RIBOSOMES Both the synthesis of secretory proteins in membrane-bound polyribosomes and the cotranslational mode of transfer across the endoplasmic reticulum membrane suggest an active role of the ribosomes. Evidence for a direct interaction of the RER membrane with ribosomes is old (Borgese et al., 1974; Adelman et al., 1973; Rolleston and Lann, 1974; Sabatini and Kreibich, 1976). It could be shown that isolated ribosomes bind specifically to rough but not smooth ER membranes. Binding was saturable and sensitive to trypsin. The interaction between ribosomes and the RER membrane is probably ionic since it is sensitive to high salt concentrations. From electron micrographs and binding data it appeared that the large ribosomal subunit contained the binding site (Sabatini and Kreibich, 1976; Unwin, 1979). However, the exit site of the nascent polypeptide chain on the ribosomes, probed by antibodies, seems to be close to the interface of the two subunits (Lake, 1983). This finding may suggest that both ribosomal subunits are in intimate contact with the membrane. It may, therefore, be more appropriate to draw the ribosome-membrane interaction as in Fig. 2 rather than as in Fig. 1. Two proteins of the RER membrane, termed ribophorins I and 11, were suggested as ribosome receptors (Kreibich et al., 1978a,b; Marcantonio et al., 1982). Their occurrence in rough ER membranes of many cell types (but not in smooth ER membranes), their close association with ribosomes, and their ability to be cross-linked to ribosomal proteins all support this hypothesis. However, direct evidence for an interaction of ribophorins with ribosomes during the translocation process is still lacking. The interaction of SRP with ribosomes is well established (Walter et al., 1981). Again the interacting component of the ribosome is unknown as yet. However, it is unlikely that SRP mediates the binding of the ribosome to the membrane. There are about 10-fold more membrane-bound ribosomes per cell than SRP and docking protein molecules (Gilmore et al., 1982b). Thus, SRP and its receptor are probably only involved in the initiation of translocation, not in the translocation itself. Direct support for this prediction comes from recent binding

1. APPLICATION OF THE SIGNAL HYPOTHESIS

23

studies (Gilmore and Blobel, 1983). By use of iodinated SRP and SRP-receptor proteins, it could be shown that (1) only SRP, but not its receptor, is bound to ribosomes, and (2) SRP is released from the ribosome when the SRP-receptor is added. These data indicate the existence of a further ribosome-binding component in the membrane and may even suggest that the signal peptide is handed over to another binding entity. The evidence for the participation of ribosomes in the translocation of secretory proteins across the inner bacterial membrane is much less convincing. Although membrane-bound ribosomes exist and appear to play a role in protein export (Randall and Hardy, 1977; Randall ef al., 1978), it is clear that there is no firm binding of ribosomes to the cytoplasmic membrane in a manner comparable to the eukaryotic RER membrane (Davis and Tai, 1980). The question whether translocation in bacteria occurs cotranslationally or posttranslationally is a matter of ongoing discussion. Davis and co-workers have claimed that nascent polypeptide chains span the inner cell membrane (Smith et al., 1977, 1979). On the other hand, completed pre-P-lactamase is found first inside the cell and later outside (Koshland and Botstein, 1982). Randall (1983) has used the accessibility of nascent chains of periplasmic proteins to externally added protease as a criterion for the translocation across the cytoplasmic membrane of E . coli. Limited proteolysis was used to identify specific regions of the growing polypeptide chains. It was concluded that translocation of the polypeptide is a late event relative to the extent of elongation. In the case of the ribose-binding protein, translocation occurred only after completion of the chain. Thus, it appears that export and elongation may not be strictly coupled in prokaryotes. Rather, entire domains of polypeptides may be translocated after their synthesis. It appears that the important results of this approach should be further supported by classic peptide mapping experiments. The procoat protein of the f l phage can be posttranslationally incorporated into membranes both in vitro and in vivo (Ito et al., 1980; Silver et al., 1981; Ohno-Iwashita and Wickner, 1983). In this case, the ribosomes appear not to be required for translocation. However, the procoat protein may be an exception because of its function during phage infection of a bacterial cell, where it is incorporated spontaneously into the target membrane (Wickner et al., 1980). It is also possible that the procoat protein predominantly uses the cotranslational mode in vivo (see also Russel and Model, 1982). In other cases, even if a completed polypeptide chain is found within the cell, the participation of the ribosomes in the translocation process is not excluded since ribosomes carrying completed chains may be functional. 3. ENERGYFOR TRANSLOCATION The source of the energy for protein translocation is still not established. For bacteria, however, some information exists. The translocation can be inhibited

24

TOM A. RAPOPORT AND MARTIN WIEDMANN

by uncouplers (Date et al., 1980a,b). Evidence was provided that not ATP but a membrane potential is required. Furthermore, using specific drugs it was excluded that the proton gradient is delivering the energy (Pages and Lazdunski, 1982; Daniels et al., 1981). Rather, the asymmetrical charges on the membrane appear to be essential. A similar situation occurs with mitochondria1 protein import (Schleyer et al., 1982; and article by Reid, this volume). The situation appears to be quite different for the RER membrane. It is doubtful whether an ion gradient can be built up in RER vesicles. Thus, the membrane potential, if it exists, must be small. Uncouplers have no effect on protein translocation (P. Walter, personal communication). Calculations by von Heijne (1981) have shown that it is probably not the translation energy which is used to “push” the growing chain across the membrane. It has been pointed out by von Heijne that the energy required for the initial translocation of the polypeptide chain is probably greater than for a subsequent residue-by-residue translocation, since for each residue incorporated into the membrane another residue would appear in the aqueous space on the luminal side of the membrane. Thus, there may be a significant energy compensation. Perhaps the energy is provided by an as yet unknown modification mechanism. For example, it is conceivable that phosphorylation or adenoribosylation activates the SRP for signal recognition and membrane binding and that a demodification would occur at the membrane, providing the energy for initiation of translocation.

D. Cleavage of Signal Peptides 1. CLEAVAGE OF SIGNAL PEPTIDESAND PROTEIN TRANSPORT Although cleavage of signal peptides is usually observed, there are exceptions. These include proteins with presumed internal signal peptides (see Section II,B ,7) and mutants of secretory and membrane proteins which are transported but not cleaved (Lin et al., 1978; Russel and Model, 1981; Koshland et al., 1982). Thus, translocation and cleavage of the signal peptide are not strictly coupled. In fact, one may raise the question why cleavage is so often observed. A possibility may be the difficulty of folding a polypeptide chain containing a long hydrophobic amino acid sequence which is without function for the mature protein. The signal peptide may also serve as membrane anchor and prevent complete detachment of a protein from the membrane (see Section III,D,2). Cleavage of the signal peptide may also be a mechanism for making the translocation process unidirectional and irreversible. In eukaryotes, processing of the precursors appears to be largely cotranslational. In some cases precursors have been found in cells pulse labeled with radioactive amino acids for very short periods of time (Habener et al., 1976;

1. APPLICATION OF THE SIGNAL HYPOTHESIS

25

Maurer and McKean, 1978; Patzelt ef al., 1978). In only one case, however, has the localization of the preprotein been determined: it was found to be miscompartmentalized on the cytoplasmic face of the RER membrane (Habener et a/., 1980). However, Hortin and Boime (1981) have shown by use of a Thr analog that cleavage of the precursors may occur posttranslationally in vivo in the correct cellular compartment. Lane et al. (1979) have provided evidence that preproteins are rapidly degraded in the cytoplasm of Xenopus oocytes. In prokaryotes processing of the precursors can be either cotranslational or posttranslational, depending on the specific protein (Josefsson and Randall, 1981a,b). In general, it appears to be a late event in the biosynthesis of secretory proteins. The cleaved-off signal peptides have been demonstrated only recently (Hussain et al., 1982). They are usually rapidly degraded after being cleaved from the precursor. The degradation of the signal peptide can be prevented in vitro in prokaryotes by addition of antipain or other peptide aldehyde inhibitors, a finding indicating that the signal peptide peptidase(s) may have a serine or cysteine residue in their active center(s) (Hussain et u f . , 1982). 2. THE SIGNALPEP~IDASE This enzyme is apparently an unusual protease with respect to its specificity. There is practically no sequence homology among different cleavage sites, yet the proteolysis is very accurate. In most cases, there is a small aliphatic amino acid residue at the C-terminus of the cleavage site (see Table ll), and this may be essential (Hortin and Boime, 1981) but is, of course, insufficient to determine specificity. Mutations which prevent the cleavage of the signal peptide map close to and on both sides of the splitting point (Lin et al., 1978; Koshland e t a / ., 1982; Russel and Model, 1981; Inouye et al., 1983a,b; and Duffaud et a/., Chapter 2, this volume). It has been proposed that a p-turn occurs near the cleavage site, thus exposing it to the signal peptidase (Steiner et al., 1980a). However, energetically, such a structure can only be marginally stable if formed by the signal peptide alone. The signal peptidase from eukaryotes has not been completely purified yet. It can be assayed after treatment of microsomes with detergent using completed preproteins as substrates (Jackson and Blobel, 1977; Kreil et al., 1977). Mollay et a/. ( 1 982) achieved a separation of the signal peptidase from signal peptidedegrading proteases. They proved that the signal peptidase is an endopeptidase. None of the common inhibitors of proteases appears to affect the enzyme. The amino acid residues involved in the active center of the signal peptidase are therefore unknown as yet. The signal peptidase is assumed to be located at the luminal side of the RER membrane. This is mainly based on the fact that even extensive treatment of

26

TOM A. RAPOPORT AND MARTIN WIEDMANN

vesicles with proteases does not affect the signal peptidase activity (Jackson and Blobel, 1977). However, on the basis of its hydrophobic nature, it is also possible that it is deeply buried in the phospholipid bilayer; perhaps even its active center is there. This assumption might explain the surprising finding by Mollay et al. (1982) that after incorporation of the enzyme into liposomes, the addition of detergent was still required for the assay, although one would have expected a more or less random orientation of the peptidase in the vesicles. In prokaryotes, it has turned out that there exist two signal peptidases, one which is specific for lipoproteins (Tokunaga et al., 1982) and one which appears to be analogous to the eukaryotic counterpart and has been termed “leader peptidase” by Wickner (1979). The former is specifically inhibited by globomycin (Inukai et al., 1978; Hussain et al., 1980). On the other hand, it appears that all exported proteins use the same translocation mechanism and that only the cleavage of the signal peptide is different. Wickner’s group has purified the prokaryotic leader peptidase, a 37K protein, to near homogeneity (Zwizinski and Wickner, 1980). They have cloned the gene (Wolfe et al., 1982) and derived the primary structure of the polypeptide from the nucleotide sequence (Wolfe et al., 1983). The enzyme appears to be anchored in the membrane with its N terminus and has its active center in the periplasmic space. It has been localized both in the inner and outer membranes (Zwizinski et al., 1981). It has been shown that the prokaryotic leader peptidase cleaves correctly eukaryotic secretory protein precursors (Talmadge et al., 1980b; Watts et al., 1983; C. Mollay and G. Kreil, personal communication). Conversely, the eukaryotic signal peptidase accurately processes pre-P-lactamase (Miiller et al., 1982). This is again evidence for the fact that the recognition of various features of signal peptides is ubiquitous in nature (see Section II,B,2). Is the signal peptidase a separate entity or an enzymatic activity of any other known component of the translocation machinery? Gilmore et al. (1982a,b) have shown that neither the SRP-receptor nor the SRP are identical with the signal peptidase. It is also unlikely that the enzyme is related to the ribophorins since the latter are sensitive to protease treatment of the RER membranes (Kreibich er al., 1978a,b). Therefore, it appears that the peptidase is a separate component of the translocation apparatus. In E . coli, the pleiotropic secretion mutants and the signal peptidase map in different regions of the genome.

E. Models for Protein Translocation-What Is Missing? Figure 2 summarizes the present view of the translocation of secretory proteins across the endoplasmic reticulum membrane. The process starts, as in the older scheme (see Fig. l), with a free ribosome. As soon as the signal peptide completely emerges from the ribosome, it is recognized by SRP and the elongation of

27

1. APPLICATION OF THE SIGNAL HYPOTHESIS

t

ribosome

SRP

t 1

signal sequence

ribosome

protein ceptor) signal peptidase

&c";i FIG.2. Schematic representation of the present view on the mechanism of protein transfer across the endoplasmic reticulum membrane. The scheme shows the early steps in the biosynthesis of a secretory protein. The translation of the mRNA starts with a free ribosome. When the signal peptide of the nascent chain emerges completely from the ribosome, it is recognized by the signal recognition particle (SRP), which may be free in the cytoplasm or bound to free ribosomes, and elongation of the polypeptide is arrested. SRP is bound much more strongly to the ribosome if it also interacts with the signal peptide. Release of the translational arrest occurs by binding of the complex to the SRPreceptor (docking protein) which is located in the RER membrane. Simultaneously, binding of the ribosome to a ribosome receptor is triggered and SRP is released from the ribosome and becomes free again. SRP cycles, therefore, as shown in the scheme during the initiation of translocation. Transfer of the polypeptide across the phospholipid bilayer presumably involves a ribosome receptor and possibly other membrane proteins (not shown). After completion of the polypeptide, the ribosome is released from the RER membrane and is disassembled into its subunits (not shown). A ribosome cycle is therefore completed (see scheme). Cleavage of the signal peptide is brought about by the signal peptidase located at the luminal side of the membrane. Modified after Walter and Blobel (I98 I b).

the polypeptide is arrested. The complex is then bound via SRP to the SRPreceptor (docking protein) present in the RER membrane. At the same time the ribosome becomes membrane bound and SRP is released from the ribosome. Initiation of the translocation process is now completed and is characterized by an SRP cycle. The actual translocation proceeds without participation of SRP and its receptor. After reaching the termination codon, the ribosome falls off the membrane and thus a ribosome cycle is completed. For the last step a protein may be required (Blobel, 1976). The signal peptide is cleaved by the signal peptidase at the luminal side of the membrane, and the peptide is further degraded. Folding of the polypeptide probably occurs after removal of the signal peptide (Lomedico et al., 1977).

28

TOM A. RAPOPORT AND MARTIN WIEDMANN

For prokaryotes the mechanism of transport is less known. It may occur in the same general manner except that the ribosomes are not bound firmly to the membrane. Also, as discussed earlier, here the membrane is energized, which appears important for the translocation process in an unknown way. Several aspects of the scheme deserve further comment. It is not certain that SRP is needed for each ribosome which starts the synthesis of a polypeptide chain. It is conceivable that SRP is only required once to build up a polyribosome attached to the RER membrane and that further ribosomes starting translation on the already membrane-bound mFWA do not need SRP for initiation of translocation. Walter and Blobel (198 1b) have proposed a regulatory function for the translational arrest: it would prevent the miscompartmentalization of proteins destined for translocation in the case that membrane binding sites are limiting. Whether a translational arrest occurs in vivo is unknown, although data by Richter and Smith (198 1) may be interpreted in favor of this assumption. They found that proteins destined for translocation across the endoplasmic reticulum membrane compete with each other in Xenopus oocytes. No competition was observed with the biosynthesis of cytoplasmic proteins. When dog pancreatic microsomes were injected into oocytes, the translocation process could be stimulated. A translational arrest is not observed in vitro in the reticulocyte system which contains low endogenous levels of SRP (Meyer et al., 1982b) even when further SRP is added (our unpublished results). Furthermore, two membrane proteins are known with uncleaved signal sequences where the translational arrest is weak or absent but SRP is nevertheless required for membrane incorporation (see Section 111,D).Therefore, it is not yet certain that the translational arrest is strictly coupled to translocation. A matter of controversy is the actual mechanism of translocation of the polypeptide chain across the membrane. Blobel and Dobberstein (1975b) have proposed that it occurs through a hydrophilic channel formed by transmembrane proteins. Theoretical arguments, on the other hand, show that the transport could proceed directly across the phospholipid bilayer (von Heijne, 1981; Engelman and Steitz, 1981). Even for hydrophilic or charged groups, the required energy would not be exceedingly high once the polypeptide has achieved a transmembrane orientation. At present there are no data which prove or disprove the existence of tunnel proteins. However, recently Ferro-Novick et al. (1983a,b) have characterized a new class of pleiotropic secretion mutants in yeast (sec53 and sec59). Their data may be interpreted to show that the translocation process is halted after shift to the nonpermissive temperature and that the protein is completed in the cytoplasm. If the transmembrane orientation of the proteins can be demonstrated, this will be good evidence for a protein component needed for the translocation process.

1. APPLICATION OF THE SIGNAL HYPOTHESIS

29

Another point of dispute is the question whether elongation of a polypeptide chain during its synthesis is strictly coupled to translocation, i.e., whether there is a residue-by-residue rather than domain transfer. It could be shown in some cases that core glycosylation at Asn-residues occurring in the lumen of the RER membranes proceeds at the time expected from the extent of elongation of the polypeptide chain (Rothman and Lodish, 1977; Glabe et al., 1980). Furthermore, disulfide bridge formation within nascent immunoglobulin light chains also indicates that the polypeptide crosses the membrane in a rather extended conformation (Bergman and Kuehl, 1979). In bacteria, things might be different, as discussed earlier (Section 11,C,2). How does the signal peptide cross the membrane? In the original scheme (Fig. I ) , it was proposed that its N terminus crosses the membrane first. Both Inouye and Halegoua (1 979) and Steiner et al. ( I 980a) have proposed a loop model, in which the N-terminal, hydrophilic, or charged amino acid residues of the signal peptide stay in the cytoplasm (see Fig. 2 ) . The fact that a signal peptide can be transposed to an internal location without loss of its function in E . coli (Talmadge et al., 1981; Rapoport et al., 1983; Colman et al., 1983) supports this assumption. The facilitating role of a positive charge at the N terminus of lipoprotein for its translocation across the cytoplasmic membrane of E . coli (Inouye et al., 1982; Vlasuk et al., 1983) has also been taken as evidence for the loop model. Steiner’s group has preliminary data showing that the extreme Nterminal part of eukaryotic signal peptides is not associated with microsomal membranes after translocation in v i m of secretory proteins (Steiner et a l . , 1980b). The role of the lipid part of the membrane for the translocation of proteins has not been extensively studied. However, it is clear that it plays at least a permissive role. Experiments in prokaryotes show that changes in the composition of the phospholipids or in their mobility influence the process of protein export (Lazdunski et al., 1979; Di Rienzo and Inouye, 1979; Halegoua and Inouye, 1979). It is possible that this is an indirect effect on membrane proteins involved in protein translocation. There are speculations that special areas of the membrane with nonbilayer structure of the phospholipids (inverted micelles) may promote the incorporation of proteins.

111.

INCORPORATION OF PROTEINS INTO MEMBRANES

A. Different Modes of Membrane Incorporation Many integral membrane proteins have large structural domains in the aqueous phase at the side of the phospholipid bilayer opposite to that where they are synthesized. The problem is therefore similar to that for secretory proteins: how

30

TOM A. RAPOPORT AND MARTIN WIEDMANN

can the polypeptide chains which contain polar and hydrophilic amino acid residues cross the phospholipid bilayer? Indeed, as will be discussed in Section III,C, these translocated membrane proteins seem to be synthesized in a way similar to that for secretory proteins: they carry signal peptides and use the same translocation machinery. The difference is, of course, that the translocation process has to stop at some point for membrane proteins. The question of how these polypeptides are retained in the membrane will be dealt with in Section III,E. For a second class of integral membrane proteins the structural domains are deeply incorporated into the phospholipid bilayer, with only small parts being in the ectoplasmic aqueous phase. This group of proteins includes many channelforming and transport proteins. It appears that in these cases the polypeptide chain spans the membrane several times. Although not immediately obvious, this class of proteins also seems to belong to the translocated ones: they also appear to have signal peptides and are again translocated by the same apparatus used by secretory proteins. It is not clear, however, whether membrane incorporation of all parts of the polypeptide chain requires this machinery. It is conceivable that some portions are indeed translocated, whereas others are only embedded (see below). Finally, there are integral membrane proteins which do not need a translocation system for their membrane incorporation. They do not contain signal peptides but have specialized domains for a direct interaction with the phospholipid bilayer. We call these proteins “embedded” and the parts of the polypeptide chain responsible for this interaction “embedding domains” or ‘‘embedding sequences.” Examples for this class of proteins are rare and will be discussed in Section II1,B. It should be noted that our definition does not exclude the possibility that embedded proteins span the membrane and have some amino acid residues in the aqueous phase at the ectoplasmic side of the membrane. However, we assume that translocation of entire folding domains across the membrane requires a translocation system triggered by signal peptides. Of course, both translocation and embedding could conceivably occur with a single polypeptide chain (Blobel, 1980). It may also be possible that a protein is first completely translocated and then embedded from the outside into the membrane.

B. Embedding versus Translocation The best-known example of an embedded protein-is cytochrome b,. It is made in free polyribosomes without precursor (Rachubinski et al., 1980; Okada et al., 1982) and appears to be posttranslationally incorporated into membranes. This is supported by the spontaneous incorporation of the purified, detergent-free poly-

1. APPLICATION OF THE SIGNAL HYPOTHESIS

31

peptide into natural and artificial membranes (Strittmatter et al., 1972; Fleming and Strittmatter, 1978; Enoch et al., 1979; Bendzko et al., 1982). The latter experiments and the fact that trypsin-treated endoplasmic reticulum membranes can incorporate cytochrome b, in a physiological manner argue for a proteinphospholipid rather than a protein-protein interaction. In keeping with this assumption, cytochrome b, is found at the endoplasmic side of many cellular membranes, including the plasma membrane, outer mitochondrial membrane, and nuclear membrane (Oshino, 1978). The higher concentration in the smooth ER may be explained by its content of transmembrane proteins with binding affinities for cytochrome b, (see Strittmatter and Dailey, 1982) or by its higher content of accessible lipid domains. Cytochrome b, consists of two domains: a hydrophilic one carrying the heme group, and a hydrophobic one for membrane attachment. The latter comprises about 35, mostly hydrophobic, residues at the C-terminus (Strittmatter et al., 1972; Takagaki et al., 1980; Kondo et al., 1979). The overall hydrophobicity of the membrane-binding segment is comparable to that of signal peptides (Bendzko et al., 1982). The two parts of cytochrome b, can be easily separated by mild treatment with proteases (Strittmatter et al., 1972; Ozols et al., 1978), a finding indicating that they fold independently. The hydrophobic tail appears to be highly structured, as indicated by CD measurements and by its resistance to guanidinium-HC1 and heat denaturations (Dailey and Strittmatter, 1978; Tajima et al., 1976; P. Bendzko and W. Pfeil, personal communication). It has a hydrophobic surface, a conclusion based on its ability to bind large amounts of detergent (Robinson and Tanford, 1975). Strittmatter has accumulated evidence that no part of the membrane-incorporated C-terminus of cytochrome b, reaches the luminal side of the phospholipid bilayer (for review, see Strittmatter and Dailey, 1982). Bendzko el al. (1982) have shown that the hydrophobic membrane anchor of cytochrome b, does not compete with the signal peptide of preproinsulin for the binding activity of RER membranes, even if present in a 1000-fold molar excess. This result indicates that cytochrome b, does not carry a C-terminal signal peptide, which, because of its location in the polypeptide chain, could not afford the translocation of the protein across the membrane. On the basis of these facts, cytochrome b, appears to be embedded into membranes with its C-terminus. Since no other part of the protein appears to be required for interaction with the membrane, the C-terminal peptide may be called an "embedding sequence." Although no other embedded protein has been studied to the extent of cytochrome b,, further proteins may follow the same mode of membrane incorporation. For instance, cytochrome b, reductase also has a hydrophobic C terminus which can be split off, thereby leaving intact the enzymatic activity,

32

TOM A. RAPOPORT AND MARTIN WIEDMANN

and which allows its spontaneous incorporation into membranes (Spatz and Strittmatter, 1973; Mihara et al., 1978; Borgese et af., 1980). The sucrase-isomaltase complex, found at the luminal surface of the microvillar membrane of the small intestine, is anchored in the plasma membrane through the N-terminal part of the isomaltase subunit (Brunner et al., 1978; Hauri et al., 1982). A carbohydrate chain is attached at residue I I and is followed by a stretch of hydrophobic amino acid residues (see Table 111) (Frank et al., 1978). These features may suggest that the complex is first secreted and then embedded from the outside into the plasma membrane, although a transmembrane orientation is also possible (Hauri et af., 1982). The basis for the distinction between embedding and translocation in specific cases has not always been sound. The best criterion for translocation of a protein is the requirement of SRP for its membrane incorporation. On the basis of this test, cytochrome P-450 carries a signal peptide (Sakaguchi et al., 1984). This conclusion is in accord with the occurrence of a hydrophobic sequence at the N-terminus (Bar-Nun et’al., 1980; Fujii-Kuriyama et af., 1982) and with the fact that the protein cannot be split into functional domains (Omura and Sato, 1967) and is not incorporated spontaneously into phospholipid bilayers (IngelmanSundberg and Glaumann, 1980). Evidence for the translocation of a polypeptide chain can also be obtained by showing competition with the translocation of secretory proteins. Failure of a protein to be incorporated posttranslationally into the RER membrane can also be taken as indirect evidence against an embedding process. On the other hand, synthesis of a protein in membrane-bound polyribosomes is not a rigorous criterion for translocation, particularly if the membrane-anchoring part is located at the N terminus. If embedding were to occur before the polypeptide chain is completed, ribosomes close to the 3’-end of the mRNA would appear to be membrane bound. For example, it is not yet clear whether cytochrome P-450reductase is embedded into or translocated across the endoplasmic reticulum membrane. It has a separate membrane-anchoring domain at its Nterminus (Black and Coon, 1982) which can be cleaved off without affecting the enzymatic activity of the hydrophilic domain (Gum and Strobel, 1979), but the protein is synthesized on membrane-bound polyribosomes (Okada et al., 1982). Since the molecular differences between embedding and signal peptides are not yet known, the knowledge of the amino acid sequence of a protein does not permit cogent conclusions. Generalizing from the knowledge of cytochrome b,, we have tentatively proposed (Bendzko et af., 1982) that for embedding of a protein there must exist a separate folded domain of the polypeptide, which is highly structured and has a hydrophobic surface. If such a globular structure approaches the phospholipid

1. APPLICATION OF THE SIGNAL HYPOTHESIS

33

bilayer, a perturbation is expected, thereby leading to spontaneous incorporation without mediation of a membrane protein.

C. Translocation Initiated by Signal Peptides Early data showed that many eukaryotic membrane proteins are synthesized in membrane-bound polyribosomes, as are secretory proteins. The similar overall pathways of the latter and plasma membrane proteins (Gumbiner and Kelly, 1982; Rotundo and Fambrough, 1980; Strous and Lodish, 1980) can also be taken as evidence for a common translocation mechanism. The first direct evidence for a signal peptide-initiated translocation of a membrane protein came from studies on the G protein of VSV (Katz et al., 1977a,b; Rothman and Lodish, 1977; Toneguzzo and Ghosh, 1977; Lingappa et al., 1978a). This transmembrane protein was shown to have a 16-residue hydrophobic sequence at its N terminus which is cleaved off during the translocation across the RER membrane. The identity of the peptide with a signal sequence of a secretory protein was proved by competition experiments in vitro (Lingappa et a / . , 1978a). Under conditions where there was no competition between the translation of the mRNAs coding for the G protein and for preprolactin, translocation of one precursor protein inhibited that of the other. Introducing the synchronized translation system, Rothman and Lodish (1977) showed that the G protein is cotranslationally incorporated into the RER membrane. Thus, the only difference between this membrane protein and a secretory protein appears to be a stop of the translocation of the former near its C-terminus. Indeed, the G protein contains a membrane-spanning segment near its C-terminus, thus leaving about 30 amino acid residues in the cytoplasm (Katz et al., 1978a,b; Rose et al.. 1980). Several other viral and cellular membrane proteins have cleavable N-terminal signal peptides and appear to follow the same way of membrane incorporation as the G protein of VSV (Dobberstein et al., 1979; Ploegh et a l ., 1979; Elder et al., 1979; Air, 1979; Porter et al., 1979; Wirth et a / . , 1977; Anderson and Blobel, 1981). The functional identity of signal sequences for secretory and membrane proteins is very convincingly exemplified by the occurrence of two forms of immunoglobulins. Secreted and membrane-bound IgM originate from the same gene by differential splicing and differ only in their C-termini (Alt et al., 1980; Rogers et al., 1980). Since the signal peptide is the same for both, it must function in the same way for secretion and membrane incorporation. Another proof for the identity was provided by artifically removing the part of

34

TOM A. RAPOPORT AND MARTIN WIEDMANN

the gene coding for the membrane-spanning segment of the hemagglutinin of the influenza virus (Gething and Sambrook, 1982; Sveda et al., 1982). When the shortened gene was introduced into mammalian cells, the protein was secreted rather than membrane incorporated. When, on the other hand, the N-terminal signal sequence was removed, neither secretion nor membrane incorporation occurred. Similar experiments with the cloned sequence of the G protein of VSV yielded identical results (Rose and Bergmann, 1982). More recently, experiments provided evidence for an identical mechanism of translocation on the molecular level both for secretory and membrane proteins. Anderson et al. (1982) showed that SRP functions in much the same way for the incorporation of the &subunit of the acetylcholine receptor into the RER membrane as it did for the translocation of preprolactin. SRP exerted a translational arrest and allowed integration of the membrane protein into the phospholipid bilayer when added to high-salt-washed membranes, which were otherwise incompetent. Signal sequences of prokaryotic membrane proteins also appear indistinguishable from those of their secretory counterparts. Accumulation of the malE-lacZ fusion product, which blocks translocation, leads to the accumulation of all precursors, be it of membrane or of secretory proteins (It0 et al., 1981). The pleiotropic secretion mutants also affect the biosynthesis of membrane proteins (see Michaelis and Beckwith, 1982). All this evidence clearly shows that the translocation process for secretory and some membrane proteins is initiated by functionally identical signal sequences. Indeed, looking at the amino acid sequences (Table 11),the hydrophobic cores of the peptides appear to be similar to those of secreted proteins.

D. Uncleaved and Internal Signal Sequences Whereas uncleaved and internal signal sequences appear to be exceptions in secretory proteins, they may be as frequent as cleaved ones in membrane proteins, at least of eukaryotes. Examples of membrane proteins synthesized without precursor include the p62 protein of the Sindbis and Semliki Forest viruses (Garoff et al., 1978; Bonatti and Blobel, 1979), opsin (Fung and Hubbell, 1978; Papermaster et al., 1980; Goldman and Blobel, 1981), cytochrome P-450 (BarNun er al., 1980), epoxide hydratase (Gonzales and Kasper, 1980), Ca2+ATPase of the sarcoplasmic reticulum (Mostov et al., 1981), band UI protein of erythrocytes (Sabban et a l . , 1981; Braell and Lodish, 1981), the major lens membrane protein (Paul and Goodenough, 1983), the y-subunit of the histocompatibility antigen (class 11) (B. Dobberstein, personal communication), and the neuraminidase of the influenza virus (Min Jou et al. 1980; Fields et al., 1981;

1 . APPLICATION OF THE SIGNAL HYPOTHESIS

35

Blok et a/., 1982). In prokaryotes, exported proteins without precursors are characterized less well but do exist (Achtman et al., 1979; Ehring et al., 1980; Nielsen et al., 1981; Michaelis and Beckwith, 1982). The fact that these membrane proteins have no precursors raises the question whether they have signal sequences at all. Indeed, in none of these examples has a signal peptide been explicitly identified. In some cases, e.g., for cytochrome P-450 (Fujii-Kuriyama et al., 1982), for the p62 precursors of Sindbis and Semliki Forest viruses (Garoff et al., 1980; Rice and Strauss, 198 I ) , and for the neuraminidase of influenza virus (Fields et al., 1981; Blok et al., 1982), the amino acid sequences are known and indicate a hydrophobic N-terminal peptide, similar to the cleaved signal peptides (see Table 111). If one takes into account that cleavage of the signal peptide is not required for membrane proteins which are normally cleaved (Lin er al., 1978), it is reasonable to assume that such hydrophobic segments are identical to signal peptides. However, definite proof is lacking in most cases. For the major lens membrane protein and for the Ca2+-ATPase it could be shown by Anderson et al. (1983) that SRP is required for membrane integration. In both cases, however, the SRP-induced arrest of translation was not detectable. A requirement of SRP for membrane integration and a SRP-induced translational arrest were found for the y-subunit of the histocompatibility antigen (B. Dobberstein, personal communication). Perhaps the arrest of translation is weaker if the signal sequence is more internally located. Nevertheless, the essential role of SRP for membrane integration is strong evidence for the existence of signal sequences somewhere in these polypeptides. As for secretory proteins, no internal signal peptide has so far been identified for a membrane protein. However, the existence of such peptides is likely. An internal signal sequence probably transports the glycoprotein El of Sindbis and Semliki Forest viruses (Fig. 3). This protein does not have a hydrophobic Nterminus but is preceded, according to the nucleotide sequence of the 26 S mRNA (Garoff et al., 1980; Rice and Strauss, 1981), by a hydrophobic peptide contained in a 6K protein. The latter has indeed been identified as a membrane protein (Welch and Sefton, 1980). It is therefore assumed that the C-terminus of the 6K protein is in fact the signal peptide for El which is cleaved off by the signal peptidase at the luminal side of the RER membrane. In a mutant of Semliki Forest virus, the precursor of E3 and E2 fails to be integrated into the membrane but the E l polypeptide is normally inserted, indicating that it has its own signal peptide (Hashimoto et al., 1981). This result also supports the conclusion that the signal sequence is internal during its function. The band I11 protein of erythrocytes has its N-terminal 450-500 amino acid residues in the cytoplasm (Fukuda et al., 1978). When a synchronized in vitro translation system was used, it was found that dog pancreatic membranes could

36

TOM A. RAPOPORT AND MARTIN WIEDMANN

FIG. 3. Scheme of the biosynthesis and membrane insertion of Semliki Forest virus structural proteins. The initial translation products of the 26 S RNA are the cytoplasmic capsid protein (C) and the membrane proteins p62, E l , and 6K. The 6K protein stays in the RER membrane while the other two membrane proteins are transported to the plasma membrane of the virus-infected cell. The p62 precursor is cleaved into E3 and E2 during the transport process. Presumed signal peptides are indicated by open cylinders, membrane-spanning segments by black ones. The p62 precursor probably contains at its N-terminus (E3 portion) an uncleaved signal sequence which is completely translocated across the membrane; the signal sequence for El is assumed to be contained in the 6K protein. Proteolytic cleavages are indicated by arrows.

be added much later than for proteins with a N-terminal signal sequence (up to the point where about 500 amino acid residues were polymerized) and translocation still occurred (Braell and Lodish, 1982b). One may, therefore, assume that in the case of band 111 the signal peptide is not at the N terminus but rather more internal-before residue 500. It is interesting that a membrane-spanning segment has been localized at this region (Fukuda et al., 1978), thus suggesting the identity of the latter with a signal peptide. The difficulty in identifying an internal signal sequence has been discussed for ovalbumin (see Section II,B,7). The existence of a stretch of hydrophobic amino acids is not a sufficient criterion. The size of a SRP-arrested protein fragment, if it occurs, may be helpful in localizing the peptide. It appears, however, most convincing when the presumed signal peptide is mutated or placed in front of a nontransported polypeptide, to prove its function. Whereas the translocation of proteins across the RER membrane and across the cytoplasmic membrane of bacteria is likely to be initiated by signal sequences and connected with the biosynthesis of the proteins, there are certainly other modes of transfer of proteins across membranes. These include not only the posttranslational protein import into mitochondria and chloroplasts which might be guided by signal peptides (see e.g., Reid, Chapter 7, this volume), but also the action of toxins on eukaryotic cells in which one of the subunits interacts with the membrane while the other penetrates into the interior of the cell (Pap-

1. APPLICATION OF THE SIGNAL HYPOTHESIS

37

penheimer, 1979). However, it may well turn out that there is a common principle involved in all cases of transfer of proteins across membranes. This common feature may be the receptor-mediated transient formation of a protein channel in the membrane through which the polypeptide is transported.

E. Membrane-Spanning Polypeptide Domains So far we have discussed the initiation of the translocation process and have shown that it is similar or identical to that of secretory proteins. In this section we consider the differences, i.e., the possible mechanism by which a membrane protein is retained in the membrane rather than completely translocated across. We start with a short consideration of the structure of the membrane-bound protein domains and turn later to mechanistic aspects of the biosynthesis. 1. SEQUENCE AND STRUCTURE OF THE POLYPEPTIDE DOMAINS

It is useful to distinguish between simple membrane segments which span the membrane once and more complex structures which involve extensive intramolecular protein-protein interactions within the phospholipid bilayer. The amino acid sequences of simple transmembrane segments have been determined for a number of proteins (some examples are given in Table 111; for a more comprehensive collection, see von Heijne, this volume). Direct sequencing of the peptides was performed after proteolysis under the assumption that the part incorporated into the membrane would be protected. Determination of the membrane-spanning part is also based on modification studies with water- or lipidsoluble reagents (see Warren, 1981). It is clear that by neither of these approaches can one determine the membrane-incorporated amino acids exactly to one residue. However, in other cases the membrane-spanning segments have only been inferred from the primary structure of the polypeptide on the basis of the criterion proposed by Segrest and Feldmann (1974). All the simple transmembrane segments consist of a long hydrophobic amino acid sequence which is followed at the cytoplasmic side by positively charged residues. Charged amino acids and amide groups (Gln and Asn) are excluded from the membrane-spanning segments except for its boundaries, where they sometimes occur but presumably reach the aqueous phase (see Table 111). These facts indicate that complete saturation of all H bonds within the membrane is essential. As discussed before (see Section II,B,6) saturation of all the H bonds within the membrane-spanning peptide implies a helical (presumably a-helicallike) conformation. One may estimate that a length of about 20 amino acid residues is required to span a membrane in an a-helical conformation. This is indeed the approximate length of membrane-spanning segments. In agreement

TABLE Iil POLYPEPI-IDE SEQUENCES OF INTERESTFOR TRANSLXATION Polypeptide

Sequence.

References

Membrane-spanningsegments at the C-terminus Immuuoglobulin pm chain E ~ E _ ~ N ~ ~ ~ ~ s ~ ~ ~ L ~ s ~ ~ Rogers _ er ~a/. (1980) K Glycophorin A ~ ~ ~ S ~ ~ I I F ~ V M A G V I G ~ ~ S Y ~ ~ - ~ ~ P ~ TomiIa and Marchesi (1975) VSV G protein ~ ~ W F S S W K S S I A S F G L n G L F L V L R V G ~ ~ Q Rose er a/. (1980) Influenza A hemagglutinin K ~ ~ K ~ Y K D w u w I s F A ~ s ~ w L ~ ~ A c Q K ~ ~ ~ ~ € ~ HMin IOU er nl. (1980) (Alvictoria/3/75) Influenza A hemagglutinin (Alaichi1268) ~DPV~SS_~Y~VUWFSFGASCFULALAVGLVF~CV~~N~~~~H Min Jou er d.(1980) Semliki Forest virus El VQF&XXX.GAFAIGAILVLW~RR-COOH Gamff er a / . (1980) Sindbis virus El FALFGGASSUllGLMIFACSMMLTS~~H Rice and Smuss (1981) Semliki Forest ~ I N SE2 GLyPAAT4ISAVVGMSLLALISlFAS~~~SK~ Gamff er a / . (1980) Sindbrs virus E2 ~PYTUAVASATVAMMIGWVAVLCAC~AI~~GCC Rice and Strauss (1981) Human histocompatibility antigen SQSTVPXVGXVAGXAVLAVWLGAW~~~SS~K~ Ploegh era/. (1981) Glycoprotein of avian sarcoma virus Czemilofsky er a/. (1980) Murim H-2K major histocompatibdity anbR ~ E ~ S T ~ I S N M A T V A V L V V L G A ~ G A W ~ T ~ Nathenson e t a / . (1981) gen Glycopmtein of rabies virus N W ~ K W L ~ A G A M I G L V L I F S ~ ~ ~ ~ - ~ E S Yelvenon era/. (1983) Membrane-spanningsegments at the N-terminus presumed to be signal peptides Influenza (Lee) neuraminidase NH,MLPST4IQ~~~TScCVLLSLYVSASLSn_LYS Shaw er a/. (1982) N H ~ ~ N Q I c - ~ ~ S _ v S ~ ~ ~ C ~ Q ~ U ~ Van Rompuy e t a / . (1982) Influenza (Victoria) neuraminidase Influenza (WSN) neuraminidase N H ~ N ~ N Q ~ G S I C M W G n S L U ~ ~ S ~ Van Rompuy er a/. (1982) Cyiochrome P-450 of ratb N H , ~ ~ L L V G F L L L L ~ S ~ K S R G ~ ~ P ~ ~ L ~ ~ ~Fujii-Kuriyama D R ~ N er a/. S (1982) ~ ~ ~ Protein segments translocated to the ectoplasmic phase Presumed signal peptides Semliki Forest virus p62 NH,SAPLITAMCVLA&ATF P P C V P C C ~ N - ~ Y ~D L ~L~AA D L T~C RYN ~ T I ~ ~ ~ Garoff er a [ . (1980) Sindbis virus p62 N H, s A A P L V T ~ c ~ - ~ ~ ~ ~ - ~ - s ~ - ~ ~ ~ ~ A u R c G Ricesands Strams ~ R (1981) s K Ovalbumin A c e t . G S I G A A S ~ C ~ V ~ L K V ~ ~ ~ Y C P ~ ~ ~ V ~ AMeek ~ Sera/. ~ (1982) Bovine opsinc NH$v&TkEF&&TaSFmYS -Papermaster er a/. (1980) Hydrophobic peptides Fusion peptide influenza HA, G L F G A I A G ~ G W E G M V I ~ G W Y G F R H Q N S E ~ ~ L K S ~ ~ ~ Ward ~ ~ and ~ ~ Dopheide (1979)

--

--

GGIGEWAV~IIILGLLLCLVWLLLVVCLPCLL~NI~RK~

_.-

Sucrase-isomaltased

A\INAFSGL~~IW.FVIVFIIATALIAVLAXXXPP

Frank et ol. (1978) ~

~~

Underlined amino acid residues are hydrophobic. Dots indicate charged amino acid residues. A plus indicates a glycosylation site. Possibly a signal peptide followed by a stop-transfer sequence. This sequence is located at the ectoplasmic side of the membrane but is not particularly hydrophobic. d The hydrophobic sequence is present in the mature protein which is located at the ectoplasmic side of the membrane and probably does not span the membrane. (Note the glycosylation site at position 11.) Q

b

K Y ~ D ~

1. APPLICATION OF THE SIGNAL HYPOTHESIS

39

with the occurrence of a helical structure in the membrane, Pro residues are rare in membrane-spanning segments (Table 111). For polypeptides spanning the membrane more than once, things might be a little different. The best-known example is bacteriorhodopsin. The complete amino acid sequence of the protein is known (Ovchinnikov et al., 1979; Khorana et ul., 1979) and a low-resolution, three-dimensional structure has been deduced from combined electron microscopic and X-ray crystallographic methods (Henderson and Unwin, 1975) and from neutron (Engelman and Zaccai, 1980) diffraction experiments. The molecule is composed of seven helical rods traversing the membrane almost perpendicularly [see, however, Jap et af. (1983) for an alternative view]. Most, but not all, of the charged amino acid residues of bacteriorhodopsin are located in the links between the helices in the aqueous phase. After fitting the amino acid sequence into the morphological model, the investigators postulated that the helices form a channel with all the charged residues oriented toward the interior away from the contact with the lipid. The nonpolar residues are directed outward to the surrounding hydrophobic environment. Such a structure has been called “inside-out’’ in comparison with soluble proteins (Engelman and Zaccai, 1980). Can the structure of bacteriorhodopsin be considered a prototype of a protein spanning the membrane several times? Many other membrane proteins have a high content of a-helix and may also be composed of helical rods. However, a completely different structure is evolving for the channel-forming proteins phoE and ompF in the outer membrane of E . coli. These proteins do not have any hydrophobic amino acid sequence longer than five residues in their mature chains (Overbeeke et af., 1983) and yet are deeply incorporated in the membrane. OmpF consists almost entirely of P-pleated sheets (Garavito and Rosenbusch, 1980; Schindler and Rosenbusch, 1981). These facts indicate that membrane proteins with folded domains within the phospholipid bilayer may have different structures. For thermodynamic reasons it is likely that shielding of charged residues from the hydrophobic environment, saturation of H bonds, and ion pair formation are basic principles in protein folding not only within a single polypeptide chain but also in the association of different chains. Obviously, many more structures of membrane proteins have to be known before generalizations can be made. It is hoped that the recent progress in the crystallization of membrane proteins (Michel and Oesterhelt, 1980) will allow the more widespread application of the powerful X-ray or neutron diffraction methods.

2. MEMBRANE-SPANNING SEGMENTS-ARETHEY STOP-TRANSFER PEPTIDES? It has been suggested (Blobel, 1980; Sabatini et al., 1982) that the translocation of a nascent chain, previously initiated by a signal sequence, is interrupted

40

TOM A. RAPOPORT AND MARTIN WIEDMANN

by a stop-transfer peptide. The latter would disintegrate the translocational apparatus, so that the C-terminal part following this sequence would be synthesized by free ribosomes and would remain in the cytoplasm. It is therefore tempting to identify the membrane-spanning amino acid sequences with stop-transfer peptides. So far this idea is supported only by circumstantial evidence. Many membrane proteins contain a cleavable N-terminal sequence and a hydrophobic membranespanning segment near the C-terminus. For some of them a cotranslational mode of membrane incorporation was demonstrated (Lingappa et al., 1978a; Wirth et al., 1977; Dobberstein et al., 1979), thus implying that initiation of translocation by the signal sequence and final membrane anchoring are consecutive events. There is little doubt that initiation of translocation and membrane incorporation are independent processes. However, so far it has not been directly shown that after synthesis of the stop-transfer peptide the ribosomes are released from the membrane. It is conceivable that the translocation machinery is not “switched off” until the polypeptide chain is completed. In fact, it is striking that many simple membrane proteins have 20-40 amino acid residues staying in the cytoplasm, which would correspond to the peptide buried within the ribosome when it has reached the termination codon. A notable exception to this rule is the secretory component which has a large C-terminal tail in the cytoplasm (Mostov et al., 1980). On the other hand, there exist extremely short tails (see Table 111) so that the hydrophobic sequence spanning the membrane must still be buried in the ribosome when the polypeptide is completed. Yost et al. (1983) have recently taken the nucleotide sequence coding for the membrane-spanning segment of immunoglobulin M, which is normally only followed by three cytoplasmic amino acid residues, and have inserted it between a sequence coding for a N-terminal portion of P-lactamase and the sequence for a-globin. After synthesis in vitro in the presence of microsomal membranes, the artificial protein had the expected orientation, with the p-lactamase part in the lumen and the P-globin part outside. This experiment clearly shows that the membrane-spanning segment does not have to be located at the C-terminus and that it carries the entire topogenic information, but it does not prove its function as a stop-transfer signal. Membrane-spanning peptides probably cannot be always stop-transfer sequences. For band 111, there is evidence that the membrane segment is identical with a signal peptide (Braell and Lodish, 1982). The neuraminidase of the influenza A virus contains a hydrophobic sequence at its N-terminus, which is presumably its signal peptide. There is no proteolytic processing involved in the biosynthesis of the protein, and the hydrophobic sequence is the membrane anchor (Fields et al., 1981; Blok et al., 1982). The aminopeptidase of the intestine has most of its polypeptide chain on the outside of the cell. A short N-

1. APPLICATION OF THE SIGNAL HYPOTHESIS

41

terminal sequence which appears to span the phospholipid bilayer serves as membrane anchor (Desnuelle, 1979). The y-subunit of the histocompatibility antigen (class 11) has 25 N-terminal amino acid residues in the cytoplasm. The following hydrophobic peptide spans the membrane and is most likely the signal sequence (B . Dobberstein, personal communication). In fact, the structural differences between signal peptides and membranespanning peptides (compare Tables I1 and 111) are not very obvious. Both contain a continuous stretch of hydrophobic amino acid residues, and many signal peptides contain at their N-terminal, presumably cytoplasmic side, charged, mainly positive residues like membrane proteins. Membrane-spanning segments appear to be more homogeneous in length than are signal peptides. There are signals as short as 15 residues, but no membrane-spanning peptide of this length has been found. The uncharged region is in general longer for the latter. Nevertheless, there are boundary cases in which signal and membrane-spanning peptides are very similar in structure. It is therefore expected (and found) that signal peptides which are not cleaved off function as membrane anchors. Given the similarities of signal peptides and membrane-spanning segments, one may assume that it is their position in a protein chain that determines their role. On the basis of the mechanism of recognition of signal peptides proposed in Section II,B,6, we would predict that the hydrophobic portion of a membranespanning segment, which is normally located at the C-terminus of a protein, assumes the role of a signal sequence when placed at the N-terminus of a polypeptide chain. This hypothesis should be testable by means of genetic engineering. Not all uncleaved signal peptides serve as membrane-spanning segments. For instance, the p62 precursors of E3 and E2 of Sindbis and Semliki Forest viruses have fairly hydrophobic seuqences at their N-termini which are presumed to be signal peptides (Garoff et al., 1978) (Table 111). They are retained in the completed p62 precursors and located in the lumen of the RER membrane. The hydrophobic sequence at the N-terminus of the isomaltase probably does not span the membrane since a preceding residue carries a carbohydrate chain which could have only been attached to the polypeptide chain inside the lumen of the intracellular membranes (Frank et al., 1978). These examples indicate that hydrophobic segments of a protein can be transferred across a phospholipid bilayer. A similar case is found for sequences in viral proteins which are involved in the fusion process between the virion and the cellular target membrane (Gething et al., 1978; Ward and Dopheide, 1979; Min Jou et al., 1980) (for instance, hemagglutinin HA2 of the influenza virus; see Table 111). The hydrophobic fusion peptides must have been transferred across the endoplasmic reticulum membrane of the host cell during their biosynthesis. The existence of uncleaved signal peptides in secretory proteins (see Section

(2)

FIG.4. Hypothetical cases of sequential membrane insertion of proteins with different topology. The schemes show the cotranslational incorporation of hypothetical polypeptides into the RER

1. APPLICATION OF THE SIGNAL HYPOTHESIS

43

II,B,7) also indicates that hydrophobic portions of a polypeptide chain can cross the phospholipid bilayer, a process which thermodynamically appears equally unfavorable as the transfer of hydrophilic parts. One may conclude that a hydrophobic sequence js not always a membrane-spanning segment. What additional features are required? One possibility is the requirement of positive charges immediately following the hydrophobic segment; another is the length of the uncharged portion of the polypeptide chain which appears to be shorter for some translocated hydrophobic segments (for instance the p62 precursors; see Table 111). The occurrence of Pro residues may also hinder a stable helix formation within the membrane. Some hydrophobic sequences may also be masked during their translocation across the membrane by interaction with other parts of the polypeptide chain in the case that entire domains of a protein are transported (Randall, 1983; see also later). For example, some hydrophobic fusion peptides become exposed only after cleavage of a precursor. Finally, it is interesting to note that carbohydrate chains are found at the N-termini of the p62 precursors of Sindbis and Semliki Forest viruses, of the isomaltase and of opsin, close to the presumed signal peptides. Possibly in these cases translocation of the signal sequences across the membrane to the luminal side is rendered irreversible by attachment of hydrophilic sugar residues at the ectoplasmic side of the membrane. 3. SEQUENTIAL INSERTION VERSUS DOMAIN INCORPORATION

Many possible membrane orientations of polytopic proteins can be explained by the assumption of alternating signal and stop-transfer peptides (Blobel, 1980; Sabatini er al., 1982). Some hypothetical examples are given in the schemes of Figs. 4 and 5.

membrane. For each case the final orientation of the protein in the membrane is given at the top. The upper side corresponds to the ectoplasmic side. The small black dots indicate the major domains of the polypeptides to both sides of the phospholipid bilayer. C and N denote the NH2- and COOHtemiini of the chains, respectively. For each case a possible order (from the N- to the C-terminus) of signal (Si) and stop-transfer (St) peptideb is given. Signal peptides can be either cleaved (Sic) or uncleaved. The latter may be further subdivided into those which are translocated across the membrane (SiNC,T)and those which span the menibrane (SiNC.NT). For each case there are shown different stages of the cotranslational incorporation of a polypeptide into the membrane. The black circles indicate the ribosomes. Cylinders denote either signal peptides or stop-transfer peptides. It should be noted that each signal sequence initiates the translocation of the succeeding part of the polypeptide chain until the process is halted by a stop-transfer peptide.

44

TOM A. RAPOPORT AND MARTIN WIEDMANN

C

N

FIG. 5 . Hypothetical examples for the sequential membrane insertion model which are more sophisticated than those in Fig. 4. Unlike the cases considered in Fig. 4, in scheme I , the signal peptide follows a stop-transfer peptide. Thus, the signal peptide must trigger the translocation of preceding parts of the polypeptide chain. It is likely that entire domains of the polypeptide are transferred across the membrane. In scheme 2, a polytopic protein is depicted with charges in its membrane-spanning segments which are oriented away from the lipid environment in the final state (“inside-out” structure). According to the sequential insertion model, the membrane spanning parts containing charged groups would be incorporated individually into the membrane in a defined order (see scheme). The incorporation of entire folded domains is an alternative to this model (see text). For symbols, see Fig. 4.

1. APPLICATION OF THE SIGNAL HYPOTHESIS

45

The assumptions inherent in this model can be summarized as follows and might be testable by experiments: 1. It is a model of sequential membrane incorporation. The topogenic sequences are decoded in a temporal order. 2. Membrane-spanning segments are functionally different: some are signal peptides, others stop-transfer peptides. 3. In the simplest formulation of the model, each topogenic sequence determines the location of the following, not of the preceding amino acid residues (see Fig. 4). In a more sophisticated version, signal sequences may also transport previously synthesized polypeptide segments (see Fig. 5). 4. Each membrane-spanning segment, or at most two consecutive ones, are incorporated individually into the membrane. This follows from the assumption that the translocation apparatus disassembles after each stop-transfer sequence. 5. Because binding of ribosomes to the RER membrane is essential for protein translocation, the ribosome should alternate between a bound and a free state during the synthesis of a polytopic protein. 6. According to the assumption of reinitiations of translocation in polytopic proteins, SRP would be needed several times. The dependence on SRP may be testable in synchronized translation experiments. As attractive as the model is, it is hardly based on experimental data. In fact, even conceptually there are some difficulties with special cases. For example, it is hard to see how a signal peptide can trigger the translocation of either the preceding or succeeding amino acid residues or possibly of both. In the case of opsin, which is synthesized without precursor and has a hydrophilic N-terminus (Table 111) located in the aqueous phase at the ectoplasmic side of the membrane (Papermaster et al., 1980), one wonders where the signal peptide might be and how it could have brought the N-terminus of the protein across the phospholipid bilayer. The final state of a membrane protein is probably the thermodynamically most stable one, as is also assumed for soluble globular proteins (Anfinsen, 1973; Privalow, 1979; Janicke, 1980; Pfeil, 1981). This may be concluded from the known structures of membrane-incorporated domains (see Section III,E, 1). Denaturation-renaturation experiments carried out for bacteriorhodopsin also support this assumption (Huang et al., 1981). It appears that in complex membrane proteins (e.g., ompF) the final state is stabilized by the interaction of several incorporated polypeptide segments which possibly may not even be perpendicular to the membrane plane, as depicted in the schemes in Figs. 4 and 5. The repeated assembly-disassembly of the translocation apparatus in the sequential insertion model, on the other hand, suggests a stepwise membrane integration of the polypeptide chain rather than the incorporation of entire structural domains. If the latter were the case, a single signal peptide would be sufficient, and stop-

46

TOM A. RAPOPORT AND MARTIN WIEDMANN

transfer sequences might not exist, although their presence would not contradict a model of domain incorporation. Recent data by Randall (1983) on the domain transfer of secretory proteins across the inner membrane of E . coli may be taken as tentative support for such a more thermodynamic model. It should be noted that the sequential insertion model may be supplemented by taking into account the presence of possible embedding sequences (Blobel, 1980). Since embedding sequences may in fact be complete folded domains, the model becomes then indistinguishable from the domain incorporation model. The proposal of a domain incorporation model has been made before by Wickner (1979) in his “membrane trigger hypothesis.” There the signal peptide merely serves to change the folding pathway of a protein, whereas according to our view, it switches on a translocation machinery. It should be noted that the asymmetrical orientation of a membrane protein is probably kinetically rather than thermodynamically determined. This assumption is supported by the fact that the natural orientation of a polypeptide in a membrane can be reproduced faithfully by in vitro translation experiments in the presence of microsomal vesicles which are leaky to small molecules (Meissner and Allen, 1981). One interesting possibility of mechanistic explanation would be the fixation within the membrane of a spanning segment of the nascent polypeptide chain in a defined orientation.

IV. PERSPECTIVES It is obvious that we are approaching a detailed knowledge of the translocation process for secretory proteins. The discoveries of SRP and of its receptor (docking protein) have provided important proof for predictions of the signal hypothesis. It is likely that other constituents of the transport machinery will be discovered in the near future, thus leading eventually to an understanding of the molecular mechanism by which a protein traverses a membrane. One may well expect within the next few years the in vitro reconstitution of the translocation process by use of purified components. Probably, whether and how a cell regulates the export of proteins will also be clarified. The knowledge on membrane proteins lags much behind that of secretory proteins, as evidenced by the existence of models and problems rather than of facts. Many more proteins, in particular complex membrane proteins such as ATPases, have to be studied in depth before firm conclusions can be drawn. Questions remain to be answered. What is the structure of membrane-spanning domains in polytopic proteins? Are there multiple signal peptides in these cases? If internal signal sequences exist, for which part of the polypeptide chain do they initiate translocation? It is likely that genetic and gene-engineering methods will greatly contribute to the progress in the field. They will provide proof of the

1. APPLICATION OF THE SIGNAL HYPOTHESIS

47

importance of a given sequence for the topography of a membrane protein. This approach could also be combined with kinetic studies in v i m using purified components of the translocation machinery and drugs or antibodies interfering at certain stages of the process. This outlook is by no means farfetched. Progress in this exciting field is so rapid that we fear the present review may be almost outdated when it appears. ACKNOWLEDGMENTS We thank G. Blobel, B. Dobberstein, G. Damaschun, R. Gilmore, W. Pfeil, I. Syllm-Rapoport, and S. M. Rapoport for critical reading of the manuscript and helpful suggestions. We are indebted to many colleagues who provided us with their unpublished results.

REFERENCES Achtman, M., Manning, P. A,, Edelbluth. C., and Herrlich, P. (1979). Export without proteolytic processing of inner and outer membrane proteins encoded by F sex factor tra cistrons in Escherichia coli minicells. Proc. Natl. Acad. Sci. U.S.A. 76, 4837-4841. Adelman, M. R., Blobel, G., and Sabatini, D. D. (1973). An improved cell fractionation procedure for the preparation of rat liver membrane-bound ribosomes. J . Cell Biol. 56, 191-205. Air, G . M. (1979). Nucleotide sequence coding for the “signal peptide” and N-terminus of the haemagglutinin from Asian (H2N2) strain of influenza virus. Virology 97, 468-472. Alt, F. W . , Bothwell, A. L. M., Knapp, M., Sieden, E., Mather, E., Koshland, M., and Baltimore, D. (1980). Synthesis of secreted and membrane-bound immunoglobulin Mu heavy chains is directed by mRNAs that differ at their 3’-ends. Cell 20, 293-301. Anderson, D. J., and Blobel, G. (1981). In vitro synthesis, glycosylation, and membrane insertion of the four subunits of Torpedo acetylcholine receptor. Proc. Natl. Acad. Sci. U.S.A. 78, 55985602. Anderson, D. J . , Walter, P., and Blobel, G. (1982). Signal recognition protein is required for the integration of acetylcholine receptor &subunit, a transmembrane glycoprotein, into the endoplasmic reticulum membrane. J . Cell Eiol. 93, 501-506. Anderson, D. J., Mostov, K. E., and Blobel, G . (1983). Mechanism of integration of de novosynthesized polypeptides into membranes: Signal-recognition particle is required for integration into microsomal membranes of calcium ATPase and of lens MP26 but not of cytochrome b5. Proc. Natl. Acad. Sci. U.S.A. 80, 7249-7253. Anfinsen, C. B. (1973). Principles that govern the folding of protein chains. Science 181, 223-230. Austen, B. M. (1979). Predicted secondary structures of amino-terminal extension sequences of secreted proteins. FEBS Lett. 103, 308-313. Baglioni, C., Bleiberg, I., and Zauderer, M. (1971). Assembly of memb,.ane-bound polyribosomes. Nature (London) New Eiol. 232, 8-12. Bar-Nun, S., Kreibich, G . , Adesnik, M., Alterman, L., Negishi, M., and Sabatini, D. D. (1980). Synthesis and insertion of cytochrome P-450 into endoplasmic reticulum membranes. P roc. Natl. Acad. Sci. U.S.A. 77, 965-969. Bassford, P. J . , Jr., Silhavy, T. J., and Beckwith, J. R. (1979). Use of gene fusion to study secretion of maltose-binding protein into Escherichia coli periplasm. J . Bacteriol. 139, 19-31, Bassuner, R., Wobus, U., and Rapport, T. A. (1984). Signal recognition particle triggers the translocation of storage globulin polypeptides from field beans (Vicia faba L. ) across mammalian endoplasmic reticulum membrane. FEBS Lett. 166, 3 14-320.

48

TOM A. RAPOPORT AND MARTIN WIEDMANN

Baty, D., Mercereau-Puijalon, 0.. Perrin, D., Kourilsky, P., and Lazdunski, C. (1981). Secretion in the bacterial periplasmic space of chicken ovalbumin synthesized in Escherichia coli. Gene 16, 79-87. Beck, E., and Bremmer, E. (1980). Nucleotide sequence of the gene ompA coding for the outer membrane protein I1 of E. coli K-12. Nucleic Acids Res. 8, 301 1-3024. Bedouelle, H . , Bassford, P. J., Fowler, A. V . , Zabin, I., Beckwith, J., and Hofnung, M. (1980). Mutations which alter the function of the signal sequence of the maltose-binding protein of Escherichia coli. Nature (London) 285, 78-8 I . Bendzko, P., Prehn, S., Pfeil, W., and Rapoport, T. A. (1982). Different modes of membrane interactions of the signal sequence of carp preproinsulin and of the insertion sequence of rabbit cytochrome 6 5 . Eur. J . Biochem. 123, 121-126. Benson, S. A,, and Silhavy, T. J. (1983). Information within the mature Lam B protein necessary for localization within the outer membrane of E . coli K 1 2 . Cell 32, 1325-1335. Bergman, L. W., and Kuehl, W. M. (1979). Formation of an intrachain disulfide bond on nascent immunoglobulin light chains. .I. Biol. Chem. 254, 8869-8876. Black, S. D., and Coon, M. I. (1982). Structural features of liver microsomal NADPH-cytochrome P-450 reductase. J . Biol. Chem. 257, 5929-5938. Blobel, G. (1976). Extraction from free ribosomes of a factor mediating ribosome detachment from rough microsomes. Biochem. Biophys. Res. Commun. 68, 1-7. Blobel, 0. (1980). Intracellular protein topogenesis. Proc. Narl. Acad. Sci. U.S.A. 77, 1496-1500. Blobel, G. (1982). Regulation of intracellular protein traffic. Cold Spring Harbor Symp. Quanf. Biol. 46,7- 16. Blobel, G., and Sabatini, D. D. (1970). Controlled proteolysis of nascent polypeptides in rat liver cell fractions. I. Location of the polypeptides within ribosomes. J . Cell Biol. 45, 130145. Blobel, G., and Sabatini, D. D. (1971). Ribosome-membrane interaction in eukaryotic cells. In "Biomembranes 2" (L. A. Manson, ed.), pp. 193-195. Plenum, New York. Blobel, G., and Dobberstein, B. (1975a). Transfer of proteins across membranes. I. Presence of proteolytically processed and unprocessed nascent immunoglobulin light chains on membrdnebound ribosomes of murine myeloma. J . Cell Biol. 67, 835-85 I. Blobel, G . , and Dobberstein, B. (1975b). Transfer of proteins across membranes. 11. Reconstitution of functional rough microsomes from heterologous components. J . Cell Biol. 67, 852-862. Blok, I., Air, G. M., Laver, W. G., Ward, C. W., Lilley, G . G . , Woods, E. F., Roxburgh, C. M., and Inglis, A. S. (1982). Studies on the size, chemical composition and partial sequence of the nueraminidase (NA) from type A influenza viruses show that the N-terminal region of the NA is not processed and serves to anchor the NA in the viral membrane. Virology 119, 109-121. Boime, I . , Boguslawski, S . , and Caine, J. (1975). Translation of a human placental lactogen mRNA fraction in heterologous cell-free systems: The synthesis of a possible precursor. Biochem. Biophys. Res. Commun. 62, 103-109. Bonatti, S . , and Blobel, G. (1979). Absence of a cleavable signal sequence in Sindbis virus glycoprotein PE2. J . Biol. Chem. 254, 12261-12264. Borgese, D., Blobel, G., and Sabatini, D. D. (1973). In vifro exchange of ribosomal subunits between free and membrane-bound ribosomes. J . Mol. Biol. 74, 415-438. Borgese, N., Mok, W., Kreibich, G., and Sabatini, D. D. (1974). Ribosome-membrane interaction: In vitro binding of ribosomes to microsomal membranes. J . Mol. Biol. 88, 559-580. Borgese, N., Pietrini, G., and Meldolesi, J. (1980). Localization and biosynthesis of NADHcytochrome 65 reductase, an integral membrane protein, in rat liver cells. 111. Evidence for the independent insertion and turnover of the enzyme in various subcellular compartments. J . Cell Biol. 86, 38-45. Braell, W. H., and Lodish, H. F. (1981). Biosynthesis of the erythrocyte anion transport protein. J . Biol. Chem. 256, 11337-11344.

1. APPLICATION OF THE SIGNAL HYPOTHESIS

49

Braell. W. H., and Lodish. H. F. (1982a). Ovalbumin utilizes a NH2-terminal signal sequence. J. B i d . Chem. 257, 4578-4582. Braell. W. H., and Lodish, H. F. (1982b). The erythrocyte anion transport protein is cotranslationally inserted into microsomes. Cell 28, 23-3 I . Brownlee. G . G . (1971). Sequence of 6 s RNA of E . coli. Nuture (London) New Biol. 229, 147149. Brunner, J . , Hauser, H., Braun, H., Wilson, K. J . , Wacker, H.. O’Neill, B., and Senienza, G. (1978). The mode of association of the enzyme complex sucrase-isomaltase with the intestinal brush border membrane. J. Biol. Chem. 254, 1821-1829. Burstein, Y . , and Schechter, I. (1978). Primary structures of N-terminal extra peptide segments linked to the variable and constant regions of immunoglobulin light chain precursors: Implications on the organization and controlled expression of immunoglobulin genes. Biochemisrry 17, 2392-2400. Carlson, M., and Botstein, D. (1982). Two differentially regulated mRNAs with different 5‘-ends encode secreted and intracellular forms of yeast invertase. Cell 28, 145-154. Chan, S. J., Keim, P., and Steiner, D. F. (1976). Cell-free synthesis of rat preproinsulins: Chardcterization and partial amino acid sequence determination. Proc. Nut/. Acud. Sci. U.S.A. 73, 1964- 1968. Chan. S. J . , Patzelt, C.. Duguid, J., Quinn, P., Labrecque, A., Noyes, B., Keim, P., Heinrikson, R . L., and Steiner, D. F. (1979). Precursors in the biosynthesis of insulin and other peptide hormones. Miami Winter Symp.. 11th 16. Chou, P. Y., and Fasman, G.D. (1978). Empirical predictions of protein conformation. Annu. Rev. Biochem. 47, 251-276. Claudio, T., Ballivet, M., Patrick, J . , and Heinemann, S. (1983). Nucleotide and deduced amino acid sequences of Torpedo cul@wnicu acetylcholine receptor gamma-subunit. Proc. Nurl. Acud. Sci. U . S . A . 80, 1111-1115. Colman, A , , and Morser, J . (1979). Export of proteins from oocytes of Xenopus luevis. Cell 17, 517-526. Colman, A,, Lane, C. D., Craig, R., Boulton, A . . Mohun, T . , and Morser, J . (1981). The intluence of topology and glycosylation on the fate of heterologous secretory proteins made in Xeizopus oocytes. Eur. J. Biochem. 113, 339-348. Coleman, J . , Inukai, M., and Inouye, M. (1983). A functional internalized signal peptide. Cell ( I n press). Czernilofsky, A. P., Levinson, A. D., Varmus, H. E., Bishop, J . M., Tischer, E., and Goodman, H. M. (1980). Nucleotide sequence of an avian sarcoma virus oncogene (SRC) and proposed amino acid sequence for gene product. Nuture (London) 287, 198-203. Dailey, H. A,, and Strittmatter, P. (1978). Structural and functional properties of the membrane binding segment of cytochrome bs. J. Biol. Chem. 253, 8203-8209. Daniels, C. J . , Bole, D. G., Quay, S . C . , and Oxender. D. L. (1981). Role for membrane potential in the secretion of proteins into the periplasm of Escherichiu coli. Proc. Natl. Acud. Sci. U.S.A. 78, 5396-5400. Date, T., Zwizinski, C., Ludmerer, S., and Wickner, W. (1980a). Mechanisms of membrane assembly: Effects of energy poisons on the conversion of soluble M 13 coliphage procoat to membrane-bound coat proteins. Proc. Nurl. Acud. Sci. U.S.A. 77, 827-831. Date, T., Goodman. J . M., and Wickner. W. (1980b). Precoat, the precursor of M 13 coat protein, requires an electrochemical potential for membrane insertion. Proc. Nut/. Acud. Sci. U.S.A. 77, 4669-4673. Davis, 8. D., and Tai, P. C. (1980). The mechanism of protein secretion across membranes. Nuture (London) 283, 433-438. Desnuelle. P. (1979). Intestinal and renal aminopeptidase: A model of a transmembrane protein. Eur. J . Biochem. 101, 1-1 I .

50

TOM A. RAPOPORT AND MARTINWIEDMANN

Devillers-Thiery, A,, Kindt, T . , Scheele, G., and Blobel, G. (1975). Homology in amino terminal sequences of precursors to pancreatic secretory proteins. Proc. Natl. Acad. Sci. U.S.A. 72, 5016-5020. Di Rienzo, J. M., and Inouye, M. (1979). Lipid fluidity-dependent biosynthesis and assembly of the outer membrane proteins of E. coli. Cell 17, 155- 161. Dobberstein, B., Garoff, H., Warren, G., and Robinson, P. J. (1979). Cell-free synthesis and membrane insertion of mouse H-2 Dd histocompatibility antigen and P2-microglobulin. Cell 17, 759-169. Dobberstein, B., Kvist, S., and Roberts, L. (1982). Structure and biosynthesis of histocompatibility antigens (H-2, HLA). Philos. Trans. R . Soc. London Ser. B 300, 161-172. Dolgikh, D., Gilmanshin, R. I., Brazhnikov, E. V., Bychkova, V. E., Semisotnov, G. V., Venyaminov, S. Yu.,and Ptitsyn, 0. B. (1981). a-Lactalbumin: Compact state with fluctuating tertiary structure? FEES Lett. 136, 31 1-315. Dunn, R., McCoy, J., Simsek, M., Majumdar, A., Chang. S. A., RajBhandary, U. L., and Khorana, G. (1981). The bacterio-rhodopsin gene. Proc. Natl. Acad. Sci. U.S.A. 78, 67446748. Ebina, Y., Kishi, F., Miki, T., Kagamiyama, H., Nakazawa, T., and Nakazawa, A. (1981). The nucleotide sequence surrounding the promotor region of colicin El gene (SOS functions, sequence homology, recA operator, transcription analysis). Gene 15, 119-126. Ehring, R., Beyreuther, K., Wright, J. K., and Overath, P. (1980). In vitro and in vivo products of E. coli lactose permease gene are identical. Nature (London) 283, 537-540. Elder, K. T., Bye, J. M., Skehel, J. J., Waterfield, M. D., and Smith, A. E. (1979). In vitro synthesis, glycosylation and membrane insertion of influenza virus haemagglutinin. Virology 95, 343-350. Emr, S. D., and Silhavy, T. J. (1980). Mutations affecting localization of an Escherichia coli outer membrane protein, the bacteriophage lambda receptor. J . Mol. B i d . 141, 63-90. Em, S. D., and Silhavy, T. J. (1982). Molecular components of signal sequence that function in the initiation of protein export. J. Cell Biol. 95, 689-696. Emr, S. D., and Silhavy, T. J. (1983). Importance of secondary structure in the signal sequence for protein secretion. Proc. Natl. Acad. Sci. U.S.A. 80, 4599-4603. Emr, S. D., Hall, M. N., and Silhavy, T. I. (1980a). A mechanism of protein localization: The signal hypothesis and bacteria. J. Cell Biol. 86, 701-71 1 . E m , S. D., Hedgpeth, J., Clement, J. M., Silhavy, T. J., and Hofnung, M. (1980b). Sequence analysis of mutations that prevent export of lambda receptor, an Escherichia coli outer membrane protein. Narure (London) 285, 82-85. E m , S. D., Hanley-Way, S., and Silhavy, T. J. (1981). Suppressor mutations that restore export of a protein with a defective signal sequence. Cell 23, 79-88. Engelman, D. M . , and Steitz, T. A. (1981). The spontaneous insertion of proteins into and across membranes: The helical hairpin hypothesis. Cell 23, 41 1-422. Engelman, D. M., and Zaccai, G. (1980). Bacteriorhodopsin is an inside-out protein. Proc. Natl. Acad. Sci. U.S.A. 77, 5894-5898. Enoch, H. G., Fleming, P. J., and Strittmatter, P. (1979). The binding of cytochrome bs to phospholipid vesicles and biological membranes. J. B i d . Chem. 254, 6483-6488. Erickson, A. H., Walter, P., and Blobel, G. (1983). Translocation of lysosomal enzyme across the microsomal membrane requires signal recognition particle. Biochem. Biophys. Res. Cornmun. 115, 275-280. Farquhar, M., Bergeron, J. J. M., and Palade, G. E. (1974). Cytochemistry of Golgi fractions prepared from rat liver. J . Cell Biol. 60, 8-25. Ferro-Novick, S., Novick, P., Field, C., and Schekman, R. (1983a). Yeast secretory mutants that block the formation of active cell surface enzymes. J . Cell Biol. 98, 35-43.

1. APPLICATION OF THE SIGNAL HYPOTHESIS

51

Ferro-Novick, S . , Hansen. W., Schauer, I.. and Schekman, R. (1983b). Genes required for completion of import of proteins into the endoplasmic reticulum in yeast. J . Cell B i d . 98, 4453. Ferro-Novick, S . , Honma, M.. and Beckwith, J. (1984). The product of gene sec C is involved in the synthesis of exported proteins in E . coli. Cell 38, 21 1-217. Fields, S . , Winter, G., and Brownlee, G. G. (1981). Structure of the ncuraminidase gene in human influenza virus A/PR/8/34. Nature (London) 290, 213-217. Finkelstein, A. V., and Ptitsyn, 0. B. (1977). Theory of protein molecule self-organization. 111. A calculating method for the probabilities of the secondary structure formation in an unfolded polypeptide chain. Biopo1,ymers 16, 525-529. Finkelstein, A. V., Bendzko, P., and Rapoport, T. A. (1983). Recognition of signal sequences. FEBS Lett. 161, 176-179. Fleming, P. J., and Strittmatter, P. (1978). The nonpolar peptide segment of cytochrome bs. Binding to phospholipid vesicles and identification of the fluorescent tryptophanyl residue. J. Biol. Chem. 253, 8198-8202. Frank, G., Brunner, J . , Hauser, H., Wacker, H., Semenza, G., and Zuber, H. (1978). The hydrophobic anchor of small-intestinal sucrase-isomaltase. N-terminal sequence of the isomaltase subunit. FEBS Lett. %, 183-188. Fraser. T. H., and Bruce, B. J. (1978). Chicken ovalbumin is synthesized and secreted by Escherichia coli. Pro(,. Natl. Acad. Sci. U.S.A. 75, 5936-5940. Fujii-Kuriyama, Y., Mizukami. Y . . Kawajiri, K., Sogawa, K., and Muramatsu, M. (1982). Primary structure of a cytochrome P-450: Coding nucleotide sequence of phenobarbital-inducible cytochrome P-450 cDNA from rat liver. Proc. Natl. Arad. Sci. U . S . A . 79, 2793-2797. Fukuda, M., Eschdat, Y . , Tarone, G . , and Marches;, V. T. (1978). Isolation and characterization of peptides derived from the cytoplasmic segment of Band 3, the predominant intrinsic membrane protein of the human erythrocyte. J. B i d . Chem. 253, 2419-2428. Fung, B . , K.-K., and Hubbell, W. L. (1978). Organization of rhodopsin in photoreceptor membranes. 2. Transniembrane organization of bovine rhodopsin: Evidence from proteolysis and lactoperoxidase-catalyzed iodination of native and reconstituted membranes. Biochemisrrv 17, 4403-4410. Garavito, R. M., and Rosenbusch, J. P. (1980). Three-dimensional crystalls of an integral membrane protein: An initial X-ray analysis. J . Cell Biol. 86, 327-329. Gamier, I., Gaye, P., Mercier, J.-C., and Robson, B. (1980). Structural properties of signal peptides and their membrane insertion. Bioclzimie 62, 23 1-239. Garoff, H., Simons, K., and Dobberstein, B., (1978). Assembly of the Semliki Forest Virus membrane glycoproteins in the membrane of the endoplasmic reticulum in vitro. J. Mol. B i d . 124, 587-600. Garoff, H.. Frischauf. A. M., Simons, K., Lehrach, H., and Delius, H. (1980). Nucleotide sequence of cDNA coding for Semliki Forest Virus membrane glycoproteins. Nature (London) 288, 236241. Gaye, P., Gautron, J. P., Mercier, J.-C., and Haze, G . (1977). Amino terminal sequence of the precursors of ovine caseins. Biochon. Biophys. Res. Commun. 79, 903-9 10. Gething, M.-J., and Sarnbrook, J. (1982). Construction of influenza haemagglutinin genes that code for intracellular and secreted forms of the protein. Nature (London) 300, 598-603. Gething, M.-J., White, J. M., and Watertield, M. D. (1978). Purification of the fusion protein of Sendai Virus: Analysis of the NH2-terminal sequence generated during precursor activation. Proc. Natl. Acad. Sci. U.S.A. 75, 2731-2740. Gething, M.-J., Bye, J., Skehel, J . , and Waterfield, M. (1980). Cloning and DNA sequence of double-stranded copies of haemagglutinin genes from H2 and H3 strains elucidates antigenic shift and drift in human influenza virus. Nature (London) 287, 301-306.

52

TOM A. RAPOPORTAND MARTIN WIEDMANN

Gilmore, R., and Blobel, G. (1983). Transient involvement of signal recognition particle and its receptor in the microsomal membrane prior to protein translocation. Cell 35, 677-685. Gilmore, R., Blobel, G., and Walter, P. (1982a). Protein translocation across the endoplasmic reticulum. I. Detection in the microsomal membrane of a receptor for the signal recognition particle. J. Cell B i d . 95, 463-469. Gilmore, R., Walter, P., and Blobel, G. (1982b). Protein translocation across the endoplasmic reticulum. 11. Isolation and characterization of the signal recognition particle receptor. J . Cell B i d . 95, 470-477. Glabe, C. G . , Hanover, J. A,, and Lennarz, W. J. (1980). Glycosylation of ovalbumin nascent chains. J. B i d . Chem. 255, 9236-9242. Goldmann, B. M., and Blobel, G. (1981). In vitrv biosynthesis, core glycosylation, and membrane integration of opsin. J. Cell B i d . 90, 236-242. Gonzales, F. J., and Kasper, C. R. (1980). In vitrv translation of epoxide hydratase messenger RNA. Bivchem. Bivphys. Res. Cvmmun. 93, 1254-1258. Gross, V., Kaiser, C., Tran-Thi, T.-A,, Schmelzer, E., Witt, I., Plummer, T. H., Jr., and Heinrich, P. C. (1983). N-terminal amino acid sequences of precursor and mature forms of alpha-Iantitrypsin. FEES Lett. 151, 201-205. Gum, J. R., and Strobel, H. W. (1979). Purified NADPH cytochrome P-450 reductase: Interaction with hepatic microsomes and phospholipid vesicles. J . B i d . Chem. 254, 4177-4185. Gumbiner, B., and Kelly, R. B. (1982). Two distinct intracellular pathways transport secretory and membrane glycoproteins to the surface of pituitary tumor cells. Cell 28, 51-59. Gundelfinger, E. D., Krause, E., Melli, H., and Dobberstein, B. (1983). The organization of the 7SL-RNA in the signal recognition particle. Nucleic Acids Res. 11, 7363-7374. Habener, J. F., Potts, J. T., and Rich, A. (1976). Pre-proparathyroid hormone. Evidence for an early biosynthetic precursor of proparathyroid hormone. J. B i d . Chem. 251, 3893-3899. Habener, J. F., Rosenblatt, M., Kemper, B., Kronenberg, H. M., Rich, A., and Potts, J. T. Jr. (1978). Pre-proparathyroid hormone: Amino acid sequence, chemical synthesis, and some biological studies of the precursor region. Prvc. Natl. Acad. Sci. U.S.A. 75, 2616-2620. Habener, J. F., Maunus, R . , Dee, P. C., and Potts, J. T., Jr. (1980). Early events in the cellular formation of proparathyroid hormone. J. Cell B i d . 85, 292-298. Hahn, V., Winkler, J., Rapoport, T. A., Liebscher, D.-H., Coutelle, Ch., andRosentha1, S. (1983). Carp preproinsulin cDNA sequence and evolution of insulin genes. Nucleic Acids Res. 11, 4541-4552. Halegoua, S., and Inouye, M. (1979). Translocation and assembly of outer membrane proteins of Escherichia coli. Selective accumulation of precursors and novel assembly intermediates caused by phenylethyl alcohol. J. Mvl. B i d . 130, 39-61. Hall, M. N., Schwartz, M., and Silhavy, T. J. (1982). Sequence information within the lamB gene is required for proper routing of the bacteriophage lambda receptor protein to the outer membrane of Escherichia coli K12. J . Mvl. B i d . 156, 93-112. Hashimoto, K., Erdei, S., Keranen, S., Saraste, J., and Kaariainen, L. (1981). Evidence for a separate signal sequence for the carboxy-terminal envelope glycoprotein El of Semliki Forest Virus. J . Virvl. 38, 34-40. Hasilik, A. (1980). Biosynthesis of lysosomal enzymes. Trends Bivchem. Sci. 5, 237-240. Hauri, H. P., Wacker, H., Rickli, E. E., Bigler-Meier, B., Quaroni, A,, and Semenza, G. (1982). Biosynthesis of sucrase-isomaltase. Purification and NH2-terminal amino acid sequence of the rat sucrase-isomaltase precursor (prosucrase-isomaltase) from fetal intestinal transplants. J . B i d . Chem. 257, 4522-4528. Hedgpeth, J., Clement, J.-M., Marchal, C., Perrin, D., and Hofnung, M. (1980). DNA sequence encoding the NHz-terminal peptide involved in transport of A-receptor, an E. coli secretory protein. Prvc. Natl. Acad. Sci. U.S.A. 77, 2621-2625.

1. APPLICATION OF THE SIGNAL HYPOTHESIS

53

Heinrich, P. C. (1982). Proteolytic processing of polypeptides during the biosynthesis of subcellular structures. Rev. Phvsiol. Biochem. Pharmacol. 93, 116- 187. Henderson, R., and Unwin, P. N. T. (1975). Three-dimensional model of purple membrane obtained by electron microscopy. Nature (London) 257, 28-30. Higgins, C. F., and Ames, G. F.-L. (1981). Two periplasmic transport proteins which interact with a common membrane receptor show extensive homology: Complete nucleotide sequences. Proc. Natl. Acad. Sci. U.S.A. 78, 6038-6042. Hortin, G . , and Boime, I. (1980). Inhibition of preprotein processing in ascites tumor lysates by incorporation of a leucine analog. Proc. Narl. Acad. Sci. U.S.A. 77, 1356-1360. Hortin, G . . and Boime, I. (1981). Miscleavage at the presequence of rat preprolactin synthesized in pituitary cells incubated with a threonine analog. Cell 24, 453-461. Huang, K.-S.. Bayley, H., Liao, M.-J., London, E., and Khorana, H. G. (1981). Refolding of an integral membrane protein. Denaturation, renaturation, and reconstitution of intact bacteriorhodopsin and two proteolytic fragments. J . Biol. Chem. 256, 3802-3809. Hussain, M., Ichihara, S., and Mizushima, S. (1980). Accumulation of glyceride-containing precursor of the outer membrane lipoprotein in the cytoplasmic membrane of Escherichiu coli treated with globomycin. J . Biol. Chem. 255, 3707-3712. Hussain, M., Ozawa, Y.,Ichihara, S . , and Mizushima, S. (1982). Signal peptide digestion in Escherichia coli. Eur. J . Biochem. 129, 233-239. Ingelmann-Sundberg, M., and Glaumann, H. (1980). Incorporation of purified components of the rabbit liver microsomal hydroxylase system into phospholipid vesicles. Biochim. Biophys. Acra 599, 417-435. Inouye, H., and Beckwith, J . (1977). Synthesis and processing of an E. coli alkaline phosphatase precursor in vitro. Proc. Nu//. Acad. Sci. U.S.A. 74, 1440-1444. Inouye, M., and Halegoua, S. (1980). Secretion and membrane localization of proteins in Escherichia coli. Crir. Rev. Biochem. 7, 339-37 I . Inouye, S., Wang, S., Sekizawa, J . , Halcgoua, S., and Inouye, M. (1977). Aminoacid sequence for the peptide extension on the prolipoprotein of the E . coli outer membrane. Proc. Natl. Acad. Sci. U.S.A. 74, 1004-1008. Inouye. S., Soberon, X . , Franceschini, T . , Nakamura, K.. Itakura, K . , and Inouye, M. (1982). Role of positive charge on the amino-terminal region of the signal peptide in protein secretion across the membrane. Proc. Nail. Acad. Sci. U.S.A. 79, 3438-3441. Inouye, S., Franceschini, T . , Sato, M., Itakura, K.. and Inouye, M. (1983a). Prolipoprotein signal peptidase of Escherichia coli requires a cysteine residue at the cleavage site. EMBO J . 2, 8791. Inouye, S., Hsu, T., Itakura, K., and Inouye, M. (1983b). Requirements for signal peptides cleavage of Escherichia roli prolipoprotein. Science 221, 59-61. Inukai, M., Nakajima, M., Osawa, M., Haneishi, T., and Arai, M. (1978). Globomycin, a new peptide antibiotic with speroplast-forming activity. 11. Isolation and physico-chemical an biological characterization. J . Anribior. 31, 421-425. Ito, K., and Beckwith, J. R. (1981). Role of the mature protein sequence of maltose-binding protein in its secretion across the E. coli cytoplasmic membrane. Cell 25, 143-150. Ito, K., Date, T., and Wickner, W. (1980). Synthesis, assembly into the cytoplasmic membrane, and proteolytic processing of the precursor of coliphage M 13 coat protein. J . Biol. Chem. 255, 2123-2130. Ito, K . , Bassford, P., and Beckwith, J . (1981). Protein localization in E. coli: Is there a common step in the secretion of periplasmic and outer-membrane proteins? Cell 24, 707-717. Jackson, R. C., and Blobel, G. (1977). Posttranslational cleavage of presecretory proteins with an extract of rough microsomes from dog pancreas containing signal peptidase activity. Proc. Nail. Acad. Sci. U . S . A . 74, 5598-5602.

54

TOM A. RAPOPORT AND MARTIN WIEDMANN

Jackson, R. C., Walter, P., and Blobel, G. (1980). Secretion requires a cytoplasmically disposed sulphydryl of the RER membrane. Nature (London) 286, 174-176. Jaenicke, R. (1980). “Protein Folding. ” Elsevier, Amsterdam. Jap, B. K., Maestre, M. F., Hayward, S. B . , and Glaeser, R. M. (1983). Peptid-chain secondary structure of bacteriorhodopsin. Biophys. J . 43, 8 1-89. Jaurin, B., and Grundstrom, T. (1981). AmpC cephalosporinase of E. coli K-12 has a different evolutionary origin from that of a-lactamases of the penicillinase type. Proc. Natl. Acad. Sri. U.S.A. 78, 4897-4901. Josefsson, L.-G., and Randall, L. L. (1981a). Processing in vivo of precursor maltose-binding protein in Escherichia coli occurs post-translationally as well as co-translationally . J. Biol. Chem. 256, 2504-2507. Josefsson, L.-G., and Randall, L. L. (1981b). Different exported proteins in E. coli show differences in the temporal mode of processing in vivo. Cell 25, 151-157. Katz, F. N., Rothman, J. E., Knipe, D. M., and Lodish, H. F. (1977a). Membrane assembly: Synthesis and intracellular processing of the vesicular stomatitis viral glycoprotein. J. Supramol. Struct. 7, 353-370. Katz, F. N., Rothman, J. E., Lingappa, V. R., Blobel, G., and Lodish, H. F. (1977b). Membrane assembly in vitro: Synthesis, glycosylation and asymmetric insertion of a transmembrane protein. Proc. Natl. Acad. Sci. U.S.A. 74, 3278-3282. Kemper, B., Habener, J. F., Mulligan, R. C., Potts, J. T., Jr., and Rich, A. (1974). Pre-proparathyroid hormone: A direct translation product of parathyroid mRNA. Proc. Natl. Acad. Sci. U.S.A. 71, 3731-3735. Khorana, H. G., Gerber, G. E., Herlihy, W . C., Gray, C. P., Anderegg, R. J., Nihei, K., and Biemann, K. (1979). Amino acid sequence of bacteriorhodopsin. Proc. Natl. Acad. Sci. U . S . A . 76, 5046-5050. Kikuchi, Y., Yoda, K., Yamasaki, M., and Tamura, G. (1981). The nucleotide sequence of the promoter and the aminoterminal region of alkaline phosphatase structural gene (PhoA) of E. coli. Nucleic Acids Res. 9, 5671-5678. Kondo, K., Tajima, S., Sato, R., and Narita, K. (1979). Primary structure of the membrane-binding segment of rabbit cytochrome bs. J . Biochem. 86, I1 19-1 128. Koshland, D., and Botstein, D. (1982). Evidence for posttranslational translocation of beta-lactamase across the bacterial inner membrane. Cell 30, 893-902. Koshland, D., Sauer, R. T., and Botstein, D. (1982). Diverse effects of mutations in the signal sequence on the secretion of beta-lactamase in Salmonella typhimurium. Cell 30, 903-914. Kreibich, G., Ulrich, B. L., and Sabatini, D. D. (1978a). Proteins of rough microsomal membranes related to ribosome binding. 1. Identification of ribophorins I and 11, membrane proteins characteristic of rough microsomes. J . Cell Biol. 77, 464-487. Kreibich, G., Freienstein, C. M., Pereyra, B. N., Ulrich, B. L., and Sabatini, D. D. (1978b). Proteins of rough microsomal membranes related to ribosome binding 11. Cross-linking of bound ribosomes to specific membrane proteins exposed at the binding sites. J . Cell Biol. 77, 488506.

Kreibich, G., Bar-Nun, S., Czako-Graham, R., Mok, W., Nack, E., Okada, Y., Rosenfeld, M. D., and Sabatini, D. D. (1980). The role of free and membrane-bound polysomes in organelle biogenesis. In “Biological Chemistry of Organelle Formation-Proceedings of the Mosbach Colloquia” (Th. Bucher, W. Sebald, and H. Weiss, eds.), pp. 147-163. Springer-Verlag, Berlin and New York. Kreil, G. (1981). Transfer of proteins across membranes. Annu. Rev. Biochem. 50, 317-348. Kreil, G., Suchanek, G., Kaschmitz, R., and Kindas-Miigge, I. (1977). The biosynthesis of melittin: From the primary product of translation to the lytic peptide. FEBS Proc., I f t h 47, 7988.

1. APPLICATION OF THE SIGNAL HYPOTHESIS

55

Lake, I . A. (1983). Evolving ribosome structure: Domains in archaebacteria, eubacteria, and eucaryotes. Cell 33, 318-319. Lane, C. D. (1981). The fate of foreign proteins introduced intoXenopus oocytes. Cell24, 281-282. Lane, C., Shannon, S., and Craig, R. (1979). Sequestration and turnover of guinea-pig milk proteins and chicken ovalbumin in Xenopus oocytes. Eur. J . Biochem. 101, 485-495. Lane, C. D., Colman, A,, Mohun, T., Morser, J., Champion, .I.Kourides, , I., Craig, R., Higgins, S . . James, T. C., Applebaum, S. W.. Ohlsson. R . I.. Pancha, E . , Houghton, M., Matthews, J., and Miflin, B . J. (1980). The Xenopus oocyte as a surrogate secretory system; the specificity of protein export. Eur. J . Biochem. 111, 225-235. Lazdunski. C . , Baty, D., and Pages, J.-M. (1979). Procaine, a local anesthetic interacting with the cell membrane. inhibits the processing of precursor forms of periplasmic proteins in Escherichia coli. Eur. J . Biochem. 96, 49-57. Lebleu. B . . Hubert, E . , Content. J . , DeWit, I . , Braude, 1. A . , and LeClercq, E. (1978). Translation of mouse interferon mRNA in Xrnopus luevis oocytes and in rabbit reticulocyte lysates. Biochem. Biophys. Res. Commun. 82, 665-673. Li, W . Y., Reddy, R., Henning, D., Epstein, P., and Bush, H. (1982). Nucleotide sequence of 7 S RNA: Homology to Alu-DNA and 4.5 S RNA. J . B i d . Chem. 257, 5136-5142. Liebke, H.. Oliver. D., Shultz, J., Silhavy. T . , Beckwith, I.. and Steitz. J. (1983). Identification of a 6 S RNA-protein complex as a possible signal recognition particle of E . coli. Cell (in press). Lin. J. I. C., Kanazawa, H., Ozols, J., and Wu, H. C. (1978). An Escheriehiu coli mutant with an amino acid alteration within the signal sequence of outer membrane prolipoprotein. Proc. Nut/. Acud. Sci. U.S.A. 75, 4891-4895. Lingappa, V. R., Katz, F. N., Lodish, H. F., and Blobel, G. (l978a). A signal sequence for the insertion of a transmembrane glycoprotein. Similarities to the signals of secretory proteins in primary structure and function. J . B i d . Chem. 253, 8667-8670. Lingappa, V. R., Shields, D., Woo, S. L., and Blobel. G . (1978b). Nascent chicken ovalbumin contains the functional equivalent of a signal sequence. J . Cell B i d . 79, 567-572. Lingappa, V . R., Lingappa, R., and Blobel, G . (1979). Chicken ovalbumin contains an internal signal sequence. Nurure (London) 281, 117- 121. Lingappa, V . R., Chaidez, J., Yost, C. S . , and Hedgepeth, J. (1984). Determinants for protein bcalization: beta-lactamase signal sequence directs globin across microsomal membranes. Proc. Nutl. Acud. Sci. U.S.A. 81, 456-460. Little, J. S . , and Widnell, C. C. (1975). Evidence for the translocation of 5’-nucleotidase across hepatic membranes in v i v a Proc. Nut/. Acud. Sci. U.S.A. 72, 4013-4017. Lodish, H. F., Braell, W . A , , Schwartz, A. L., Strous, G. J. M., and Zilberstein, A . (1981). Synthesis and assembly of membrane and organelle proteins. Inr. Rev. Cytol. Suppl. 12, 247307. Lomedico, P., Chan, S. J . , Steiner, D. F., and Saunders, G . F. (1977). Immunological and chemical characterization of bovine preproinsulin. J . B i d . Chem. 252, 797 1-7978. Mach. B . , Faust, C . , and Vassalli, P. (1973). Purification of 14 S niesenger RNA of immunoglobulin light chain that codes for a possible light-chain precursor. Proc. Null. Acud. Sci. U.S.A. 70, 451-455. McKean, D. J., and Maurer, R. A. (1978). Complete amino acid sequence of the precursor region of rat prolactin. Biochemistry 17, 52 15-52 19. Majzoub, J . H., Rosenblatt, M., Fennick, B., Marnus, R . , Kronenberg, H. M., Potts, J . T. J r . , and Habener, J . F. (1980). Synthetic prepropdrathyroid hormone leader sequence inhibits cell-free processing of placental parathyroid and pituitary prehormones. J . B i d . Chem. 255, 1147811483. Malkin. L. I . , and Rich, A. (1967). Partial resistance of nascent polypeptide chains to proteolytic digestion due to ribosomal shielding. J . Mu/. B i d . 26, 329-346.

56

TOM A. RAPOPORT AND MARTINWIEDMANN

Marcantonio, E. E., Grebenau, R. C., Sabatini, D. D., and Kreibich, G. (1982). Identification of ribophorins in rough microsomal membranes from different organs of several species. Eur. J . Biochem. 124, 217-222. Maurer, R. A,. and McKean, D. J . (1978). Synthesis of preprolactin and conversion to prolactin in intact cells and a cell-free system. J . Biol. Chem. 253, 6315-6318. Mechler, B., and Vassalli, P. (1975). Membrane-bound ribosomes of Myeloma cells. 111. The role of messenger-RNA and the nascent polypeptide chain in the binding of ribosomes to membranes. J . Cell Biol. 67, 25-37. Meek, R. L., Walsh, K., and Palmiter, R. D. (1982). The signal sequence of ovalbumin is located near the NH2-terminus. J . Biol. Chem. 257, 12245-12251. Meissner, G., and Allen, R. (1981). Evidence for two types of rat liver microsomes with differing permeability to glucose and other small molecules. J . Biol. Chem. 256, 6413-6422. Mercereau-Puijalon, O., Royal, A., Cami, B.. Garapin, A., Krust, A., Cannon, F., and Kourilsky, P. (1978). Synthesis of an ovalbumin-like protein by Escherichia coli K 12 harbouring a recombinant plasmid. Nature (London) 275, 505-5 10. Meyer, D. I . , and Dobberstein, B. (1980a). A membrane component essential for vectorial translocation of nascent proteins across the endoplasmic reticulum: Requirements for its extraction and reassociation with the membrane. J . Cell Biol. 87, 498-502. Meyer, D. I., and Dobberstein, B. (1980b). Identification and characterization of a membrane component essential for the translocation of nascent proteins across the membrane of the endoplasmic reticulum. J . Cell B i d . 87, 503-508. Meyer, D. I., Louvard, D., and Dobberstein, B. (1982a). Characterization of molecules involved in protein translocation using a specific antibody. J . Cell Biol. 92, 579-583. Meyer, D. I., Krause, E., and Dobberstein, B. (1982b). Secretory protein translocation across membranes-the role of the “docking protein. ” Narure (London) 297, 647-650. Michaelis, S . , and Beckwith, I. (1982). Mechanism of incorporation of cell envelope proteins in Escherichia coli. Annu. Rev. Microbiol. 36, 435-465. Michel, H., and Oesterhelt, D. (1980). Three-dimensional crystals of membrane proteins: Bacteriorbodopsin. Proc. Natl. Acad. Sci. U.S.A. 77, 1283-1285. Mihara, K . , Sato, R., Sakakibara, R., and Wada, H. (1978). Reduced nicotinamide adenine dinucleotide-cytochrome bS reductase: Location of the hydrophobic, membrane-binding region at the carboxyl-terminal end and the masked amino terminus. Biochemisrry 17, 2829-2834. Milstein, C., Brownlee, G. G., Hamson, T. M., and Mathews, M. B. (1972). A possible precursor of immunoglobulin light chains. Nature (London) New Biol. 239, 117-120. Min Jou, W., Verhoeyen, M., Devos, R., Saman, E., Fang, R., Huylebrueck, D., Fiers, W., Threlfall, G., Barber, C., Carey, N., and Emtage, S. (1980). Complete structure of hemagglutinin gene from the human influenza A/Victoria/3/75 (H3 N2) strain as determined from cloned DNA. Cell 19, 683-696. Mizuno, T., Chou, M.-Y., and Inouye, M. (1983). DNA sequence of the promoter region of the ompC gene and the amino acid sequence of the signal peptide of pro-ompC protein of Escherichia coli. FEBS Lett. 151, 159-164. Mollay, C., W a s , U . , and Kreil, G. (1982). Cleavage of honeybee prepromelittin by an endoprotease from rat liver microsomes: identification of intact signal peptide. Proc. Natl. Acad. Sci. U.S.A. 79, 2260-2263. Moreno, F., Fowler, A. V., Hall, M., Silhavy, T. J . , Zabin, S., and Schwartz, M. (1980). A signal sequence is not sufficient to lead P-galactosidase out of the cytoplasm. Narure (London) 286, 356-359. Mostov, K. E., Kraehenbuhl, J.-P., and Blobel, G . (1980). Receptor-mediated transcellular transport of immunoglobulin: Synthesis of secretory component as multiple and larger transmembrane forms. Proc. Natl. Acad. Sci. U.S.A. 77, 7257-7261.

1 . APPLICATION OF THE SIGNAL HYPOTHESIS

57

Mostov, K. E., DeFoor. P., Fleischer. S . . and Blobel, G. (1981). Co-translational membrane integration of calcium pump protein without signal sequence cleavage. Nulure (Lorzdon) 292, 87-88. Movva, N. R., Nakamura, K., and Inouye, M. (1980). Amino acid sequence of the signal peptide of omp A protein, a major outer membrane protein of Escherichiu coli. J . Biol. Chem. 255, 2729. Miiller, M., Ibrahimi, I . . Chang, C. N., Walter, P.. and Blobel, G. (1982). A bacterial secretory protein requires signal recognition particle for translocation across mammalian endoplasmic reticulum. J . B i d . Chem. 257, 1 1860- I 1863. Mutoh, W., Inokuchi, K . , and Mizushima, S. (1982). Amino acid sequence of the signal peptide of onip F, a major outer membrane protein of E . u i l i . FEBS Letr. 137, 17 I - 174. Nathenson, S. G . , Uehara, H., and Ewenstein. B. M. (1981). Primary structural analysis of the transplantation antigens of the murine H-2 major histocompatibility complex. Annu. Rev. Biochem. 50, 1025-1052. Nielsen, J., Hansen, F. G., Hoppe, J., Fried, P., and von Meyenburg, K. (1981). The nucleotide sequence of the atp genes coding for the Fo subunits a. b. c, and F I subunit 6 of the membrane bound ATP synthase of Escherichia c d i . M o l . Gm. Gener. 184, 33-39. Ohno-lwashita, Y., and Wickner, W. (1983). Reconstitution of rapid and asymmetric assembly of MI3 procoat protein into liposomes which have bacterial leader peptidase. J . B i d . Chem. 258, 1895-1900. Okada, Y., Frey, A. B., Guenthner, T. M., Oesch, F.. Sabatini, D. D., and Kreibich, G. (1982). Studies on the biosynthesis of microsomal membrane proteins. Eur. J . Biochem. 122, 393-402. Oliver, D. B., and Beckwith, J. (1981). E . coli mutant pleiotropically defective in the export of secreted proteins. Cell 25, 765-772. Oliver, D. B., and Beckwith, I. (1982). Regulation of a membrane component required for protein secretion in Escherichia coli. Cell 30, 31 1-319. Omura, T . , and Sato, R . (1967). Isolation of cytochromes P-450 and P-420. In “Methods in Enzymology” (R. W. Estabrook and M. E. Pullman, eds.), Vol. 10, pp. 556-561. Academic Press, New York. Oshino, N. (1978). Cytochrome bS and its physiological significance. Pharmacol. Ther. A . 2, 477515. Ovchinnikov, Y., Abdulaev, N . , Feigiva, M., Kiselev, A., and Lobanov, N. A. (1979). The structural basis of the functioning of bacteriorhodopsin: An overview. FEBS Lett. 100, 219224. Overbeeke, N., Bergmanns, H., van Mansfeld, F.. and Lugtenberg, B. (1983). Complete nucleotide sequence of pho E, the structural gene for the phosphate limitation inducible outer membrane pore protein of Escherichia coli K 12. J . Mol. B i d . 163, 513-532. Oxender, D. L., Anderson, J. J., Daniels, C. J., Landick, R., Gunsalus, R. P., Zurawski, G . , and Yanofsky, C. (1980). Amino-terminal sequence and processing of the precursor of the leucinespecific binding protein and evidence for conformational differences between the precursor and the mature form. Pruc. Narl. Acad. Sci. U.S.A. 77, 2005-2009. Ozols, I., Gerard, C., and Nobrga, F. G . (1978). Proteolytic cleavage of horse liver cytochrome h5: Primary structure of the heme-containing moiety. J . Biol. Chem. 251, 6767-6774. Pages, J. M., and Lazdunski, C. (1982). Maturation of exported proteins in Escherichia coli. Requirement for energy, site and kinetics of processing. Eur. J . Biochem. 124, 561-566. Palade, G . (1975). Intracellular aspects of the process of protein synthesis. Science 189, 347-358. Palmiter, R. D., Gagnon, J., Ericson, L. H., and Walsh, K. A. (1977). Precursor of egg white lysozyme. J . Biol. Chem. 252, 6386-6393. Palmiter, R. D., Gagnon, J., and Walsh, K. A. (1978). Ovalbumin: A secreted protein without a transient hydrophobic leader sequence. Proc. Narl. Acad. Sci. U.S.A. 75, 94-98.

58

TOM A. RAPOPORT AND MARTINWIEDMANN

Palmiter, R. D., Davidson, J . M., Gagnon, J., Rowe, D. W., and Bernstein, P. (1979). NH2terminal sequence of the chick pro a l(1) chain synthesized in the reticulocyte lysate system. J. Biol. Chem. 254, 1433-1436. Palmiter, R. D., Thibodeau, S. N., Rogers, G., and Boime, I. (1980). Cotranslational sequestration of egg white proteins and placental lactogen inside membrane vesicles. Ann. N . Y. Acad. Sci. 343, 192-209. Papermaster, D. S., Burstein, Y., and Schechter, I. (1980). Opsin mRNA isolation from bovine retina and partial sequence of the in vitro translation product. Ann. N.Y. Acad. Sci. 343, 347355. Pappenheimer, A. M. (1979). Interaction of protein toxins with mammalian cell membranes. Microbiology, 187- 192. Patzelt, C., Labreque, A. D., Duguid, J . R., Caroll, R. J., Keim, P. S., Heinrikson, R. L., and Steiner, D. F. (1978). Detection and kinetic behavior of preproinsulin in pancreatic islets. Proc. Natl. Acad. Sci. U.S.A. 75, 1260-1264. Paul, L. D., and Goodenough, D. A. (1983). In vitro synthesis and membrane insertion of bovine MP26, an integral protein from lens fiber plasma membrane. J. Cell Biol. 96, 636-638. Perlman, D., Halvorson, H. O., and Cannon, E. (1982). Presecretory and cytoplasmatic invertase polypeptides encoded by distinct mRNAs derived from the same structural gene differ by a signal sequence. Proc. Natl. Acad. Sci. U.S.A. 79, 781-785. Pfeil, W. (1981). The problem of the stability of globular proteins. Mol. Cell. Biochem. 40, 3-28. Ploegh, H. L., Cannon, L. E., and Strominger, J. L. (1979). Cell-free translation of the mRNAs for the heavy and light chains of HLA-A and HLA-B antigens. Proc. Narl. Acad. Sci. U.S.A. 76, 2213-2277. Ploegh, H. L., Orr, H. T., and Strominger, J. L. (1981). Major histocompatibility antigens: The human (HLA-A,-B,-C) and murine (H-2K, H-2D) class I molecules. Cell 24, 287-299. Porter, A. G., Barber, C., Carey, N. H., Hallewell, R. A., Threlfall, G., and Emtage, J. S. (1979). Complete nucleotide sequence of an influenza virus haemagglutinin gene from cloned DNA. Nature (London) 282, 47 1-477. Prehn, S., Tsamaloukas, A,, and Rapoport. T. A. (1980). Demonstration of specific receptors of the rough endoplasmic membrane for the signal sequence of carp preproinsulin. Eur. J. Biochem. 107, 185-195. Prehn, S . , Niimberg, P., and Rapoport, T. A. (1981). A receptor for signal segments of secretory proteins in rough endoplasmic reticulum membranes. FEBS Leu. 123, 79-84. Privalov, P. V. (1979). Stability of proteins. Adv. Protein Chem. 33, 167-241. Ptitsyn, 0. B., and Finkelstein, A. V. (1983). Theoryof protein secundary structure and algorithm o f its prediction. Biopolymers 22, 15-25. Rachubinski, R. A,, Verma, D. P. S., and Bergeron, J. J . M. (1980). Synthesis of rat liver microsomal cytochrome bs by free ribosomes. J. Cell Biol. 84, 705-716. Randall, L. L. (1983). Translocation of domains of nascent periplasmic proteins across the cytoplasmic membrane is independent of elongation. Cell 33, 231-240. Randall, L. L., and Hardy, S. J . S. (1977). Synthesis of exported proteins by membrane-bound polysomes from Escherichia coli. Eur. J . Biochem. 75, 43-53. Randall, L. L., Hardy, S. J. S . , and Josefsson, L. G. (1978). Precursors of three exported proteins in Escherichia coli. Proc. Natl. Acad. Sci. U.S.A. 75, 1209-1212. Rapoport, T. A. (1981). Intracellular compartmentation and secretion of carp proinsulin synthesized in Xenopus oocytes. Eur. J . Biochem. 115, 665-669. Rapoport, T. A,, Huth, A., Prosch, S., Bendzko, P., Kaariainen, L., Bassuner, R., Manteuffel, R., and Finkelstein, A. (1983). Transport of proteins and signal recognition, I n “Protein Synthesis, Translational and Post-Translational Events” (Abraham, A. K., Eikhom, T. S., and Pryme, I. F., eds.) pp. 81-99. Humana, Clifton, New Jersey.

1. APPLICATION OF THE SIGNAL HYPOTHESIS

59

Redman, C. M., and Sabatini, D. D. (1966). Vectorial discharge of peptides released by puromycin from attached ribosomes. Proc. Nutl. Acud. Sci. U . S . A . 56, 608-615. Redman, C. M., Siekevitz, P., and Palade, G . E. (1966). Synthesis and transfer of amylase in pigeon pankreatic microsomes. J . Eliol. Chem. 241, 1150-1 158. Rice, C. M., and Strauss, J. H. (19811. Nucleotide sequence ofthe 26.5 mRNA of Sindbis virus and deduced sequence of the encoded virus structural proteins. Proc. Null. Acad. Sci. U.S.A. 78, 2062-2066. Richter, J. D., and Smith, L. D. (1981). Differential capacity for translation and lack of competition between mRNAs that segregate to free and membrane-bound polysomes. Cell 27, 183-191, Robinson, N. C., and Tanford, C. (1975). The binding of desoxycholate, Triton X-100,sodium dodecyl sulfate and phosphatidylcholine vesicles to cytochrome bs. Biochemistry 14, 365-378. Rogers, I., Early, P., Carter, C., Calame, K., Bond, M., Hood, L., and Wall, R. (1980). Two mRNAs with different 3’ ends encode membrane-bound and secreted forms of immunoglobulin p chain. Cell 20, 303-312. Roggenkamp, R.. Kusterniahn-Kuhn, B., and Hollenberg, C. P. (1981 ). Expression and processing of bacterial beta-lactamase in the yeast Succharomyces cerevisiae. Proc. Nail. Acud. Sci. U.S.A. 78, 4466-4470. Rollestone, F. S., and Lann, T. Y. (1974). Dissociation constant of 60 S ribosomal subunit binding to endoplasmic reticulum membrane. Biochcm. Biophys. Res. Commun. 59, 467-473. Rose, J. K., and Bergmann, J. G . (1982). Expression from cloned cDNA of cell-surface secreted forms of the glycoprotein of vesicular stomatitis virus in eucaryotic cells. Cell 30, 753762. Rose. J. K., Welch, W. J., Sefton, B. M., Esch, F. S., and Ling, N. C. (1980). Vesicular stomatitis virus glycoprotein is anchored in the viral membrane by a hydrophobic domain near the COOH terminus. Proc. Null. Acad. Sci. U.S.A. 77, 3884-3888. Rosenblatt, M., Beaudette, N. V., and Fasman, G . D. (1980). Conforniational studies of the synthetic precursor-specific region of preproparathyroid hormone. Proc. Nail. Acad. Sci. U.S.A. 77, 3983-3987. Rothman. J. E., and Lodish, H. F. (1977). Synchronised transmembrane insertion and glycosylation of a nascent membrane protein. Nature (London) 269, 775-780. Rotundo, R. L., and Fambrough, D. M. (1980). Secretion of acetylcholinesterase: Relation to acetylcholine receptor metabolism. Cell 22, 595-602. Russel, M., and Model, P. (1981). A mutation downstream from the signal peptidase cleavage site affects cleavage but not membrane insertion of phage coat protein. Proc. Narl. Acad. Sci. U.S.A. 78, 1717-1721. Russel, M., and Model, P. (1982). Filamentous phage pre-coat is an integral membrane protein: Analysis by a new method of membrane preparation. Cell 28, 177-184. Sabatini, D. D., and Blobel, G . (1970). Controlled proteolysis of nascent polypeptides in rat liver cell fractions. 11. Location of the polypeptides in rough microsomes. J . Cell Biol. 45, 146- 157. Sabatini, D. D., and Kreibich, G . (1976). Functional specialization of membrane-bound ribosomes in eukaryotic cells. I n “The Enzymes of Biological Membranes” (A. Martonosi, ed.), Vol. 2, pp. 531-579. Plenum, New York. Sabatini, D. D., Kreibich, G . , Morimoto, T., and Adesnik, M. (1982). Mechanisms for the incorporation of proteins in membranes and organelles. J . Cell B i d . 92, 1-22. Sabban, E., Marchesi, V., Adesnik, M., and Sabatini, D. D. (1981). Erythrocyte membrane proteins, Band 3: Its biosynthesis and incorporation into membranes. J . Cell B i d . 91, 637-646. Sakaguchi, M., Mihara, K., and Sato, R. (1984). Signal recognition particle is required for cotranslational insertion of cytochrome P-450 into microsomal membranes. Proc. Natl. Acad. Sci. U.S.A. 81, 3361-3364. Schaller, H.. Beck, E., and Takanami, M. (1978). In “The Single Stranded DNA Phages” (D.

60

TOM A. RAPOPORT AND MARTINWIEDMANN

Denhardt, D. Dresslev, and D. Ray, eds.), pp. 139-153. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Schechter, I. (1973). Biologically and chemically pure mRNA coding for mouse immunoglobulin Lchain prepared with the aid of antibodies and immobilized oligothymidine. Proc. Natl. Acad. Sci. U.S.A. 70, 2256-2260. Schechter, I., McKean, D. J., Gujer, R., and Terry, W. (1975). Partial amino acid sequence of the precursor of immunoglobulin light chain programmed by mRNA in vitro. Science 188, 160162. Scheele, G . , Jacoby, R., and Carne, T. (1980). Mechanism of compartimentation of secretory proteins: Transport of exocrine pancreatic proteins across the microsomal membrane. J . Cell Biol. 87, 61 1-628. Schindler, H., and Rosenbusch, J . P. (1981). Matrix protein in planar membranes: Clusters of channels in a native environment and their functional reassembly. Proc. Nail. Acad. Sci. U.S.A. 78, 2302-2306. Schleyer, M., Schmidt, B., and Neupert, W. (1982). Requirement of a membrane potential for the posttranslational transfer of proteins into mitochondria. Eur. J. Biochem. 125, 109- 116. Schmidt, M. F. G. (1982). Acylation of proteins-a new type of modification of membrane glycoproteins. Trends Biochem. Sci. 7, 322-324. Seeburg, P., Shine, J . , Martial, J. A,, Baxter, J . P., and Goodman, H. M. (1977). Nucleotide sequence and amplification in bacteria of the structural gene for rat growth hormone. Nature ( London) 270, 486-494. Seeburg, P. H., Shine, J., Martial, J. H., Ivarie, R. D., Morris, J. H., Ullrich, A,, Baxter, J . D., and Goodman, H. M. (1978). Synthesis of growth hormones by bacteria. Nature (London) 276, 795-798. Segrest, I . J . , and Feldman, R. J . (1974). Membrane proteins: Amino acid sequence and membrane penetration. J . Mol. Biol. 87, 853-858. Shaw, M . W., Lamb, R. A , , Erickson, B. W., Briedis, D. J., and Choppin, P. W. (1982). Complete nucleotide sequence of the neuraminidase gene of influenza B virus. Proc. Natl. Acad. Sci. U.S.A. 79, 6817-6821. Shultz, J., Silhavy, T. J., Berman, M. L., Fiil, N., and E m , S . D. (1982). A previously unidentified gene in the spc operon of Escherichia coli K12 specifies a component of the protein export machinery. Cell 31, 227-235. Siginioto, K., Sugisaki, H., Okamoto, T., and Takanami, M. (1977). Studies on bacteriophage fd DNA, IV. The sequence of mRNA for the major coat protein gene. J. Mol. Biol. 111,487-507. Silhavy, T. J . , Shuman, H. A., Beckwith, J., and Schwartz, M. (1977). Use of gene fusion to study outer membrane protein localization in Escherichia coli. Proc. Natl. Acad. Sci. U.S.A. 74, 541 1-5415. Silver, P., Watts, C., and Wickner, W. (1981). Membrane assembly from purified components. I. Isolated M 13 procoat does not require ribosomes or soluble proteins for processing by membranes. Cell 25, 341-345. Smith, W. P., Tai, P. C., Thompson, R. C., and Davis, B. D. (1977). Extracellular labeling of nascent polypeptides traversing the membrane of Escherichia coli. Proc. Nail. Acad. Sci. U.S.A. 74, 2830-2834. Smith, W. P., Tai, P. C., and Davis, B. D. (1979). Extracellular labeling of growing secreted polypeptide chains in Bacillus subiilis with diazoiodosulfanilic acid. Biochemistry 18, 198-202. Spatz, L., and Strittmatter, P. (1973). A form of reduced nicotinamide adenine dinucleotidecytochrome b5 reductase containing both the catalytic site and an additional hydrophobic membrane-binding segment. J . Biol. Chem. 248, 793-799. Steiner, D. F., Quinn, P. S., Chan, S. J., Marsh, J., and Tager, H. S. (1980a). Processing mechanism in the biosynthesis of proteins. Ann. N.Y. Acad. Sci. 343, 1-16.

1. APPLICATION OF THE SIGNAL HYPOTHESIS

61

Steiner, D. F., Quinn, P., Patzelt, C., Chan, S . . Marsh, J . , and Tager, H. (1980b). Proteolytic cleavage in the posttranslational processing of proteins. In “Cell Biology-A Comprehensive Treatise”,(D. Prescott and L. Goldstein, cds.) Vol 4 , pp. 175-202. Academic Press, New York. Stoffel, W., Blobel, G., and Walter, P. (1981). Synthesis in v i m and translocation of apolipoproteins A1 across microsomal vesicles. Eur. J . Eiochem. 120, 519-522. Strauss, A. W.. Bennett, C. A,. Donohue, A. M., Rodkey, J . A., Boime, I . , and Alberts, A. W. (1978). Conversion of rat pre proalbumin to proalbumin in iitru by ascites membranes. Demonstration by NHz-terminal sequence analysis. J . Eiol. Chem. 253, 6270-6274. Strittmatter, P., and Dailey, H. A. (1982). Essential structural features and orientation of cytochrome b, in membranes. In “Membranes and Transport” (A. N. Martonosi. ed.), Vol. I , pp. 71-82. Plenum, New York. Strittmatter. P.. Rogers, M. J.. and Spatz, L. (1972). The binding of cytochrome bS to liver microsomes. J . Eiol. Chem. 247, 7188-7194. Strous, G . J . A. M., and Lodish, H. F. (1980). Intracellular transport of secretory and membrane proteins in hepatoma cells infected by vesicular stomatitis virus. Cell 22, 709-717. Suchanek. G., Kindas-Mugge, J., and Kreil, G. (1975). Translation of honeybee promelittin messenger RNA. Formation of a larger product in a mammalian cell-free system. Eur. J . Biochem. 60, 309-315. Suchanek. K . . Kreil, G . , and Hermodson. M. A. (1978). Amino acid sequence of honeybee prepromelittin synthesised in vitro. Proc. Natl. Acud. Sci. U.S.A. 75, 701-704. Sutcliffe, J . G. ( 1 978). Nucleotide sequence of ampicillin resistance gene of Escherichia coli plasmid pBR 322. Proc. Nuil. Acad. Sci. U.S.A. 75, 3737-3741. Sveda, M. M . , Markoff. L. J . , and Lai, C.-J. (1982). Cell surface expression of the influenza virus haemagglutinin requires hydrophobic carboxy-terminal sequences. Cell 30, 649-656. Swan, D., Aviv, H., and Leder, P. (1072). Purification and properties of biologically active messenger RNA for a myeloma light chain. Pro(. Nuil. Acud. Sci. U.S.A. 69, 1967-1971. Szczesna, E., and Boime, I. (1976). rnRNA-dependent synthesis of autentic precursor to human placental lactogen: Conversion to its mature hormone form in ascites cell-free extracts. Proc. Nut/. A c ~ d Sci. . U . S . A . 73, 1179-1183. Tajima, S . , Enomoto, K.-I., and Sato, R. (1976). Denaturation of cytochrome bS by guanidine hydrochloride: Evidence for independent folding of the hydrophilic and hydrophobic moieties of the cytochrome molecule. Arch. Eiochem. Eiophys. 172, 90-97. Takagaki, Y., Gerber, G . E., Nihei, K., and Khorana, H. G. (1980). Amino acid sequence of the membraneous segment of rabbit liver cytochrome bs. Methology for separation of hydrophobic peptides. J . Eiol. Chem. 255, 1536-1541. Talmadge, K., Stahl, S . , and Gilbert, W. (1980a). Eukaryotic signal sequence transports insulin antigen in Escherichiu coli. Proc. Nail. Acad. Sci. Sci. U.S.A. 77, 3369-3373. Talmadge, K., Kaufman, J . , and Gilbert, W. (1980b). Bacteria mature preproinsulin to proinsulin. Proc. Natl. Acad. Sci. U.S.A. 77, 3988-3992. Talmadge, K., Brosius, J . . and Gilbert, W. (1981). An “internal” signal sequence directs secretion and processing of proinsulin in bacteria. Nature (London)294, 176-178. Tanford, C. (1980). “The Hydrophobic Effect.” Wiley, New York. Taniguchi, T . , Guarente. L., Roberts, T. M., Kirnelman. D., Douhan, J., 111, and Ptashne, M. ( 1980). Expression of the human fibroblast interferon gene in Escherichia coli. Proc. Nurl. Acad. U . S . A . 77, 5230-5233. Thibodeau, S . N., and Walsh, K. A. (1980). Processing of precursor proteins by preparations of oviduct microsomes. Ann. N.Y. Acad. Sci. 343, 180-190. Thibodeau. S . N., Lee, D. C., and Palmiter, R. D. (1978). Identical precursors for serum transferrin and egg white conalbumin. J . B i d . Chem. 253, 3771-3774.

62

TOM A. RAPOPORTAND MARTIN WIEDMANN

Tokunaga, M., Loranger, J. M., Wolfe, P. B., and Wu, H. C. (1982). Prolipoprotein signal peptidase in Escherichia coli is distinct from the MI3 procoat protein signal peptidase. J. Biol. Chem. 257, 9922-9925. Tomita. M., and Marchesi, V. T. (1975). Amino-acid sequence and oligosaccharide attachment sites of human erythrocyte glycophorin. Proc. Nail. Acad. Sci. U.S.A. 72, 2964-2968. Toneguzzo, F., and Ghosh, H. P. (1977). Synthesis and glycosylation in viiro of glycoprotein of vesicular stomatitis virus. Proc. Narl. Acad. Sci. U.S.A. 74, 1516-1520. Tonegawa, J., Maxam, A. M., Tizard, R., Bernard, O., and Gilbert, W. (1978). Sequence of a mouse germ-line gene for a variable region of an immunoglobulin light chain. Proc. Narl. Acud. Sci. U.S.A. 75, 1485-1489. Ullu, E., Murphy, S . , and Melli, M. (1982). Human 7SL RNA consists of a 140 nucleotide middlerepetitive sequence inserted in an Ah-sequence. Cell 29, 195-201. Unwin, P. N. T. (1979). Attachment of ribosome crystals to intracellular membranes. J. Mol. Biol. 132, 69-84. Van Rompuy, L., MinJou, W . , Huylebroeck, D., and Fiers, W. (1982). Complete nucleotide sequence of a human influenza neuraminidase gene of subtype N2 (A/Victoria/3/75). J. Mol. Biol. 161, 1-11. Villa-Komaroff, L., Efstratiadis, A., Broome, S., Lomedico, P., Tizard, R., Naber, S. P., Chick, W. L., and Gilbert, W. (1978). A bacterial clone synthesizing proinsulin. Proc. Natl. Acad. Sci. U.S.A. 75, 3727-3731. Vlasuk, G. P., Inouye, S., Ito, H., Itakura, K., and Inouye, M. (1983). Effects of the complete removal of basic amino acid residues from the signal peptide on secretion of lipoprotein in Escherichia coli. J. Biol. Chem. 258, 7141-7148. Von Heijne, G. (1980). Transmembrane translocation of protein. A detailed physico-chemical analysis. Eur. J. Biochem. 103, 431-438. Von Heijne, G. (1981). On the hydrophobic nature of signal sequences. Eur. J. Biochem. 116,419422. Von Heijne, G. (1983). Patterns of amino acids near signal-sequence cleavage sites. Eur. J. Biochem. 133, 17-21. Von Heijne, G., and Blomberg, C. (1979). Trans-membrane translocation of proteins. (The direct transfer model.) Eur. J. Biochem. 97, 175-181. Walter, P., and Blobel, G . (1980). Purification of a membrane-associated protein complex required for protein translocation across the endoplasmic reticulum. Proc. Narl. Acad. Sci. U.S.A. 77, 7112-7116. Walter, P., and Blobel, G. (1981a). Translocation of proteins across the endoplasmic reticulum. 11. Signal recognition protein (SRP) mediates the selective binding to microsomal membranes of invirro-assembled polysomes synthesizing secretory protein. J. Cell Biol. 91, 55 1-556. Walter, P., and Blobel, G. (1981b). Translocation of proteins across the endoplasmic reticulum. 111. Signal recognition protein (SRP) causes signal sequence-dependent and site-specific arrest of chain elongation that is released by microsomal membranes. J. Cell Biol. 91, 557-561. Walter, P., and Blobel, G. (1982). Signal recognition particle contains a 7 s RNA essential for protein translocation across the endoplasmic reticulum. Nurure (London) 299, 691-698. Walter, P., and Blobel, G. (1983a). Subcellular distribution of signal recognition particle and 7SLRNA determined with polypeptide specific antibodies and cDNA probe. J. Cell Biol., 97, 1693- 1699. Walter, P., and Blobel, G. (1983b). Disassembly and reconstitution of signal recognition particle. Cell 34, 525-533. Walter, P., Jackson, R. C., Marcus, M. M., Lingappa, V., and Blobel, G. (1979). Tryptic dissection and reconstitution of translocation activity for nascent presecretory proteins across microsomal membranes. Proc. Nail. Acad. Sci. U.S.A. 76, 1795-1799.

1. APPLICATION OF THE SIGNAL HYPOTHESIS

63

Walter, P., Ibrahimi, I., and Blobel, C. (1981). Translocation of proteins across the endoplasmic reticulum. I . Signal recognition protein (SRP) binds to in virro assembled polysomes synthesizing secretory protein. J. Cell B i d . 91, 545-550. Ward, C. W., and Dopheide. T. A. (1979). Primary structure of the Hong Kong (H,) haemagglutinin. Br. Med. Bull. 35, 51-56. Warren, G. (1981). Membrane proteins: Structure and assembly. In “Membrane Structure” (Finean/Michell, eds.). Ch. 6, pp. 215-257. Elsevier, Amsterdam. Warren, G., and Dobberstein, B. (1978). Protein transfer across microsomal membranes rcassembled from separated membrane components. Nature (London) 273, 569-57 I . Watts, C., Wickner, W., and Zimmerniann, R. (1983). M I 3 procoat and pre-immunoglobulin share processing specificity but use different membrane receptor mechanisms. Proc. Nor/. Acad. Sci. U.S.A. 80, 2809-2813. Welch, W. J., and Sefton, B. M. J. (1980). Characterization of a small, nonstructural polypeptide present late during infection of BHK cells by Semliki Forest Virus. J . Virol. 33, 230-237. Wickner, W. (1979). Assembly of proteins into biological membranes. The membrane trigger hypothesis. Annu. Rev. Biochem. 48, 23-45. Wickner. W., Ito, K., Mandel, G . , Bates, M., Nokelainen, M., and Zwizinski, C. (1980). The three lives of M 13 coat protein: A virion capsid, an integral membrane protein and a soluble cytoplasmic proprotein. Ann. N . Y. Acad. Sci. 343, 384-390. Wilson, V. G . , and Hogg, R. W. (1980). The NH,-terminal sequence of a precursor form of the arabinose binding protein. J . Biol. Chem. 255, 674556750, Wirth, D. F., Katz, F., Small, B., and Lodish. H. F. (1977). How a single Sindbis virus mRNA directs the synthesis of one soluble protein and two integral membrane glycoproteins. Cell 10, 253-263. Wolfe, P. B., Silver, P., and Wickner, W. (1982). The isolation of homogeneous leader peptidase from a strain of Escherichia coli which overproduces the enzyme. J . B i d . Chem. 257, 78987902. Wolfe, P. B., Wickner, W., and Goodman, J . (1983). Sequence of the leader peptidase genc of Escherichia coli and the orientation of leader peptidase in the bacterial envelope. J . Biol. Chem. 258, 12073- 12080. Yamada, M., Miki, T., and Nakazawa, A. (1982a). Translocation of colicin El through the cytoplasmic membrane of Escherichia coli. FEES Len. 150, 465-468. Yamada, M.. Ebina, I . , Miyata, T . , Nakazawa. T . , and Nakazawa, A. (1982b). Nucleotide sequence of the structural gene for colicin E l and predicted structure of protein. Proc. Nurl. Acud. Sci. U.S.A. 79, 2827-2831. Yelverton, E., Norton, S . , Obijeski, F. I., and Coeddel, D. V. (1983). Rabies virus glycoprotein analogs: Biosynthesis in Escherichia coli. Science 219, 614-620. Yost, C. S . , Hedgpeth, J . , and Lingappa, V. R. (1983). A stop transfer sequence confers predictable transmembrane orientation to a previously secreted protein in cell-free systems. Cell 34, 759766. Zehavi-Willner, T.. and Lane, C. (1977). Subccllular compartmentation of albumin and globin made in oocytes under the direction of injected messenger RNA. Cell 11, 683-693. Zwizinski, C., and Wickner, W. (1980). Purification and characterization of leader (signal) peptidase from Escherichia coli. J . Biol. Chem. 255, 7973-7977. Zwizinski, C., Date, T.. and Wickner, W. (1981). Leader peptidase is found in both the inner and outer membranes of Escherichiu d i . J . B i d . Chem. 256. 3593-3597.

This Page Intentionally Left Blank

CURRENT TOPICS IN MEMBRANES AND TRANSPORT, VOLUME 24

Chapter 2 Structure and Function of the Signal Peptide GUY D.DUFFAUD, SUSAN K. LEHNHARDT, PAUL E. MARCH, AND MASAYORI INOUYE Department of Biochemisrry State University of New York af Stony Brook Stony Brook, New York

I . Introduction. ................................................ 11. Signal Peptid ................................................ A. General Considerations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

C. Loop Model.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Experimental Ap ..................

......................

A.

Signal Peptidases.. . . . . . . . . . . . . . . . . . . . . .

1.

65 66

.......... in . . . . . . . . . . 81 . . . . . . . . . . . . 81

.........................

95

INTRODUCTION

Proteins that are to be secreted to the outside of the cytoplasmic membrane have to be translocated across the hydrophobic membrane from a hydrophilic environment inside of the cell. These secretory proteins are produced on polyribosomes in the cytoplasm. It has been well established that the secretory 65

Copynghl 0 1985 by Academlc Press. Inc All right, of reproduction in any form reserved ISBN 0-12-153324-7

66

GUY D. DUFFAUD ET AL.

proteins are produced from their secretory precursors, which have an extra peptide extension at their amino-terminal ends called the signal peptide or leader peptide. The signal peptide is generally composed of approximately 20 amino acid residues and is able to direct protein translocation through the membrane. It appears that all information necessary to initiate protein secretion resides in the structure of the signal peptide. Many secretory proteins have been characterized and the amino acid sequences of their signal peptides have been determined. In this article we describe the structural as well as the functional aspects of the signal peptide and discuss how the signal peptide is able to guide protein translocation through the cytoplasmic membrane.

II. SIGNAL PEPTIDE

A. General Considerations Existence of precursors for secretory proteins was first demonstrated by Milstein et d . (1972). They showed that immunoglobulin light chains are produced from precursors of slightly larger molecular weights. They suggested that a short amino acid sequence at the amino terminus of a precursor protein provides a signal which is required for the formation of membrane-bound polyribosomes as well as for the secretion of the protein across the membrane. The signal hypothesis, proposed by Blobel and Dobberstein (1975a,b), offered a more precise description for translocation steps of secretory proteins. In the prokaryotic system, the existence of the signal peptide was first shown for the major outer membrane lipoprotein (Halegoua et al., 1976; Inouye et al., 1977). Since then the existence of the signal peptide has been shown for many other prokaryotic secretory proteins. A comparison of the general structure of different signal peptides demonstrates that, independent of their origin, they share more characteristics than expected from such a diverse population of proteins (Inouye and Halegoua, 1980; Michaelis and Beckwith, 1982; Perlman and Halvorson, 1983; von Heijne, 1983). A list of the amino acid sequences of all the signal peptides determined at present for prokaryotic secretory proteins is given in Fig. 1. When the structures of these signal peptides are analyzed, one can find a few common features. These features are less obvious in the eukaryotic signal peptides. Therefore, in the present article, we focus our attention on common features found in the prokaryotic signal peptides and discuss their possible functional roles in protein secretion.

B. Common Features of the Prokaryotic Signal Peptide When the amino acid sequence of the prolipoprotein signal peptide was first determined, it was pointed out that its signal peptide has several unique features

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

67

(Inouye el al., 1977). Since then, 34 signal peptides of bacterial secretory proteins have been determined (Fig. 1). These sequences support the original rules proposed for the structure of the signal peptide (Inouye et al., 1977; Inouye and Halegoua, 1980); these rules have been extended to the following five common features. 1. One to three positively charged amino acid residues in the amino-terminal region of the signal peptide (amino-terminal basic region) 2. A long hydrophobic sequence consisting of 14-20 amino acid residues following the amino-terminal basic region (hydrophobic domain) 3. In most signal peptides, the presence of one to two proline and/or glycine residue(s) within the hydrophobic domain 4. The presence of a serine and/or threonine residue(s) within the hydrophobic domain, dividing it into two regions of different degrees of hydrophobicity 5 . An alanine or glycine residue at the carboxy terminus of the signal peptide (cleavage site). In the following discussion the signal peptides are numbered negatively starting from the left of the cleavage site . The first amino acid to the left of the cleavage site is designated -1 (see Fig. 1). 1. AMINOTERMINAL REGION

In every known prokaryotic signal peptide (Fig. l), there is, without exception, one to five positively charged amino acid residues (lysine and/or arginine) located at the amino-terminal portion (see amino acids circled in Fig. 1). Most of the signal peptides have two basic amino acids (80% of the sequences listed). These basic amino acid residues are usually adjacent. However, when they are separated, it is by amino acids from a specific group: proline, glycine, threonine, or serine residues. The position of the innermost basic residue varies from - 15 (closest to the cleavage site) to -21 (farthest from the cleavage site) in the case of gram-negative bacteria (the cleavage sites of some gram-positive bacteria signal peptides have not been well established). There are two cases for which basic residues are found in the hydrophobic domain: lysine at -2 in alkaline phosphatase and lysine at -4 in a-amylase. Infrequently, histidine residues can be found at positions -2 or -4, but they are likely to be uncharged at physiological pH. It should also be pointed out that no glutamic or aspartic acid residues are found in signal peptides. What is the role of positive charges in the amino terminal region of the signal peptide? The initial interaction between the signal peptide and the cytoplasmic membrane may be carried out by the positively charged amino-terminal region. This interaction may be facilitated by ionic interaction with the inner surface of the cytoplasmic membrane negatively charged as a result of the presence of

GRAM-NEGATIVE BACTERIA Outer membrane proteins Lipoprotein ( 1 )

Met~Ala~LeuValLeuGlyAiaValIleLeuGly~LeuLeuAlaGlyCysSer

Lipoprotein (S. murcescens) (2)

My&n ~ ~ L e u V a l L e u GAlaValIleLeuGl l y y ~ ~ u L e ~ l aCysSe~ G l y

Lipoprotein (E. umylovoru) (3)

MetAsn @Th@

J. + I

-20

***

-1c)

***

+I

***

**I

+I

-20

-20

LeuValLeuGly AlaValIleLeuGiy ~ L e u L e Aul a G l y~sSer ***

***

+I

MetGly ~Ser~IleValLeuGlyAlaVaiValLeuAlaSerAlaLeuLeuAiaGlyCysSer

Lipoprotein (M.morganii) (4)

*I*

+I

-19

Met~Ala~IleValLeuGlyAlaValIleLeuAla~~ly~uLeuA~aG~yCysSer

Lipoprotein ( P . mirubilis) (5)

***

-21

+I

Met@

OmpA protein (S. dysenteriue) (7)

Me~~~AIaIleAlaIleThrVaIAlaLeuAiaGlyPheA~a~V~AlaG~~i~~ *** +I

~ThrAlaIleAlaIleAlaValAlaLeuAlaGlyPheAla~ValAlaGlnAlaAlaPro ***

-21

-25

%

***

OmpA protein (6)

Lambda receptor (8)

+I

MetMetIleThrLeu~~~uPro~uAlaValAlaValA~aAlaG~yValMet~rAlaGl~~~e~laValAsp *** *** +I -21

PhoE protein (9) -22

Met~~SerThrLeuAlaLeuValValMetGiyIleValAla~A~a~rValG~nAiaAiaGlu *** +1

M e t M e t m ~AsnIleLeuAlaValIleValProAlaLeuLeuV~AlaGly~AlaAs~aAlaGiu *** *** fl

OmpF protein (10)

-21

Met@

OmpC protein (11)

Val~ValLeuSerLeuLeuValProAlaLeuLeuVal~aGlyAl~i~s~laA~aGlu *** ***

Periplmmic proteins -21

+I

Met~GlnSerThrIleAlaLeuAlaLeuLeuProLeu~u~e~ProVal~LysAl~g~ *** *** +I

Alkaline phosphatase (12)

- I9

MetPhe@ThrThrLeuCysAlaLeuLeuIleT&AlaSerCysSerThrPheAlaAlaF'ro

AmpC j3-lactamase (13)

+I

-23

MetSerIleglnHisPhe@ValAlaLeuIleProPhePheAlaAlaPhe~LeuProVaiPheAiaHisPro

TEM j3-lactamase (14)

r**

-23

Arabinose-binding protein (1 5)

***

+I

M e t H i s mP h e T h r ~ A l a L e u A l a A l ~ I l e G l y L e u A l a A l a V a l M e t ~ ~ l n ~ A l ~ e t A l a G l u A s n

-23

Galactose-binding protein (IS) - 25

Ribose-binding protein (16)

+I

M e t A s n B ~ValLeuThrLeuSerAlaValMetAla~MetLeuPheGlyAlaAlaAlaHisAl~l~sp *** +I

MetA s n M e B ~ L e u A l a T h r L e u V a l S e r A l a V a l A l ~ u ~ A l a ~ V a l ~ r A l a A s n A l a M e t A l ~ y s A s p *I*

+I

-22

Met~@ThrValLeuAlaLeuSerLeuLeuIleGlyLeuGlyAla~AlaAla~rTyrAlaAlaLeu

Lysine-arginine-ornithine binding protein ( I 7)

***

***

+I

- 22

Met@@

Histidine-binding protein (17)

LeuAlaLeuSerLeuSerLeuValLeuAIaPhe~rAla~Al~laPheAl~AlaIle

-23

+I

M e t ~ A l a A s n A l a ~ ~ l l e I l e A l a G l y* *M * etIleAlaLeuAlaIle~rHis~AlaMetAlaAspAsp

Leucine-binding protein ( I 8)

+I

-23

AlaLeuLeuAlaGlyCysIleAlaLeuAlaPhe~rAsnMetAlaLeuAla

MetAsnIle@Gly@

Isoleucine-valine binding protein (19)

I**

+I

- 26

Maltose-binding protein (20)

z?

Met~Ile~ThfirlyAla~IleLeuAlaLeuSerAlaLeuThrThrMetMetPhe~rAla~~aLeuAlaLyslle

Viral proteins +I

Major coat protein (phage M13) (21)

M:t@

~SerLeuValLeu~AlaSerValAlaValAla~LeuVal~Met~u~PheA~aAlaGlu -IR

+I

LeuLeuPheAlaIleProLeuValValProPheTyr~rHisSerAlaGlu

Met@@

Minor coat protein (phage M13) (22)

***

***

Enterotoxins -21

ti

M e t A s n m Val@CysTyrValLeuPheThrAlaLeuLeuSerSerLeuTyrAlaHisGlyAlaPro

Heat-labile toxin. subunit B (23)

+I

-18

AsnlleThrPhellePhePheIle~uLeuAla~r~oLeuTyrAlaAsnGly

Met@

Heat-labile toxin. subunit A (24)

***

-21

MetIle@Leu@

Cholera toxin. subunit B ( V . cholerae) (25)

t i

PheGlyValPhePheThrValLeuLeu~rAlaTyrAlaHisGlyThrPro ***

-18

+I

MetVal~IleIlePheValPhePheIlePheLeu~rPhe~TyrAlaAsnAsp

Cholera toxin, subunit A ( V . cholerae) (25)

FIG. 1 .

continues onto

mxt

page

GRAM-POSITIVE BACTERIA -26

Penicillinase ( B . licheniformis) (26)

Met@

+I

LeuTrpPheSerThrLeum Leu(@@

A l a A l a A l a V a l L e u L e u P h e ~ V aAl l a G l y C y s A l a +I

- 24

P-Lactamase (S. aureus) (27)

Me~~LeuIlePheLeuIleValIleAlaLeuValLeuSerAla~sAsn~Asn~rHisAl~ysGl~

a-Amylase (B. amy/o/iquefaciens)(28) -31

+I

M e t I l e G l n m @@~ T h r V a l S e r P h e ~ L e u V a l L e u M e t C y s T h r L e u L e u P h e V *** al~rLeu~o~e~Lys~rAlaValAsn -41

a-Amylase (B. subrilis) (29)

MetPheAla~~Ph~ThrSerLeuLeuhoLeuPheAlaGlyPhe~uLeuLeuPheT~LeuVa~uAlaGly~oAl~la~a~rA~a *** *** ****** -9

+I

Glu~AlaAsnLysSerAsnGluLeuThr Diphtheria toxin (C. diphtheriae) (30) -32

+I

ValLeuVal ~ G l y T y r V a l V a l S e r ~ L e u P h e A l a S e r I l e L e u I l e G l y ~ a L e u L e u G ~ y I l e G l y A l a ~ o ~ o ~ r A l ~ i s A l a G l y A l a *** *** *** ****** Protein A (S. aureus) (31) -36

Leu@@@

+I

AsnIleTyrSerIle@@L e u G l y V a l G l y I l e A l a S e r V a l T h r L e u G l y ~ ~ u L e ~ I l e ~ ~ l y G ~ ~ V a l ~ ~ o ~ ~ l a A s n A ~ ~ ~ a G l *** I** *** *** *****I

FIG. 1. Amino acid sequences of prokaryotic signal peptides. Classification has been made according to the type of bacteria and final localization of the protein. Specific amino acids thought to be important to the structure of signal peptides are pointed out, according to the rules discussed in the text. The positively charged residues at the amino terminus are circled. The glycine and proline residues within the hydrophobic domain are underlined with asterisks. The serine, threonine, and cysteine residues that locate near or at position -6 are underlined. The cleavage site is shown by an arrow. The carboxy-terminal amino acid of the signal peptide is designated as - 1. The cleavage site of a-amylase of Bacillus subtilis has not been exactly determined and the two possibilities are indicated (-9 and + 1). References: (1) Nakamura and Inouye (1 979); (2) Nakamura and Inouye (1 980); (3) Yamagata et a/. (1 981); (4) Huang et a/. (1 983); ( 5 ) Ching and Inouye (unpublished);(6) Movva et al. (1980); (7) Braun and Cole (1982); (8) Hedgpeth et al. (1980); (9) Overbeeke era/. (1983); (10) Mutoh e t a / . (1982); ( 1 1) Mizuno er al. (1983); (12) Inouye et al. (1982); (13) Jaurin and Grundstrom (1981); (14) Sutcliffe (1978); (15) Scripture and Hogg (1983); (16) Groarke el a/. (1983); (17) Higgins and Ames (1981); (18) Oxender er al. (1980); (19) Landick and Oxender (1982); (20) Bedouelle era/. (1980); (21) Sugimoto et a/. (1977); (22) Schaller er a/. (1979); (23) Dallas and Falkow (1980); (24) Spicer and Noble (1982); (25) Mekalanos et al. (1983); (26) Neugebauerer a/. (1981);(27) McLaughlin era/. (1981 ); (28) Palva e t a / . (1981); (29) Ohmura e t a / . (1983); (30) Greenfield e t a / . (1983); (31) Lofdahl e t a / . (1983).

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

71

phosphatidylglycerol. The role of the positive charges at the amino terminus has been investigated for the signal peptide of the Escherichia coli prolipoprotein by S. Inouye et al. (1982) and Vlasuk et a / . ( 1983), and these experiments will be discussed later in this article. It should also be noted that in 20 out of 34 cases the amino terminus includes at least one amino acid known to destabilize a-helices (such as serine, glycine, or proline).

2. HYDROPHOBIC DOMAIN The second notable feature of the signal peptide is the presence of a series of hydrophobic residues after the amino-terminal basic region. The length of this hydrophobic domain is quite variable, from 14 to 20 amino acids long. The most frequent residues are alanine, valine, leucine (and less frequently isoleucine), phenylalanine, tryosine, and methionine (tryptophan has not been found). The distribution of these amino acids within the hydrophobic domain is random, no position being favored for a given residue. This domain is most likely to form a particular conformation, essential for the function of the signal peptide.

3. GLYCINE AND/OR PROLINE RESIDUES I N THE HYDROPHOBIC DOMAIN Glycine and proline are known to be involved in the 6-turn structure of proteins. Thus, their locations in a peptide may help predict its structure. Within the central region of the hydrophobic domain there is very often one glycine or one proline residue (27 out of 34). There is no apparent preference for either glycine or proline, except perhaps that when more than one is present in a signal peptide it tends to be the same one (10 cases). The exact role of these amino acids in the hydrophobic domain is not clear at present. Glycine residues at positions -7 and - 12 of the E . coli prolipoprotein have been substituted with valine residues or deleted to study their role in the signal peptide (Inouye ef al., 1984). These mutations will be discussed later.

4. THREONINE AND/OR SERINE RESIDUES In the hydrophobic domain there is almost always a serine and/or a threonine (with a cysteine residue in one case). The positions of these residues are in general between -3 and -6 (-6 being most favored, as can be seen in Fig. 1). These residues appear to divide the hydrophobic domain into two portions: an extremely hydrophobic domain of approximately 10 hydrophobic amino acids at the center of the signal peptide, and a less hydrophobic portion of 5 or 6 amino acid residues at the cleavage site. The structural function of a serine or threonine residue is not well understood yet. However, it has been suggested that they may be involved in the formation of a specific conformation, as will be discussed later.

72

GUY D. DUFFAUD ET AL.

5 . THE CLEAVAGE SITE

The most striking characteristic is the nature of the amino acid at position - I . This position is filled by alanine in 23 out of 34 cases, glycine in 9 cases, and serine in the remaining 2 cases. It should be noted that they all are amino acids with a small side chain. Interchangeability between glycine and alanine residues has been demonstrated for prolipoprotein (Inouye et al., 1983a) and for plactamase (J. Knowles, personal communication). This rather strict requirement at position - 1 may reflect signal peptidase substrate specificity. The other positions around the cleavage site are not as highly conserved as position - 1. Nevertheless there are some notable features. Frequently (17 out of 34 cases), position -2 is occupied by an amino acid possessing a large or charged side chain (histidine in 6 cases, phenylalanine or tryosine in 6 cases, glutamine or asparagine in 5 cases). Position 1 is occupied in 20 instances by a small side chain amino acid (alanine, cysteine, glycine). However, this is not exclusive and amino acids such as arginine, histidine, lysine, glutamic acid, aspartic acid, asparagine, leucine, and threonine are also found at this position (see Fig. 1). For lipoprotein, a cysteine at + 1 was shown to be necessary for processing (Inouye ef al., 1983b). In position $2, an acidic or a proline residue is found 17 times. Finally, it should be noted that the signal peptide is very unique as an endopeptidase substrate since it appears to have stringent requirements for a secondary structure at the cleavage site in addition to a specific primary structure requirement. Therefore, the nature of the sequence at the cleavage site will be essential to achieve the formation of the required specific conformation at the cleavage site.

+

6. SIGNAL PEPTIDESOF THE SAMEPROTEIN FROM DIFFERENT BACTERIA Comparison of the signal peptides of the same protein in different species yield several interesting conclusions concerning the necessity of certain structural features. The most detailed study has been carried out on the signal peptides of the prolipoprotein of five different species of enterobacteria: E . coli, Serratia mrcescens, Erwinia amylovora, Morganella morganii, and Proteus mirabilis (see Fig. 1). The signal peptides of the prolipoproteins from S. marcescens and E. amylovora are identical and differ by only two residues in the amino-terminal region (- 18, -19) from the signal peptide of the E . coli prolipoprotein. More changes are observed in the signal peptide of M . morganii and P . mirabilis lipoprotein. In spite of these changes, all the prolipoproteins can be processed in E. coli to form the fully modified mature lipoprotein assembled in the outer membrane. Therefore, the changes found in the signal peptides are considered to be permissible mutations in terms of the function of the signal peptide. Nevertheless, with a more detailed examination, it is possible to notice that these

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

73

various lipoproteins conserve many features. There are always two basic amino acid residues at the terminal region, and the innermost basic residue is always at position - 16. The sequence at the cleavage site from position -4 to - 1, LeuLeu-Ala-Gly is conserved, whereas the amino acid residue at position -5 appears to be less stringent (it can be threonine, alanine, or glycine). The amino acid at position -7 can also be alanine or glycine. The remaining sequence from position - 15 to -8 is structurally highly conserved and consists of hydrophobic amino acid residues. The signal peptides for the two OmpA proteins from E . coli and Shigelfa dysenteriae are highly conserved and differ only at one position, - 13 (Fig. 1). For the signal peptides of the E . coli and the Vibrio cholerae enterotoxins, the only significant similarities are observed at the cleavage site (-2, - 1, + 1) and at the positively charged amino termini (Fig. 1). Although only six sequences are known so far, the signal peptides of grampositive bacteria appear to be longer than those for gram-negative bacteria (see Fig. 1). These signal peptides are at least 24 amino acid residues long, 4 amino acids longer than the average size of the gram-negative bacterial signal peptides. However, it should be noted that in some cases the cleavage sites of grampositive secretory proteins have not been definitively established.

7. COMPARISON WITH THE EUKARYOTK SIGNALPEP~IDE From the analysis of 78 eukaryotic signal peptides, von Heijne (1983; see also this volume) found a common pattern of amino acids near the signal peptide cleavage site. He observed that certain amino acid residues appeared to be preferred at specific positions. At positions - 1 and -3 small or neutral residues (glycine, alanine, serine, and threonine) were found very frequently (95% at position -1 and 68% at position -3), whereas at positions + 1 and -2 these numbers dropped (26 and 15%, respectively). Similar observations were described by Pearlman and Halvorson ( 1983), who used 39 eukaryotic and prokaryotic signal peptides to study the possibility of a common pattern for signal peptides. The positive charges of the amino-terminal region in the prokaryotic signal peptides are not as common in the eukaryotic signal peptides, in which the number of positive charges varies from 0 to 2. Also, in eukaryotes there is occasionally a negatively charged amino acid, usually within the last amino acids of the hydrophobic section. At position + 1, charged or polar residues were found in 40% of the sample. Otherwise, the rules discussed above seem to apply to a certain extent to the eukaryotic signal peptides. These differences between the prokaryotic and eukaryotic signal peptides may be attributed to the fact that eukaryotes have evolved a more complex secretory machinery. It is possible that localization of a secretory protein among the cellular compartments might depend on particular features of the eukaryotic signal peptides.

74

GUY D. DUFFAUD ET AL.

C. Possible Structure of the Signal Peptide The common features described above are considered to be important in determining specific structure(s) that the signal peptide will adopt during the process of translocation of the secretory precursor. Pearlman and Halvorson (1983) predicted the secondary structure of the signal sequence. Besides the potential pturn preceding the cleavage site, they calculated a higher probability for a psheet structure for the hydrophobic portion by using the rules proposed by Chou and Fasman (1978), which enable one to predict possible secondary structures from the primary structure of a protein. They argued that a P-sheet structure is more appealing in that it could interact more easily than an helical structure with the membrane lipids and other membrane proteins. According to their analysis, formation of an a-helical structure in this region was possible, but less likely. However, the size of the sample used for these calculations (15 sequences) limits the scope of these calculations. In another study Rosenblatt et al. (1980) investigated the secondary structure of synthetic preproparathyroid hormone signal peptide by measuring the circular dichroism spectra of this peptide in either aqueous or nonpolar solvents. Results showed that p-sheet structure was predominant in aqueous buffer (43%), whereas in a nonpolar medium it was nonexistent and there was a sharp rise of a-helical structure (46%). This observation suggests that a structure for a signal peptide dynamically changes during the process of translocation across the membrane. As pointed out by Rosenblatt et al. (1980), when a signal peptide changes its conformation from a P-sheet to an a-helix, it results in a shortening of the length of the signal peptide, a change which might result in pulling the precursor protein through the membrane. Another example of the importance of a-helical structure in the signal peptide is suggested in the LamB protein. E m and Silhavy (1983) investigated revertants of a mutant having a 12-bp deletion in the region coding for the signal peptide for the LamB protein. As a result of the mutation, amino acid residues in positions from - 16 to - 11 of the signal peptide were deleted, thereby blocking secretion of the protein (see Fig. 4). Analysis of the pseudorevertants from this deletion mutant revealed the substitution in one case of a proline by a leucine at position - 13 (corresponding to a position - 17 in the wild-type sequence) and in the other case of a glycine by a cysteine at position -9 (see Fig. 4). Prediction of the secondary structure showed that in the original mutant the hydrophobic domain contained a region of random coil which was replaced in each of the pseudorevertants by an a-helical conformation. They attributed recovery of a functional signal peptide to the regaining of a-helicity in the hydrophobic core region. It is thought that the most stable structure for a peptide spanning the membrane is an a-helical conformation (Guidotti, 1977). Such a conformation in the signal peptide might be required for its interaction with the membrane hydrophobic domain.

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

75

As pointed out earlier, however, protein secretion is a dynamic process, and a signal peptide most likely experiences drastic changes in its conformation during the secretory process. A specific conformation is possibly required for each step of translocation.

111.

MODELS FOR PROTEIN SECRETION

A. Signal Hypothesis Milstein et al. (1972) were the first to demonstrate that a protein destined for export across a membrane (the IgG K light chain) was initially synthesized as a higher molecular weight precursor. They postulated that this precursor protein contained an amino-terminal peptide extension not found in the mature protein. The function of this extension was to signal that this protein would be secreted. Schechter (1973) characterized the amino-terminal extension of the IgG K light chain precursor protein. It was shown that this extension was indeed at the amino terminus of the protein and that it consisted of 20 amino acids. Surprisingly, it had a very high content of leucine (12 residues). The hydrophobicity conferred by the leucine residues was proposed to aid in the association of the nascent polypeptide with the endoplasmic reticulum (Schechter, 1973; Schechter et al., 1975). These initial observations and others led to the signal hypothesis (Blobel and Sabatini, 197 1; Blobel and Dobberstein, 1975a,b). The signal hypothesis as first put forth stated that the signal peptide emerging from the ribosome facilitates interaction with a complex protein pore within the membrane. Subsequent cotranslational, linear extrusion of the growing polypeptide through the membrane pore relies on the binding of the ribosome to this pore. Several mechanisms of translocation initiated and sustained by the signal peptide have been advanced. With the discovery of two elements of the eukaryotic export machinery which assure delivery of a secretory protein to the membrane, the signal hypothesis was revised. These elements are the SRP (signal recognition particle), an 11 S ribonucleoprotein consisting of six different polypeptides and a 7 S RNA (Walter and Blobel, 1982), and the SRP-receptor (Gilmore et al.. 1982) or “docking protein” (Meyer et al., 1982). The SRP inhibits translation of secretory (but not cytoplasmic) proteins when microsomes are not present in a cell-free system (Walter and Blobel, 1981a,b; Walter et al., 1981). The 72,000-MW, membrane-bound, SRP-receptor protein must be present in the microsomes to allow release of SRP translation inhibition (Gilmore et al., 1982; Meyer et uf., 1982). The model for protein secretion which encompasses these elements states that SRP binds to the nascent polypeptide of a secretory protein, possibly at the hydrophobic core of the signal peptide,

76

GUY D. DUFFAUD ET AL.

and halts translation. The SRP-polyribosome complex then interacts with the SRP-receptor at the endoplasmic reticulum, the SRP is displaced, translation recommences, and the polypeptide is linearly extruded through the membrane. The SRP, SRP-receptor, and polyribosomes are not present in stoichiometric amounts. The interaction of these components may therefore be transitory and necessary only for attachment of the ribosome to the membrane because it has been indicated that once translation starts at the membrane-associated ribosome neither SRP nor SRP-receptor remain bound to either the ribosome or the nascent polypeptide (Gilmore and Blobel, 1983). Whether the ribosome is directly associated with a membrane protein during the secretory process or whether the nascent polypeptide chain by itself is sufficient to maintain the functional association between the ribosomes and the membrane is not certain. A class of endoplasmic reticulum integral membrane glycoproteins, the ribophorins, have been shown to bind to ribosomes (Kreibich et al., 1978a,b; Rodriguez-Boulan et al., 1978). The possibility that these proteins may be involved with ribosomemembrane interactions during secretion has not been proved. These considerations of protein translocation across a membrane assume that the interaction of the protein with the membrane is initiated by the nascent signal peptide-ribosome-SRP complex and that subsequent translocation of the protein is vectorial and cotranslational. The case for this initial reaction is well documented for eukaryotic systems where the steps are defined and temporally distinct. What actually happens at the moment of protein translocation in eukaryotes is not entirely clear, but translocation is believed to proceed cotranslationally. Bacterial translocation systems are not as well defined. Several proteins are known to be important as part of a translocation machinery. They are not well characterized with regard to function (for review, see Silhavy et al., 1983, and Section V). Bacterial proteins destined for export have been found to be synthesized on membrane-bound polyribosomes (Smith et al., 1977, 1979; Randall and Hardy, 1977). These results suggest the existence of a system similar to that found in eukaryotes. However, it is not known at present when and how polyribosomes become associated with the cytoplasmic membrane. A strong analogy between prokaryotic and eukaryotic secretion machineries is therefore precluded.

B. Membrane Trigger Hypothesis It is clear that in bacteria not all secretory proteins are cotranslationally translocated (Wickner, 1979; Koshland and Botstein, 1982; Randall, 1983). Proteins fully synthesized before crossing the cytoplasmic membrane will most likely fold in a manner such that association of the signal peptide with the membrane and linear extrusion of the remainder of the molecule is impossible. A model proposed to account for posttranslational translocation is the membrane trigger

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

77

hypothesis (Wickner, 1979, 1980). This model was first proposed to account for the mechanism of M I3 phage procoat insertion into the cytoplasmic membrane. It was shown that M 13 procoat protein was made on cytoplasmic polyribosomes, posttranslationally translocated, and cleaved to coat protein (Wickner, 1979, 1980, 1983). This hypothesis states that the signal peptide is necessary for the attainment of a particular conformation. For MI3 procoat protein, the entire protein would be involved, whereas for larger proteins this particular conformation may encompass only a domain of a nascent polypeptide. Once this conformation is achieved, interaction of the protein or protein domain with the cytoplasmic membrane results in a further conformational change, thereby allowing interaction between the hydrophobic residues of the protein and the hydrophobic portion of the bilayer. The signal peptide is thus important in allowing the protein to have alternate conformations, depending on its environment. Cleavage of the signal peptide would abolish this ability and serve to fix the protein in its appropriate location (Wickner, 1979). The driving force would be the membrane proton motive force rather than a complex secretion machinery (Date et al., 1980a,b; Zimmerman et ul., 1982). This hypothesis is based on the data showing that M I3 procoat synthesis appears to occur free in the cytoplasm and not on membrane-bound ribosomes (It0 et al., 1979, 1980). In vitro studies have shown that processing and assembly of M13 procoat protein requires only procoat, a lipid bilayer, and purified signal peptidase (Silver et al., 1981; Watts et al., 1981). This has been taken to indicate that M 13 procoat protein can posttranslationally insert into a lipid bilayer in the proper manner, without the aid of a secretory apparatus. Another secreted protein, pro-OmpA, has been shown to be capable of posttranslational translocation in vitro (Zimmerman and Wickner, 1983). Many other polypeptides are known to be able to spontaneously associate with membranes in vitro (for review, see Wickner, 1980). These examples have been taken as evidence to support the membrane trigger hypothesis. However, whether or not this has any bearing on the mechanism employed in vivo remains to be seen. Under optimal conditions, secretory proteins are rarely accumulated as precursors. This occurs only when the system is perturbed in some way. Association of accumulated precursor proteins and other purified membrane proteins with membranes both in vivo and in vitro attests to the strong affinity of secretory proteins for the lipid bilayer. It may not, however, reflect the mechanism by which secretory proteins are normally translocated across biomembranes. Two additional models which require minimal membrane components have been proposed. A direct transfer mechanism was advanced by von Heijne and Blomberg (1979) in which the signal peptide adopts an a-helix as it emerges from the ribosome. Because of its hydrophobic nature, the signal peptide inserts into the hydrophobic portion of the membrane. The ribosome is then attached to the surface of the membrane by a ribosome binding protein, thus allowing the

78

GUY D. DUFFAUD ET AL.

remainder of the nascent polypeptide to be pushed through the membrane as translation proceeds. Engelman and Steitz (1981) proposed that when a signal peptide adopted the proper conformation-the helical hairpin-the remainder of the protein could be secreted without the aid of additional components.

C. Loop Model The models described in the preceding sections were proposed for the overall process of secretion. The signal hypothesis, as originally proposed, implied that the signal peptide passes linearly through the membrane (Blobel and Dobberstein, 1975a,b; see Fig. 2). In this model (linear model), the highly charged amino-terminal region of the signal peptide would first have to enter the lipid bilayer and pass through to the outside. At the same time the hydrophobic domain of the signal peptide would also have to pass through the membrane. In particular, hydrophobic interactions would form between the hydrophobic domain of the signal peptide and the lipid bilayer. These interactions are considered to be extremely stable; therefore, breaking them in order to push the signal peptide to the outside of the membrane would require a great deal of energy. How can the positively charged amino terminal region of the signal peptide cross the hydrophobic lipid bilayer of the cytoplasmic membrane? How can the hydrophobic domain of the signal peptide emerge through the hydrophobic lipid bilayer to the outside of the cytoplasmic membrane? To accommodate these serious problems in the linear model, the loop model has been proposed (Inouye et al., 1977, 1979; Inouye and Halegoua, 1980). This model is based on the common features observed in the primary structure of prokaryotic signal peptides A

0 (OUTSIDE

___)

( INSIDE)

FIG. 2. Loop model and linear model for the translocation of secretory proteins across the membranes (Inouye and Halegoua, 1980). (A) Loop model; (B) linear model. Solid portions represent the basic positively charged region at the amino terminus of the signal peptide. The following dotted portion represents the hydrophobic domain. The white portion corresponds to the mature protein. The cleavage site is indicated by a small arrow. The broken dotted portion indicates the degraded signal peptide.

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

79

and specifically addresses the role of the signal peptide in the initial steps of protein secretion. In the loop model, instead of the amino terminal section of the signal peptide going across the membrane, it interacts with the inner surface of the cytoplasmic membrane, which has negative charges at physiological pH because of phosphatidylglycerol (Fig. 2A). Next, as the peptide elongates, the hydrophobic domain of the signal peptide is progressively inserted into the cytoplasmic membrane by hydrophobic interactions with the lipid bilayer. During this process, a loop is formed by the hydrophobic domain, which is extended stepwise into the lipid bilayer. Eventually, the cleavage site of the signal peptide becomes exposed to the outer surface of the cytoplasmic membrane, while the positively charged amino-terminal section remains on the inside surface of the cytoplasmic membrane (see Fig. 2A). The proline or glycine usually found in the hydrophobic domain may play an important role in enabling the peptide to form a loop at the proper position. If the hydrophobic domain were fully extended as P-sheet structure, its total length would be approximately 50 A, long enough to extend through the lipid bilayer, thus exposing the cleavage site to the outside surface of the membrane. On the other hand, if it forms an a-helical structure, the signal peptide would be 25 A long, and the cleavage site would be buried in the hydrophobic region of the lipid bilayer where it could interact with the signal peptidase (Inouye, 1979).

EVIDENCE FOR

THE

LOOPMODEL

Since both the loop model and the linear model address a very specific role and function for the signal peptide in protein secretion, one can easily make several predictions according to the models. For example, in the loop model, the cleaved signal peptide should remain in the cytoplasmic membrane after cleavage, while in the linear model the cleaved signal peptide should be undetectable in the cytoplasmic membrane. Furthermore, the loop model predicts that a functional signal peptide could be internalized, while the linear model predicts that signal peptides should always be at the amino terminal end of a secretory protein. In this section, we discuss supporting evidence for the loop model. a. Location of the Cleaved Signal Peptide. In a cell-free assay system for the prolipoprotein signal peptidase, Hussain et al. ( I982a) used envelope fractions prepared from globomycin-treated E . cofi cells. (Globomycin is an antibiotic that inhibits the cleavage of the signal peptide of lipoprotein; see Section V.) They observed that the cleaved prolipoprotein signal peptide remained in the membrane fraction, as predicted by the loop model. In this assay system, the prolipoprotein signal peptidase was extremely stable and active up to 80°C. A second enzymatic activity, responsible for digestion of the cleaved signal peptide, was not active at this temperature. The cleaved signal peptide thus accumu-

80

GUY

D. DUFFAUD ET AL.

lated in the envelop fraction at 80°C, but not at 37°C (Hussain et a l . , 1982a). In the presence of protease inhibitors, prolipoprotein was quantitatively converted to mature lipoprotein and the signal peptide was accumulated in the membrane fraction (Hussain et al., 1982b). The appearance of the signal peptide in the membrane is readily explained by the loop model. b. Orientation of a Precursor Protein in the Membrane. The loop model predicts that if the signal peptide cleavage reaction were inhibited, the resulting precursor protein would not be exported to the outer surface of the membrane, but would be anchored in the membrane by its uncleaved signal peptide. This prediction is supported by studies on the orientation of prolipoprotein accumulated in the envelope of globomycin-treatedE . coli cells. Inukai et al. (1979) first observed that prolipoprotein produced in the presence of globomycin was capable of being covalently bound to the peptidoglycan layer, a finding indicating that, even though the signal peptide was not cleaved, the mature portion of the protein was translocated through the cytoplasmic membrane. These results were extended by Ichihara et al. (1982), who showed that the bound form of prolipoprotein, like the wild-type mature lipoprotein, was covalently attached through its carboxy-terminal lysine to the peptidoglycan layer. It was also shown, by employing subcellular fractionation of E . coli cell envelopes, that prolipoprotein was localized in the cytoplasmic membrane after digestion of the peptidoglycan layer with lysozyme. Inukai and Inouye (1983) further investigated prolipoprotein accumulated in the presence of globomycin and found that lysine residues at positions - 19 and - 16 in the amino-terminal basic region of the signal peptide could be labeled with [3H]dinitrophenylflu~r~benzene only when cells were disrupted by sonication. These results can easily be interpreted according to the loop model as follows: the lysine residues in the amino-terminal basic region are associated with the inner surface of the cytoplasmic membrane, while the major part of the protein is translocated across the cytoplasmic membrane. Therefore, the lipoprotein portion of the precursor molecule starts to assemble in the outer membrane in the absence of signal peptide cleavage. c. Internal Signal Peptide. The loop model, unlike the linear model, does not require that the signal peptide be located at the amino terminus of a secretory protein. It has been demonstrated that the prolipoprotein signal peptide is fully functional even when it is internalized by constructing P-lactamase-prolipoprorein or P-galactosidase-prolipoprotein hydrid proteins (Coleman and Inouye, unpublished data; discussed in Section IV and Fig. 3). Such an observation cannot be explained by the linear model. d. Requirement of p-Turn Structure in the Signal Peptide. The requirement for a p-turn structure for secretion of prolipoprotein was indicated by Vlasuk et al. (1984). Mutants were constructed by oligonucleotide-directed site-specific mutagenesis which decreased the probability for p-turn in the hydrophobic domain (see Section IV). Secretion was inhibited when the probability for p-turn was reduced in the hydrophobic domain.

81

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

Hcoo

Lipid

COOH OUT

FIG. 3. Translocation of lipoprotein with an internalized signal peptide across the cytoplasmic membrane. The hatched area is the fused P-galactosidase. See Fig. 2 for other symbols. The fused protein is inserted in the cytosolic membrane via the hydrophobic domain (dotted portion). Then the prolipoprotein segment is translocated and subsequently processed to mature lipoprotein according to the loop model as illustrated in Fig. 2 (see also Fig. 6).

In conclusion, evidence presented in this section clearly demonstrates that the signal peptide is not secreted in a linear fashion. On the other hand, the loop model attempts to account for the common features of the signal peptide and is consistent with experimental results discussed in this section.

IV. EXPERIMENTAL APPROACHES To assess the functional role of the signal peptide, two approaches have been taken. Both draw on the use of mutations which alter the functions of the signal peptide. The first approach is to isolate and to characterize signal peptide mutants which are defective in secreting proteins across the membrane. The second method is the directing of specific mutations in a signal peptide of a protein in question. In this approach one can design mutations to obtain a specific alteration in the structure of the signal peptide.

A. The Genetic Approach LamB protein functions in maltose and maltodextrin transport (Silhavy et al., 1977; Luckey and Nikaido, 1980) and is the receptor for phage X (Randall-

82

GUY D. DUFFAUD ET AL

Hazelbauer and Schwartz, 1973). LamB protein has a signal peptide of 25 amino acid residues (see Figs. 1 and 4) (Randall et al., 1978; Hedgpeth et al., 1980; Marchal et al., 1980). Mutants have been isolated which are unable to transport LamB protein to the outer membrane because of defects in the signal peptide (Emr et al., 1978, 1980). These series of mutations were well characterized and were located, with a few exceptions, in the signal peptide (Fig. 4). Emr et al. (1980) showed that of 15 mutants isolated, 8 were the result of single base changes in the coding region for the signal peptide. Three of them were identical mutations which altered the amino acid residue at position - 11 from alanine to glutamic acid. In three other cases, residue -7 was changed from methionine to arginine. In the remaining two single base changes, residue - 12 was changed from valine to aspartic acid and residue -10 from alanine to glutamic acid (see Fig. 4). Besides these single point mutations, six deletion mutants were also isolated. Two had deletions within the signal peptide and four had deletions which originated in the signal peptide and extended into the mature protein (Fig. 4). A study according to the rules proposed by Chou and Fasman (1978) of the conformational changes arising from these last four deletions shows that the configuration of the new amino-terminal region does not resemble the original signal peptide. Also, there is no amino acid sequence that is likely to be the equivalent of a new cleavage site. These results show clearly that alteration of the hydrophobic region of the LamB signal peptide, either by insertion of a charged amino acid or by deletion, inhibits protein translocation across the cytoplasmic membrane. A 12-bp deletion mutant in the signal peptide was further utilized to study the function of the hydrophobic region of the signal peptide. This mutant lacks four of seven amino acid residues in the hydrophobic domain and thus was ideal to find pseudorevertants, since true reversion to the wild type was impossible (Emr and Silhavy , 1983). Analysis of pseudorevertants revealed that two separate point mutations led to amino acid substitutions: a glycine residue at position -9 was replaced with a cysteine residue, and in the other, a proline residue at position - 13 (- 17 of wild type) was replaced with a leucine residue (see Fig. 4). These secondary mutations in pseudorevertants were considered to restore a-helical structure in the hydrophobic domain, which was lost in the 12-bp deletion mutant (Emr and Silhavy, 1983). This result suggests that the a-helical conformation of the hydrophobic domain may be important in translocation of proteins across the cytoplasmic membrane. Other export-defective mutants tend to support the above observation. For instance, secretion of MalE protein (maltose binding protein), normally exported to the periplasmic space, becomes defective by a number of mutations in its signal peptide. Mutations which have been analyzed are between residues - 17 and -8 of the signal peptide, once again within the hydrophobic domain (Bedouelle el al., 1980) (see Fig. 4). In each mutation the hydrophobic region is disrupted. Leucine at position -17 was altered to a proline residue; alanine at - 13 to glutamic acid; threonine at - 11 to lysine; and methionine at -8 and -9

I

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

a3

to arginine. These mutants have been shown to lack any striking changes in conformation when analyzed by the rules proposed by Chou and Fasman (1978). The fact that all the mutants allow growth on maltose minimal medium (to a greater or lesser extent always slower than wild type) indicates that mutations in the signal peptide have varying effects. The mutations most detrimental to export of MalE protein were those located at positions -8 and -9. In both cases mutations reduce the probability of P-sheet structure at this region when compared to wild-type protein. The significance of this is unclear. Another periplasmic protein, PhoA (alkaline phosphatase), also has been utilized in a similar type of study. In one mutant, the amino acid residue at position - 14 was changed from leucine to glutamine. This mutant allowed PhoA transport at 16% of the wild-type level. In contrast, replacing residue -7 (leucine) with arginine abolished export of the protein (Michaelis et al., 1983). When the conformations of the signal peptides of all the PhoA (and MalE mutants described above) were analyzed according to the rules of Chou and Fasman (1978), the insertion of a charged amnio acid causes little change in the conformation of the hydrophobic domain of the signal peptide, as has been reported for the LamB protein (Emr et al., 1980). Thus, the deficiency in export of those mutants in which a charged amino acid was introduced may be due to the insertion of that charge in the hydrophobic domain but not to a change in the conformation of the signal peptide. These results indicate that the hydrophobic domain of the signal peptide should not contain any charged amino acid residues in order to remain functional in transport. A similar example can be found in the outer membrane lipoprotein. When glycine at position -7 of the signal peptide was replaced with an aspartic acid residue, the mutant prolipoprotein was accumulated in the cytoplasmic membrane (Lin er al., 1978). When this mutant prolipoprotein gene was inserted into an inducible expression vector and was expressed together with the wild-type lipoprotein gene, the association of the mutant prolipoprotein with the cytoplasmic membrane was severely inhibited and it was mainly accumulated in the cytoplasm (Lee et al., 1983). This indicates that the mutant prolipoprotein utilizes the same secretion machinery as the wild-type lipoprotein, but much less efficiently. Talmadge et al. (1980) have found that fusion proteins produced from hybrid genes of the first half of the signal peptide of p-lactamase and the last third of the rat preproinsulin signal peptide are not effectively transported to the periplasm and remain in the cytoplasm. The intact signal peptide from either the bacterial p-lactamase gene or the eukaryotic preproinsulin gene were sufficient to allow limited transport of the fused preproinsulin gene to the periplasmic space. This fact indicates that the fused signal peptide was functionally inadequate (Talmadge et al., 1980). Possibly this was because the fusion resulted in a defective hydrophobic domain containing charged amino acid residues (Talmadge et al., 1980).

LAMBDA RECEPTOR PROTEIN -25

-15

-20

-5

- 10

-1

+I

Met-Met-Ile-Thr-Leu-Arg-Lys-Leu-Pro-Leu-Ala-Val-Ala-Val-Ala-Ala-Gly-Val-Met-Ser-Ala-Gln-Ala-Met-Ala-Val Point murations (1)

Met-Met-lle-Thr-Leu-Arg-Lys-Leu-Pro-Leu-Ala-Val-Ala-~-Ala-Ala-Gly-Val-Met-Ser-Ala-Gln-Ala-Met-Ala-Val Met-Met-Ile-Thr-Leu-Arg-Lys-Leu-~-Leu-Ala-Val-Ala-Val-~-Ala-Gly-Val-Met-Ser-Ala-Gln-Ala-Met-Ala-Val Met-Met-Ile-Thr-Leu-Arg-Lys-Le~-Pro-Leu-Ala-Val-Ala-Val-Ala@-Gly-Val-Met-Ser-Ala-Gln-Ala-Met-Ala-Val Met-Met-Ile-Thr-Leu-Arg-Lys-Leu-Pro-Leu-Ala-Val-Ala-Val-Ala-Ala-Gly-Val@-Ser-Ala-Gln-Ala-Met-Ala-Val Deletions ( 1) -17

-12

Met-Met-Ile-Thr-Leu-Arg-Lys-Leu-Pro~tVal-Ala-AlaCly-Val-Met-Ser-Ala-Gln-Ala-Met-Ala-Val-Asp-Phe-His-Gly-Tyr-Ala -17

-4

[Gin-Ma-Met-Ala-Val-Asp-Phe-His-Gly-TyrAla

Met-Met-Ile-Thr-LeuArg-Lys-Leu-Pro]

-'

Met-Met-Ile-Thr-Leu-Arg-Lys-Leu-Pro-Leu-Ala-Val-Ala-Val-Ala-Ala-Gly-Val 1

Met-Met-Ile-Thr-LeuArg-Lys -

Met-Met-Ile-Thr-LeuArg-Lys Met-Met-Ile-Thr-Leu-Arg

1

9

1

7

I

' ~-~~g-Leu-Gly-Am-Gl"-~ys-Glu-Thr-T~ -ProtThr-Gly-Ala-Gln-Ser-Lys-Tyr-Arg-Leu-Gly +68 -~~Val-Ala-Gln-Gln-Asn-Asp-Trp-Glu-Ala-Th + I50 -&.+ly-Gly-Ser-Ser-Ser-Phe-Ala-Ser-Asn-Asn

Pseudorevertants (2) - 12

-17

Met-Met-lle-Thr-Leu-Arg-Lys-Leu-~o~~~Val-Ala-Ala-Gly-Val-Met-Ser-Ala-Gln-Ala-Met-Ala-Val Met-Met-Ile-Thr-Leu-Arg-Lys-Leu-Pro~~Val-Ala-Ala-~-Val-Met-Ser-Ala-Gln-Ala-Met-Ala-Val Met-Met-Ile-Thr-Leu- A r g - L y s - L e u - ~~ t V a l - A l a - A l a - G l y - V a l - M e t - S e r - A l a - G l n - A l a - M e t - A l a - V a l MALTOSE-BINDING PROTEIN -26

-20

-15

- 10

-5

-I

tl

Met-Lys-Ile-Lys-Thr-Gly-Ala-Arg-Ile-Leu-Ala-~u-Ser-Ala-Leu-~r-Thr-Met-Met-Phe-Ser-Ala-Ser-Ala-Leu-Ala-Lys Mutants (3) Met-Lys-Ile-Lys-Thr-Gly-Ala-Arg-Ile-~-Ala-Leu-Ser-Ala-Leu-~r-Thr-Met-Met-Phe-Ser-Ala-Ser-Ala-Leu-Ala-Lys Met-Lys-Ile-Lys-Thr-Gly-Ala-Arg-Ile-Leu-Ala-Leu-Ser-~-Leu-Thr-Thr-Met-Met-Phe-Ser-Ala-Ser-Ala-~u-Ala-Lys -Thr-Met-Met-Phe-Ser-Ala-Ser- Ala-Leu-Ala-Lys Met-Lys-Ile-Lys-Thr-Gly-Ala-Arg-Ile-Leu-Ala-Leu-Ser-Al~-Leu~ Met-Lys-Ile-Lys-Thr-Gly-Ala-Arg-Ile-Leu-Ala-Leu-Ser-Ala-~u-Thr-Thr-@)-Met-Phe-Ser-Ala-Ser-Ala-Leu- Ala-Lys Met-Lys-Ile-Lys-~r-Gly-Ala-Arg-1le-Leu-Ala-Leu-Ser-Ala-Leu-Thr-Thr-MetAla-Ly s ALKALINE PHOSPHATASE

~-Phe-Ser-Ala-Ser- Ala-Leu-

-2 I

- 15

-10

-5

-1

+ I

Met-Lys-Gln-Ser-Thr-Ile-Ala-Leu-Ala-Leu-Leu-Pro-Leu-Leu-Phe-Thr-Pro-Val-Thr-Lys-Ala-Arg Mutants (4) Met-Lys-Gln-Ser-Thr-Ile-Ala-~-Ala-Leu-Leu-Pro-Leu-Leu-Phe-T~-Pro-Val-Thr-Lys-Ala-Arg Met-Lys-Gln-Ser-Thr-Ile- Ala-Leu- Ala-Leu-Leu-Pro-Leu-~-Phe-Thr-Pro-Val-Thr-Lys-Ala-Arg

FIG. 4. Signal peptides of export-defective proteins. Signal peptide mutants. defective in secretion, are listed. Three types of mutants have been studied for LamB protein: point (circled); deletion (boxed); and pseudorevertants arising from a point mutation of one deletion mutant. Amino acids resulting from deletion mutations are included inside the boxes. The signal peptide mutants of maltose-binding protein (MalE) and alkaline phosphatase (PhoA) are all point mutations, which are indicated by a circle. The effect of these mutations is discussed in the text. References: ( I ) Emr ef a / . (1980); (2) Emr and Silhavy (1983); (3) Bedouelle er a / . ( I 980); (4)Michaelis ef ol. (1983).

86

GUY D. DUFFAUD ET AL.

B. Oligonucleotide-Directed Site-Specific Mutagenesis in the Slgnal Peptide The second approach is site-specific mutagenesis with the use of synthetic oligonucleotides as mutagens. This approach has been used extensively to study the functions of the signal peptide of the E. cofi outer membrane lipoprotein. The lipoprotein is the most abundant protein in E. cofi and one of the most extensively investigated membrane proteins (see review by Inouye, 1979). As described earlier, prolipoprotein contains a signal peptide consisting of 20 amino acid residues. The amino-terminal structure of the mature lipoprotein is unique, consisting of a glycerylcysteine [S-(propane-2’,3’-diol)-3-thioaminopropionic acid] to which two fatty acids are linked by two ester linkages and one fatty acid by an amide linkage. Its gene has been cloned in an inducible vector so that its expression can be controlled. Thus, lipoprotein provides an ideal model system for the study of secretory proteins and, in particular, the functions of the signal peptide. In principle, one can construct any mutation such as a base substitution, an insertion, and a deletion, in any part of a gene by oligonucleotide-directedsitespecific mutagenesis. In fact, such mutations have been introduced in various regions of the prolipoprotein signal peptide, as listed in Fig. 5. In this section, we discuss these mutants according to the regions in the signal peptide. 1 . THEAMINO-TERMINAL REGION

The first group of mutants are designed to investigate the role of the positively charged amino-terminal region of the signal peptide, which, as discussed above is considered to play an important role in the initial step of protein translocation across the membrane. For this purpose, the positive charges due to two lysine residues in the amino-terminal region were systematically altered from +2 to 1, 0, - 1,and -2 as shown in Fig. 5A. Among the seven mutants thus obtained, mutants, I1 ( + l ) , I2 ( + l ) , I3 (0) (Inouye et al., 1982), I5 ( + I ) , and I6 (0) (Vlasuk et al., 1983) were found to export and process their respective prolipoproteins as efficiently as wild type, although the rate of synthesis was reduced in all cases. In mutants 11, 12, and 15, this result was not surprising, since there was one remaining positive charge at the amino-terminal region. However, the results for mutants I3 and I6 were rather surprising since the net charge at the amino terminal region in these mutants became zero. In particular, in the case of mutant 16 there were no positively charged amino acids at its amino-terminal region (see Fig. 5A). In contrast to the mutants described above, when the net charges at the aminoterminal region became negative, striking effects were observed. In both I4 (- 1) and I7 (-2) mutants, significant accumulation of unmodified prolipoprotein was detected in the cytoplasmic fraction when the cells were pulse-labeled with

+

Wild-type signal sequence - 20

\vr

- I5

-10

Met- Lys - Ala -Thr -Lys -Leu-Val-Leu- Gly -Ala-Val-lle-Leu- Gly

-5

-

-1

+I

Ser - Thr -Leu- Leu -Ala- Gly - Cys-

(+I)

A Mutants affecting the global charge of the amino terminus Met- Lys -@-Thr- Lys -Leu-Val-Leu- Gly -Ala-Val-Ile-Leu- Gly - Ser - Thr -Leu- Leu -Ala- Gly - CysI 1 (D3) I2 (A2) - Ala -Thr- Lys -Leu-Val-Leu- Gly -Ala-Val-Ile-Leu- Gly - Ser - Thr -Leu- Leu -Ala- Gly - Cys13 (A2,D3) ~~~~~-~ s -Thr- Lys -Leu-Val-Leu- Gly -Ala-Val-lle-Leu- Gly - Ser - Thr -Leu- Leu -Ala- Gly - CysMet- lu - As -Thr- Lys -Leu-Val-Leu- Gly -Ala-Val-Ile-Leu- Gly - Ser - Thr -Leu- Leu -Ala -Gly - CysI4 (E2,D3) I5 (N5) Met- Lys - Ala -Thr- s -Leu-Val-Leu- Gly -Ala-Val-Ile-Leu- Gly - Ser - Thr -Leu- Leu -Ala- Gly - CysM e t - 8 - Ala - T h r - g - L e u - V a I - L e u - Gly -Ala-Val-Ile-Leu- Gly - Ser - Thr -Leu- Leu -Ala- Gly - CysI6 (A2,N5) I7 (E2,D3,N5) Met- GI -@-Thrs -Leu-Val-Leu- Gly -Ala-Val-lle-Leu- Gly - Ser - Thr -Leu- Leu -Ala- Gly - CysB Substitution of the glycine residues Met- Lys - Ala -Thr- Lys -Leu-Val-Leu- Ser - Thr -Leu- Leu -Ah- Gly - CysA1 (V9) - Ser - Thr -Leu- Leu -Ala- Gly - CysMet- Lys - Ala -Thr- Lys -Leu-Val-LeuA2 (A9) Met- Lys - Ala -Thr-Lys -Leu-Val-Leu- Ser - Thr -Leu- Leu -Ala- Gly - CysBI (V14) - Ser - Thr -Leu- Leu -Ala- Gly - CysMet- Lys - Ala -Thr- Lys -Leu-Val-LeuB2 (A14) - Ser - Thr -Leu- Leu - A h Gly - CysAlBl (V9,V14) Met- Lys - Ala -Thr- Lys -Leu-Val-Leu- Ser - Thr -Leu- Leu -Ala- Gly - CysA1B2 (V9.AI4) Met- Lys - Ala -Thr- Lys -Leu-Val-Leu- Ser - Thr -Leu- Leu -Ala- Gly - CysA2B1 (A9,V14) Met- Lys - Ala -Thr- Lys -Leu-Val-Leu- Ser - Thr -Leu- Leu -Ala- Gly - CysA2B2 (A9,A14) Met- Lys - Ala -Thr- Lys -Leu-Val-LeuC Substitution of the serine-threonine residues Met- Lys - Ala -Thr- Lys -Leu-Val-Leu- Gly -Ala-Val-Ile-Leu- Gly -@Thr -Leu- Leu -Ala- Gly - CysHI (A15) Met- Lys - Ala -Thr- Lys -Leu-Val-Leu- Gly -Ala-Val-lle-Leu- Gly - Ser - Ala -Leu- Leu - A h Gly - CysH2 (A16) Met- Lys - Ala -Thr- Lys -Leu-Val-Leu- Gly -Ala-Val-Ile-Leu- Gly -@Ala -Leu- Leu -Ala- Gly - CysH3 (A15.AI6) D Mutants affecting the cleavage site Met- Lys - Ala -Thr- Lys -Leu-Val-Leu- Gly -Ala-Val-lle-Leu- Gly - Ser - Thr -Leu- Leu -Ala -@CysC1 (A20) Met- Lys - Ala -Thr- Lys -Leu-Val-Leu- Gly -Ala-Val-Ile-Leu- Gly - Ser - Thr -Leu- Leu -Ala - Gly -@C2 ((321) Met- Lys - Ala -Thr- Lys -Leu-Val-Leu- Gly -Ala-Val-Ile-Leu- Gly - Ser - Thr -Leu- Leu -Ala CysC3 (820) Met- Lys - Ala -Thr- Lys -Leu-Val-Leu- Gly -Ala-Val-Ile-Leu- Gly - Ser - Thr -Leu- Leu -Ala - Gly -@-Cys C4 (2OG21) Met- Lys - Ala -Thr- Lys -Leu-Val-Leu- Gly -Ala-Val-Ile-Leu- Gly - Ser - Thr - L e u - o - A l a - Gly - CysC5 (A.18)

8

-0-

FIG.5 . Prolipoprotein signal peptide mutants obtained by oligonucleotide-directed site-specific mutagenesis. The altered amino acids are circled. The deletions are symbolized by an empty circle. The mutants are named in two ways: a letter and a number code which is used in the text, and, in parenthesis, a systematic nomenclature proposed here. This nomenclature follows four rules: ( 1 ) numbers always refer to the wild-type amino acid positions (in the positive number system); (2) substitution is indicated by the single letter code name of the new amino acid followed by the position at which the subtitution takes place; (3) deletion is indicated as for substitution but using A instead; and (4) insertion is expressed as the inserted amino acid(s) sandwiched by two numbers corresponding to the amino acids between which the insertion occurs.

aa

GUY D. DUFFAUD ET AL.

[35S]methionine (S. Inouye et al., 1982; Vlasuk et al., 1983). These soluble precursor molecules were then posttranslationally translocated across the cytoplasmic membrane and processed to the fully modified mature lipoprotein. The rate of translocation for mutant I7 was slower by a factor of two than that for mutant 14. The posttranslational translocation of cytoplasmic prolipoprotein mutant I7 was inhibited by carbonylcyanide-m-chlorophenylhydrazone(CCCP). Thus, this process may require a transmembrane electrochemical potential similar to the posttranslational insertion of the phage M13 coat protein into the cytoplasmic membrane (Date et al., 1980a,b). At present, it is not known how the positively charged amino-terminal region of prolipoprotein participates in the initial stages of secretion. The results obtained from these seven amino-terminal mutants clearly indicate that a positive charge at the amino-terminal region of prolipoprotein is not an absolute requirement for its secretion. However, whether the presence of such a charge in this region facilitates this process could not be determined from the present experiments. In contrast, the presence of a net negative charge in this region severely affects the rate of translocation. It is not known whether this effect on translocation is directly due to the charge at the amino-terminal region as predicted by the loop model or instead to a secondary effect of these mutations on the level of protein synthesis. In this regard, it should be noted that the positively charged amino-terminal region is required for maximal prolipoprotein synthesis since the rate of prolipoprotein synthesis was reduced in all of the amino-terminal mutants studied thus far. This reduced synthesis does not appear to be due to the formation of more stable secondary structures within the region of the mRNA corresponding to the amino-terminal region as was shown for a pro-LamB aminoterminal mutant (Schwartz et al., 1981). Instead, the decreased level of prolipoprotein synthesis appears to be related to the net charge present at the aminoterminal region with mutants having a net negative charge (i.e., mutants I4 and 17) being the lowest. The exact influence of the amino-terminal charge on the translational efficiency of these mutants is not known at this time, although a possible coupling between translation and translocation, mediated through systems similar to those described for eukaryotic secretory protein biosynthesis (Walter and Blobel, 1981b) or those described in a model proposed by Hall and Schwartz (1982) and Hall et al. (1983) may be involved. 2. THE HYDROPHOBIC DOMAIN a. Glycine Residues. As seen earlier, one of the common features of the signal peptide is that in general there is one or more glycine and/or proline residues in the hydrophobic domain. In the case of the signal peptide of prolipoprotein, there are two glycine residues at positions -7 and - 12 (see Fig. 5). It has been shown that replacement of glycine at position -7 with aspartic acid

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

89

prevented the modification with glycerol or fatty acids and cleavage of the signal peptide (Wu et al., 1977). However, this mutation did not affect the secretion of the mutant prolipoprotein, which was found primarily in the outer membrane fraction (Wu et al., 1977; Lee et al., 1983). To study in more detail the importance of these residues in the processing and secretion of prolipoprotein, eight mutants were constructed; in these mutants the glycine residues were systematically either deleted or substituted with a valine residue, as shown in Fig. 5B (Inouye er a / . , 1984). Analysis of these mutants revealed that only the deletion of glycine at position -7 resulted in an accumulation of glyceride-modified prolipoprotein in the membrane fraction. The other individual mutants could not be distinguished from wild type with respect to their secretion and processing. The prolipoprotein accumulated in mutant B2 could be chased, in the presence of excess nonradioactive methionine, to mature lipoprotein, thus indicating that the deletion of glycine at position -7 alters the efficiency of signal peptide cleavage by the signal peptidase, possibly through a conformational change in this region of the signal peptide. The same phenotype observed in mutant B2 was also observed in the double mutant A1B2 in which glycine at -7 was changed to valine and glycine at - 12 was deleted (Fig. 5B). However, the effect of deleting glycine at -7 could be compensated for by the deletion of glycine at - 12, as evidenced by the normal secretion of mutant A2B2. By eliminating both glycine residues without changing the length of the hydrophobic core, e.g., mutant AlBl (Fig. 5B), we have definitively shown that the glycine residues per se are not required for normal secretion since the expression of this mutant prolipoprotein could not be distinguished from wild type. From these results, however, it is not easy to conclude the functional roles of the glycine residues in the lipoprotein signal peptide. The “helix-bending” property of the glycine residue may be important at position -7 but not at - 12. The question of how the deletion of glycine at - I2 supresses the inhibitory effect of the deletion of glycine at -7 remains to be answered. The results obtained from these mutants may indicate that the distance of the cleavage site from a putative recognition site within the signal peptide is important in the rapid and efficient cleavage of the signal peptide. Further experiments will be required to definitively assign a role, if any, for the glycine residues in the prolipoprotein signal peptide and to determine whether a recognition site does indeed exist. b. Serine and Threonine Residues. The lipoprotein signal peptide has serine and threonine residues located at positions -6 and - 5 , respectively (Fig. 5). To investigate the importance of this well-conserved feature of the signal peptide in the secretion of prolipoprotein (see Fig. I ) , these residues were systematically changed to alanine, as shown in Fig. 5C. These mutations had no detectable effects on the secretion of their respective prolipoproteins as compared to wild type when examined at 37°C (Vlasuk et al.,

90

GUY D. DUFFAUD ET AL.

1984). However, when analyzed at 42"C, mutants H1 and, to a greater extent, H3 (Fig. 5C) were found to accumulate prolipoprotein in the membrane fraction, whereas mutant H2 and wild type did not. Subsequent analysis indicated that the accumulated prolipoprotein in both mutants H1 and H3 was unmodified prolipoprotein. This accumulated prolipoprotein was found to be slowly converted to mature lipoprotein in both mutants. In addition, the glyceride-modified prolipoprotein was not detected in these mutants, a finding indicating that the glyceride modification step in the conversion to mature lipoprotein was slow followed by a rapid cleavage of the signal peptide. The conversion of the accumulated membrane-bound prolipoprotein to the mature lipoprotein was inhibited by CCCP. The membrane-bound unmodified precursor accumulted in mutant H3 is believed to be in the cytoplasmic membrane fraction, anchored via its signal peptide. The fact that it is slowly converted to mature lipoprotein could reflect the fact that it is not yet accessible to the modification and/or processing enzymes, i.e., not translocated across the membrane bilayer. This may explain why in both wild type and mutant H3 this precursor persists in the presence of CCCP, which has been proposed to inhibit the translocation event (Wickner, 1983). Alternatively, the mutation in mutant H3 may be inhibiting the enzymes responsible for the lipid modification, a condition which would also result in the slow processing of the unmodified membrane-bound precursor. Which of these two processes is actually occurring is currently being investigated. In addition to the membrane-bound unmodified precursor, mutant H3 accumulated a small fraction of unmodified prolipoprotein in the cytoplasmic fraction which appeared to be posttranslationally translocated to the membrane fraction. This process was also inhibited by CCCP in a manner similar to that observed for the negatively charged amino-terminal mutants described above. This indicates that replacement of serine and threonine has the twofold effect of causing a slow association of the precursor with the cytoplasmic membrane followed by a slow translocation and/or modification. Whether this interpretation of the results is correct remains to be seen. However, it is clear that both the serine and, to a lesser extent, the threonine residues (located at positives -6 and -5, respectively) in the prolipoprotein signal peptide play a role in the secretion and processing of prolipoprotein. It should be noted that changing serine at -6 to alanine results in a significant change in the conformation, specifically the P-turn structure of the signal peptide in the region of the cleavage site, as determined by the rules developed by Chou and Fasman (1978). Replacement of threonine at -5 does not affect the conformation significantly; however, replacement of both serine and threonine with alanine completely abolishes this p-turn structure. Whether this correlation between the secondary structure of prolipoprotein and its secretion is real remains to be proved. Alternatively, the functional hydroxy groups of these amino acids may be important as a recognition site for enzymes involved in the secretory and/or

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

91

processing steps. Future studies involving such mutants should lead to a more definitive answer regarding their function. SITE 3. THE CLEAVAGE

Although signal peptidases can be classified as endopeptidases like trypsin and chymotrypsin, they are considered to have a more stringent requirement for their proteolytic activity. The specificities of endopeptidases such as trypsin and chymotrypsin are determined by the amino acid residues at the cleavage site. In contrast, the signal peptidase requires not only specific amino acid residues but also a specific secondary structure or conformation at the cleavage site. As discussed earlier, alanine is found at positive - 1 in almost all signal peptides, prolipoprotein being the one exception. The cleavage site of the prolipoprotein signal peptide has glycine at position - I , although it can be replaced with alanine without any significant effect on secretion and processing of prolipoprotein (Inouye et al., 1983a). The specificity for the amino acid residue at position + I of almost all signal peptides has been shown to be much less stringent. However, for prolipoprotein the requirement of the glyceride-modified cysteine residue at the cleavage site has been demonstrated (Inukai et al., 1978; Hussain et al., 1980; Ichihara et al., 1981; Tokunaga et al., 1982). This was also shown more directly and unambiguously by the construction of a lipoprotein mutant in which the cysteine residue was replaced with a glycine residue (mutant C2 in Fig. 5D; Inouye et al., 1983b). It was found that the mutant prolipoprotein was accumulated in the outer membrane without any detectable mature lipoprotein. The most interesting cleavage-site mutants are those which demonstrate its stringent requirement of secondary structure. As mentioned earlier, the replacement of the glycine residue at position - I with an alanine residue (mutant C2) had no effect on the function of the signal peptide (Inouye et al., 1983a). Surprisingly, when the glycine residue at position - 1 was deleted, the mutant prolipoprotein was neither modified nor processed (mutant C3; see Fig. 5D; Inouye et al., 1983a). The potential cleavage site of mutant C3 prolipoprotein is still Ala-Cys, since the amino acid residue at position -2 is alanine. Therefore, the major requirement at the cleavage site in this mutant appears to be the secondary structure, not the primary structure. Likewise, mutant C5, in which the leucine residue at position -3 was deleted, behaves in much the same fashion (see Fig. 5D; unpublished data). This mutant prolipoprotein could be modified and processed to produce the mature, fully modified lipoprotein. However, both modification and processing were very slow. As discussed earlier, mutant B2 prolipoprotein, in which the glycine residue at position -7 was deleted (Fig. 5B) was also slowly processed. This may be due to the effect of the deletion mutation on the conformation at the cleavage site. When an extra glycine residue was inserted at the cleavage site (mutant C4, unpublished results; Fig. 5D),modification and processing were not altered.

92

GUY D. DUFFAUD ET AL.

Chou and Fasman rules predict a very high probability of @-turnstructure at the prolipoprotein cleavage site. Whether this has any functional significance for the signal peptide remains to be shown. It is particularly interesting, therefore, to introduce mutations in the mature lipoprotein (instead of inside the signal peptide) which result in defective modification and/or processing of prolipoprotein. C. Steps of Lipoprotein Secretion From the analysis of the mutants described above, we are able to dissect the steps required for secretion, modification, and processing of prolipoprotein. Figure 6 schematically illustrates these steps. A model based on this analysis is presented in this section. 1. STEPA: MEMBRANE ASSOCIATION

For wild-type prolipoprotein, as soon as the amino-terminal basic region emerges from the ribosome, it interacts with the inner surface of the cytoplasmic

COTRANSLATIONAL

SECRETION

--

--2-

a

e

. L

8

0

Globomycin

M

~

~

~

~ Translocation ~ ~ i o

I-mutants A , % - mutants

I POSTTRANSLATIONAL

H-mutants

nModification

Cleavage

C-mutants 8-mutants

0

MaturationAssembly

C-mutants

1 DE~~~~~NSLATIONAL

FIG. 6 . Translocation and assembly of lipoprotein. Symbols utilized: Solid area, the positively charged amino-terminalregion of the signal peptide; dotted area, the hydrophobicdomain; clear area, lipoprotein; and oval with two dots, the glyceride group linked to the amino terminal cysteine residue (+ I). The smaller and the larger circle represent, respectively, the 30 S and 50 S ribosomes. See text for details.

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

93

membrane (Step A). This step is most likely to be facilitated in the cell by a specific secretory machinery or complex like the eukaryotic SRP. As will be discussed later, there is enough genetic evidence to support this notion. This step is drastically slowed when the charge of the amino-terminal region is changed from + 2 to a negative charge (mutants 14 and 17). The positive charge in this region may be required for the efficient interaction of the ribosome with an SRPlike complex or for the effective association of a ribosome-SRP-like complex with a “docking protein” in the cytoplasmic membrane. In any event, the negative charge in this region appears to block a necessary interaction and results in the accumulation of soluble unmodified prolipoprotein in the cytoplasm. In eukaryotes, translation is arrested until the SRP-ribosome complex binds to the docking protein. In bacterial secretion, arrest by an SRP-like component may not be as tight as in eukaryotic systems. Consequently, even when there is no interaction between a SRP-like complex and a docking protein in prokaryotes, translation can still proceed. The positive charge ( + 2 ) in the amino-terminal basic region appears to be important for efficient, coordinated production of prolipoprotein. The substantial reduction of the synthesis of prolipoprotein in all the mutants in the amino-terminal basic region may therefore be due to the loss of this coordination, even in the case of the mutants which still retain one positive charge in this region (11, 12, and 15). In the case of mutants I4 (- 1) and 17 (-2), soluble mutant prolipoproteins are able to be secreted posttranslationally. This process appears to be slower with the more negatively charged mutant (17). It is important to notice that a protein can be secreted either cotranslationally or posttranslationally, depending upon the structure of the signal peptide (see Fig. 6). It is not certain at present whether this posttranslational translocation can occur without any components. However, this process appears to depend on a mechanism which can be blocked by CCCP.

2. STEPB: TRANSLOCATION Translocation of lipoprotein follows immediately after Step A. The energy required for its translocation has not yet been characterized. If the translocation is coupled or synchronized with the elongation of the peptide and the ribosome is somehow attached to the membrane, then the energy for peptide bond formation may be sufficient for translocation. However, at least for mutants such as HI and H3 which show slow translocation, translation and translocation of the peptide across the membrane are not coupled. This results in the accumulation of prolipoprotein attached to the inner surface of the cytoplasmic membrane (see Fig. 6 ) . This portion is then slowly translocated by a mechanism which is again CCCP-sensitive. Such delayed cotranslational translocation is thought to occur for several secretory proteins in E . coli (Josefsson and Randall, 198 I;Randall, 1983). This type of translocation may be possible for certain secretorjr proteins without the need of additional energy (von Heijne and Blomberg, 1979; Engelman and Steitz, 1981).

94

GUY D. DUFFAUD ET AL

3. STEPC: MODIFWATION

In the case of prolipoprotein, the modification of the cysteine residue at position 1 with glyceride is a prerequisite for the cleavage of the signal peptide by the signal peptidase. This modification reaction appears to require a very specific secondary structure at the cleavage site, as can be seen in the analysis of mutant B2 (deletion of glycine at position -7), C3 (deletion of glycine at position -l), and C5 (deletion of leucine at position -3) (see Fig. 5).

+

4. STEPD: CLEAVAGE The signal peptide of wild-type prolipoprotein is cleaved immediately after its modification. This cleavage reaction appears to occur during the translocation of the peptide, since no accumulation of the modified prolipoprotein is observed. The enzyme responsible for this cleavage, the prolipoprotein signal peptidase or signal peptidase 11, seems to require not only a specific amino acid sequence but also specific secondary structure at the cleavage site. Globomycin, a cyclic peptide antibiotic, specifically inhibits this enzyme, possibly working as a substrate analog. Therefore, globomycin treatment of cells results in the accumulation of glyceride-modified prolipoprotein anchored in the outer surface of the cytoplasmic membrane through the signal peptide. A part of the lipoprotein portion appears to be associated with the outer membrane as well. The carboxy terminus of prolipoprotein has been shown to be linked to the peptidoglycan, thus clearly demonstrating that the entire lipoprotein portion is completely translocated across the cytoplasmic membrane (Inukai ef al., 1979; Ichihara ef al., 1982). Upon cleavage of the signal peptide, it is immediately digested by the signal peptide peptidase (Hussain et al., 1982a,b), which removes a potentially hazardous component from the membrane and helps recirculation of amino acid residues for protein synthesis.

5. STEPE: MATURATION AND ASSEMBLY After the cleavage of the signal peptide, another fatty acid is linked to the free amino group, thereby forming an amide linkage. The fully modified mature lipoprotein thus synthesized is then assembled into the outer membrane. It should be emphasized that the modification and, more importantly, the cleavage of the signal peptide is not absolutely required for translocation (or secretion) of the lipoprotein across the cytoplasmic membrane. As pointed out earlier, mutant prolipoproteins such as C1 and C2 can be still translocated across the cytoplasmic membrane and are found associated with outer membrane. Removal of the signal peptide must be required for lipoprotein to be normally assembled in the outer membrane and to be fully functional. The function of the internalized signal peptide of prolipoprotein can be explained in the same way: A long stretch of the peptide is connected at the amino-

I

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

95

terminal end of the internalized signal peptide (see Fig. 3). This amino-terminal extra peptide can be folded to form a specific conformation before the signal peptide starts to interact with the membrane. Such a structure may prevent the signal peptide from interacting with the membrane. Therefore, it should be kept in mind that the internalized signal peptide may not be functional all the time.

V.

COMPONENTS INTERACTING WITH THE SIGNAL PEPTIDE IN BACTERIA

A. Signal Peptidases In eukaryotes several factors are known to interact with the signal peptide (Section III,A), whereas in prokaryotes the only component which has been clearly demonstrated to interact with the signal peptide is the signal peptidase. In E. cofi it has been shown that there are at least two signal peptidases, signal peptidase I (SPase I) and signal peptidase I1 (SPase 11). SPase II is apparently specific for the signal peptide of the prolipoproteins, whereas SPase I is considered to be required for cleavage of the signal peptides of all other secretory precursors in E. coli. SPase I was first characterized with the use of M13 procoat protein as a substrate and was originally referred to as leader peptidase (Zwizinski and Wickner, 1980). However, in this article it will be referred to as signal peptidase I (SPase I). The gene coding for SPase I (lepl) has been characterized and sequenced (Date and Wickner, 1981; Wolfe er al., 1983). The fepl gene has been mapped between purl and nadB at 54-55 minutes of the E. cofi chromosome (Silver and Wickner, 1983) and has been shown to be essential (Date, 1983). Initial experiments indicated that SPase I is located in both the cytoplasmic and outer membranes (Zwizinski et al., 1981). More recent experiments (Wolfe et a f . , 1983) have shown that in an E. cofi strain overproducing SPase I only 10% of the SPase 1 was found in the outer membrane. It is not clear whether SPase I in the outer membrane has any functional significance. As mentioned, the gene coding for SPase I (lepl)has been sequenced, and the predicted amino acid sequence agrees with the amino acid analysis obtained from purified SPase 1 (Wolfe et al., 1983). Also, the calculated molecular weight (34,994) was comparable to the observed molecular weight (37,000). Tryptic digestion of spheroplasts, followed by anti-SPase I immunoprecipitation, showed that SPase I spans the cytoplasmic membrane. The majority of the polypeptide chain was exposed to the outer surface, and the amino-terminal portion was anchored to the membrane. It is interesting to note that SPase I itself does not possess an amino-terminal signal peptide (Wolfe et al., 1983). It has been demonstrated that SPase I cleaves a variety of secretory precursor proteins (Wickner, 1983).

96

GUY D. DUFFAUD ET AL.

The signal peptidase for prolipoprotein (SPase 11) has not yet been purified. One of the striking characteristics of SPase I1 is its substrate specificity. Processing of prolipoprotein occurs only after glyceride modification of the cysteine residue at the cleavage site (Hussain et al., 1980; Tokunaga et al., 1982). When cells were treated with globomycin, the glyceride-modified prolipoprotein with an uncleaved signal peptide accumulated in the membrane fraction (Inukai et al., 1978; Hussain et al., 1980). This precursor form of prolipoprotein could be converted to mature lipoprotein in an in vitro system which was sensitive to globomycin (Tokunaga et al., 1982). The fact that only prolipoproteins, but not any other secretory precursor proteins, are accumulated in the globomycin-treated cells clearly indicates the existence of a specific signal peptidase for prolipoproteins. This fact is further demonstrated by the isolation of a temperaturesensitive SPase 11 mutant (Yamagata et al., 1982, 1983). 8. Evidence for Other Components

Clear demonstration of the requirement of SRP and docking protein for protein secretion in eukaryotic cells (Section III,A) prompted a search for similar machinery in E . coli. It is important to point out that a prokaryotic signal peptide was shown to be functional in a eukaryotic system (Muller et af., 1982). Similarly, a eukaryotic signal peptide was demonstrated to be partially functional in a prokaryotic system (Talmadge, et al., 1980). This mutual exchangeability of signal peptides between prokaryotes and eukaryotes, however limited, is very suggestive, but not necessarily supportive, of the existence in prokaryotic cells of components resembling SRP and docking protein. Requirement of protein components for protein secretion in prokaryotes has been suggested. In Bacillus subtilis, a protein of molecular weight 64,000, peripherally associated with the inside of the cytoplasmic membrane, has been shown to be involved in the attachment of ribosomes to the cytoplasmic membrane (Horiuchi et al., 1983). In the absence, but not in the presence, of ribosomes, this protein was sensitive to proteases and bound antibody prepared against it. On the other hand, genetic studies of mutants affecting secretion have led to the discovery of other cellular factors involved in secretion in E . coli. These mutants can be divided into two categories: mutants defective in export of normal proteins and mutants allowing the export of defective secretory proteins. Examples of the first category are the secA and secB mutations. In the secA mutants of E . coli (Oliver and Beckwith, 1981), growth at 42°C was inhibited, and precursors of MalE (maltose-binding)protein, OmpF protein, and LamB were accumulated. Mutants of the secB gene were defective in the localization of the MalE and OmpF proteins. The effects of this mutation were enhanced when a secA mutation was co-introduced (Kumamoto and Beckwith, 1983). The product of

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

97

the secA gene has been identified as a protein of molecular weight 92,000 and is apparently essential for normal growth. The secA gene product is a protein peripherally associated with the cytoplasmic membrane of E. cofi (Oliver and Beckwith, 1982). These properties suggest homology to the eukaryotic SRP. However, the fact that some periplasmic proteins are normally secreted in secA mutants (Oliver and Beckwith, 1981) raises the possibility that at least two secretory pathways exist in E. coli. The second category is the prl (protein localization) mutations. These were identified in studies on LamB proteins which were no longer transported to the outer membrane because of mutations in the hydrophobic core of their signal peptides ( E m et a f . , 1980, 1981). Export of these LamB mutants was restored by second site mutations in the three prl loci: A , B, and C (Emr ef al., 1981). These mutations have been recently reviewed (Silhavy er a f . , 1983) and will only be summarized here. The most extensively studied is theprfA mutant, which has been shown to restore translocation and processing of many export-defective signal peptide mutations (Emr and Bassford, 1982). The gene encodingprlA has been shown to map in the spc operon ( E m et al., 1981) but does not code for any known ribosomal proteins (Schultz et al., 1982). It is thus very likely that the prlA gene product is involved in the coupling between protein translation and export (Ito et a f . , 1983; Silhavy et a f . , 1983; Schultz et a f . , 1982). The remaining mutants, prlB and prlC, have not been well characterized. The prlB mutant functions only to restore translocation of LamB signal peptide mutants. In contrast toprlA mutants, it does not restore processing ( E m et a f . , 1981). Mutants in the prlC gene have been shown to have much the same phenotype as the prfA suppressors although they are mapped to different positions on the E . cofi chromosome (Emr et a f . , 1981). Further suggestions that additional protein components are involved in secretion comes from a study of the recently discovered fep operon of E . cofi, which contains the gene coding for SPase I (lepl) (Date and Wickner, 1981; Wolfe et a f . , 1983). Date and Wickner (1 98 1) discovered that the promoter region of lepl was located at least 1700 bp upstream of fepl, a finding indicating that a second gene (lepA) was under control of the same promoter. The DNA sequence of lepA has confirmed this and shows that its product is a very basic protein consisting of 598 amino acid residues (P. March and M. Inouye, unpublished results). Since the fepA gene product is synthesized together with SPase 1, it is also most likely a component of the secretion machinery. VI.

CONCLUSION

As one can see from this article, the precise molecular mechanism of protein secretion has not yet been determined. However, it is clear that the signal peptide contains within its short sequence the information necessary for translocation of a

98

GUY D. DUFFAUD ET AL.

protein across a membrane. To understand the function of the signal peptide, we must visualize it as going through a series of dynamic events. Most likely, as the signal peptide emerges from the ribosome, its conformation continually changes in response to the components with which it interacts. With the advances in recombinant DNA technology and biochemistry, it should not be long before we are able to have a more complete understanding of how the signal peptide functions during protein translocation across the membrane and the nature of its interactions with the components of the secretory machinery.

REFERENCES Bedouelle, H., Bassford, P. J., Fowler, A. V., Zabin, I., Beckwith, J., and Hofnung, M. (1980). Mutations which alter the function of the signal sequence of the maltose binding protein of Eschericia coli. Nature (London) 285, 78-8 I . Blobel, G., and Dobberstein, B. (1975a). Transfer of proteins across membranes. I. Presence of proteolytically processed and unprocessed nascent immunoglobulin light chains on membranebound ribosomes of murine myeloma. J . Cell Biol. 67, 835-851. Blobel, G., and Dobberstein, B. (1975b). Transfer of proteins across membranes. 11. Reconstitution of functional rough microsomes from heterologous components. J. Cell Biol. 67, 852-862. Blobel, G., and Sabatini, D. D. (1971). Ribosome membrane interaction in eukaryotic cells. In “Biomembranes” (L. A. Manson, ed.), pp. 193-195. Plenum, New York. Braun, G . , and Cole, S. T. (1982). The nucleotide sequence coding for major outer membrane protein OmpA of Shigellu dysenteriae. Nucleic Acid Res. 10, 2367-2378. Chou, P. Y.,and Fasman, G. D. (1978). Empirical predictions of protein conformation.Annu. Rev. Biochem. 47, 251-276. Dallas, W. S . , and Falkow, S. (1980). Amnio acid sequence homology between cholera toxin and E. coli heat-labile toxin. Nature (London) 288,499-500. Date, T. (1983). Demonstration by a novel genetic technique that leader peptidase is an essential enzyme of Escherichia coli. J . Bacteriol. 154, 76-83. Date, T., and Wickner, W. (1981). Isolation of the Escherichiu coli leader peptidase gene and effect of leader peptidase overproduction in viva Proc. Nad. Acad. Sci. U.S.A. 78, 6106-6110. Date, T., Zwizinski, C., Ludmerer, S., and Wickner, W. (1980a). Mechanisms of membrane assembly: Effects of energy poisons on the conversion of soluble MI3 colipage procoat to membrane-bound coat protein. Proc. Nurl. Acud. Sci. U.S.A. 77, 827-831. Date, T., Goodman, 1. M., and Wickner, W. (1980b). Procoat, the precursor of M13 coat protein, requires an electrochemical potential for membrane insertion. Proc. Natl. Acud. Sci. U.S.A. 77, 4669-4673. Emr, S. D., and Bassford, P. J. (1982). Localization and processing of outer membrane and periplasmic proteins in Escherichia coli strains harboring export-specific suppressor mutations. J . B i d . Chem. 257, 5852-5860. Emr, S. D., and Silhavy, T. J. (1983). Importance of secondary structure in the signal sequence for protein secretion. Proc. Natl. Acad. Sci. U.S.A. 80, 4599-4603. Em, S. D., Schwartz, M., and Silhavy, T. J. (1978). Mutations altering the cellular localization of the phage A receptor, an Escherichia coli outer membrane protein. Proc. Natl. Acad. Sci. U.S.A. 75, 5802-5806. Emr, S. D., Hedgpeth, J., Clement, J., Silhavy, T. J., and Hofnung, M. (1980). Sequence analysis of mutations that prevent export of A receptor, an Escherichiu coli outer membrane protein. Nature (London) 285, 82-85.

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

99

Emr, S. D., Havley-Way, S.. and Silhavy, T. J. (1981 ). Suppressor mutations that restore export of a protein with defective signal sequence. Cell 23, 79-88. Engelman, D. M., and Steitz, T. A. (1981). The spontaneous insertion of proteins into and across membranes: The helical hairpin hypothesis. Cell 23, 41 1-422. Gilmore, R., and Blobel, G. (1983). Transient involvement of signal recognition particle and its receptor in the microsomal membrane prior to protein translocation. Cell 35, 677-685. Gilmore, R.. Blobel, G., and Walter, P. (1982). Protein translocation across the endoplasmic reticulum. I. Detection in the microsomal membrane of a receptor for the signal recognition particle. J. Cell Eiol. 95, 463-469. Greenfield, L., Bjorn, M. J., Horn, G., Fong, D., Buck, G. A., Collier, R. J., and Kaplan, D. A. (1983). Nucleotide sequence of the structural gene for diphteria toxin carried by corynebacteriophage. Proc. Natl. Acad. Sci. U.S.A. 80, 6853-6857. Groarke, J. M.. Mahoney, W. C., Hope, J. N., Furlong, C. E., Robb, F. T., Zalkin, H., and Hermodson, M. A. (1983). The amino acid sequence of D-ribose-binding protein from E. coli K12. J . Eiol. Chem. 258, 12952-12956. Guidotti, G. (1977). The structure of intrinsic membrane proteins. J. Supramol. Srrucr. 7,489-497. Halegoua, S . , Sekizawa, J., and Inouye, M. (1976). A new form of structural lipoprotein of outer membrane of Escherichia coli. J. B i d . Chem. 252, 2324-2330. Hall, M. N., and Schwartz, M. (1982). Reconsidering the early steps of protein secretion. Ann. Microbial. (Inst. Pasreur) 133A, 123- 127. Hall, M. N., Gabay, J., and Schwartz, M . (1983). Evidence for a coupling of synthesis of an outer membrane protein in Escherichia coli. EMEO J . 2, 15-19. Hedgpeth, J., Clement, J.-M., Marchal, B . , Perrin, D., and Hofnung, M. (1980). DNA sequence encoding of NH2-teminal peptide involved in transport of A receptor, an E. coli secretory protein. Proc. Natl. Acad. Sci. U.S.A. 77, 2621-2625. Higgins, C. F., and Ferro-Luzzi Ames, G . (1981). Two periplasmic transport proteins which interact with a common membrane receptor show extensive homology: Complete nucleotide sequence. Proc. Narl. Acad. Sci. U.S.A. 78, 6038-6042. Horiuchi, S.. Tai, P. C., and Davis, B. D. (1983). A 64-kilodalton membrane protein of Bacillus subtilis covered by secreting ribosomes. Proc. Narl. Acad. Sci. U.S.A. 80, 3287-3291. Huang, Y.-X., Ching, G . , and Inouye, M. (1983). Comparison of the lipoprotein gene among Enterobacteriaceae: DNA sequence of Morganella morganii lipoprotein gene and its expression in Escherichia coli. J. Eiol. Chem. 258, 8139-8145. Hussain, M., Ichihara, S . , and Mizushima, S . (1980). Accumulation of glyceride-containing precursor of the outer membrane lipoprotein in the cytoplasmic membrane of Escherichia coli treated with globomycin. J. B i d . Chem. 255, 3707-3712. Hussain. M.. Ichihara, S., and Mizushima, S. (1982a). Mechanism of signal peptide cleavage in the biosynthesis of the major lipoprotein of the Escherichia coli outer membrane. J . Eiol. Chem. 257, 5177-5182. Hussain, M., Ozawa, Y., Ichihara, S., and Mizushima, S. (1982b). Signal peptide digestion in Escherichia coli: Effect of protease inhibitors on hydrolysis of the cleaved signal peptide of the major outer-membrane lipoprotein. Eur. J. Eiochem. 129, 233-239. Ichihara, S., Hussain, M., and Mizushima, S. (1981). Characterization of new membrane lipoproteins and their precursors of Escherichia colr. J. Eiol. Chem. 256, 3125-3129. Ichihara, S . , Hussain, M., and Mizushima, S. (1982). Mechanism of export of outer membrane lipoproteins through the cytoplasmic membrane in Escherichia coli.J. Eiol. Chem. 257, 495500. Inouye, H., Barnes, W., and Beckwith, I. (1982). Signal sequence of alkaline phosphatase of E . coli. J. Eacteriol. 149, 434-439. Inouye, M. (1979). What is the outer membrane? I n “Bacterial Outer Membranes: Biogenesis and Function” (M. Inouye, ed.), pp. 1-12. Wiley, New York.

100

GUY D. DUFFAUD ET AL.

Inouye, M., and Halegoua, S. (1980). Secretion and membrane localization of proteins in Escherichia coli. CRC Crit. Rev. Biochem. 7, 339-371. Inouye, M., Pirtle, R., Pirtle, I., Sekizawa, J., Nakamura, K., Di Rienzo, J., Inouye, S., Wang, S., and Halegoua, S . (1979). Mechanism of translocation of the outer membrane proteins of Escherichia coli through the cytoplasmic membrane. Microbiology 34-37. Inouye, S., Wang, S., Sekizawa, J., Halegoua, S., and Inouye, M. (1977). Aminoacid sequence for the peptide extension on the prolipoprotein of the Escherichia coli outer membrane. Proc. Narl. Acad. Sci. U.S.A. 74, 1004-1008. Inouye, S., Soberon, X., Franceschini, T., Nakamura, K., Itakura, K., and Inouye, M. (1982). Role of positive charge on the amino terminal region of the signal peptide in protein secretion across the membrane. Proc. Natl. Acad. Sci. U.S.A. 79, 3438-3441. Inouye, S.,Hsu, C., Itakura, K., and Inouye, M. (1983a). Requirements for signal peptide cleavage of Escherichia coli prolipoprotein. Science 221, 59-61. Inouye, S., Franceschini, T., Sato, M., Itakura, K., and Inouye, M. (1983b). Prolipoprotein signal peptidase of E. coli requires a cysteine residue at the cleavage site. EMBO J . 2, 87-91. Inouye, S., Vlasuk, G. P., Hsuing, H., and Inouye, M. (1984). Structural flexibility in the hydrophobic region of the Escherichia coli prolipoprotein signal peptide. J. Biol. Chem. 259, 37293733. Inukai, M., and Inouye, M. (1983). Association of the prolipoprotein accumulated in the presence of globomycin with the Escherichia coli outer membrane. Eur. J . Biochem. 130, 27-32. Inukai, M., Nakajima, M., Osawa, M., Haneishi, T., and Arai, M. (1978). Globomycin, a new peptide antibiotic with spheroplast-forming activity. 11. Isolation and physio-chemical and biological characterization. J. Antibiof. 31, 421-425. Inukai, M., Takeuchi, M., Shimizu, K., and Arai, M. (1979). Existence of the bound form of prolipoprotein in Escherichia coli B cells treated with globomycin. J. Bacterial. 140, 10981101. Ito, K., Mandel, G., and Wickner, W. (1979). Soluble precursor of an integral membrane protein: synthesis of procoat protein in Escherichia coli infected with bacteriophage M13. Proc. Natl. Acad. Sci. U.S.A. 76, 1199- 1203. Ito, K., Date, T., and Wickner, W. (1980). Synthesis, assembly into the cytoplasmic membrane, and proteolytic processing of the precursor of coliphage M13 coat protein. J . B i d . Chem. 255, 2 133-2130. Ito, K., Wittekind, M., Nomura, M., Shiba, K., Yura, T . , Miura, A., and Nashimoto, H. (1983). A temperature-sensitive mutant of E . coli exhibiting slow processing of exported proteins. Cell 32, 789-797. Jaurin, B., and Grundstrom, T. (1981). ampC cehalosporinase of E. coli K-12 has a different evolutionary origin from that of p-lactamase of the penicillinase type. Proc. Narl. Acad. Sci. U.S.A. 78, 4897-4901. Josefsson, L.-G., and Randall, L. L. (1981). Different exported proteins in E. coli show differences in the temporal mode of processing in vivo. Cel/ 25, 151-157. Koshland, D., and Botstein, D. (1982). Evidence for post-translational translocation of p-lactamase across the bacterial inner membrane. Cell 30, 893-902. Kreibich, G., Ulrich, B. L., and Sabatini, D. D. (1978a). Proteins of rough microsomal membranes related to ribosome binding. I. Identification of ribophorins I and 11, membrane proteins characteristic of rough microsomes. J . Cell Biol. 77, 464-487. Kreibich, G., Freientstein, C. M., Pereyra, B. N., Ulrich, B. L., and Sabatini, D. D. (1978b). Proteins of rough microsomal membranes related ribosome binding 11. Cross-linking of bound ribosomes to specific membrane proteins exposed at binding sites. J . Cell Biol. 77, 488-506. Kumamoto, C. A., and Beckwith, J. (1983). Mutations in a new gene, secB, cause defective protein localization in Escherichia coli. J . Bacreriol. 154, 253-260. Landick, R., and Oxender, D. L. (1982). Membr. Transp. 2.

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

101

Lee, N.. Yamagata, H., and Inouye, M. (1983). Inhibition of secretion of a mutant lipoprotein across the cytoplasmic membrane by the wild-type lipoprotein of the Escherichia coli outer membrane. J . Bacteriol. 155, 407-4 I I . Lin, J. J. C., Kanawaza, H., Ozols, J., and Wu, H. C. (1978). An Escherichia coli mutant with an amino acid alteration within the signal sequence of outer membrane prolipoprotein. Proc. Natl. Acad. Sci. U.S.A. 75, 4890-4895. Lofdahl, S., Guss, B., Uhlen, M., Philipson, L., and Lindberg, M. (1983). Gene for Staphylococco protein A. Proc. Natl. Acad. Sci. U.S.A. 80, 697-701. Luckey, M., and Nikaido, H. (1980). Specificity of diffusion channels produced by phage A receptor protein of Escherichia coli. Proc. Nail. Acad. Sci. U.S.A. 77, 167-171. McLaughlin, J. R., Murray, C. L., and Rabinowitz, J . C. (1981). Unique features in the ribosome binding site sequence of the gram-positive Stuphylococcus aureus p-lactamase gene. J . Biol. Chem. 256, 11283-1 1291. Marchal. C., Perrin, D., Hedgpeth, J., and Hofnung, M. (1980). Synthesis and maturation of A receptor in Escherichia coli: in vivo and in vitro expression of gene LamB under lac promoter control. Proc. Natl. Acad. Sci. U.S.A. 77, 1491-1495. Mekalanos, J. J., Swartz, D. J., Pearson, G. D. N . , Harford, N., Groyne, F., and deWilde, M. ( 1983). Cholera toxin genes: Nucleotide sequence, deletion analysis and vaccine development, Nature (London) 306, 551-557. Meyer. D. I., Krause, E., and Dobberstein, B. (1982). Secretory protein translocation across membranes-the role of the ‘docking protein’. Nature (London) 297, 647-650. Michaelis, S., and Beckwith, J. (1982). Mechanism of incorporation of cell evelope proteins in Escherichia coli. Annu. Rev. Microbiol. 36, 435-465. Michaelis, S., Inouye, H., Oliver. D., and Beckwith, J . (1983). Mutations that alter the signal sequence of alkaline phospbatase in Escherichia coli. J. Bucteriol. 154, 366-374. Milstein, C., Brownlee, G . G., Harrison, T. M., and Mathews, M. B. (1972). A possible precursor of immunoglobulin light chains. Nature (London) 239, 1 17-120. Mizuno, T., Chou, M., and Inouye, M. (1983). DNA sequence of the promoter region of the ompC gene and the amino acid sequence of the signal peptide of pro-ompC protein of E . coli. FEBS Lett. 151, 159-164. Movva, N. R., Nakamura, K., and Inouye, M. (1980). Amino acid sequence of the signal peptide of OmpA protein: A major outer membrane protein of E . coli. J. Biol. Chem. 255, 27-29. Muller, M., Ibrahimi, I., Chang, C. N., Walter, P., and Blobel, G. (1982). A bacterial secretory protein requires signal recognition particle for translocation across mammalian endoplasmic reticulum. J. Biol. Chem. 257, 11860-11863. Mutoh, N., Inokuchi, K., and Mizushima, S . (1982). Amino acid sequence of the signal peptide of OmpF, a major outer membrane protein of E. coli. FEBS Lett. 137, 171-174. Nakamura, K., and Inouye. M. (1979). DNA sequence of the gene for the outer membrane lipoprotein of E . colic An extremely AT-rich promoter. Cell 18, 1109-1 117. Nakamura, K . , and Inouye, M. (1980). DNA sequence of the Serratia marcescens lipoprotein gene. Proc. Nut/. Acad. Sci. U.S.A. 77, 1369-1373. Neugebauer, K., Sprenojel, R.. and Schaller, H. (1981). Penicillinase from Bucillus lichenijorrnis: Nucleotide sequence of the gene. Nucleic Acids Res. 9, 2577-2588. Ohmura, K . , Yamazaki, H., Takeichi, Y., Nakayama, A., Otozai, K., Yamane, M., Yamasaki, M., and Tamura, 0 . (1983). Nucleotide sequence of the promoter and NHz-terminal signal peptide of Bacillus suhtilis a-amylase gene cloned in PUB 110. Biochem. Biophys. Res. Commun. 112, 678-683. Oliver, D. B., and Beckwith, J. (1981). E . coli mutant pleitropically defective in the export of secreted proteins. Cell 25, 765-772. Oliver, D. B., and Beckwith, J . (1982). Regulation of a membrane component required for protein secretion in Escherichia coli. Cell 30, 31 1-319.

102

GUY D. DUFFAUD ET AL.

Overbeeke, N., Bergmans, H., von Mansfeld, F., and Lugtenberg, B. (1983). Complete nucleotide sequence of phoF, the structural gene for the phosphate limitation inducible outer membrane pore protein of E. coli K-12. J . Mol. Biol. 163, 513-532. Oxender, D. L., Anderson, J. J., Daniels, C. J., Landick, R., Gunsalus, R. P., Zurowski, G., and Yamojski, C. (1980). Amino-terminal sequence and processing of the precursor of the lencine specific binding protein, and evidence for conformational differences between the precursor and the mature form. Proc. Natl. Acad. Sci. U.S.A. 77, 2005-2009. Palva, I., Pettersson, R. F., Kalkkinen, N., Lehtovaara, P., Sarvas, M., Soderlund, H., Takkinen, K., and Kaariainen, L. (1981). Nucleotide sequence of the promoter and NH2-terminal signal peptide region of the a-amylase gene from Bacillus amyloliquegacious. (1981). Gene 15, 4351. Perlman, D., and Halvorson, H. (1983). A putative signal peptidase recognition site and sequence in eukaryotic and prokaryiotic signal peptides. J. Mol. Biol. 167, 391-409. Randall, L. L. (1983). Translocation of domains of nascent periplasmic proteins across the cytoplasmic membrane is independent of elongation. Cell 33, 231-240. Randall, L. L., and Hardy, S. J. S. (1977). Synthesis of exported proteins by membrane bound polysomes from Escherichia coli. Eur. J . Biochem. 75, 43-53. Randall-Hazelbauer, L., and Schwartz, M. (1973). Isolation of the phage lambda receptor from Escherichia coli K-12. J . Bacteriol. 116, 1436- 1444. Randall, L. L., Hardy, S. J . S., and Josefsson, L . G . (1978). Precursors of three exported proteins in Escherichia coli. Proc. Natl. Acad. Sci. U.S.A. 75, 1209-1212. Rodriguez-Boulan, E. R., Sabatini, D. D., Pereyra, B. N., and Kreibich, G. (1978). Spatial orientation of glycoproteins in membranes of rat liver rough microsomes. 11. Transmembrane disposition and characterization of glycoproteins. J. Cell Biol. 78, 894-909. Rosenblatt, M., Beaudette, N. V., and Fasman, G. D. (1980). Conformational studies of the synthetic precursor-specific region of preproparathyroid hormone. Proc. Natl. Acad. Sci. U.S.A. 77, 3983-3987. Schaller, H., Beck, E., and Takanami, M. (1979). Sequence and regulatory signals of the fiamentous phage genome. In “The Single Stranded DNA Phages” (D. Denhart, D. Dressler, and D. Ray, eds.), pp. 139-153. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Schechter, I. (1973). Biologically and chemically pure mRNA coding for a mouse immunoglobulin L-chain prepared with the aid of antibodies and immobilized oligothymidinge. Proc. Natl. Acad. Sci. U.S.A. 70, 2256-2260. Schechter, I . , McKean, D. J., Guyer, R., and Terry, W. (1975). Partial amino acid sequence of the precursor of immunoglobulin light chain programmed by messenger RNA in vitro. Science 188, 160- 162. Schultz, J., Silhavy, T. J., Berman, M. L., Fiil, N., and Emr, S. D. (1982). A previously unidentified gene in the spc operon of Escherichia coli K12 specifies a component of the protein export machinery. Cell 31, 227-235. Schwartz, M., Roa, M., and Debarboville, M. (1981). Mutations that effect LamB gene expression at a posttranscriptional level. Proc. Natl. Acad. Sci. U.S.A. 78, 2937-2941. Scripture, J. B., and Hogg, R. W. (1983). The nucleotide sequences defining the signal peptides of the galactose-binding protein and the arabinose-binding protein. J. Biol. Chem. 258, 10853. Silhavy, T. J . , Shuman, H. A., Beckwith, J . , and Schwartz, M. (1977). Use of gene fusions to study outer membrane protein localization inf Escherichia coli. Proc. Natl. Acad. Sci. U.S.A. 74, 541 1-5415. Silhavy, T. J., Benson, S. A., and Emr, S. D. (1983). Mechanisms of protein localization. Microbiol. Rev. 47, 313-344. Silver, P., and Wickner, W. (1983). Genetic mapping of the Escherichia coli leader (signal) peptidase gene (lep): A new approach for determining the map position of a cloned gene. J . Bacteriol. 154, 569-572.

2. STRUCTURE AND FUNCTION OF SIGNAL PEPTIDE

103

Silver, P., Watts, C., and Wickner, W. (1981). Membrane assembly from purified components. I . Isolated MI3 procoat does not require ribosomes or soluble proteins for processing by membranes. Cell 25, 341-345. Smith, W. P., Tai, P. C., Thompson, R. C., and Davis, B. D. (1977). Extracellular labeling of nascent polypeptides traversing the membrane of Escherichia coli. Proc. Nail. Acad. Sci. U.S.A. 74, 2830-2834. Smith, W. P., Tai, P. C., and Davis, B. D. (1979). Extracellular labeling of growing secreted polypeptide chains in Bacillus subtilis with diazoiodosulfanilic acid. Biochemistry 18, 198-202. Spicer, E. K . , and Nobel. J. A. (1982). Escheric,hiu coli heat-labile enterotoscin. J. B i d . Chem. 257, 5116-5121. Sugimoto, K., Sugisaki, H., Okamoto, T . , and Takanami, M. (1977). Studies on bacteriophage fd DNA. J. Mol. Biol. 111, 487-507. Sutcliffe, I . G. (1978). Nucleotide sequence of the ampicillin resistance gene of E. coli plasmid PBR322. Proc. Natl. Acad. Sci. U.S.A. 75, 3737-3741. Talmadge, K . , Kaufman, J., and Gilbert, W. (1980). Bacteria mature preproinsulin to proinsulin. Proc. Nail. Acad. Sci. U.S.A. 77, 3988-3992. Tokunaga, M., Tokunaga, H., and Wu, H. C. (1982). Post-translational modification and processing of Escherichia coli prolipoprotein in vitro. Proc. Natl. Acad. Sci. U.S.A. 79, 2255-2259. Vlasuk, G. P., Inouye, S., Ito, H., Itakura, K . , and Inouye, M. (1983). Effects of the complete removal of basic amino acid residues from the signal peptide on secretion of lipoprotein in Escherichia coli. J. Biol. Chem. 258, 7141-7148. Vlasuk, G. P., Inouye, S., and Inouye, M. (1984). Effects of replacing serine and threonine residues within the signal peptide on the secretion of the major outer membrane lipoprotein of Escherichia coli. J. Biol. Chem. 259, 6195-6200. von Heijne, G. (1983). Patterns of amino acids near signal-sequence cleavage sites. Eur. J. Biochem. 133, 17-21. von Heijne, G . , and Blomberg, C. (1979). Transmembrane translocation of protein. Eur. J. Biochem. 97, 175-181. Walter, P., and Blobel, (3. (1981a). Translocation of protein across the endoplasmic reticulum. 11. Signal recognition protein (SRP) mediates the selective binding to microsomal membranes of in vitro-assembled polysomes synthesizing secretory protein J. Cell Biol. 91, 55 1-556. Walter, P., and Blobel, G. (1981b). Translocation of protein across the endeplasmic reticulum. 111. Signal recognition protein (SRP) causes signal sequence dependent and site-specific arrest of chain elongation that is released by microsomal membranes. J. Cell Biol. 91, 557-561. Walter, P., and Blobel, G. (1982). Signal recognition particle contains a 7s RNA essential for protein translocation across the endoplasmic reticulum. Nature (London) 299, 691-698. Walter, P., Ibrahimi, I., and Blobel, G. (1981). Translocation of proteins across the endoplasmic reticulum. I. Signal recognition protein (SRP) binds to in vitro-assembled polysomes synthesizing secretory protein. J. Cell Bid. 91, 545-550. Watts, C., Silver, P., and Wickner, W. (1981). Membrane assembly from purified components. 11. Assembly of MI3 procoat into liposomes reconstituted with purified leader peptidase. Cell 25, 347-393. Wickner, W. (1979). The assembly of proteins into biological membranes: the membrane trigger hypothesis. Annu. Rev. Biochem. 48, 23-45. Wickner, W. (1980). Assembly of proteins into membranes. Science 210, 861-871. Wickner, W. (1983). M13 coat protein as a mode of membrane assembly. T.I.B.S. 8, 90-94. Wolfe, P. B., Wickner, W . , and Goodman, J. M. (1983). Sequence of the leader peptidase gene of Escherichia coli and the orientation of leader peptidase in the bacterial envelope. J. Biol. Chem. 258, 12073-12080. Wu, H. C., Hou, C., Lin, J. J. C., and Yem, D. W. (1977). Biochemical characterization of a mutant lipoprotein of Escherichia coli. Proc. Natl. Acad. Sci. U.S.A. 74, 1388-1392.

104

GUY D. DUFFAUD ET AL.

Yamagata, H., Nakamura, K., and Inouye, M. (1981). Comparison of the lipoprotein gene among the enterobacterioceae. J . Biol. Chem. 256, 2194-2198. Yamagata, H., Ippolito, C., Inukai, M., and Inouye, M. (1982). Temperature sensitive processing of outer membrane lipoprotein in an Escherichia coli mutant. J . Bacteriol. 152, 1163-1 168. Yamagata, H., Daishima, K . , and Mizushima, S. (1983). Cloning and expression of a gene coding for the prolipoprotein signal peptidase of Escherichia coli. FEBS Lett. 158, 301-304. Zimmermann, R . , and Wickner, W. (1983). Energetics and intermediates of the assembly of protein OmpA into the outer membrane of E. coli. J. Biol. Chem. 258, 3920-3925. Zimmerman, R., Watts, C., and Wickner, W. (1982). The biosynthesis of membrane-bound MI3 coat protein: Engergetics and assembly intermediates. J . Biol. Chem. 257, 6529-6536. Zwizinski, C., and Wickner, W. (1980). Purification and characterization of leader (signal) peptidase from Escherichia coli. J. Biol. Chem. 255, 1913-1971. Zwizinski, C., Date, T., and Wickner, W. (1981). Leader peptidase is found in both the inner and outer membranes of Escherichia coli. J. Biol. Chem. 256, 3593-3591.

CURRENT TOPICS IN MEMBRANES AND TRANSPORT. VOI.UME 24

Chapter 3 The Use of Genetic Techniques to Analyze Protein Export in Escherichia coli W A S A . BANKAITIS,' J . PATRICK RYAN, BETH A . RASMUSSEN,2 AND PHILIP J . BASSFORD, JR. Department of Microbiology and Immunology School of Medicine University of North Carolina Chapel Hill, North Carolina

I. Introduction. . . ... ................................... 11. The Use of Gene Fusions to Study Protein Export in E. coli . . . . . . A. Gene Fusion Technology: The Logic behind This Approach.. . B. malE-lacZ Protein Fusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. lamB-lacZ Protein Fusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Intragenic Information Specifying Protein Export. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Interpretation of Signal Sequence Mutations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. lntragenic Suppressor Mutations for malEAl2-18. ........................ C. Additional lntragenic Information Specifying Initiation of MBP Export.. . . . . . . D. Export-Specific Information Contained within the Mature LamB Protein.. . . . . . iV. Components of the E. coli Protein Export Machinery . . . . . . . . . . . . . . . . . . . . . . . . . .

Defective Signal Peptides C. The Coupling of Protein Synthesis and Export.. . . . . . . . . . . . . . . . . . . . . . . . . . . ......................................... References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

106

120 124 124

127 130 133 135

142 145 146

IPresent address: Division of Biology, California Institute of Technology, Pasadena, California 91125.

*Present address: Department of Molecular Biology, Princeton University, Princeton, New Jersey 08544. 105

Cupyright Q 19115 by Academic Press. Inc. All rights of reproduction in any k r m reserved. ISBN 0-12-153324-7

106

W A S A. BANKAlTlS ET AL.

1.

INTRODUCTION

The gram-negative bacterium Escherichiu coli lacks any kind of internal, membrane-bounded organelles. Thus, the task of localizing proteins to their correct, final destination is somewhat simpler than that faced by a eukaryotic cell. Still, various proteins are exported to one of three extracytoplasmic compartments: (1) The cytoplasmic or inner membrane includes a number of proteins involved in nutrient transport, energy metabolism, chemotaxis, lipid biosynthesis, and other processes. (2) The outer membrane contains relatively fewer proteins, but certain of these constitute major cellular proteins. Many outer membrane proteins function in the transport of various hydrophilic substances across this barrier by either active or passive mechanisms and were initially identified by their properties as receptors for different bacteriophages andlor colicins. (3) The periplasm, an aqueous space bounded by the cytoplasmic and outer membranes, contains two general classes of proteins. First, a number of different binding proteins are involved in transport of various nutrients and may also serve as chemoreceptors. Second, a number of degradative enzymes, such as alkaline phosphatase and ribonuclease, are found to be periplasmically localized. In this respect, the periplasm has been thought of as an evolutionary precursor to lysosomes of higher cells. The localization of proteins to the periplasm and outer membrane of E. coli is a process that seems analogous to protein localization in eukaryotic cells. These envelope proteins, the synthesis of which occurs on ribosomes located within the confines of the cytoplasmic membrane, must be exported to specific sites beyond that membrane barrier. Studies during the last few years have indicated that the initial steps in the translocation of proteins to the periplasm and outer membrane of E. coli are very similar to those described for exported proteins synthesized on the rough endoplasmic reticulum of eukaryotic cells. For example, exported proteins in E . coli are synthesized on polysomes bound to the cytoplasmic membrane (Randall and Hardy, 1977; Smith, 1980). Also, many periplasmic and outer membrane proteins have been shown to be synthesized with amino-terminal signal peptides that are very similar in character to signal peptides of eukaryotic exported proteins (see articles by Duffaud et al. and von Heijne, this volume). The evolutionary conservation of the protein export process is further illustrated by the fact that E. coli exports to the periplasm various eukaryotic secretory proteins whose structural genes have been engineered to be expressed in this organism. These include chicken ovalbumin (Fraser and Bruce, 1978) and rat preproinsulin (Talmadge et ul., 1980). In fact, the latter protein is correctly processed by E . coli to proinsulin. Whereas the biochemical analysis of protein export in E. coli has generally lagged behind similar efforts in various eukaryotic systems, the use of sophisticated genetic techniques in E. coli provides a major contribution to this field. It is not our intention to present here a comprehensive review of the literature con-

3. PROTEIN EXPORT IN E. coli

107

cerning the various aspects of protein localization in E. coli. Several excellent reviews on this subject have recently been published (Michaelis and Beckwith, 1982; Silhavy et al., 1983; Randall and Hardy, 1984). Rather, in this article we will describe certain genetic approaches that have been successfully employed in studies on the mechanisms of protein localization in this organism.

II. THE USE OF GENE FUSIONS TO STUDY PROTEIN EXPORT IN E. coli

A. Gene Fusion Technology: The Logic behind This Approach A classic genetic approach to the understanding of protein localization in E . coli calls for the isolation of mutants unable to export one or more proteins to their proper extracytoplasmic compartment. Analysis of such mutants could reveal ( 1) the properties inherent to extracytoplasmic proteins that facilitate correct localization; (2) the domains of these proteins within which such specific export signals are located; and (3) the nature of the cellular protein export machinery, if indeed any such specialized apparatus exists. The difficulty with a direct application of such a general strategy becomes apparent when one attempts to devise a selection scheme that can distinguish between mutants unable to properly localize a particular protein and mutants unable to synthesize a biologically active form of that protein. Both classes of mutants would be expected to exhibit the same null phenotype. In practice, mutations affecting just export must be of a very specific nature and, hence, are considerably rarer than mutations affecting synthesis or activity. This dilemma has been somewhat fortuitously solved by the development of general genetic techniques for the isolation of specific fusions of the lacZ gene encoding the cytoplasmic protein P-galactosidase to virtually any E . coli gene. Of particular value to those interested in protein export is the capability of obtaining gene fusions programming the synthesis of hybrid proteins having at their amino termini an amino-terminal fragment derived from an exported protein and at their carboxy termini enzymatically active P-galactosidase. The actual construction of such “protein fusion” strains, by either in vivo or in vitro methods, is beyond the scope of this article and will not be described here (see Casadaban, 1976; Casadaban et al., 1980; Guarente et al., 1980; Weinstock et al., 1983; Silhavy and Beckwith, 1983). Such fusion strains have proved useful in two different ways. First, a series of protein fusions of P-galactosidase to a particular exported protein can be constructed such that amino termini of various lengths, derived from the exported protein, are attached to an essentially constant P-galactosidase moiety. The enzymatic and antigenic properties of the P-galactosidase portion of the hybrid protein provide a simple biochemical tag for the amino-terminal exported protein

108

VYTAS A. BANKAlTlS ET AL.

FIG. 1. SDS-PAGE of whole cell extracts of malE-lucZ protein fusion strains. Whole cell extracts were prepared and electrophoresis performed as described in Bassford er al. (1979). (a) Class I fusion, uninduced; (b) class I fusion, induced; (c) class I1 fusion, uninduced; (d) class I1 fusion, induced; (e) class IJI fusion, uninduced; (f)class 111 fusion, induced; (g) class IV fusion, uninduced; (h) class IV fusion, induced; (i) class V fusion, uninduced; (j) class V fusion, induced. The hybrid proteins are easily identified as maltose-inducible proteins of high molecular weight (small, filled arrows). In addition, the class V fusion strain exhibits several additional maltose-inducible protein bands that are apparently breakdown products of the hybrid protein (small, open arrows). A number of new protein bands are discerned when maltose-sensitive fusion strains (classes I1 through V) are induced with maltose (large arrows, on right). These correspond to precursors of normal exported

3. PROTEIN EXPORT IN E. coli

109

fragment. Correlation of the subcellular location of a given hybrid protein to the size of the heterologous amino terminus provides information, with some cautious interpretation, about the regions within a polypeptide that function in determining the ultimate cellular destination of the exported protein. Second, the unusual properties of certain protein fusion strains have enabled direct selection of E. coli mutants producing export-defective proteins or exhibiting pleiotropic defects in general protein export. Gene fusion technology for studying protein localization has been extensively applied to several E. coli exported proteins. In this article we concentrate chiefly on one of these, the periplasmic maltose-binding protein. For comparison, studies with an outer membrane protein, the LamB protein, are presented in a more abbreviated form.

B. ma/€-/acZ Protein Fusions 1. INITIAL CHARACTERIZATION The periplasmic maltose-binding protein (MBP), molecular weight 38,500, is an essential component of the E. coli maltose transport system. It is encoded by the malE gene, one of five genes whose products are involved in maltose uptake. These genes constitute the mafB region of the E. coli chromosome and are organized into two positively controlled operons that diverge from a common promoter region. It is important to note that the expression of these operons is maltose inducible. (For an excellent recent review of the maltose transport system of E . cofi, see Hengge and Boos, 1983.) Five classes of mafE-fucZ protein fusion strains (designated classes I through V) have been isolated and characterized (Bassford et al., 1979). The different classes can be distinguished on the basis of the location of the fusion joint between the malE and fucZ genes and also on the basis of the molecular weights of the corresponding hybrid proteins as determined by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 1). The end points within facZ are very nearly the same for each fusion (Brickman er a l . , 1979). Thus, the molecular weight differences discerned for the various hybrid proteins are a direct reflection of the contribution of the MBP to the hybrid product. The hybrid proteins vary in size, from one that includes only a very small amino-terminal portion of the MBP (class I) to one that lacks only a very small carboxyl-terminal portion of the MBP (class V). The native MBP is initially synthesized with an amino-terminal signal peptide proteins that accumulate under these conditions. See text for additional details. This figure was originally published in Bassford et al. (1979) and is reproduced here with permission of the authors and the American Society for Microbiology.

110

W A S A. BANKAlTlS ET AL.

of 26 residues (Bedouelle et al., 1980) that can be processed cotranslationally or posttranslationally (Josefsson and Randall, 1981). Both genetic and biochemical evidence indicate that the four larger malE-lacZ hybrid proteins include intact MBP signal peptides (Bassford et al., 1979). The class I hybrid protein includes only the first 14 residues of this structure (Bedouelle et al., 1980). It was originally anticipated that those hybrid proteins having intact MBP signal peptides at their amino termini might very well be secreted into the periplasm. Such a result would have provided strong support for the hypothesized role of the signal peptide in initiating protein export. However, based on two different operational criteria for defining periplasmic proteins (Neu and Heppel, 1965), secretion of hybrid protein into the periplasm could not be demonstrated for any of the malE-lacZ fusion strains (Bassford et al., 1979). Further investigation revealed that the presence of an intact MBP signal peptide does influence the cellular localization of the hybrid proteins. The class I hybrid protein, lacking the complete signal peptide, clearly fractionates as a cytoplasmic protein, the same as native E. coli P-galactosidase. However, a significant proportion of the larger hybrid proteins fractionate with the cytoplasmic membrane. In general, the larger the hybrid protein, the greater the proportion of total hybrid protein bound to the membrane (Bassford et al., 1979). It was suggested that secretion of the larger malE-lacZ hybrid proteins commences in the same manner as for the native MBP, i.e., the amino-terminal signal peptide initiates the cotranslational transfer of the nascent protein through the cytoplasmic membrane. However, at some point in the transfer process, amino acid sequences within the carboxyl-terminal P-galactosidase moiety of the hybrid protein abort the export process, thereby leaving the hybrid protein embedded in the cytoplasmic membrane. A variety of evidence (see below) supports this model. The failure to obtain the secretion of any of the malE-lucZ hybrid proteins into the E. coli periplasm is an important finding. At one time, it had been anticipated that the secretion of virtually any soluble protein could be genetically engineered by simply attaching to it the appropriate signal(s). This work, and the subsequent findings of Moreno et al. (1980) with l a d - l a c Z hybrid proteins, indicates that this is most probably not the case. It appears that there are constraints regarding those amino acid sequences that can move through membranes via the usual protein secretory route. 2. THE “MALTOSE-SENSITIVE” PHENOTYPE The induction of high-level synthesis of certain hybrid protein classes has been observed to elicit an inhibitory effect on growth of the E. coli host cell. This unusual property has been termed the maltose-sensitive (Mals) phenotype and is characteristic of strains carrying class I1 through class V malE-lacZ protein

111

3. PROTEIN EXPORT IN E. coli

fusions (Bassford et al., 1979). Upon the addition of maltose to such cultures, synthesis of the hybrid protein is induced. Cell division is inhibited concomitantly with the appearance of high levels of P-galactosidase enzyme activity. Microscopic inspection of such moribund cultures reveals that the cells actually filament under these conditions, elongating to several times their normal length. Eventually, some cell lysis becomes evident. The direct correlation of the MalS phenotype with production of the hybrid proteins is illustrated by the experiment shown in Fig. 2. Such a relationship suggests it is the synthesis of the malE-lacZ

-

P

300-

?

Y

t

t

100-

v)

z

W 0 -1

50-

U

0

+ a 0

.-

E

+Moltore

-

1

f'

L-q;

107

i

i

I

? \

?

,A ~'*;o

\

\

'+I20 ' *I80 '+240 2

0

TIME ( m i n )

FIG. 2. The effect of maltose on the growth and viability of the maltose-sensitive malE-lucZ protein fusion strain PB72-47. Cells growing logarithmically in glycerol minimal medium were induced for expression of mu/ genes by addition of 0.2% maltose at 0 minutes. (A) The effect of maltose on optical density of culture (expressed in Klett units). (B) The effect of maltose on cell viability (circles) and the induction of P-galactosidase activity (triangles). Open symbols: control culture. Filled symbols: maltose-induced culture. Figure was originally published in Bassford ei al. (1979) and is reproduced here with permission of the authors and the American Society for Microbiology.

112

W A S A. BANKAlTlS ET AL.

hybrid protein itself that is directly responsible for the physiological effects that define the MalS phenotype. Several lines of evidence support this contention. First, elimination of hybrid protein synthesis restores normal growth properties to MalS fusion strains. When a series of known lacZ nonsense mutations are recombined into the malE-lacZ hybrid gene of the class IV fusion strain PB72-47 (the resultant recombinants lack P-galactosidase activity and become Lac - ), the response to maltose differs, depending on the relative position of the chain-terminating mutation. Only those nonsense mutations mapping early in lacZ, and therefore in close proximity to the malE-lacZ fusion joint, totally eliminate the MalS phenotype. When the chain termination mutation maps late in the lacZ gene, there is virtually no relief of maltose sensitivity. Nonsense mutations mapping in the middle of lacZ have an intermediate effect. Introduction of a nonsense suppressor allele that permits these strains to complete synthesis of the hybrid protein in all cases restores the original MalS phenotype (Bassford et a l . , 1979). Second, the severity of maltose sensitivity is a direct function of the amount of MBP attached to P-galactosidase and, thus, correlates with the tendency of the hybrid protein to associate with the cytoplasmic membrane. The class I fusion strain is totally maltose insensitive, whereas the class IV and V strains are exquisitely MalS. The class I1 and 111 strains exhibit a MalS phenotype between these two extremes (Bassford et al., 1979). These results also further indicate that maltose sensitivity is a consequence of the cell’s attempt to export the malElacZ hybrid proteins. As mentioned above, this process is ultimately unsuccessful because of the cell’s inability to translocate the P-galactosidase portion of the hybrid protein across the cytoplasmic membrane. The marked position effect observed with the different lacZ nonsense mutations in relieving maltose sensitivity (presumably by permitting passage of the truncated hybrid protein) strongly suggests that there is not just one single site within the 9-galactosidase polypeptide that aborts export; rather, it appears to result from the cumulative effect of multiple incompatible sites. The most convincing evidence that the MalS phenotype is a protein exportrelated phenomenon derives from observations concerning the export of normal envelope proteins following maltose induction of hybrid protein synthesis. If, several hours after induction of hybrid protein synthesis, whole cells of a MalS strain are prepared and analyzed by SDS-PAGE, a number of novel protein bands can be discerned (see Fig. 1). The appearance of these novel proteins is totally dependent on the fusion strain being MalS (Bassford et a l . , 1979) and is representative of accumulation of the higher molecular weight, unprocessed precursors of normally exported proteins. Under these conditions, precursors have been identified for a number of envelope proteins, including the periplasmic wild-type MBP and alkaline phosphatase, and the major outer membrane proteins LamB, OmpA, OmpC, OmpF, and the murein lipoprotein (Ito and Beck-

3. PROTEIN EXPORT IN E. coli

113

with, 1981). Precursors for two cytoplasmic membrane proteins have also been demonstrated (Herrero et al., 1982). There has yet to be identified an exported protein in E. coli that does not accumulate in its precursor form under these conditions. In fact, induction of MalS strains has been used to experimentally demonstrate processing of exported proteins (Ito and Beckwith, 1981; Dodd et al., 1984) as well as to obtain proteins in their precursor form for experimental purposes (Ito, 1982; Tokunaga et al.. 1982). Studies have shown that precursor accumulation is the esrliest manifestation of the MalS phenotype. The inhibition of normal processing by high-level hybrid protein synthesis has been interpreted to suggest that insertion of the rnalE-lacZ hybrid protein into the cytoplasmic membrane eventually occupies all the sites through which exported proteins exit the cytoplasm. Presumably, it is this general block in protein export that results in the inhibition of cell division and eventual cell death. Thus, these results indicate that export of all of the various envelope proteins may involve some common site and that there must be a finite number of such sites. Using a kinetic analysis of precursor accumulation under conditions of high-level hybrid protein synthesis, it has been estimated that an E. coli cell has roughly 2 x lo4 such sites in the cytoplasmic membrane (Ito et al., 1981).

3. A SELECTION FOR MUTANTS DEFECTIVE I N MBP EXPORT

As noted above, it appears that the E . coli cell’s attempt to secrete the rnalElacZ hybrid protein is directly responsible for the MalS phenotype exhibited by certain protein fusion strains. It was reasoned that maltose-resistant (MalR)Lac derivatives of a MalS strain would include mutants no longer attempting hybrid protein export. The requirement for Lac+ ensures synthesis of the hybrid protein. If this condition is not demanded in the MalR selection, an overwhelming majority of the mutants obtained are defective in synthesis of the hybrid protein. These are usually due to either nonsense mutations mapping within the hybrid gene or loss of the entire fusion by homologous recombination (Bassford et al., 1979). This observation serves to underscore, as discussed above, the problem with conventional selection schemes aimed at obtaining export-defective mutants and illustrates the utility of gene fusions for this purpose. Beginning with the class IV fusion strain PB72-47, a number of spontaneous, independent mutants were obtained that continue to grow normally in the presence of maltose and that synthesize a rnalE-lacZ hybrid protein retaining pgalactosidase activity (Bassford and Beckwith, 1979). The behavior of the hybrid protein is as predicted, i.e., the hybrid protein is primarily found in the cytoplasm. Induced levels of hybrid protein synthesis in these MalR mutants are comparable to those observed in the MalS parental strain, a finding indicating that maltose resistance is not simply the consequence of reduced expression of +

114

VYTAS A. BANKAlTlS ET AL.

the hybrid gene. In every case, the mutation responsible for the MalR phenotype mapped early in the malE portion of the malE-lacZ hybrid gene. These results suggested that the malR mutants are producing malE-lacZ hybrid proteins with export-defective signal peptides. Since the mutations obtained in the MalR selection map within the malE portion of the hybrid gene, they can be recombined into an otherwise wild-type malE gene, where their effects on MBP export can be investigated. Such recombinants have been obtained, and these exhibit M d - phenotypes on maltose tetrazolium indicator agar, thereby demonstrating that the ability of these strains to utilize maltose as a carbon source is impaired. Each of the mutants still grows on maltose minimal medium, albeit less efficiently than a wild-type strain. In all, five distinct phenotypes were recognized on the basis of reduced maltose utilization. In addition, since each of the mutants spontaneously reverts to Mal+ , it appeared that these represent single point mutations in the malE gene. These conclusions are confirmed by DNA sequencing (see later). The MBP produced by the parental and Mal- strains has been analyzed by radioimmunoprecipitation with rabbit anti-MBP serum, SDS-PAGE, and autoradiography (see Fig. 3). Under these radiolabeling conditions, only mature MBP can be detected in the immune precipitate obtained from the wild-type parent. Although some apparently authentic mature MBP can be precipitated from each of the M a l t mutants, each of these strains is seen to accumulate significant amounts of unprocessed, precursor MBP (preMBP). The fraction of the total MBP that is precipitated as preMBP is directly related to the strength of the Mal- phenotype and, therefore, the severity of the export defect. For example, the strongest export-defective mutants, 18-1 and 19-1, accumulate greater than 95% of the MBP as precursor. In contrast, only about 60% of the MBP produced by the weakest of the mutants, 16-1, accumulates as preMBP. Fractionation studies have confirmed that the preMBP accumulated by these export-defective mutants is found in the cytoplasm or loosely associated with the cytoplasmic membrane (Bassford and Beckwith, 1979; P. Bassford, unpublished observations). In all cases, the mature MBP produced by these strains is periplasmic. The kinetics of MBP export in these strains has also been investigated (Ryan et al., 1985). In the wild-type strain, mature MBP appears in the periplasm extremely rapidly following its synthesis, a result that is entirely consistent with the observed cotranslational processing of this protein (Josefsson and Randall, 1981). In marked contrast, the appearance of mature MBP in the periplasm in the export-defective mutants exhibits a half-time of 5 minutes or more, following a 15-second pulse-label with [35SJmethionine. The experimental evidence suggests that these slow kinetics result from a decrease in the rate of preMBP translocation across the cytoplasmic membrane rather than from simply delayed processing. In fact, the preMBP produced by these strains can be processed in vitro as efficiently as the wild-type preMBP.

3. PROTEIN EXPORT IN E coli

115

FIG.3 . Immune precipitation of radiolabeled maltose-binding protein from solubilized extracts of ma//?+ and malE signal sequence mutants. Immune precipitation of MBP with rabbit anti-MBP serum, SDS-PAGE, and autoradiography was performed as described (Bankditis et al., 1984). Note that under the radiolabeling conditions employed, the ma//?+ strain yields only mature MBP. MBP precipitated from the malEA12-18 strain is found almost exclusively in its truncated precursor form. MBP precipitated from the five ma/E point mutants is found in varying ratios of intact precursor and mature form. See text for additional details.

Since the isolation and characterization of these original MBP export-defective mutants (Bassford and Beckwith, 1979; Bedouelle et al., 1980), a similar selection scheme has been employed to isolate additional mutants. Many of these have proved to be repeats of the original isolates, but several classes of malE deletion mutations were also obtained (see below). In addition, it was of interest to obtain mutants exhibiting only minor MBP export defects. It was anticipated that a comparison of mutational alterations in the MBP eliciting both major and minor defects in export would provide additional insight into the nature of those signals required for efficient MBP secretion into the periplasm. Such mutants were obtained by selecting for MalRLac+ derivatives of a strain carrying a class I11 malE-lacZ protein fusion, PB62-37. As discussed, although strain PB62-37 is maltose sensitive, it is markedly less sensitive than the class IV fusion strain PB72-47. This demonstrates that high-level synthesis of the hybrid protein produced by strain PB62-37 is somewhat less disruptive to the cell. Thus, it was reasoned that MalRLac+ mutants of this strain might represent new classes of mutations different from those obtained with strain PB72-47, because of the less stringent selection employed. A number of MalRLac+ mutants of strain PB62-37 have been isolated and characterized (V. Bankaitis and P. Bassford, manuscript in preparation). Once again, the responsible mutations are all found to map early in the malE portion of the hybrid gene and have been recombined into a wild-type malE gene. Although certain of these recombinants exhibit Ma1 - phenotypes identical to those mu-

116

V M A S A. BANKAlTlS ET AL.

tants obtained in the original PB72-47 selection, four of the new mutations proved to be phenotypically silent when crossed in cis to malE+, i.e., the recombinants remain Ma1 . This indicates that the mutations are different from those described above, and strongly suggest that they do not significantly alter the export competence of the MBP. The MBP produced by these mutants has been immune precipitated and analyzed by SDS-PAGE. Only one of these mutants, harboring a mutation designated malE14-2, accumulates a significant amount of preMBP (approximately 50% of the total synthesized). The presence of a small amount of preMBP in the precipitates obtained from the three remaining mutants demonstrates that MBP export in these strains is somewhat less efficient than wild type. DNA sequence analysis reveals that each of these mutations is unique (see below). In Section III,A, we provide an explanation as to why certain mutations can render the class I11 malE-lacZ protein fusion strain MalR and yet be almost phenotypically silent when introduced into an intact malE gene. +

IN THE MBP SIGNAL PEPTIDE 4. MUTATIONAL ALTERATIONS

It had been suspected from the outset that the MBP export-defective mutations alter the amino-terminal signal peptide of this protein. The primary amino acid sequence of the MBP signal peptide was determined by a combination of DNA and protein sequencing techniques (Bedouelle et al., 1980). It exhibits a structure typical of most eukaryotic and prokaryotic signal peptides that have been so analyzed (see chapters by Duffaud et al. and von Heijne, this volume). The MBP signal peptide is 26 amino acids in length and consists of two distinct regions, the hydrophilic segment and the hydrophobic core (Fig. 4A). The hydrophilic segment is composed of the eight amino-terminal residues, of which three are positively charged; the remaining 18 residues comprise the hydrophobic core, a region totally devoid of charged residues. Note that the ultimate residue is an alanine, the amino acid typically found at the processing site of prokaryotic nonlipoprotein signal peptides (Perlman and Halvorson, 1983). DNA sequence analysis demonstrated that the five Ma1 - phenotypic classes, originally isolated by selecting MalR Lac+ derivatives of the class IV fusion strain PB72-47, define five unique mutational alterations in the MBP signal peptide (Bedouelle et al., 1980). Each of these alters the hydrophobic core (Fig. 4B). The mutation designated malE1O-1 substitutes a proline for leucine at residue 10. Since proline has the property of interrupting a-helices in polypeptide chains, it was proposed that an alteration in the secondary structure of the MBP signal peptide is responsible for the export defect elicited by this mutation (see Section 111,A). The remaining four mutations introduce a charged residue, either acidic or basic, into the central portion of the core, a finding indicating that the hydrophobic nature of this region is essential for proper signal peptide func-

Processing Site

A

1

2

3

4

5

6

7

8

9 10 11 12 13 14 15 16 17 18

19 20

21 22 23 24 25 2 6 1 1

2

3

4

5

met lys ile lys thr gly ala arg ile leu ala leu ser ala leu thr thr met met phe s e r ala ser ala leu ala lys ile glu glu gly

B

c

4 4

4

4

pro

glu

lys

arg arg

10-1

14-1

16-1

18-119-1

4

glu EzzzBZB21bpv-

C A 323

D

a12-18

11-3

p

v 327b

4 4

4

arg pro

pro

10-2 11-1

14-2

...

._..

4 ser 11-2

FIG.4. Primary amino acid sequence of the wild-type MBP signal peptide and the alterations resulting from various malE signal sequence mutations. ( A ) The wild-type MBP signal peptide and the first six resldues of the mature MBP. The processing site is shown. (B) MBP signal peptide alterations resulting from the strong malE point mutations obtained in the initial selection (Bassford and Beckwith, 1979; Bedouelle er al., 1980). (C) MBP signal peptide alterations resulting from the unique m l E point and deletion mutations obtained in a subsequent selection (Bankaitis and Bassford. 1984b). (D) MBP signal peptide alterations resulting from the selection employing the class 111 malE-lacZ fusion strain PB62-37 (Bankaitis and Bassford. 1984b). Single amino acid alterations in the wild-type sequence are indiated by arrows; residues removed from the signal peptide as a result of deletion mutations are indicated by shaded bars. The corresponding mnlE allelic designations are also given. See text for additional details.

118

VYTAS A. BANKAlTlS ET AL.

tion. Further support for this contention is derived from an analysis of how each mutation affects MBP export efficiency. The two arginine-for-methionine substitutions (malEl8-1 and malE19-1) are the strongest mutations, in terms of MBP export, and clearly represent the most radical alterations from the standpoint of hydrophobicity. In contrast, the substitution of lysine for the neutral amino acid threonine at residue 16 is a more moderate alteration in terms of hydrophobicity and has the least impact on MBP export. Although these conclusions seem reasonable, any interpretation of how these strong export-defective alterations elicit their effects on MBP export must consider properties other than just hydrophobicity. Consider that, of the 18 amino acids that constitute the hydrophobic core, most can be replaced by a charged residue as a result of a single base change in the DNA coding region. Yet, isolation of numerous, independently obtained export-defective mutants, beginning with the class IV fusion strain PB72-47, has repeatedly yielded these same four mafE signal sequence mutations. This observation suggests that the particular residues altered by these mutations (numbers 14, 16, 18, and 19) play a more critical role in the initiation of MBP export. A similar subset of four key residues exists in the signal peptide of the LamB protein. Some possibilities for a functional role for these particular residues will be discussed in a later section. Employing a slightly more complicated version of the same selection scheme with strain PB72-47, several additional mafE signal sequence mutations have been obtained more recently (V. Bankaitis and P. Bassford, manuscript in preparation). One of these substitutes glutamic acid for alanine at residue 1 1 of the MBP signal peptide (Fig. 4C). Unlike the four mutations just described, this mutation (designated mafEl1-3)introduces a charged residue fairly early in the hydrophobic core. Perhaps for this reason, it has less effect on MBP export; approximately 50% of the MBP precipitated is found in its precursor form. Several deletion mutations in the malE gene also have been obtained. One of these is a 21-bp deletion, designated mafEA12-18, whose end points lie entirely within the DNA sequence encoding the hydrophobic core of the MBP signal peptide. This deletion fuses malE codons 11 and 18 at their respective wobble bases, thereby removing residues 12 through 18 from the MBP signal peptide (Fig. 4C). In terms of MBP export, mafEA12-18 is the strongest signal sequence mutation yet isolated. Less than 1% of the MBP synthesized in a mafEA12-18 strain is exported to the periplasm and processed (see Fig. 3). Somewhat surprisingly, an E. cofi malEA12-18 mutant can still utilize maltose as sole carbon source, albeit very inefficiently. This deletion is of particular interest in that three of the four critical residues of the MBP signal peptide identified above are missing in the truncated signal peptide. Another deletion isolated, mafEA323, removes the last 20 residues of the MBP signal peptide and the first 89 residues of the mature protein. Not surprisingly, this mutation results in a nonfunctional MBP, and a mafEA323 strain is unable to utilize maltose as a carbon source. This

3. PROTEIN EXPORT IN E. coli

119

internal malE deletion has helped define a site within the mature MBP that facilitates MBP secretion (see Section 111,C). Three of the four mutations obtained using the class 111 fusion strain PB62-37 are substitutions of one uncharged residue for another (Fig. 4D). Note that the introduction of proline at residue 1 1 (malE11-f)has a much less dramatic effect on MBP export than that elicited by a conversion of leucine to proline at the immediately preceding residue (malE1O-1). Also, note that the substitution of proline for alanine at residue 14 (malE14-2) is the alteration that causes the strongest export defect among the unique mutations obtained in this selection. Still, this alteration is considerably less detrimental to signal peptide function than introduction of a glutamic acid residue at the same position (malE14-1).The one mutation introducing a charged residue into the signal peptide, malE1O-2, results in only a slight reduction in MBP export efficiency. Thus, it would seem that the strength of the export defect resulting from introduction of a charged residue into the hydrophobic core is a function of position. Those that are introduced into the center of the core cause the strongest export defects, whereas similar changes toward the amino-terminal boundary of the core exhibit considerably weaker defects. Further discussion of these various signal sequence alterations is presented in a subsequent section.

5. PROCESSING OF malE-lac2 HYBRID PROTEINS Before concluding this section, we would like to briefly mention several additional points concerning the synthesis of malE-lac2 hybrid proteins in E. coli. As previously discussed, it is believed that secretion of those hybrid proteins having an intact MBP signal peptide is initiated in a manner analogous to the native MBP. Presumably, the amino terminus of the nascent chain interacts with the cytoplasmic membrane. The polypeptide chain is cotranslationally extruded through the membrane into the periplasm until that point in the transfer process where the carboxyl-terminal P-galactosidase moiety blocks further passage. Added support for this proposal stems from recent observations that the malE-lacZ hybrid proteins are exposed on the outer surface of the cytoplasmic membrane and proteolytically processed (Rasmussen et a f . , 1984). The experimental evidence for this is as follows: 1. Two distinct forms of these hybrid proteins can be detected under certain conditions of radiolabeling and SDS-PAGE. The difference in the apparent molecular weights of these two forms is very slight, which is to be expected if these very large proteins differ by only 26 amino acid residues. 2. Only the higher-molecular-weight (i.e., unprocessed) forms are detected when the hybrid proteins have a defective signal sequence that prevents their insertion into the cytoplasmic membrane. 3. High-level synthesis of these hybrid proteins, which eventually results in a

120

VYTAS A. BANKAlTlS ET AL.

total block in general protein export, causes a concomitant and progressive accumulation of the hybrid protein in its higher molecular weight form. 4. The processed form of the hybrid protein, but clearly not the unprocessed form, is accessible to externally added proteinase K in E . coli spheroplasts, indicating that at least a portion of the processed hybrid protein has been translocated across the cytoplasmic membrane. Processing of the hybrid protein has been observed for each of the classes of malE-lacZ protein fusion strains except class I. Thus, although the hybrid protein produced by a class 11 fusion strain includes the MBP signal peptide but very little of the mature protein (probably fewer than 15 residues), this is apparently sufficient to lead the amino-terminal portion of this protein through the cytoplasmic membrane. Two interesting differences have been noted between the processing of the classes I1 and III hybrid proteins and that observed for the larger classes IV and V hybrid proteins. First, under the most favorable conditions for export (i.e., immediately after maltose induction but prior to manifestation of the MalS phenotype), only a modest fraction of the class I1 and class I11 hybrid proteins synthesized are processed, whereas essentially 100% of the larger hybrid proteins are seen to be processed under identical conditions. This is certainly consistent with the observation discussed earlier that classes IV and V fusion strains exhibit a significantly more severe MalS phenotype than that exhibited by classes I1 and 111 fusion strains. Second, pulse-chase experiments have revealed that processing of the classes I1 and I11 hybrid proteins is a relatively slow event, occurring posttranslationally and exhibiting a half-time of several minutes. In marked contrast, processing of the larger hybrid proteins is achieved very rapidly, probably cotranslationally. The added significance of these observations will be presented in Section III,C.

C. IarnB-lac2 Protein Fusions 1 . PROPERTIES AND COMPARISON TO malE-lacZ FUSIONS

Gene fusion technology has also been applied with great success to the study of how the L a d protein is exported to the outer membrane of E . coli. This work, from the laboratory of T. Silhavy and co-workers, is described in detail in several recent reviews ( E m et a l . , 1980a; Emr and Silhavy, 1982; Silhavy et a f . , 1983) and will only be summarized here. The LamB protein is encoded by another gene in the malB region, lamB. It constitutes a major outer membrane protein for cells that are grown in the presence of maltose, it is essential for import of maltodextrins across the outer membrane (Szmelcman and Hofnung, 1975), and it serves as the cell surface receptor for certain bacteriophages, most notably coliphage A (Randall-Hazelbauer and Schwartz, 1973). The analysis of

3. PROTEIN EXPORT IN €. coli

121

lad-lacZ protein fusions has provided ( 1 ) an alternative to the MBP for studying the features of a secreted polypeptide that are required for efficient initiation of export, and (2) an initial insight as to the nature and location of signals inherent to an outer membrane protein that serve to distinguish it from a protein destined for the periplasm. The behavior of lamB-lacZ protein fusions is, in some instances, very similar to that observed for malE-lacZ fusions. However, certain lamB-lacZ fusions exhibit a behavior that is strikingly unique. Four classes of lamB-lacZ fusions were initially identified on the basis of the amount of lamB DNA contained within the hybrid gene (Hall et 41.. 1982b). Protein sequence analysis of representative class I (61-4) and class I1 (52-4) lamB-lacZ hybrid proteins has defined precisely the position of their respective fusion joints (Moreno et al., 1980). The class I protein includes only the first four residues of the LamB protein signal peptide. The class I1 hybrid protein includes the entire signal peptide and the first 15 residues of the mature LamB protein. These two proteins fractionate as soluble, cytoplasmic species, although the class I1 protein exhibits a slight interaction with the cytoplasmic membrane (Moreno et al., 1980). Neither the class I nor the class I1 hybrid proteins elicit any obvious stress to the host cell when produced at high levels. The representative class 111 lamB-lacZ hybrid protein (42- 1) includes an intact signal peptide and approximately the first 173 of 421 residues of the mature LamB protein (Benson and Silhavy, 1983). A class 111 fusion strain exhibits a MalS phenotype that is essentially identical to that shown by class IV and class V malE-lacZ fusion strains. However, whereas a malE-lacZ hybrid protein cannot be secreted into the periplasm, approximately 30-40% of the 42-1 lamB-lacZ hybrid protein is incorporated into the outer membrane. The remainder of the hybrid protein population is evenly distributed between the cytoplasmic membrane and the cytoplasm (Hall er al., 1982b). An additional phenotype of class 111 fusion strains is their temperature-sensitive (ts) ability to utilize lactose as sole carbon source. The strains grow slowly on lactose at 30°C but not at all at 37°C (Emr and Silhavy, 1982). The molecular basis for this ts phenotype is not understood, but it seems to be related to hybrid protein export. This additional phenotype has proved useful for selecting certain classes of lamB signal sequence mutations (see later). The representative class IV hmB-lacZ fusion protein (42- 12) includes an intact signal peptide and approximately the first 240 residues of the mature LamB protein and is incorporated into the outer membrane at an efficiency of 85 to 90% (Hall et al., 1982b). Class 1V fusion strains, although still Mals, are markedly less so than class 111 lamB-lacZ fusion strains. Presumably, this reduction in sensitivity, not observed in the analogous malE-lacZ fusion strains, is a direct result of increased efficiency in hybrid protein export. Interestingly, although

122

VYTAS A. BANKAlTlS ET AL.

both class 111 and class IV lumB-lucZ fusion strains produce approximately equal amounts of their respective hybrid proteins, the latter strains exhibit induced pgalactosidase enzymatic activities that are one to two orders of magnitude lower than those measured in the former. For this reason, class IV fusion strains are Lac-. The low specific activity of the class IV hybrid protein appears to be a consequence of its localization. Tetramerization of the hybrid monomer is required for p-galactosidase activity. Efficient export of the class IV hybrid monomer to the outer membrane endows a membrane-bound state to the monomer that may preclude efficient tetramerization. IN THE LamB 2. MUTATIONAL ALTERATIONS PROTEINSIGNALPEFTIDE

The phenotypic properties of the class 111 luml-luci! fusion strains provide two different selection schemes for mutants producing an export-defective LamB protein. The first of these is to isolate MalRLac+ derivatives of the 42-1 fusion strain. The second involves simply selecting for Lac derivatives of the 42- 1 fusion strain at 37°C. This latter sdection does not demand a MalR phenotype and rests on the prediction that any export defect will result in altered localization of at least some of the hybrid monomer. A consequence of such altered localization (i.e., internalization in the cytoplasm) should allow for greater efficiency of hybrid protein tetramerization and subsequent expression of a Lac phenotype by the mutant strain. To date, 36 MalRLac and 12 Lac derivatives of the 42- 1 fusion strain have been characterized in detail. Each of the responsible mutations map early in the luml portion of the hybrid gene and result in failure to export the hybrid protein from the cytoplasm ( E m and Silhavy, 1982). Twenty-six of the mutations obtained by selecting MalRLac , but just one of the mutations obtained in the Lac+ selection, confer a typical LamB- phenotype [resistance to phage A (AR) and inability to utilize maltodextrins (Dex- )] when recombined into lumB . DNA sequence analysis reveals that the various mutations fall into a number of different classes, including both point mutations and deletions ( E m et ul., 1980b). All of the mutations alter the signal peptide of the LamB protein, as shown in Fig. 5 . Although there is little primary sequence homology to the MBP signal peptide, the 25 residue LamB signal peptide is another typical example of such structures. The hydrophilic segment includes two basic residues followed by the 18 residues comprising the hydrophobic core. Each of the four point mutations obtained introduces a charged residue into the core and results in a strong export defect. In all cases, greater than 95% of the LamB protein synthesized is found in the cytoplasm as unprocessed precursor (preLamB protein) (Emr and Silhavy, 1982; +

+

+

+

+

+

Processing Site

A

B

1

2

3

4

5

6

7

8

9

10 11 12 13 14 15 16 17 18 19

20 21 22 23 24 251 1

2

3

4

5

met met ile thr leu arg lys leu pro leu ala val ala val ala ala gly val met ser ala gln ala met ala val asp phe his gly

4 4 4

4

asp glu glu S7 1 S70 S99

arg

S69

1 2 b p m S78 36b

S60

4 C

arg S96

c

asp s73

FIG. 5. Primary amino acid sequence of the wild-type LamB protein signal peptide and the alterations resulting from various lamB signal sequence mutations. (A) The wild-type LamB protein signal peptide and the first six residues of the mature LamB protein. The processing site is shown. (B) LamB signal peptide alterations resulting from strong lamB point and deletion mutations ( E w e r a/.. 1980b). (C) Alterations in the LamB protein signal peptide that are phenotypically “silent” (Emr and Silhavy, 1982). Single amino acid alterations in the wild-type sequence are indicated by arrows; residues removed from the signal peptide as a result of deletion mutations are indicated by shaded bars. The corresponding la& allelic designations are also given. See text for additional details.

124

VYTAS A. BANKAlTlS ET AL.

Silhavy et ul., 1983). Two of the deletion mutations remove DNA sequences internal to the signal peptide coding region. The export defect resulting from the 12-bp deletion (lumBS78)is similar to that produced by the four point mutations. The 36-bp deletion (lumBS60) removes a full two-thirds of the core and results in a LamB protein that is totally export defective. With this mutant, it has not been possible to detect any LamB protein localized to the outer membrane by either biochemical or exquisitely sensitive biological assays for LamB function, even under fully induced conditions (Emr et al., I98 1). The remaining 10 mutations obtained in the MalRLac selection and 1 1 of the 12 mutations obtained by selecting Lac+ derivatives at 37°C are found to be phenotypically silent when recombined into lumB (Emr and Silhavy, 1982). These mutants accumulate only minute quantities of preLamB protein (less than 2% of the total synthesized), a finding indicating that the LamB protein produced by these strains has no major export defect. DNA sequencing reveals that these mutations fall into two classes, each representing unique alterations in the LamB protein signal peptide. Both classes represent point mutations that convert the glycine at residue 17 to either an acidic or basic residue (Emr and Silhavy, 1982). These mutations provide a striking example of the importance of the position at which a charged residue is introduced into the hydrophobic core vis-8-vis its effect on export. As for MBP, a defined subset of residues within the LamB protein signal peptide is demonstrated to be critical for initiation of export. Introduction of charged amino acids at residues 14, 15, 16, and 19 results in strong export defects. However, introduction of charged amino acids at residue 17 has virtually no effect on export. +

+

111.

INTRAGENIC INFORMATION SPECIFYING PROTEIN EXPORT

A. Interpretation of Signal Sequence Mutations Analysis of point mutations for both the MBP and LamB protein signal peptides identifies a subset of four positions within the hydrophobic core at which introduction of a charged residue is severely detrimental to protein export. The importance of this finding is underscored by two general observations. First, these mutations have been isolated on multiple, independent occasions, even though other codons within the region can be converted by a single base change to introduce charged residues. Second, characterization of those few mutations that do introduce charged residues at other positions within the hydrophobic core demonstrates that such alterations do not result in major signal peptide dysfunction.

3. PROTEIN EXPORT IN E. coli

125

Significantly, all other strong export-defective mutations (i.e., deletions) in some manner affect the residues constituting these subsets. The malEA12-18 mutation removes three of the four key residues; lamBS60 removes the entire region. These deletions represent the strongest export-defective, yet still exportspecific, mutations obtained for their respective signal peptides. The lamBS78 mutation probably affects the subset within the LamB protein signal peptide indirectly by altering protein secondary structure. This deletion is predicted to prevent nucleation of an a-helical structure through the center of the hydrophobic core, thereby causing the subset residues to be included in an abnormal random coil conformation (Emr and Silhavy, 1983). Support for this contention is obtained by the analysis of intragenic mutations suppressing lamBS78. Two such mutations have been characterized ( E m and Silhavy, 1983). Both are predicted to restore an a-helical conformation through this region. The only MBP signal peptide alteration that has been shown to have a major effect on protein export without introducing a charged residue into the hydrophobic core introduces a proline at residue 10. Although not previously discussed, the rules of Chou and Fasman (1978) for predicting protein secondary structure indicate that the first 6 residues of the wild-type MBP signal peptide probably exist in a random coil whereas the remaining 20 residues, including the entire hydrophobic core, are predicted to assume an a-helical conformation. While this mutation, malElO-1, does not affect the predicted secondary structure of the MBP subset region, it is predicted to lend a more extensive random coil component to the extreme amino terminus of the core. Bedouelle and Hofnung (198I ) have suggested that this effect may bring the hydrophilic segment of the MBP signal peptide in close proximity to the hydrophobic subset region by a fold-back mechanism. The net result of such an association may be similar to introducing a charged residue into this region. Two general proposals concerning the role of the region defined by the subset mutations have been suggested, Although the two are certainly not mutually exclusive, one ascribes primarily a structural role, whereas the other postulates a major role in recognition by the cellular protein export machinery. We will initially address the first of these. As previously noted, it is the introduction of charged residues at specific sites within the hydrophobic core that results in a strong export defect. Substitution of an uncharged residue at one of these sites has significantly less effect (e.g., compare mutations malE14-1 and malE14-2). Also, substitution of a charged residue for a hydrophobic residue (e.g., malEI8-1) results in a stronger export defect than substitution of a charged residue for a neutral polar residue (e.g., malE16-I). These results are consistent with hydrophobicity being the major feature disrupted by the strong exportdefective point mutations. Deletions removing portions of the hydrophobic core would have a similar effect.

126

VYTAS A. BANKAlTlS ET AL.

Bedouelle and Hofnung (198 1) have applied predictive rules for protein structure to a number of wild-type and mutant signal peptides. They have attempted to relate the physical length of the hydrophobic core to efficiency of export initiation. A parameter termed hydrophobic axis length (HAL) has been defined and proposed to constitute a measure of signal peptide functionality. They conclude that the hydrophobic core of a functional signal peptide must exist in a periodic structure (a-helix or P-strand) and have a minimum length of 18 8,. This value of 18 8, has been designated the threshold HAL (tHAL). Signal peptides with a HAL of at least 18 8, are predicted to be functional, whereas those that fall below the tHAL are predicted to be nonfunctional. The notion of HAL and its relationship to signal peptide function is an attractive one. The initiation of protein export may require a direct interaction of the hydrophobic core with the lipid bilayer. Inouye et al. (1977) have proposed that the hydrophobic core “loops” into the lipid bilayer following the initial interaction of the charged amino-terminal portion with the membrane. The stringent requirement for a minimum length core may reflect a distance the signal peptide must penetrate the membrane in order to successfully initiate protein translocation across this barrier. Most of the malE and lamB signal sequence mutations exerting strong export defects depress HAL to a value less than tHAL. The two exceptions are mutations malEIO-1 and malE14-I. A possible mechanism for the effect produced by malEIO-1 is noted above. Analysis of malE signal mutations that result in only minor export defects (Fig. 4D)is also consistent with the HAL hypothesis. These four mutations do not depress HAL to a value below tHAL. Furthermore, the two classes of intragenic suppressor mutations of lamBS78 are predicted by the HAL hypothesis (Bedouelle and Hofnung, 1981). However, other predictions of this hypothesis are clearly not borne out. For instance, the introduction of charged residues anywhere within the region of hydrophobic core residues 15 through 19 of either the MBP or LamB signal peptide would be predicted to produce a strong export defect. Yet, introduction of either a basic or acidic residue for glycine at position 17 of the LamB protein signal peptide has only a minor impact on export. Of similar significance is malEI4-I. The MBP signal peptide encoded by this malE allele exhibits a HAL equal to tHAL, yet it is very inefficient in terms of facilitating MBP export. In fact, malEl4-1 is a stronger mutation than malEl6-I, a mutation that depresses HAL to a value below tHAL. An alternative proposal concerning the putative role of the residues identified by strong malE and lamB signal sequence point mutations suggests that these residues constitute a “recognition site” mediating the interaction of the signal peptide with the cellular protein export machinery (Emr and Silhavy, 1982; Silhavy et al., 1983). The virtual indifference of the LamB protein signal peptide to the introduction of a charged residue at position 17 is consistent with such a

3. PROTEIN EXPORT IN E. coli

127

proposal. In addition, the strong export-defective mutations are suppressed in a allele-specific manner by extragenic mutations that identify genes encoding components of the normal secretion machinery (see below), These results provide convincing genetic evidence indicating that the subset region interacts directly with protein components of the E . coli export apparatus. Deletion mutations removing part or all of this region (e.g., mnlEA12-18 and lumBS60) or that alter the normal secondary structure through this region (e.g., lamBS78) are envisioned to disrupt or totally eliminate recognition. Mutations that do not affect these key residues have only minor effects on recognition. Weak malE and lnmB signal sequence mutations are considered to fit into this latter category. Finally, it is of interest to note that each of the malE and lumB signal sequence mutations discussed above affects the hydrophobic core of their respective signal peptides and leaves the hydrophilic segments intact. One might conclude from these data that the hydrophilic segment does not have an important role in initiating protein export. However, this seems not to be the case. M. Inouye and co-workers (Inouye et al., 1982) have shown that mutational alterations in the hydrophilic segment of the E . coli lipoprotein signal peptide can affect both synthesis and export of this protein to the outer membrane. Also, Hall et al. (1982a, 1983) have described a mutation in the lamB gene that converts the arginine at residue 6 of the signal peptide to a serine. Although this mutation does not appear to have an effect on LamB protein export per se, it does significantly reduce the translation efficiency of the lamB gene. As discussed below in Section IV,C, this and some additional data suggest that there may be an obligate coupling of protein export and translation.

B. lntragenic Suppressor Mutations for rnalEA12-18 The isolation of mutations that result in an export-defective signal peptide allows, in turn, a selection for second-site mutations that phenotypically suppress the effects of the first mutation by restoring some degree of export-competence to the protein in question. Such suppressor mutations can be extragenic, i.e., map in a gene other than that encoding the export-defective protein. Several classes of such extragenic suppressor mutations will be discussed in Section IV,B. Such suppressor mutations can also be intragenic and would be expected to restore certain features to the defective signal peptide that render it more efficient in facilitating protein export. The use of two complementary genetic strategies, provided by the initial isolation of export-defective mutations, followed by the subsequent isolation of intragenic suppressor mutations, constitutes a very powerful method for identifying the important functional features of a signal peptide. As previously discussed, such an approach was instrumental in formulating mechanisms for how lumBS78 exerts a strong export defect.

128

VYTAS A. BANKAITIS ET AL.

Beginning with an E. coli strain harboring maLEAI2-18, intragenic suppressor mutations have been obtained by simply selecting for mutants better able to utilize maltose as sole carbon source (Bankaitis et al., 1984). In addition to representing the strongest malE signal sequence mutation, malEA12-18 is of particular interest because it alters the MBP signal peptide in two ways. First, it decreases the length of the hydrophobic core from 18 to 11 residues. Second, it excises three of the four residues constituting the critical subset for the MBP signal peptide. Since malEAI2-I8 eliminates a large portion of the putative MBP signal peptide recognition site, it was reasoned that analysis of intragenic suppressor mutations would provide evidence that either supports or questions the validity of the recognition site concept. Six different intragenic malEAI2-18 suppressor mutations have been isolated and characterized. Each of these increases the proportion of the total MBP synthesized that is correctly localized and processed. As shown in Fig. 6 , the three strongest suppressor mutants exhibit an MBP export efficiency that is virtually indistinguishable from wild type, indicating that the altered signal peptide must mediate MBP secretion in a manner analogous to the wild-type signal peptide. Although there remains a considerable MBP export defect in the three weakest suppressor mutants, export is still clearly more efficient than in the malEA12-18 parental strain. DNA sequence analysis reveals that five of the six suppressor mutations result

FIG. 6. Immune precipitation of MBP from intragenic mnlEA12-18 suppressor strains. Representative strains were radiolabeled, solubilized, and the MBP precipitated as described in Bankaitis et al. (1984). Precipitates were analyzed by SDS-PAGE and autoradiography. The mature MBP precipitated from a mdE+ strain is shown in lane A. Lane B shows the MBP precipitated from a mnlE18-l mutant strain that provides a marker for full-sized pre-MBP. It is to be compared with the truncated pre-MBP shown in lane C that was precipitated from a mnlEA12-18 strain. Lanes D through I show precipitates obtained from representatives of each of the six classes of intragenic mnIEA12-I8 suppressor mutants, shown in the following order: lane D, R6; lane E, R5;lane F, R4; lane G,R3; lane H, R2; and lane I, R1. The corresponding changes in the primary sequence of the MBP signal peptide for each of the suppressor mutants is given in Fig. 7. See text for additional details.

129

3. PROTEIN EXPORT IN E. coli

Processing Site 1 2 3 4 5 6 7 8 9 10 11 19 20 21 22 23 24 25 261 1 2 3 met lys ile lys thr gly ala arg ile leu ala met phe ser ala ser ala leu ala lys ile glu

1: CYS

R3

4

5

6

7

I\

leu ala met R1

8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

glu gly l y s leu val ile trp ile asn gly a s p lys gly tyr asn gly leu ala glu val R6 FIG.

7 . Alterations in the MBP signal peptide resulting from intragenic suppression of malEAI2-

18. The truncated MBP signal peptide resulting from the malEA12-18 deletion mutation is shown,

along with alterations in this structure rresulting from intragenic suppressor mutations. Alterations resulting from substitution mutations are indicated by a downward-pointing arrow. Alterations resulting from genetic duplications are indicated by an upward-pointing arrow. See text and Bankaitis et al. (1984) for additional details.

in alterations within the MBP signal peptide (Fig. 7). Each of these appears to lengthen the truncated hydrophobic core by one of three mechanisms: (1) Two mutations (R1 and R4) add additional residues to the core by duplicating preexisting codons within this region. (2) Two suppressor mutations (R2 and R3) extend the amino-terminal boundary of the hydrophobic core by substituting an uncharged residue for the arginine residue at position 8. ( 3 ) The mutation designated R5, a substitution of valine for alanine at residue 1 1 , is the only suppressor mutation predicted to alter the secondary structure of the MBP signal peptide. Neither malEAI2-I8 nor any of the other intragenic suppressors are predicted to have any effect on secondary structure, except for the number of residues comprising the a-helix. However, in the R5 suppressor mutant, the pentapeptide centered around the valine at residue 1 1 is predicted to assume a P-strand, rather than an a-helical, conformation. Since the axial residue length for peptides in a P-strand is greater than for peptides in an a-helix (3.6 vs 1.5 A), the net effect of this mutation may be to physically extend the hydrophobic core without actually increasing the number of residues that comprise it. The analysis of these five intragenic suppressor mutations provides additional evidence that hydrophobicity, particularly the absolute length of the hydrophobic core, is a major determinant of signal peptide function. This analysis also lends further insight into the role of the critical subset in the initiation of protein export,

130

VYTAS A. BANKAlTlS ET AL.

particularly in relation to the recognition site concept. It is apparent that a functional MBP signal peptide can be reconstituted from a deletion mutant by several different mechanisms and that it is nearly as efficient as the wild-type structure in facilitating MBP export. The suppression is achieved without reconstituting a signal peptide whose primary amino acid sequence resembles that of the wild-type through the region removed by the malEA12-18 deletion. This is consistent with the general lack of obvious primary sequence homology that exists between the various known signal sequences of either prokaryotic or eukaryotic origin (see articles by Rapoport and Wiedmann and by Duffaud et al., this volume). Consequently, it has been proposed that the strong position effect observed upon introduction of a charged residue into the hydrophobic core identifies the amino acid residues forming the initial contact point with the hydrophobic environment through which the signal peptide must penetrate (Bankaitis et al., 1984). Mutations introducing charges at these positions or altering the secondary structure normally assumed through this region would be particularly deterimental to signal peptide function. Finally, the least efficient intragenic malEA12-18 suppressor mutation encountered in this selection (designated R6) results in a single amino acid substitution at residue 19 of the mature MBP. It remains unclear how such an alteration, fairly far removed from the signal peptide, serves to suppress malEA12-18. This result indicates that information for initating MBP secretion may be contained within the mature protein. As discussed below, there may in fact be several regions in the mature MBP and LamB proteins that have an active role in the protein localization process.

C. Additional lntragenic Information Specifying Initiation of MBP Export There is overwhelming evidence that a functional signal peptide is absolutely essential for the efficient initiation of protein export. However, it seems that the full complement of information required for this process may not be contained within the signal peptide. Data suggesting the existence of information in the mature portion of the MBP involved in export initiation are obtained from several sources. First, as mentioned in the preceding section, an intragenic malEA12-18 suppressor mutation has been isolated that alters residue 19 of the mature MBP. Second, studies with protein fusion strains have shown that, although class 11 through class V malE-lacZ hybrid proteins each include an intact MBP signal peptide, the efficiency with which the cell attempts to export the hybrid protein increases with the size of the mature MBP moiety. Third, Randall (1983) has shown that the nascent MBP is not translocated across the cytoplasmic membrane and processed until approximately 80% of the polypeptide chain has been translated.

131

3. PROTEIN EXPORT IN E. coli

Recent studies have provided additional evidence indicating the existence of protein export information, at the level of export initiation, located within the mature MBP (Bankaitis and Bassford, 1984). It was found that the high level synthesis of MBP having an export-defective signal peptide markedly interferes with the export of normal envelope proteins. This interference phenomenon has no obvious effects on the growth properties of the cell. However, interference can be demonstrated in pulse-chase experiments by comparing the rate of export for various envelope proteins (e.g., OmpA protein) in wild-type strains versus strains harboring mulE signal sequence mutations. Strains were grown on maltose to induce high-level synthesis of the wild-type or export-defective MBPs. Under identical conditions, only the mulE signal sequence mutant accumulates significant amounts of pulse-labeled envelope proteins in their precursor forms (Fig. 8). These wild-type precursors are slowly processed to the mature species. Since the export-defective MBP is not itself translocated, presumably the step in normal protein export that is being interfered with must be an early one, i.e., prior to translocation. Additional experiments have established the following points: (1) The severity of the interference is generally directly proportional to the rate of synthesis of export-defective MBP. ( 2 ) Suppression of the export block on the mutant MBP, by either intragenic or extragenic mutations, relieves the interference for export of wild-type proteins in a manner that is directly related to the strength of suppression. (3) It is not the accumulation of mutant preMBP in the cytoplasm that is responsible for the interference. Rather, this effect is exerted by the concomitant synthesis of the export-defective MBP and suggests that interference is a cotranslational process.

FIG. 8. The effect of export-defective MBP on export of the OmpA protein. Cells were radiolabeled for 15 seconds with [3sS]methionine and chased for 20 seconds with excess cold methionine. The chase was terminated with ice-cold trichlorodcetic acid, and the OmpA protein was immune precipitated and analyzed by SDS-PAGE and autoradiography. In a malE strain, the radiolabeled OmpA protein is primarily found in its mature form (mOmpA). In strains harboring various malE signal sequence point mutations, a significant proportion of the OmpA protein precipitated is found in its precursor form (pOmpA). The precursor is slowly converted to the mature form in these strains (Bankaitis and Bassford, Jr., 1984). See text for additional details. +

132

W A S A. BANKAlTlS ET AL.

These data have been interpreted to indicate that MBP lacking a functional signal sequence-in one instance a substantial portion of which has been removed (malEAI2-18)-is nevertheless recognized by the cell as an exported protein and at least transiently enters the secretory pathway. Since the MBP cannot exit this pathway via the normal route, it may occupy some component of the pathway for some abnormal period of time. This in turn causes a temporal disruption in the normal functioning of the cellular secretion machinery, an event that interferes with the normal traffic of export-competent proteins that utilize a common branch of the localization pathway. The observed interference in wild-type protein export has provided an assay for determining the region of the export-defective preMBP recognized by the secretion apparatus. The malEA12-18 mutation has been recombined into the hybrid gene of representative classes 11, 111, and IV malE-lacZ protein fusion strains, and the ability of the hybrid protein produced by each strain to interfere with export of OmpA protein and wild-type MBP was analyzed. Only the malEA12-18 derivative of the class IV fusion exhibits the interference, a finding indicating that the interfering region must reside on that part of the MBP included in the class IV hybrid protein but not the class 111 hybrid protein (i.e., between residues 23 and 189 of the mature MBP). To localize the region more precisely, a deletion mutation removing from residue 7 of the signal peptide through residue 89 of the mature MBP (malEA323, see Fig. 4C) was recombined into the hybrid gene of the class IV fusion strain. The resultant hybrid protein is also capable of exerting this interference phenomenon. Thus, based on these various experimental observations, it is currently believed that there is a region between residues 89 and 189 of the mature MBP that harbors information for initiation of export and to identify the MBP as a secreted protein (V. Bankaitis and P. Bassford, unpublished observations). The existence of a region toward the center of the mature MBP that helps to mediate the cotranslational initiation of this protein would explain several observations. First, as previously mentioned, only the classes IV and V malE-lacZ hybrid proteins are efficiently recognized by the cell as exported proteins. Also, only these larger hybrid proteins appear to be cotranslationally processed. In contrast, a significant proportion of the class 11 and class 111 hybrid proteins synthesized apparently never leaves the cytoplasm, and the detectable processing of these hybrid proteins occurs posttranslationally. Second, the existence within the mature MBP of an “internal export signal” could explain why malE signal sequence mutations have been isolated that prevent the attempted export of a class 111 hybrid protein but which have very little effect when introduced into an intact malE gene (see Section II,B,3), Since the class 111 hybrid protein lacks this internal export information and is only inefficiently exported, it may be much easier to alter the signal peptide of this protein in such a way as to render it totally export incompetent. Third, Randall ( 1 983) has reported that cotranslational

3. PROTEIN EXPORT IN E. coli

133

translocation and processing of the MBP only occurs after approximately 80% of the nascent chain has been translated. This finding has been intrepreted to indicate that folding of the nascent chain into a “translocation-competent” domain is required for translocation. However, Ito and Beckwith (1981) have reported that MBP nonsense fragments only one-third the length of the wild-type MBP can be inefficiently translocated to the surface of the cytoplasmic membrane and processed. Taken together, all these experiments strongly suggest that there is a region near the center of the mature MBP that, while not absolutely required for MBP export, functions to ensure that the export process is timely and efficient. Further mutational analysis of the mulE gene, particularly the isolation of mutations that leave the signal peptide intact but alter downstream export information, should define more precisely the nature of this site. Evidence is presented below that a similar site may exist near the center of the mature LamB protein.

D. Export-Specific Information Contained within the Mature LamB Protein Although the export of periplasmic and outer membrane proteins appears to involve one or more common steps, the pathways for localization of these proteins to their respective destinations must diverge at some point. Gene fusion technology has provided some insight as to how this sorting may occur. A key observation of relevance to this process is the demonstration that class IV lumBlucZ hybrid proteins are efficiently localized to the outer membrane. In contrast, localization of hybrid proteins to the periplasm has never been detected. Based on these observations, a model explaining the sorting of periplasmic and outer membrane proteins was proposed (Silhavy el ul., 1979; E m el al., 1980a). It postulates the existence of additional information within the mature portion of the pre-LamB protein that serves as the sorting signal. This signal has been termed the dissociation sequence. It was proposed that, soon after that portion of the nascent chain containing this dissociation sequence emerges from the ribosome, the ribosome dissociates from the cytoplasmic membrane and no further transport of the nascent chain through the membrane occurs. Completion of translation of the entire LamB protein yields a polypeptide embedded within and spanning the membrane. The LamB protein is subsequently translocated to the outer membrane via diffusion through sites of cytoplasmic membrane-outer membrane adhesion, or perhaps by some kind of carrier vesicle. According to this model, the LamB protein would actually reach the outer membrane without being completely extruded through the cytoplasmic membrane. Thus, lamBlucZ hybrid proteins that include an intact dissociation sequence are efficiently localized to the outer membrane. The dissociation step occurs prior to the pgalactosidase moiety entering the membrane; thus, this portion of the hybrid

134

W A S A. BANKAlTlS ET AL.

protein does not have the opportunity to abort the export process. Hybrid proteins that have initiated export but lack an intact dissociation sequence, i.e., those encoded by class 111 lamB-lac2 fusions and classes 11, 111, IV, and V malE-lucZ fusions, continue to be extruded, eventually jam the export site, and precipitate the MalS response. From their initial studies with lamB-Lac2 protein fusions, Silhavy and coworkers concluded that there were probably three regions of the LamB protein required for efficient localization of this protein to the outer membrane. First, the amino-terminal signal peptide is required to initiate the export process. Second, some region between residues 15 and 173 of the mature protein must be involved, since class I11 hybrid proteins are at least inefficiently incorporated into the outer membrane. Third, a region between residues 173 and 240 must also be required, since only class IV hybrid proteins are efficiently incorporated into the outer membrane. To further define these latter two regions, which together must constitute the so-called dissociation sequence, a series of internal, in-frame deletions were generated in vitro in the LamB-lacZ hybrid gene of the class IV fusion 42-12 (Benson and Silhavy, 1983). It was found that deletions that collectively removed residues 70 through 220 have no effect on hybrid protein export. Neither do these deletions, when recombined into lamB , have any effect on export of the LamB protein itself. Thus, one of the sites involved in LamB protein export must reside between residues 15 and 70 of the mature protein. Two deletions were generated that significantly reduce the efficiency of hybrid protein export. Analysis of these deletions, which could not be recombined into lamB , indicate that the second site was encoded by sequences very near the lamB-lac2 fusion joint in the hybrid gene. This corresponds to approximately residues 235-240 in the mature l a d protein. This site may be analogous to the site in the middle of the mature MBP thought to be required for efficient initiation of export of this protein, since the synthesis of export-defective LamB protein also interferes with normal protein export in a manner identical to that observed by synthesis of export-defective MBP (Bankaitis and Bassford, 1984). Experiments locating the region of the LamB protein responsible for the interference phenomenon have not yet been done. It is also important to note that the lamB signal sequence mutations described in Fig. 5C that are phenotypically silent when introduced into the lamB gene were isolated using a class 111 lamBlacZ protein fusion strain. Since the class 111 hybrid protein lacks this later internal export signal, these mutations may be similar in nature to the class of phenotypically silent malE signal sequence mutations previously discussed. Recent experiments have more precisely identified residues near the amino terminus of the mature LamB protein involved in export. As previously mentioned, a strain harboring the class IV lamB-lac2 fusion 42-12 is phenotypically Lac- as a result of the inability of the hybrid protein to form enzymatically active tetramers in the outer membrane. Lac+ derivatives of this fusion strain +

+

3. PROTEIN EXPORT IN E. coli

135

were isolated and scored for maltose sensitivity (Benson et ul., 1984). It was reasoned that a MalS phenotype in the Lac+ derivatives of the fusion strain would indicate that hybrid protein export was still being initiated. However, the Lac phenotype would require a decrease in efficiency of hybrid protein export to the outer membrane. This selection has yielded a number of large deletion mutations and reveals that: +

1. A hybrid protein that includes an intact signal sequence and 27 residues of the mature LamB protein remains cytoplasmic. The corresponding fusion strain is MalR. 2. A hybrid protein that includes 39 residues of the mature LamB protein is inserted into the cytoplasmic membrane but is not translocated to the outer membrane. The corresponding fusion strain in this case is MaF. 3. When the hybrid protein includes the first 49 residues of the mature LamB protein, the protein is translocated to the outer membrane with an efficiency similar to class 111 lumB-lacZ hybrid proteins. Again, the corresponding strain is Mals. These studies convincingly demonstrate that the initiation of LamB protein export requires, in addition to an intact signal sequence, information contained within the first 39 residues of the mature protein. In addition, specific localization to the outer membrane requires information residing between residues 39 and 49 of the mature protein. The latter corresponds to a region of sequence homology noted to exist among various major outer membrane proteins (Nikaido and Wu, 1984) and has been termed by Benson et al. (1984) the “outer membrane signal. ”

IV. COMPONENTS OF THE E. coli PROTEIN EXPORT MACHINERY We have described in some detail instances where the synthesis of certain hybrid proteins or export-defective proteins can have a generalized effect on normal protein export in E . coli. These results have been interpreted to indicate the existence of a specific protein export machinery in the cell whose normal function is disturbed by interaction with these abnormal proteins. Although it has been suggested that the cell does not require a specific apparatus to facilitate protein export (von Hiejne and Blomberg, 1979; Engelman and Steitz, 1981), it now appears that there are a number of proteins synthesized by the cell specifically for this purpose. Over the last several years, a number of new genetic loci have been identified that are thought to encode proteins involved in protein export. Certain of these loci were identified by obtaining mutants that are pleiotropically defective in protein export; others were identified by selecting for

136

VYTAS A. BANKAlTlS ET AL.

mutants that restore the proper localization of export-defective proteins. In most cases, these selection schemes were derived directly or indirectly from studies with gene fusions. In this section, we describe those genetic loci whose identification has emerged from these studies. Genes identified using other techniques, e.g., fep (Date and Wickner, 1981; Date, 1983) and lsp (Regue e?al., 1984), are beyond the scope of this article and will not be discussed.

A. Mutations Resulting in Pleiotropic Defects in Protein Export The isolation of MalRLac derivatives of MalS malE-facZ protein fusion strains provided a number of malE signal sequence mutations. Disappointingly, these studies failed to yield mutations unlinked to the hybrid gene that might have produced pleiotropic defects in protein export. It could well be that any mutational alteration in the export machinery sufficient to alleviate the MalS phenotype would be lethal to the cell because of a generalized defect in normal protein export. However, by altering the selection, it did prove possible to utilize protein fusion strains to obtain such unlinked mutations. An early observation concerning the class IV malE-lac2 fusion strain PB72-47 was its Lac - phenotype resulting from its abnormally low uninduced level of P-galactosidase (Bassford et al., 1979). In contrast, all the MalR derivatives of PB72-47 grown under noninducing conditions exhibited some hundredfold greater activity and were phenotypically Lac +. The low level of P-galactosidase activity of the parental fusion strain probably reflects the inability of the hybrid protein monomers to form enzymatically active tetramers when inserted into the cytoplasmic membrane at a low density. Oliver and Beckwith (1981) chose to exploit the depressed P-galactosidase activity of strain PB72-47 in an effort to isolate unlinked mutations affecting the membrane localization of the hybrid protein. It was reasoned that by selecting for Lac (i.e., strains exhibiting increased basal levels of P-galactosidase activity) one would obtain mutants in which some proportion of the hybrid protein is internalized and tetramerization is favored. It was estimated that internalization of as little as 5- 10%of the hybrid protein would be sufficient to produce a Lac+ phenotype. Although mutants producing an export-defective hybrid protein were expected in this selection, it was thought that mutants exhibiting a partial, general protein export defect might also be recovered. A number of Lac i. derivatives of strain PB72-47 were selected at 30°C. Two of the mutants tested proved to be temperature sensitive (ts) for growth and remained Mals; and each harbored a single mutation, genetically unlinked to the malE-lac2 hybrid gene, that was responsible for both the Lac+ and ts phenotypes. Further investigation revealed that, when shifted to a temperature not +

+

137

3. PROTEIN EXPORT IN E. coli

permissive for growth, these mutants accumulate in their cytoplasm the precursors for a number of exported proteins, including the periplasmic MBP, ribosebinding protein, and alkaline phosphatase, and the outer membrane LamB and OmpF proteins. A small amount of precursor accumulation can be detected when these mutants are grown at 30°C, a finding indicating a partial export defect at this permissive temperature (predicted by the selection). Interestingly, the export of certain proteins seems unaffected in these mutants at all temperatures tested. The two mutations obtained in this manner defined a new genetic locus mapping at 2.5 minutes on the E . coli chromosome that was designated secA. Oliver and Beckwith (1982a) were able to clone the secA gene and identify its product as a 92,000-dalton polypeptide. They subsequently constructed a secA-lucZ protein fusion strain that produced a hybrid protein that included a substantial amino-terminal portion of the SecA protein (Oliver and Beckwith, 1982b). It was found that antiserum raised against the purified hybrid protein precipitates the SecA protein from extracts prepared from both secA and secAts strains. In this manner, they were able to demonstrate that the SecA protein is a minor cellular protein (approximately 500- 1000 copies per cell) that appears to be loosely bound to the cytoplasmic side of the cytoplasmic membrane. Such a cellular location is certainly consistent with the SecA protein having a role in protein export. Furthermore, it was found that the synthesis of SecA protein greatly increases in a secAts mutant strain shifted to the nonpermissive temperature. A similar increase in SecA protein was observed in a MalS mulE-lacZ fusion strain induced for hybrid protein synthesis. Since both of these conditions stress the protein export capacity of the cell, it was suggested that the SecA protein, as well as other components of the cell's protein export machinery, may be regulated in response to the secretion needs of the cell. Further screening of Lac+ mutants derived from strain PB72-47 yielded a second class of unlinked mutations that result in pleiotropic defects in protein secretion. These mutations also define a new genetic locus, designated secB, located at 80.5 minutes on the E . coli linkage map (Kumamoto and Beckwith, 1983). None of the secB mutants obtained in this manner exhibit a conditionallethal phenotype and, perhaps for this reason, these strains accumulate in the cytoplasm only relatively small amounts of the various exported proteins. Also, the spectrum of exported proteins affected by secB mutations appears to be somewhat narrower than that seen for secA mutants. For example, secB mutants accumulate small but significant amounts of pre-MBP and pre-OmpF protein. However, in contrast to secA mutants, none of the secB mutants tested accumulate the precursor for ribose-binding protein. This may indicate that the secB product, which has not yet been identified, is only involved in the export of a certain subset of proteins. Alternatively, this may be a property of the individual secB alleles encountered in this selection. Kumamoto and Beckwith ( 1983) also constructed several secAtssecB double +

138

W A S A. BANKAlTlS ET AL.

mutants, using several different combinations of secArs and secB alleles. It was found that, in general, the double mutant strains grow less well at elevated temperatures than do the parental secArs secB+ strains. In such double mutant backgrounds, significantly more pre-MBP accumulates in the cytoplasm than that observed when only a single mutant allele is present. Indeed, the increased accumulation is noticeably greater than can be accounted for by an additive effect of the two mutations. This synergism in the double mutant background is even more impressive in terms of pre-OmpF protein accumulation. Here, depending on the given allelic combination, the OmpF protein can be found almost exclusively in its precursor form. The allele specificity observed in the synergistic effects of different combinations of secA and secB alleles (i.e., the fact that certain combinations of mutations result in a stronger export defect than other combinations) may be a very important observation. Such allele specificity has been used to argue for a direct interaction between two mutationally altered proteins (for example, see Jarvik and Botstein, 1973; Parkinson and Parker, 1979; Botstein and Maurer, 1982). Although the SecB protein has yet to be identified, it may well be that the SecA and SecB proteins directly interact while mediating their respective roles in protein export. Specific roles for these proteins are discussed in Section IV,C.

8. Extragenic Suppressor Mutations That Restore Export of Proteins with Defective Signal Peptides As previously discussed, the failure of the cell to properly localize proteins having a defective signal peptide may result from the inability of the cellular protein export machinery to efficiently recognize these as exported proteins. It was reasoned that it might be possible to obtain extragenic suppressor mutations that would permit a cell to properly localize these export-defective proteins. Presumably, such mutations would reside in one or more loci encoding cellular components of the protein export machinery and would alter these components in such a manner that would restore some degree of recognition to a protein with a mutant signal sequence. Beginning with an E . coli strain harboring the lamES6O signal sequence deletion mutation (see Fig. 5 ) , Emr et al. (1981) isolated a number of LamB+ pseudorevertants by selecting for mutants that could utilize maltodextrin as a carbon source. Each of the mutants obtained concomitantly reacquired sensitivity to bacteriophage A. This indicated that these mutants export at least some LamB protein to the outer membrane, despite the fact that a significant portion of the LamB protein signal peptide is missing in the parental strain. Since a deletion mutation obviously cannot revert to wild type and since such a major portion of the LamB signal peptide had been removed, it was felt that phenotypically

3. PROTEIN EXPORT IN E. coli

139

LamB revertants would most likely result from mutations unlinked to the lamB gene. This proved to be the case for each of the mutants obtained. Extragenic suppressor mutations for lamBS60 were isolated in three different genetic loci. These were designated prlA, B, and C (prl for protein localization). The prlA locus includes the great majority of suppressor mutations, and fine structure genetic mapping assigned this locus to the Pspc operon of E. coli. At the time these mutations were first described, all of the genes assigned to the P,pc operon were known to encode ribosomal proteins. Hence, the isolation of prlA mutations mapping within this operon indicated that the prlA product is a ribosomal protein. This finding was very exciting since it strongly suggested that the ribosome can play an active role in determining the cellular location of proteins. However, later results demonstrated that the prlA locus was located promoter-distal to the last known gene in the PbPCoperon. Thus, it appears that prlA represents a new gene whose product has yet to be identified (Shultz et al., 1982), although an open reading frame of 443 codons has been implicated (Cerretti ef al., 1983). It is now believed that the prlA product is not a ribosomal protein. Still, the fact that a gene encoding a protein thought to be a component of the cellular protein export machinery is included in the Pspc operon emphasizes the close relationship between protein synthesis and export (see below). The various prlA mutations isolated as suppressors of lamBS60 were also found to suppress other lamB signal sequence mutations (Emr et al., 1981). Subsequently, these same mutations were found to suppress both malE and phoA signal sequence mutations ( E m and Bassford, 1982; Michaelis et al., 1983). Suppression of the mutant phenotype is manifested by the proper localization of a greater fraction of the export-defective protein than that discerned in the corresponding prl+ strain. Within the limit of resolution of the SDS-PAGE systems employed, it appears that proteins exported as a result of prlA-mediated suppression are correctly processed. The one exception found is the lumBS60 product which is exported to the outer membrane without processing, indicating that information required for processing may have been removed from the signal peptide by this particular deletion mutation. It is important to note that suppression by prlA appears to be very export specific. There is no evidence that prlA mutations result in translational misreading, or that they exhibit any other kind of generalized suppressor activity. Rather, these mutations appear to specifically suppress only those lamB and malE mutations that alter the signal sequence in a manner such as to produce an export-defective protein. There are several additional points that should be noted regarding prlA suppressor alleles: +

1 . The various prlA alleles obtained exhibit a wide range of suppression efficiencies. Certain of the suppressor mutations are particularly efficient at suppressing signal sequence mutations. The suppression pattern exhibited by one

140

VYTAS A. BANKAlTlS ET AL.

of the stronger prlA mutations in combination with five different malE signal sequence point mutations is illustrated in Fig. 9. 2. With regard to suppression efficiency, a fair degree of allele specificity has been demonstrated in the suppression of signal sequence mutations by prlA alleles, i.e., certain prlA alleles suppress certain signal sequence mutations more efficiently than others. As previously discussed in the case of secAsecB double mutants, such allele specificity strongly implies a direct interaction between two mutant proteins. Thus, the prlA gene product may interact directly with the signal sequence to initiate protein export through the cytoplasmic membrane. 3. The demonstration that signal sequence alterations in both an outer membrane and a periplasmic protein can be compensated for by mutations at prlA further indicates that there may be common components involved in the export machinery for proteins destined for different regions of the cell envelope. 4. E . coli strains harboring prlA mutations manifest no obvious growth defects and exhibit no apparent defect in the export of wild-type LamB protein, MBP, or other envelope proteins. This might suggest that prlA mutations have not greatly altered the normal export pathway or possibly that prlA-mediated export is not accomplished via a route that is employed in these strains to any significant degree. Evidence that the prlA product is a component of the normal protein export pathway, and a possible role for this protein in the pathway, are presented below.

In comparison toprlA, relatively little has been learned concerning the remaining two classes of suppressor mutations obtained in the original selection, those

FIG. 9. Immune precipitation of MBP and pre-MBP from various prl+ and prlA4 strains. Radiolabeled MBP was precipitated from various strains and analyzed by SDS-PAGE and autoradiography, as described in E m and Bassford (1982). The relevant malE and prl genotypes are given above each lane. Suppression of the ma6E signal sequence mutation by the prlA4 allele is indicated by an increase in the ratio of mature MBP to pre-MBP precipitated from the prL44 strain when compared to theprl+ strain. Figure was originally published in Emr and Bassford (1982) and is reproduced here by permission of the authors.

141

3. PROTEIN EXPORT IN E. coli

designated prlB and prlC. The single prlB allele isolated is unusual in several respects, including the fact that it suppresses only lamB signal sequence mutations and that export of LamB protein is achieved without any apparent processing (Emr and Bassford, 1982). Furthermore, the prlB suppressor is a deletion within the gene encoding the periplasmic ribose-binding protein. It seems almost certain that suppression in this case does not result from an alteration of some component of the export machinery (Silhavy et al., 1983). Two prlC alleles have been isolated. Both exhibit some suppression of lamB and malE signal sequence mutations, and prlC-mediated export of LamB protein and MBP results in apparently correct processing ( E m and Bassford, 1982). The mutations have been mapped on the E. coli chromosome to a site between 69 and 71 minutes (Silhavy et al., 1983). Attempts are currently underway to further characterize this locus. Extragenic export-specific suppressor mutations have also been obtained by selecting for phenotypic revertants of malE signal sequence mutations. The isolation of phenotypic Ma1 revertants beginning with a strain harboring the malEA12-18 signal sequence mutation has already been described. It was found that the mutation responsible for suppression of malEA12-18 mapped outside malE in 16 revertants obtained (Bankaitis and Bassford, 1985). Furthermore, it was established that 15 of these suppressor mutations map at prlA. When the ability of these new prlA alleles to suppress various malE signal sequence mutations was examined, several were found to be demonstrably more efficient than any of the prl alleles obtained as suppressors of lamBS60. Interestingly, three of the new prlA alleles failed to exhibit detectable suppression of lamBS60. The one remaining extragenic suppressor mutation that did not map at any of the previously characterized prl loci was designated prlDl. It was found that prlD maps at 2.5 minutes on the E. coli chromosome, a location placing it very close to the secA gene. However, complementation studies indicate that prlD and secA are distinct genetic loci. Immune precipitation of pre-MBP and mature MBP from various malEprlD double mutants demonstrates that the prlDl mutation is a fairly weak suppressor of malE signal sequence mutations. A slight increase in the proportion of mature MBP precipitated is evident only in the case of malEA12-18 and two of five malE point mutations. It was also found that prlDl weakly suppresses at least one lamB signal sequence mutation. In addition, it was noted that LamB protein export in several of the lamB signal sequence mutants was actually reduced following introduction of the prlDl allele. Again, such allele specificity strongly suggests a direct interaction between the signal peptide of the MBP or LamB protein and the as-yet-unidentified prlD gene product. The kinetics of prlA and prlD-mediated MBP export have recently been investigated (Ryan et al., 1985). The appearance of mature MBP in the periplasm in strains harboring a malE signal sequence mutation and an extragenic suppressor allele is significantly slower than in a malE strain. Following a 10-second pulse +

+

142

VYTAS A. BANKAlTlS ET AL.

with [35S]methionine, virtually all of the labeled MBP precipitated is found in the precursor form. Following varying times of chase with unlabeled methionine, a slow conversion of pre-MBP to mature MBP is observed, exhibiting a half-time of 5 minutes or more, depending on the particular malE signal sequence mutation andprl suppressor allele employed. As with other experiments of this kind, all of the evidence indicates that this observation represents slow translocation of the MBP across the cytoplasmic membrane rather than simply delayed processing of the pre-MBP in the periplasm. As in the case of the various prlA alleles, the prlDl mutation has no apparent effect on export of wild-type MBP, LamB protein, or other envelope proteins. This observation, coupled with the markedly different export kinetics mentioned above, might suggest that suppression mediated by either the prlA or prlD gene product does not involve the usual protein export route. However, it was found that strains harboring both prlDl and certain prlA alleles exhibit severe growth defects. Pulse-labeling experiments have shown that such double mutant strains accumulate significant amounts of the precursor forms of several normal exported proteins specifically investigated, including the MBP, LamB protein, and OmpA protein (Bankaitis and Bassford, 1985). These results lead one to conclude that prlA and prlD do encode components of the cell’s primary protein export pathway. Apparently, the cell has difficulty accommodating two alterations in its export machinery. Furthermore, the fact that this general protein export defect is only observed with certain combinations of prlA and prlD alleles serves as another example of allele specificity. This would suggest that the prlA and prlD gene products not only specifically interact with the signal peptide but that they interact with one another as well (see the next section). Finally, Ito et al. (1983) have reported the isolation of a conditional lethal ts mutant that manifests a severe protein export defect at the nonpermissive temperature. Aware of a possible active role for ribosomes in the export process, these researchers employed localized mutagenesis to obtain ts mutants in the region of the Pspc operon of E . coli. These were then screened for mutants that accumulated preMBP at the nonpermissive temperature. One such mutant was obtained and designated secY. It has been demonstrated that this mutant, when shifted to the nonpermissive temperature, accumulates the precursor for several additional envelope proteins as well. It seems very likely that this mutation is affecting the activity of the prlA gene product, thus demonstrating an essential cellular function for the PrlA protein.

C. The Coupling of Protein Synthesis and Export Although secA is an essential gene, Oliver and Beckwith (1982b) were able to obtain chain-terminating nonsense (amber) mutations in secA by employing

3. PROTEIN EXPORT IN E. coli

143

strains harboring unlinked amber suppressor mutations. In a strain having both a mutation and a ts amber suppressor allele ( s d . ' ) , synthesis of the SecA protein is abruptly terminated when cells are shifted to the nonperniissive temperature of 42"C, and the cells eventually die. Somewhat surprisingly, it was found that at several hours after the temperature shift, at a time when total protein synthesis is as yet largely unaffected, there is a selective termination in MBP synthesis. The small amount of MBP that is synthesized under these conditions is found in its precursor form. These results have been interpreted to indicate that the SecA protein is required for both synthesis and export of the MBP. As previously mentioned, mutational alterations in the hydrophilic segment of the E . coli lipoprotein and LamB protein signal peptides can affect the efficiency of translation of these proteins. The observation that functional SecA protein is apparently required for synthesis of the MBP is further evidence that the synthesis of exported proteins may be directly coupled to and dependent upon cotranslational export. Recently, Ferro-Novick ef al. (1984) described the isolation of extragenic suppressors of a secA'" mutant that permit normal growth and protein export at 37°C. Two such suppressor mutations were found to map at 68.5 minutes on the E . coli chromosome and were designated secC. These mutations result in a cold-sensitive phenotype, even when introduced into strains that are secA . It was found that secC mutants are defective in the synthesis of exported proteins but not cytoplasmic proteins when shifted to the nonpermissive growth temperature of 23"C, an observation suggesting that the secC gene product may also somehow be tied into the E . coli protein export pathway. Interestingly, both secC mutations exhibit a high degree of allele specificity in their ability to suppress different secAfJ alleles, again strongly suggesting a direct interaction between the secA and secC gene products. The direct coupling of protein synthesis and export in E . coli is highly analogous to the situation demonstrated to exist in eukaryotic cells, where a cytosolic complex of molecular weight 250,000, composed of six distinct proteins and one 7 S RNA molecule, functions by selectively arresting translation of nascent secretory proteins upon the emergence of the signal peptide from the large ribosomal subunit (Walter and Blobel, 1980, 1981). This complex has been termed the signal recognition particle (SRP). Interaction of this translation-arrested ribosome-SRP complex with a membrane receptor, the "docking protein" (molecular weight 72,000), on the surface of the rough endoplasmic reticulum releases this block and couples further synthesis of the nascent chain to translocation across the membrane (Meyer et a/., 1982) (see article by Rapoport and Wiedmann, this volume). There is reason to believe that a similar mechanism may function in prokaryotic cells to obligately couple protein synthesis and export. The high degree of evolutionary conservation in the protein export process shown to exist between prokaryotic and eukaryotic cells has already been pointed out. Also, secAU"

+

144

W A S A. BANKAlTlS ET AL.

Muller et al. (1982) have demonstrated that the bacterial secreted protein plactamase, when translated in an in vitro system, requires SRP for translocation across mammalian endoplasmic reticulum. Silhavy et al. (1983) have speculated that a system analogous to the SRP-docking protein system operates in bacteria, and they suggested that SecA protein is the E . coli equivalent of the docking protein. Such an assignment is consistent with the known cellular location of the SecA protein on the cytoplasmic membrane. Furthermore, the observation that a xecAammfSstrain cannot synthesize MBP at the nonpermissive temperature can be explained by the inability of the SRP to interact with the membrane, thereby imposing a permanent block on the translation of exported proteins. Likewise, SecB protein, which probably directly interacts with SecA protein, may be the component of the putative bacterial SRP that interacts with the SecA protein. By the same reasoning, the SecC protein, which also may directly interact with the SecA protein, may be the component of the bacterial SRP that interacts with the ribosome to arrest translation of nascent exported proteins. Although neither the prlA nor prlD gene products have been identified, one can speculate that these two proteins also may be components of a putative E . coli SRP-equivalentthat directly interact with the signal peptide to inhibit further translation of the nascent exported protein. The genetic evidence strongly suggests that these two proteins interact with both the signal peptide and each other. However, if this is the case, then why do both prlA- and prlD-mediated protein export exhibit markedly different kinetics than wild-type proteins exported via the normal route? One possible explanation is that, although the suppressor mutations restore recognition of the defective signal peptides at a very early step in the export process, the defective signal peptide cannot function normally in later steps in the process, particularly the translocation step. Finally, if translation is directly coupled to protein export, then how is it that export-defective proteins are synthesized and can accumulate in the cytoplasm? Presumably, mutational alterations in the signal peptide that prevent protein export must also prevent interaction of the nascent chain with any SRP-like complex. Alternatively, the defect in protein export could be manifested at some point in the process after translation has been restored at the membrane surface. Two lines of evidence indicate that the former is the case: 1. Kumamoto et al. (1984) have shown that a SecAamsufSstrain that also harbors a malE signal sequence mutation continues to synthesize pre-MBP following a shift to the nonpermissive temperature. Similarly, Ferro-Novick et af. (1984) have shown that export-defective MBP continues to be synthesized following the shift of a cold-sensitive secC mutant to the nonpermissive temperature. These results clearly indicate that synthesis of export-defective MBP is not subject to the block in translation that is exerted on the wild-type MBP. 2. Randall and Hardy (1977) demonstrated that the MBP and other envelope proteins of E . coli are synthesized on membrane-bound polysomes. It was re-

145

3. PROTEIN EXPORT IN E. coli

cently demonstrated that export-defective MBP and alkaline phosphatase are primarily synthesized on free polysomes in the cytoplasm (Rasmussen and Bassford, 1985). This result indicates that the site of synthesis of an exported protein in E. coli is determined by the export competence of the signal peptide. Furthermore, this result provides strong support for our current concept of the signal hypothesis. The failure of the nascent chain to interact with SRP would be predicted to prevent association of the polysome with the membrane and to allow for uninterrupted synthesis of the protein in the cytoplasm.

V.

SUMMARY

The genetic studies described in this article were initiated several years ago in Jon Beckwith’s laboratory. The original aim was to determine whether or not one could export a cytoplasmic protein, P-galactosidase, to the periplasm or the outer membrane of E . coli by attaching to it an amino terminus derived from the MBP or LamB protein. Except in the case of class IV lamB-lac2 fusions in which the hybrid protein is efficiently incorporated into the outer membrane, this simple premise proved incorrect. However, it was quite clear that export of certain classes of malE-lacZ and lamB-lacZ hybrid proteins is initiated, subsequently allowing the use of such fusions to identify regions within the MBP and the LamB protein that have a role in the export process. Furthermore, the unusual properties of particular fusion strains provided selections for both signal sequence mutations and unlinked mutations with pleiotropic effects on protein export. The signal sequence mutations in turn have permitted the isolation of both intragenic and extragenic suppressor mutations that have yielded additional insight into the protein export process. The secC locus was identified as an extragenic suppressor of a secArJ mutation. Most recently, we have found that certain combinations of extragenic prl suppressor mutations in the same strain can result in strains with severe growth defects. We anticipate that the isolation and characterization of phenotypically “healthy” revertants, currently underway in our laboratory, should provide additional useful information. Certainly, most of the “fallout” from the fusion strains was unforeseen when the initial experiments were planned, but the experience further illustrates the utility of genetic approaches in advancing our understanding of complex biological phenomena. The major findings obtained from the studies described in this chapter can be briefly summarized: 1. The isolation and characterization of signal sequence mutations in both the malE and lamB genes provided the genetic proof for the proposed essential role of this structure in the initial steps of protein secretion. From an analysis of such mutations, and of intragenic suppressors of these mutations, it is apparent that the absolute length of the hydrophobic core is a major determinant of signal

146

VYTAS A. BANKAITIS ET AL.

peptide function. In addition, critical subsets of residues in the MBP and LamB protein signal peptides have been identified, although the precise function of these residues remains unknown. 2. The evidence strongly suggests that several regions within the mature MBP and LamB protein have an important role in protein export. For both of these proteins, there appears to be a site early in the mature protein that is essential for export, and a second site near the center of the polypeptide chain that is required for efficient localization. Also, it appears that information in the mature protein determines the ultimate cellular destination of the exported protein, be it periplasm or outer membrane. 3. Clearly, it is not sufficient to attach an exported protein to a nonexported protein to assure its export from the cytoplasm. Experimental evidence strongly indicates that there are multiple regions within the P-galactosidase polypeptide that are not compatible with passage through the cytoplasmic membrane. 4. Several genes encoding components of the cellular protein export machinery have been identified. Although the protein product of just one of these (secA) has been identified, one can suggest functions for the various gene products from the genetic evidence and our understanding of protein export in eukaryotic cells. Also, it appears that synthesis of these proteins is regulated in response to the secretion requirements of the cell. 5. Finally, it appears that, under normal circumstances, the synthesis of an exported protein may be obligately coupled to its export. Thus, in this and other respects, protein export in E . coli and eukaryotic cells share many similarities. ACKNOWLEDGMENTS We thank our colleagues in the laboratories of Jon Beckwith and Tom Silhavy for preprints, and we thank Drs. Paul Ray and Spencer Benson for critical reading of the manuscript. Also, many thanks to Ms. Shirley Alston for typing the manuscript. Work in the authors’ laboratory has been supported by Grant A1 17292 from the National Institute of Allergy and Infectious Diseases. REFERENCES Bankaitis, V. A., and Bassford, P. J., Jr. (1984). The synthesis of export-defective proteins can interfere with normal protein export in Esrherichia coli. J . B i d . Chem. 259, 12193-12200. Bankaitis, V . A., and Bassford, Jr., P. J. (1985). A proper interaction between at least two components is required for efficient export of proteins to the Escherichia coli cell envelope. J . Bacteriol. 161, 169- 178. Bankaitis, V. A., Rasmussen, B. A., and Bassford, P. J., Jr. (1984).lntragenic suppressor mutations that restore export of maltose binding protein with a truncated signal peptide. CeN37,243-252. Bassford, P . , and Beckwith, J. (1979). Escherichia coli mutants accumulating the precursor of a secreted protein in the cytoplasm. Nature (London) 277, 538-541. Bassford, P. J., Jr., Silhavy, T. J., and Beckwith, J . R. (1979). Use of gene fusions to study secretion of maltose-binding protein into Escherichia coli periplasm. J . Bacteriol. 139, 19-3 1. Bedouelle, H., and Hofnung, M. (1981).Functional implications of secondary structure and analysis

3. PROTEIN EXPORT IN E. coli

147

of wild-type and mutant bacterial signal peptides. In “Membranes Transport and Neuroreceptors” (D. Oxender, ed.), pp. 309-343. Liss, New York. Bedouelle, H., Bassford, P. J., Jr., Fowler, A. V., Zabin, I . , Beckwith, J., and Hofnung, M. (1980). Mutations which alter the functions of the signal sequence of the maltose binding protein of Escherichia coli. Nature (London) 285, 78-81. Benson, S. A,, and Silhavy, T. J. (1983). Information within the mature LamB protein necessary for localization to the outer membrane of E. coli K-12. Cell 32, 1325-1335. Benson, S. A,, Bremer, E., and Silhavy, T. J. (1984). Intragenic regions required for LamB export. Proc. Natl. Acad. Sci. U.S.A. 81, 3830-3834. Botstein, D., and Maurer, R. (1982). Genetic approaches to the analysis of microbial development. Annu. Rev. Gene/. 16, 61-84. Brickman, E., Silhavy, T. J., Bassford, P. J., Jr., Shuman, H. A., andBeckwith, J. R. (1979). Sites within gene lacZ of Escherichia coli for formation of active hybrid P-galactosidase molecules. J . Bacteriol. 139, 13-18. Casadaban, M. J. (1976). Transposition and fusion of the lac operon to selected promoters in E . coli using bacteriophage A and F . J. Mol. Biol. 104, 541-555. Casadaban, M. J., Chou, J . , and Cohen, S. N. (1980). In vitro gene fusions that join an enzymatically active P-galactosidase segment to amino-terminal fragments of exogenous proteins: Escherichia coli plasmid vectors for the detection and cloning of translational initiation signals. J. Bacteriol. 143, 971-980. Cerretti, D. P., Dean, D., Davis, G. R., Bedwell, D. M., and Nomura, M. (1983). The spc ribosomal protein operon of Escherichia coli; Sequence and cotranscription of the ribosomal protein genes and a protein export gene. Nucleic Acids Res. 11, 2599-2616. Chou, P. Y., and Fasman, G. D. (1978). Empirical predictions of protein conformation. Annu. Rev. Biochem. 47, 251-276. Date, T. (1983). Demonstration by a novel genetic technique that leader peptidase is an essential enzyme of Escherichia coli. J. Bacteriol. 154, 76-83. Date, T., and Wickner, W. (1981). Isolation of the Escherichia coli leader peptidase gene and effects of leader peptidase overproduction in viwo. Proc. Narl. Acad. Sci. U.S.A. 78, 6106-6110. Dodd, D. C., Bassford, P. J., Jr., and Eisenstein, B. 1. (1984). Dependence of secretion and assembly of type 1 fimbrial subunits of Escherichia coli on normal protein export. J . Bacteriol. 159, 1077-1079. E m , S. D., and Bassford, P. J., Jr. (1982). Localization and processing of outer membrane and periplasmic proteins in Escherichia coli strains harboring export-specific suppressor mutations. J . Biol. Chem. 257, 5852-5860. Emr, S. D., and Silhavy, T. I . (1980). Mutations affecting localization of an Escherichia coli outer membrane protein, the bacteriophage h receptor. J. Mol. Biol. 141, 63-90. E m , S. D., and Silhavy, T. J. (1982). Molecular components of the signal sequence that function in the initiation of protein export. J . CeN Biol. 95, 689-696. Emr, S. D., and Silhavy, T. J. (1983). Importance of secondary structure in the signal sequence for protein secretion. Proc. Narl. Acad. Sci. U.S.A. 80, 4599-4603. Emr, S. D., Hall, M. N., and Silhavy, T. J. (1980a). A mechanism of protein localization: The signal hypothesis and bacteria. J. Cell Biol. 86, 701-71 1. Emr, S. D., Hedgepeth, J., Clement, J.-M., Silhavy, T. J . , and Hofnung, M. (1980b). Sequence analysis of mutations that prevent export of phage lambda receptor, an Escherichia coli outer membrane protein. Narure (London) 285, 82-85. Em, S. D., Hanley-Way, S . , and Silhavy, T. J. (1981). Suppressor mutations that restore export of a protein with a defective signal sequence. Cell 23, 79-88. Engelman, D. M., and Steitz, T . A. (1981). The spontaneous insertion of proteins into and across membranes: The helical hairpin hypothesis. Cell 23, 41 1-422.

148

W A S A. BANKAlTlS ET AL.

Ferro-Novick, S., Honma, M., and Beckwith, J. (1984). The product of geneSecC is involved in the synthesis of exported proteins in E. coli. Cell 38, 211-217. Fraser, T. H., and Bruce, B. 1. (1978). Chicken ovalbumin is synthesized and secreted by Escherichiu coli. Proc. Nutl. Acud. Sci. U.S.A. 75, 5936-5940. Guarente, L.,Lauer, G., Roberts, T. M., and Ptashne, M. (1980). Improved methods for maximizing expression of a cloned gene: A bacterium that synthesizes rabbit P-globin. Cell 20, 543553. Hall, M. N., Gabay, J., Debarbouille, M., and Schwartz, M. (1982a). A role for mRNA secondary structure in the control of translation initiation. Nature (London) 295, 616-618. Hall, M. N., Schwartz, M., and Silhavy, T. J. (1982b). Sequence information within the lumB gene is required for proper routing of the bacteriophage A receptor protein to the outer membrane of Escherichiu coli K-12. J . Mol. Biol. 156, 93-112. Hall, M. N., Gabay, J., and Schwartz, M. (1983). Evidence for a coupling of synthesis and export of an outer membrane protein in Escherichiu coli. EMBO J . 2, 15-19. Hengge, R., and Boos, W. (1983). Maltose and lactose transport in Escherichiu coli: Examples of two different types of concentrative transport systems. Biochim. Biophys. Acru 737, 443478. Herrero, E., Jackson, M., Bassford, P. J., Sinden, D., and Holland, I. B. (1982). Insertion of a MaIE P-galactosidase fusion protein into the envelope of Escherichiu coli disrupts biogenesis of outer membrane proteins and processing of inner membrane proteins. J . Bucteriol. 152, 133139. Inouye, S., Wang, S., Sekizawa, J., Halegoua, S., and Inouye, M. (1977). Amino acid sequence for the peptide extension on the prolipoprotein of the Escherichiu coli outer membrane. Proc. Nutl. Acud. Sci. U.S.A. 74, 1004-1008. Inouye, S., Soberon, X., Franceschini, T., Nakamura, K., Itakura, K., and Inouye, M. (1982). Role of positive charge on the amino-terminal region of the signal peptide in protein secretion across the membrane. Proc. Nutl. Acud. Sci. U.S.A. 79, 3438-3441. Ito, K. (1982). Purification of the precursor form of maltose-binding protein, a periplasmic protein of Escherichiu coli. J . Biol. Chem. 257, 9895-9897. Ito, K., and Beckwith, J. R. (1981). The role of the mature protein sequence of maltose binding protein in its secretion across the E. coli cytoplasmic membrane. Cell 25, 143-150. Ito, K., Bassford, P. J., Jr., and Beckwith, J. (1981). Protein localization in E. coli: Is there a common step in the secretion of periplasmic and outer membrane proteins? Cell 24, 707-717. Ito, K., Wittekind, M., Nemura, M., Shiba, K., Yura, T., Miura, A., and Nashimoto, H. (1983). A temperature-sensitive mutant of E. coli exhibiting slow processing of exported proteins. Cell 32, 789-797. Jarvik, J., and Botstein, D. (1973). A genetic method for determining the order of events in a biological pathway. Proc. Nutl. Acud. Sci. U.S.A. 70, 2046-2050. Josefsson, L.-G., and Randall, L. L. (1981). Different exported proteins in E. coli show differences in the temporal mode of processing in vivo. Cell 25, 151-157. Kumamoto, C., and Beckwith, J. (1983). Mutations in a new gene, secB, cause defective protein localization in Escherichiu coli. J. Bucteriol. 154, 253-260. Kumamoto, C. A,, Oliver, D. B., and Beckwith, J. (1984). Signal sequence mutations disrupt feedback between secretion of an exported protein and its synthesis in Escherichiu coli. Nature (London) 308, 863-864. Meyer, D. I., Krause, E., and Dobberstein, B. (1982). Secretory protein translocation across membranes-the role of the “docking protein.” Nature (London) 297, 647-650. Michaelis, S., and Beckwith, J. (1982). Mechanism of incorporation of cell envelope proteins in Escherichiu coli. Annu. Rev. Microbiol. 36, 435-464. Michaelis, S., Inouye, H., Oliver, D., and Beckwith, J. (1983). Mutations that alter the signal sequence of alkaline phosphatase in Escherichiu coli. J. Bucteriol. 154, 366-374.

3. PROTEIN EXPORT IN E. coli

149

Moreno, F., Fowler. A. V . , Hall, M., Silhavy, T. J., Zabin, I., and Schwartz, M. (1980). A signal sequence is not sufficient to lead P-galactosidase out of the cytoplasm. Nature (London) 286, 356-359. Muller, M . , Ibrahimi, I . , Chang, C. N., Walter, P., and Blobel, G. (1982). A bacterial secretory protein requires signal recognition particle for translocation across mammalian endoplasmic reticulum. J. Eiol. Chem. 257, I 1860- I 1863. Neu, H. C., and Heppel, L. A. (1965). The release of enzymes from Escherichiu coli by osmotic shock and during the formation of spheroplasts. J . B i d . Chem. 240, 3685-3692. Nikaido, H.,and Wu, H. C. P. (1984). Amino acid sequence homology among the major outer membrane proteins of Escherichiu coli. Proc. Narl. Acad. Sci. U . S . A . 81, 1048-1052. Oliver, D. B., and Beckwith, J. (1981). E. coli mutant pleiotropically defective in the export of secreted proteins. Cell 25, 765-772. Oliver, D. B., and Beckwith, J. (19824). Identification of a new gene (secA) and gene product involved in the secretion of envelope proteins in Escherichiu coli.J. Eacteriol. 150, 686-69 I . Oliver, D. B., and Beckwith, J. (1982b). Regulation of a membrane component required for protein secretion in Escherichiu coli. Cell 30, 3 1 1-3 19. Parkinson, J. S . , and Parker, S. R. (1979). Interaction of the cheC and cheZ gene products is required for chemotactic behavior in Eschen’chia coli. Proc. Nufl. Acad. Sci. U.S.A. 76, 2390-2394. Perlman, D., and Halvorson, H. 0. (1983).A putative signal peptidase recognition site and sequence in eukaryotic and prokaryotic signal peptides. J. Mu/. Eiol. 167, 391-409. Randall, L. L. (1983). Translocation of domains of nascent periplasmic proteins across the cytoplasmic membrane is independent of elongation. Cell 33, 23 1-240. Randall, L. L., and Hardy, S. J. S . (1977). Synthesis of exported proteins by membrane-bound polysomes from Escherichia coli. Eur. J . Biochem. 75, 43-53. Randall, L. L., and Hardy, S. J. S. (1984). Export of protein in bacteria: Dogma and data. In “Modern Cell Biology” (B. Satir, ed.), Vol. 3. pp. 1-20. Liss, New York. Randall-Hazelbauer, L., and M. Schwartz (1973). Isolation of the phage lambda receptor from Escherichiu coli K-12. J . Bacteriol. 116, 1436-1446. Rasmussen, B. A , , and Bassford, Jr., P. 1. (1985). Both linked and unlinked mutations can alter the intracellular site of synthesis of exported proteins of Escherichiu coli. J. Bacferiol. 161, 258264. Rasmussen, B. A,, Bankaitis, V. A , , and Bassford, P. J., Jr. (1984). Export andprocessingof MalELac2 hybrid proteins in Escherichia coli. J. Bacteriol. 160, 612-617. Regue, M., Remenick, J., Tokunaga, M., Mackie, G. A,, and Wu, H. C. (1984). Mapping of the lipoprotein signal peptidase gene ([up).J. Bucteriol. 158, 632-635. Ryan, J . P., Ray, P., and Bassford, P. I . , Jr. (1985). In preparation. Shultz, J., Silhavy, T. J., Berman, M. L., Fiil, N., and Emr, S. D. (1982). A previously unidentified gene in the spc operon of Escherichia coli K-12 specifies a component of the protein export machinery. Cell 31, 227-235. Silhavy, T. J . , and Beckwith, J. (1983). Isolation and characterization of mutants ofEscherichia coli K-12 affected in protein localization. I n “Methods in Enzymology” (S. Fleischer and B. Fleischer, eds.), Vol. 97, pp. 11-40, Academic Press, New York. Silhavy, T. J., Bassford, P. J., Jr., and Beckwith, J. R. (1979). A genetic approach to the study of protein localization in Escherichiu culi. In “Bacterial Outer Membranes: Biogenesis and Functions” (M. lnouye, ed.), pp. 203-254. Wiley, New York. Silhavy, T. J., Benson, S. A,, and Emr, S. D. (1983). Mechanisms of protein localization. Microb i d . Rev. 47, 313-344. Smith, W . P. (1980). Cotranslational secretion of diphtheria toxin and alkaline phosphatase in vifro: Involvement of membrane protein(s). J. Bacferiol. 141, 1142-1 147. Szmelcman, S . , and Hofnung, M. (1975). Maltose transport in Escherichiu coli K-12: Involvement of the bacteriophage lambda receptor. J. Eucteriol. 124, 1 12- 1 18.

150

V M A S A. BANKAlTlS ET AL.

Talmadge, K., Kaufman, J., and Gilbert, W. (1980). Bacteria mature preproinsulin to proinsulin. Proc. Natl. Acad. Sci. U.S.A. 77, 3988-3992. Tokunaga, M., Tokunaga, H., and Wu, H. C. (1982). Post-translational modification and processing of Escherichia coli prolipoprotein in vitro. Proc. Natl. Acad. Sci. U.S.A. 79, 2255-2259. von Heijne, G., and Blomberg, C. (1979). Transmembrane translocation of proteins. Eur. J. Biochem. 97, 175-181. Walter, P., and Blobel, G. (1980). Purification of a membrane-associated protein complex required for protein translocation across the endoplasmic reticulum. Proc. Nutl. Acad. Sci. U.S.A. 77, 7112-7116.

Walter, P., and Blobel, G. (1981). Translocation of proteins across the endoplasmic reticulum 111. Signal recognition particle (SRP) causes signal sequence-dependent and site-specific arrest of chain elongation that is released by microsomal membranes. J . Cell B i d . 91, 557-561. Weinstock, G. M., Berman, M. L., and Silhavy, T. J. (1983). Chimeric genetics with P-galactosidase. In “Gene Amplification and Analysis” (T.S. Papas, M. Rosenberg, and J. G. Chirikjain, eds.), Vol. 3, pp. 28-64. Elsevier, Amsterdam.

CURRENT TOPICS IN MEMBRANES AND TRANSPORT. VOLUME 24

Chapter 4

Structural and Thermodynamic Aspects of the Transfer of Proteins into and across Membranes GUNNAR VON HElJNE Research Group for Theoretical Biophysics Department of Theoretical Physics Royal Institute of Technology Stockholm. Sweden

1. introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............. 11. How Hydrophobic Is Hydrophobic’? . . . . . . . . . . . . . . . . . . . . . . . . . . ............. A. The Hydrophobicity Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............. B. Scales of Hydrophobicity.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............. C. Hydrophobicity Scales and Membrane Proteins . . . . . . . . . . . . . . . . . . . . . . . . . . 111. The Signal Sequence: A Sequence of Signals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. What a Signal Sequence D o e s . . . . . . . . . . . . . . . . . . . . . . . . . . . ............. ............. B. What a Signal Sequence Looks Like. . . . . . . . . . . . . . . . . . . . . . C. What Experimental Manipulations Can Do to Signal Seque IV. The Transmembrane Segment: Making Friends wit ..................... A. Sequence and Composition. . . . . . . . . . . . . . . . .... B. Conformation within the Membrane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V . Protein Export: Rules of the Game . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Models of Protein Export. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. “Something Is Rotten in the State of Denmark”. . . . . . . . . . . . . . . . . . . . . . . . . . VI . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

151 152

152 153 156

157 157 158

163 167 169 169 170 173 174

I. INTRODUCTION How do proteins pass through or integrate into membranes? The “signal hypothesis,” i.e., the hypothesis that proteins destined for export or for the cell membrane are synthesized with a transient N-terminal extension that initiates a process of cotranslational translocation across either the cytoplasmic membrane 151

Copynght 0 1985 by Academic Pres,. In‘ All nghh of rcproduclion In any form reserved ISBN 0-12-153324 7

152

GUNNAR VON HEIJNE

(in prokaryotes) or the endoplasmic reticulum (in eukaryotes) (Blobel and Dobberstein, 1975), implies that transmembrane and wholly exported proteins share the same basic export machinery. Thus, understanding the biogenesis of membrane proteins requires an understanding of protein export at large. Experimental results with bearing on the problem of protein export are pouring out of many laboratories at a high rate (for recent reviews, see Kreil, 1981; Strauss and Boime, 1982; Michaelis and Beckwith, 1982). In the midst of this wealth of experimental data, work of a more theoretical nature can sometimes yield new insight and provide inspiration for the future. Thus, in this article I shall concentrate on information gleaned from the ever-accumulating amino acid sequence data, both in terms of sequence as such, and in terms of hydrophobicity-this elusive catchword that crops up in almost any discussion of protein structure and protein-membrane interactions. In the first section, I review the knotty hydrophobicity concept and the various hydrophobicity scales that circulate in the literature. The two subsequent sections deal with signal sequences and transmembrane segments, thus preparing the ground for a discussion of global models of protein export and membrane protein biogenesis in Section V .

II. HOW HYDROPHOBIC IS HYDROPHOBIC?

A. The Hydrophobicity Concept The concept of a “hydrophobic effect,” i.e., a tendency for nonpolar molecules or parts of molecules in aqueous solution to aggregate to reduce the nonpolar surface area exposed to water, has an immediate intuitive appeal, and it has been a central idea in many attempts to come to grips with the thermodynamics of protein structure ever since Kauzmann (1959) coined the term “hydrophobic bond. ’ ’ Beyond this heuristic and qualitative level, however, giving a precise definition of “hydrophobicity” has been no easy task. Clearly, it must be related to quantum mechanical properties of both solute and solvent, e.g., dipole-dipole interactions and hydrogen bonding, as well as to entropic effects stemming mainly from the relatively well-ordered water layer around a nonpolar surface (Finney et al., 1980)-all difficult problems in their own right. Thus, “hydrophobicity” is a complex concept, and there is no unambiguous way to amve at a numerical scale expressing the hydrophobicity of various molecules, such as amino acid residues in proteins. Two main avenues have been tried: (1) the empirical, experimental approach, whereby one calculates the Gibbs’ free energy of transfer of an amino acid residue between water and some nonpolar phase from a measurement of the distribution coefficient between the

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

153

two phases; and (2) various statistical approaches whereby one observes the statistical propensity of a given type of residue to be either buried inside or exposed on the surface of proteins with known X-ray structures. These two kinds of hydrophobicity scales will now be dealt with in turn.

B. Scales of Hydrophobicity 1. EMPIRICAL SCALES The seminal paper in this field is by Nozaki and Tanford (197 1). They measured the solubility of a series of amino acids in water and ethanol or dioxane and calculated the free energy of transfer from the respective solubilities. To arrive at a scale useful when dealing not with isolated amino acids but with whole proteins, they subtracted the free energy of transfer for glycine, giving a hydrophobicity scale for transfer of amino acid side chains from organic solvent to water (Table I). As a result of experimental difficulties, no values were reported for the charged residues. Segrest and Feldmann ( 1974) later renormalized the Nozaki-Tanford scale (with zero and unity defined by glycyl and alanyl, respectively), yielding the widely used “hydrophobicity index” scale. A second variation on the NozakiTanford scale has been presented by Jones (1975). Bull and Breese (1974) took a somewhat different approach. They studied the variation in surface tension with amino acid concentration and converted this into a distribution coefficient of the amino acid between the solution and the surface and then into a free energy of transfer from solution to surface. Subtracting the value for glycine, they obtained a scale which correlates well with the NozakiTanford scale. Finally, Wolfenden et al. (1979) have measured the distribution coefficient of amino acid side chains alone (i.e., hydrogen to represent Gly, isobutane to represent Leu, etc.) between water and vapor phases. Residues charged at pH 7 were rendered neutral by measuring in an appropriate buffer, and the values were later corrected to pertain to pH 7. The results were presented in the form of a “hydration potential” scale giving the free energies of transfer of amino acid side chains from the vapor phase to neutral aqueous solution.

2. STATISTICAL SCALES With the ever-increasing number of solved protein structures, statistical studies of the local environment of individual amino acid residues have become possible. Also, hybrid schemes relating the so-called accessible surface area (calculated from X-ray structures, see below) to one or another measure of hydrophobicity have found extensive use.

TABLE I HYDROPHOBICITY SCALES~ ~~

Residue Ala CYS

ASP Glu Phe GlY His Ile LYS

Leu Met Asn pro Gln

Nozaki

~

Segrest

0.5 -

1.o 0.0

2.5 0.0 0.5 2.6

0.0

5.0 1.O

5.0 -

1.8 1.3

-

3.5 2.5 -1.5 1.5 -1.0

-0.3 0.4 1.5 3.4 2.3

-0.5 0.5 3.0 6.5 4.5

’4%

Ser Thr Val Trp

TYr

~~~~~

~

~

~

~~~~~~~

Jones

Bull

Wolfenden

Manavalan

Janin

Robson

Argos

von Heijne

0.9 1.5 0.7 0.7 2.9 0.1 0.9 3.2 1.6 2.2 1.7 0.1 2.8 0.0 0.9 0.1 0.1 1.9 3.8 2.7

-0.2 -0.5 -0.2 -0.3 -2.3 0.0 -0.1 -2.3 -0.4 -2.5 -1.5 0.1 -1.0 0.2 -0.1 -0.4 -0.5 -1.6 -2.0 -2.2

1.9 -1.2 -10.9 -10.2 -0.8 2.4 - 10.2 2.2 -9.5 2.3 -1.5 -9.7 -9.4

13.0 14.6 10.9 11.9 14.0 12.4 12.2 15.7 11.4 14.9 14.4 11.4 11.4 11.8 11.7 11.2 11.7 15.7 13.9 13.4

0.3 0.9 -0.6 -0.7 0.5 0.3 -0.1 0.7 -1.8 0.5 0.4 -0.5 -0.3 -0.7 - 1.4 -0.1 -0.2 0.6 0.3 -0.4

-1.0

1.6 1.2 0.1 0.2 2.0 0.6 0.3 1.7 0.2 2.9 3.0 0.3 0.8 0.5 0.5 0.8 0.9 1.1 1.1 0.7

-1.0

-5.1 -4.9 2.0 -5.9 -6. I

2.1 -1.2 -0.7 2.8 0.0 1.1 4.0 -0.9 2.0 1.8 -0.7 0.4 -0.1 0.3 -1.2 -0.5 1.4 3 .O 2.1

-1.5 7.4 5.9 -3.4 0.0 3.4 -2.5 4.2 -2.4 -2.7 2.9 3.3 2.4 11.3 1.5 0.9 -2.0 -2.0 1.1

” Various hydrophobicity scales, see text. The Nozaki, Bull, Wolfenden, Manavalan, Janin, Robson, and von Heijne scales are in kcaI/mol; the others are dimensionless.

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

155

Applying Jones’s scale to a sample of known protein structures, Manavalan and Ponnuswamy (1978) defined the “surrounding hydrophobicity” of a given residue as the sum of the hydrophobicities of all residues within an 8-A radius sphere around the residue in question. Averaging over all residues of this type in the sample, they came up with an “average surrounding hydrophobicity” scale that relates not to the intrinsic hydrophobicity of the residue itself but to the hydrophobicity of its average surrounding in the protein sample. Clearly, the correlation between this scale and Jones’s is not very good. X-ray structures have also been used to obtain statistical “partition coefficients” for amino acids between the predominantly nonpolar interior and the more polar surface of proteins (Janin, 1979). Calculating the “accessible surface area” (i.e., the surface area exposed to solvent) for each residue in a sample of proteins and defining as “buried” all residues with an accessible surface area of less than 20 A*, a hydrophobicity scale was constructed by defining the free energy of transfer from the inside to the outside of a protein as AC = RT Inf, wherefis the ratio of the number of buried to the number of accessible residues of the given type. This scale also does not correlate particularly well with the Nozaki-Tanford scale. A similar scale, based on an analysis in terms of buried and exposed residues, has been presented by Robson and Osguthorpe (1979). As a final variation on this theme, with particular relevance to the study of membrane-bound proteins, consider the recent work by Argos er al. (1982b). Using a data base consisting of those segments from membrane-bound proteins thought to bind to and/or span the membrane, these workers calculated “membrane-buried preference parameters” by dividing the fraction of a given type of residue in the membrane-bound sample by the fraction observed in a sample of soluble proteins. The values were not converted into free energies, but they were claimed to be useful in predicting membrane-buried parts of the protein from sequence data. One or more of these scales have also been adjusted, averaged, or otherwise molded to yield scales with an optimal predictive value in a given context, e.g., in attempts to find antigenic determinants (i.e., exposed parts), interior portions, or membrane-spanning segments in a given protein sequence (Hopp and Woods, 1981; Kyte and Doolittle, 1982). As noted above, “hybrid schemes” also exist where the hydrophobicity of a given type of residue is calculated from its accessible surface area. This was pioneered by Chothia (1974), who observed a close correlation between the Nozaki-Tanford hydrophobicity scale and the accessible surface areas calculated for the various amino acids when put in the tripeptide Ala-X-Ala. Each square angstrom of accessible surface area was found to correspond to 24 cal/mol of hydrophobic free energy. Polar groups forming hydrogen bonds to water were

156

GUNNAR VON HEIJNE

about 1 kcal/mol less hydrophobic than nonpolar ones, but with this correction the relation 1 A2 = 24 cal/mol was still valid. This relation, with a gain in hydrophobic free energy of some 25 cal/mol per A2 of surface area removed from contact with water, has been widely used in many subsequent studies of the thermodynamics of protein folding. In particular, and bringing us closer to the field of protein-membrane interactions, it was used by von Heijne and Blomberg (1979) and von Heijne (1981a) to construct a hydrophobicity scale giving the free energy difference for the transfer of a residue originally in a helix in water to a helix in a nonpolar phase lacking hydrogen-bonding capacity. [Recently, it has been suggested that the values for Ser and Thr should be reduced by 2.5 kcal/mol to - 1.O and - 1.6 kcal/mol, respectively, since these side chains can make hydrogen bonds to the main chain backbone (Capaldi et al., 1983).] Independently, a very similar line of reasoning led Engelman and Steitz (1981) to propose almost exactly the same scale. Before leaving the subject, a word of caution is warranted, though. Chothia’s relation, although extensively employed, is fraught with difficulties, and it cannot be said to be well understood theoretically. For good discussions and criticisms, see Lesk and Chothia (1980) and Tanford (1979). The belief in a single hydrophobicity concept, valid under all circumstances, has also been questioned (Charton and Charton, 1982).

C. Hydrophobicity Scales and Membrane Proteins It is not at once obvious which scales to prefer when dealing with proteinmembrane interactions. Recently, however, Argos et al. (1982b) conducted an extensive investigation to find out which scales can be used most profitably to predict membrane-buried segments of membrane-bound proteins. Bacteriorhodopsin,with its seven well-defined membrane-spanning helices, was used as a model case. It was found that the “hydration potential” (Wolfenden et al., 1979) and the “buried transfer free energy” (von Heijne, 1981a) were the best hydrophobicity measures for membrane-buried amino acids, the latter scale having been explicitly designed to pertain to residues in a membranous environment. At this point, it may also be noted that there is some confusion as to how hydrophobicity scales should be used to locate likely membrane-spanning segments. Unfortunately, one of the most widely used methods, that of Kyte and Doolittle (1982), uses a seven-residue moving average to calculate a “hydropathy profile” for the protein, with peaks over a certain height being predicted as good candidates for transmembrane segments. In my opinion, however, it seems more reasonable to average over a length of chain that corresponds to the dimension of the structure looked for; i.e., if one is searching for membrane-spanning segments that have a typical length of about 20 residues (see below), one should

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

157

use a 20-residue moving average. If, on the other hand, one is trying to find antigenic determinants, typically located in exposed bends, averaging over five to seven residues seems more appropriate. Adjusting the averaging-length to the structure one wants to see will serve to reduce the noise level and make identification easier. In conclusion, it seems clear that, since most of the hydrophobicity scales in the literature agree in broad terms (be they empirical or statistical), most scales will yield similar results in any particular application. For obvious reasons, the scale used in the remainder of this chapter will be the author’s own “buried transfer free energy ”-some a posteriori justification for this choice being found in the work of Argos et al. (1982b).

111. THE SIGNAL SEQUENCE: A SEQUENCE OF SIGNALS A. What a Signal Sequence Does Early in the 1970s, Milstein ct al. (1972) noticed that immunoglobulin light chain made in vitro was longer by about 15 amino acids at the amino terminus than the form isolated in vivo. This immediately made sense in terms of a model for protein export put forward by Blobel and Sabatini (1971), in which it had been postulated that the information required to initiate export of the chain resides in the amino-terminal region. A few years later, the more elaborate signal hypothesis was presented (Blobel and Dobberstein, 1975), stating in effect that an amino-terminal signal sequence guides the ribosome to an export site on the membrane. Since then, and particularly after the advent of recombinant DNA techniques, a large number of signal sequences have been sequenced (see Section III,B), and their mode of action has been progressively clarified. So far, two basic functions have been clearly assigned to these sequences. First, they initiate export, possibly by binding to the so-called signal recognition particle (Walter and Blobel, 1982) (a protein-RNA complex found only in eukaryotes so far, but see Kumamoto et al., 1984, for indications of a prokaryotic counterpart), which halts translation on cytoplasmic ribosomes programmed to synthesize exported proteins until an unoccupied export site is found. Second, the signal sequence also guides its own proper removal from the mature chain, although this step does not seem to be an obligatory requirement for export (Lin et al., 1978). In the remainder of this section, the structural properties of signal sequences are reviewed with these basic functions in mind.

158

GUNNAR VON HEIJNE

B. What a Signal Sequence Looks Like 1. THE CLEAVAGE SITE Almost from the outset it was recognized that signal sequences were surprisingly variable in amino acid sequence. Grossly, they were shown often to have a positively charged N-terminus, followed by a stretch of uncharged, mostly hydrophobic residues and ending in a residue with a short side-chain such as Ala, Ser, Gly, and Thr. Conceivably, export initiation may be accomplished through some more or less unspecific hydrophobic interaction between the signal sequence, the signal recognition particle, and the membrane, but the highly precise cleavage reaction involved in removing the signal sequence from the mature protein has been harder to reconcile with the high sequence variability. Only recently, when the number of known sequences has grown sufficiently to allow for statistically meaningful studies, has it become possible to shed some light on this problem. Austen and Ridd (1981), using a sample of 38 sequences, noticed that Ala is abundant not only at the cleavage site (position - 1, cf. Fig. 1) but also in position -3, where, in addition, charged residues seemed to be absent. In a more thorough study (von Heijne, 1983), based on a sample of 78 eukaryotic signal sequences, it was shown that the whole region between positions -5 and 1 is subject to strong selection (Table 11). Small, neutral residues are abundant in positions -1 and -3 but are rare in -2. Conversely, aromatic residues, charged residues, and large, polar residues are virtually absent from positions - 1 and -3 and are relatively abundant in -2. Pro is absent from positions -3 to + 1 but is quite common in - 5 ; and Gly is predominantly found in positions -1 and -4. Finally, the proportion of hydrophobic residues increases markedly at position -6. These observations made it possible to propose a scheme for predicting cleav-

+

\

-8-5-

FIG. 1. Proposed disposition of a “minimal” signal sequence positioned for cleavage from the mature protein. The peptide spans the membrane as an “8 + 5” helix + sheet structure. Small, neutral residues in position - 1 and - 3 fit into a pocket in the cleavage enzyme, thereby defining the cleavage site between positions - 1 and + 1.

159

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

TABLE I1 DISTRIBUTION OF RESIDUESNEAR CLEAVAGE SITESO Number of residues in position ~~

Group

-6

-5

-4

-3

-2

-1

+I

f 2

~

I I1 Ill IV V G

I1 0 0 42 21 0

P

I

I I

0

I

1 19 22 21

0 22 52 1

16

6

0

6

6 6 7 27

0

21 9 13 29 9

0 0 1 0 57

3 0

18

6 25 14 13 18 2

16 9 4

0

0

I1

7 15

If

Number of residues of different physicochemical character in positions -6 to + 2 in a sample of 78 eukaryotic signal sequences, cf. Fig. 1. The cleavage site is located between positions - I and + 1. I (aromatic residues): Phe, His, Trp, Tyr; I1 (charged residues): Asp, Glu, Lys, Arg; 111 (large, polar residues): Asn, Gln; IV (hydrophobic residues): Phe, Ile, Leu, Met, Val; V (small, neutral residues): Ala, Cys, Ser, Thr; G: Gly; P: Pro. N.B., Phe is included in both groups I and IV since it is aromatic (thus excluded from positions - 1 and -3) as well as hydrophobic (thus abundant in the hydrophobic core).

age sites from a knowledge of the primary sequence alone, a scheme that was applied with fair success to the sample under study. Further patterns of nonrandom amino acid usage in a region around the cleavage site that can be understood on the basis of a (-3,- I)-recognition site have also been found (von Heijne, 1984a). Other, so far less successful, approaches aimed at finding the molecular basis for the cleavage specificity have used secondary structure-prediction methods to locate probable sites (Austen, 1979; Garnier et al., 1980; Bedouelle and Hofnung, 1981). 2. THE HYDROPHOBIC COREAND

THE

N-TERMINAL REGION

The central hydrophobic core is the most outstanding signature of a signal sequence. In eukaryotes, it is exceptionally rich in Leu, whereas Ala dominates in prokaryotes, making prokaryotic signal sequences a little less hydrophobic (von Heijne, 198la). Beyond the strongly hydrophobic character, however, no specific structural motifs have yet been discovered in the central core. Among the signal sequences listed in von Heijne (1983), no hydrophobic core is shorter than eight residues. It is not always obvious how to delineate the core region; as an operational definition, we take the core to be the sum total of all eight-residue stretches with an overall hydrophobicity within 20% of the hydrophobicity calculated for the most hydrophobic eight-residue stretch found in the

160

GUNNAR VON HEIJNE

signal sequence in question (choosing another cutoff value, e.g., hydrophobicity within 10 or 15% of the maximum, makes only a small difference in the delineation of the cores.) With this definition, 90% of all cores in the sample are between 8 and 12 residues long. The N-termini of the core regions all are at least 13 residues distant from the cleavage site, with 70% falling between positions - 13 and - 17 (Fig. 2). The core C-termini are distributed with a marked preference for position -6. Finally, a recent study of the distribution of charged residues in the polar Nterminal region (von Heijne, 1984b) indicates that eukaryotic and prokaryotic signal sequences are very similar also with regard to net N-terminal charge and positioning of the positively charged residues, if it is assumed that the initiator Met in prokaryotes remains formylated (and hence uncharged) throughout the functional lifetime of the signal sequence. The emerging picture of the signal sequence, then, is that of a peptide 15-25 amino acids long, with a short positively charged N-terminal region; a strongly hydrophobic core, typically 8-12 residues long; and a more polar region 5-7 residues long, defining the cleavage site. A “minimal” signal sequence should

i

1 Position relative to clearage site

FIG. 2. Distribution of positions of core N- (hatched) and C-termini relative to the position of the clevage site (between - 1 and + 1) in a sample of 81 eukaryotic signal sequences. The three cases with N-termini at position -9 all have acceptable alternative cores with N-termini beyond position - 13.

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

161

-4t FIG.3 . The mean hydrophobicity (calculated with the “buried transfer free energy” scale) and standard deviation for each position in a sample of 81 eukaryotic signal sequences. The cleavage site is between positions - 1 and + 1.

+

thus conform to an “8 5” pattern, with an eight-residue hydrophobic core possibly responsible for initiating export, followed by a five-residue region conferring cleavage specificity (Fig. 1). This pattern is also present in a plot of the mean hydrophobicity in each position in a sample of signal sequences lined up with coincident cleavage sites (Fig. 3). The problem of the secondary structure of signal sequences has also generated much interest and speculation but very little experimental data. A synthetic peptide modeled after the precursor-specific N-terminal extension of preproparathyroid hormone (including the signal sequence) has been shown by CD analysis to be a mixture of a-and p-structure in an aqueous buffer and predominantly a in a nonpolar solvent (Rosenblatt et al., 1980). Theoretical predictions of secondary structure also indicate high a- and P-potentials (Austen, 1979; Gamier et al., 1980); and an energy minimization study of an immunoglobulin signal sequence predicted the central hydrophobic core to be a-helical (Pincus and Klausner, 1982). These results should be interpreted with great care, since nothing is known about the environment that the signal sequence finds itself in during initiation of export and cleavage: in an aqueous environment the classic methods for predicting secondary structure work reasonably well, whereas in a nonpolar membrane the situation may be quite different, as illustrated by the study of Argos et al. (1982b). If the hydrophobic core of the signal sequence does indeed interact directly with the membrane interior, then a great energetic premium would be put on maximizing the number of intrachain hydrogen bonds, almost certainly forc-

162

GUNNAR VON HEIJNE

ing this part of the sequence into a helical conformation, quite irrespective of any contrary results from one or another secondary structure prediction methods. It may be noted that a stretch of eight helical residues followed by five residues in an extended p-like conformation, as in Fig. 1 , is just about enough to span the approximately 30-8, wide central nonpolar part of a membrane. In addition, recent mutation studies have suggested that an a-helical conformation may be important for the proper functioning of the core in initiating export ( E m and Silhavy, 1982).

C. What Experimental Manipulations Can Do to Signal Sequences Various forms of experimental fiddling with signal sequences largely corroborate the picture given in the preceding section. The importance of the central hydrophobic core for initiating export has been repeatedly demonstrated through the isolation and characterization of export-defectivemutants and hybrid proteins (Bedouelle et al., 1980; Emr et al., 1980; E m and Bassford, 1982; Talmadge and Gilbert, 1982), in which deletions or point mutations destroying the unbroken stretch of hydrophobic residues have been shown to abolish or at least greatly impair export and cleavage. Interestingly, in many cases the effect of the mutation can be partly overcome by unlinked mutations mapping in the major ribosomal gene cluster ( E m et al., 1981). Similarly, by substituting a more polar leucine analog for the normal Leu residues in the Leu-rich preprolactin signal sequence, Hortin and Boirne (1980) were able to prevent export and cleavage. A threonine analog incorporated into position - 1 in the preprolactin signal sequence did not prevent export, however, but resulted in cleavage taking place at position -4 rather than at - 1 (Hortin and Boime, 1981). Recently, it was also discovered that the signal sequences in some human interferon molecules synthesized by yeast cells are cleaved at multiple sites, only one of which corresponds to the “normal” one (Hitzeman et al., 1983). A few mutants in which cleavage but not export is affected have also been isolated. In the lipoprotein of Escherichia coli, replacing a Gly at position -7 with an Asp gives this effect (Lin et al., 1978), and a similar situation obtains for a coat protein of a mutant phage M13 with Leu instead of Glu at position +2, i.e., downstream from the cleavage site (Russel and Model, 1981). A couple of cleavage mutants have also been isolated in the P-lactamase of Salmonella typhirnuriurn (Koshland et al., 1982; for a discussion, see von Heijne, 1984a). The cleavage reaction is also sensitive to the electrochemicalmembrane potential, specifically to its electrical component (Date et al., 1980; Daniels et al., 1981; Pagks and Lazdunski, 1982).

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

163

Finally, the positive charges on the amino-terminal extremity of the signal sequence seem to have a pronounced influence on the export process, at least in prokaryotes. For instance, decreasing the net charge in the E . coli lipoprotein signal sequence from +2 to -2 considerably inhibits the rate of export but does not abolish cleavage (Inouye et al., 1982; Vlasuk et al., 1983). Concluding this section, I would like to suggest that the hydrophobic core, besides functioning in initiating export (possibly by binding to the signal recognition particle), may also serve to define a “window” of possible processing sites, the final choice (or set of choices) within this window being dependent on how well the various potential cleavage sites conform to the “ideal” structure discussed above.

IV. THE TRANSMEMBRANE SEGMENT: MAKING FRIENDS WITH LIPIDS A. Sequence and Composition Proteins can bind to membranes in many ways. One useful distinction is between “intrinsic” and “extrinsic” membrane proteins (Singer and Nicolson, 1972; Capaldi and Green, 1972), denoting, respectively, proteins spanning the nonpolar hydrocarbon interior of a membrane, and proteins only associated with the (inner or outer) surface of the membrane. Here, we will deal only with the former category, which further can be subdivided into “simple” intrinsic proteins spanning the membrane only once and “complex” intrinsic proteins with multiple membrane-spanning segments. In the following discussion of protein-membrane interactions inside the nonpolar central part of the membrane, the “simple” transmembrane segments will be in focus since they do not involve the extensive intramolecular proteinprotein contacts typical of the “complex” proteins such as bacteriorhodopsin (see below). A list of “simple” eukaryotic transmembrane segments is presented in Table 111. A few of these are N-terminal sequences that probably have the dual role of initiating export as well as anchoring the mature protein to the membrane, but the majority are found in the C-terminal parts of the molecules, with no other known function than to provide membrane attachment. Thus, this last group (1 to 22) is the sample of choice for further study. So far, all “simple” transmembrane segments that have been sequenced consist of a long stretch of hydrophobic residues immediately followed on the Cterminal side by a cluster of positively charged, basic residues. These charges most likely bind to the negatively charged inner (cytoplasmic) surface of the

TABLE III MEMBRANE-SPANNING SEGMENTS~ A

0)

P

1

2 3 4 5

6 7 8 9 10 11

12 13 14 1.5 16

17 18

D S P S K K I LRR*I G Y S 1 L L I T G I V G A M V G F V 1 K VK*F L T V T T S Y F L S L L F L V I F T S V V S S F I W K V K * F L T V A A S Y C V S L L F L S I F I T K VK*l F T V F G S Y L L T L L F L A V F T C E V V N R K K I GK*M I L I I G V V I G V L G V F L G L A C K G G T A R R R M K * M V F A V V A G T V 1 A A G L V V L V A K G G T N R R R M K * M V F A V V A G T V I A A G L V V L V A G K G G T N R R R K * M V F A M V A G L I V V A G L V V P V A G K G G T N R R R K * M V F A V V A G I I A M A G L V G L V A V G G K G G S S K R R*C M V A A V V A G X V V V A X V A X G A 1 R D K V F Q V L R*N L I C P G F L L I M L L V I L P G M I L K I C L H I G VR*L V L F L G I I L G I I F F F S A I S S I R K P A K R T RR*H M W Y V I G C 1 V L A A L L S G G V A D R I F S R S K YK*L Y L L T C V L A L A T I S L L A T C F L A Y P T L C K SR*A A V L M Y C S A F I S 1 L A L L S M G RR*L G I C T V V V L V L 1 A G I A F A G L Y P T L C E R R AK*C A C L V A V T V G I M M A V T A S A V RR*T S T L M M S C A F I M L G I I L L S S

L*T I E I E S F H H A*T T W L N E F G E I * T T W L G D L E G M*T P W L G N E E E M*V N E K T E P L L V*T A M N S V T S P V * T G Q H Y V H C T 1*1 V T N T K T S S * I V M Y S D T S P V*X G V X P V T S Q T*S I L T T F W P S K*W S S F W G E V L G*A I L G M N N P T T*G T N E L C K Q P V*V A S V T A A P Y G*G S I K Q V W S L A*L I T Y V P H R H A*G G F L A F L W S

A E D L T P Y P P S R E A P L A Y W

19 20 21 22 23 24 25

4

g

I C I N C R I N G K*Q C I T C R M N G N K * V Q K S E P R N A RR*C Q T P E S R N V RR*C M N P N Q K * I M N P N QK*I M L P S TV*Q

C C W C I I T

A I T T T T

W F M M I

M V L L G I G L T L

I L S F S S

F G L G V A F I L I I L V S L I C M L L T S

L I V M T V G

V A L L I V G

V L G A A G V

C L L L L F I M A T L A T I C I 1 S L L S

F C G A F L L

C S A A L I Y

S A T S M L V

I G M L Q Q S

A F L L I I A

F*S I W S * F W L V*Y K G V*Y K G A*I L V G*N I I S*L S Y

L I W I V D W N P W N P

D K L L T T V T S I W I L L Y S

K Y G G L S D

G S D D F H S V L Y G L L H

“ A compilation of “simple” eukaryotic transmembrane segments. The C-terminus is to the left, except for entries 23-25, which are N-terminal segments. Tentative membrane-buried parts are enclosed between * symbols. Amino acids are given in the one-letter code, i.e., A, Ala; C , Cys; D, Asp; E, Glu; F, Phe; G , Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; Y, Tyr; X, unknown. 1, Human erythrocyte glycophorin (Segrest and Jackson, 1977); 2, mouse immunoglobulin p, chain (Rogers er a/., 1980); 3, mouse immunoglobulin y l chain (Tyler et a / . , 1982); 4, mouse immunoglobulin D chain (Cheng et al., 1982); 5. mouse E, I antigen (McNicholas et al., 1982); 6 , mouse H-2Kb histocompatibility antigen (Ploegh eta/., 1981);7 , mouse H-2Kh histocompatibility antigen (Reyes et a/., 1982); 8, mouse H-2d histocompatibility antigen (Kvist eral., 1981); 9, mouse Ld histocompatibility antigen (Moore e r a / . , 1982); 10, human HLA-B7 histocompatibility antigen (Ploegh et al.. 1981); 1 1 . Moloney leukemia virus Prl5E (Green et a / . , 1981); 12, vesicular stomatitis virus glycoprotein (Rose ef a / . , 1980); 13, herpes simplex virus D glycoprotein (Watson et al., 1982); 14, adenovirus E3115 glycoprotein (Persson e t a / ., 1980): 15, Semliki Forest virus E2 protein (Garoff et a / . . 1980); 16, Semliki Forest virus E l protein (Garoff et a / . , 1980); 17, Sindbis virus E2 protein (Rice and Strauss, 1981); 18. Sindbis virus E l protein (Rice and Strauss. 1981); 19, influenza (Victoria) HA2 protein (Min Jou e t a / . , 1980): 20. influenza (FPV) HA2 protein (Min Jou et al.. 1980); 21, rabies virus (CVS) glycoprotein (Yelverton et a / . , 1983); 22, rabies virus (ERA) glycoprotein (Anilionis et a / . , 1981); 23, influenza (Victoria) neuraminidase (Van Rompuy e t a / . , 1982); 24. influenza (WSN) neuraminidase (Van Rompuy et a / . , 1982); 25. influenza (Lee) neuraminidase (Shaw et a/., 1982).

166

GUNNAR VON HEIJNE

membrane (Segrest and Jackson, 1977) and, hence, make a good point of reference for aligning the segments. With such an alignment, the mean hydrophobicity for each position can be calculated (Fig. 4). From the figure, we may tentatively identify a contiguous hydrophobic stretch 21 residues long. In this region, no particular variations in the mean hydrophobicity comparable to those seen in the signal sequence sample (Fig. 3) can be discerned. The undifferentiatedcharacter of the hydrophobic stretch is also clearly seen in the distribution of individual amino acids (data not shown). All the charged amino acids, together with Asn, Gln, His, and Trp, are excluded from the central parts of the hydrophobic region. Occasionally, however, they do appear near the N- or C-terminus of the 21-residue stretch, close enough to make it possible for their polar groups to reach outside the nonpolar region (cf. von Heijne, 1981a). Pro is also exceedingly rare in the central region. However, Ser and Thr are quite common, possibly because, although polar, their side chains can form hydrogen bonds to the peptide backbone (Capaldi et al., 1983). All other residues are more or less evenly distributed within the segments. Thus, these 21-residue transmembrane segments very much look like “lengthened and inverted” signal sequence cores, with charged C-termini and contiguous hydrophobic regions. One clear difference, though, is that there is no preference for Leu residues comparable to that found among the signal sequence cores, thus making the mean hydrophobicity per residue a little less in the transmembrane segments (this holds also for the N-terminal segments; 23-25).

Position

FIG. 4. The mean hydrophobicity (kcal/mol) and standard deviation for each position in the sample of “simple” transmembrane segments shown in Table 111 (entries 1-22). The putative membrane-buried region is indicated by arrows.

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

167

B. Conformation within the Membrane The three-dimensional structure of membrane-spanning segments is only well characterized in one case, bacteriorhodopsin, which is an integral membrane protein of the “complex” variety. This protein consists of a bundle of seven slightly tilted a-helices extending through the membrane, coming together to form a cylinder with a hydrophobic surface facing the membrane interior and a central polar region (Henderson and Unwin, 1975; Engelman and Zaccai, 1980). A similar structure has been proposed for two maize zein proteins (Argos et al., 1982a). As yet, no intrinsic membrane protein of the “simple” kind with an intact transmembrane segment has been crystallized, but many lines of evidence point toward a similar a-helical structure in the membrane-spanning part (Segrest and Jackson, 1977; Capaldi et al., 1983). Indeed, a segment 21 residues long (cf. Fig. 4) will form a 30-A a-helix, just the right length for reaching through the nonpolar part of a membrane. This also matches well with the observation that the membrane protects a stretch of chain approximately 21 residues long from peptidase digestion during synthesis of exported proteins on membrane-bound ribosomes (Smith et al., 1978). From a thermodynamic point of view, the ahelix should be the structure of lowest free energy for an isolated piece of chain in a nonpolar environment simply because it maximizes the number of intrachain hydrogen bonds. Given that most or all transmembrane segments in “simple” integral membrane proteins extend through the membrane as a-helices, it has been of interest to determine whether these helices are uniformly hydrophobic or show some kind of bias toward “sidedness,” i.e. a clustering of polar residues on one side of the helix, with a corresponding nonpolar face opposite. This, of course, is the situation in bacteriorhodopsin and similar “complex” integral proteins, as well as in some extrinsic membrane proteins such as the plasma lipoproteins (Segrest and Jackson, 1977) which form so-called amphipathic helices, but the evidence is less conclusive for the “simple” proteins. Nevertheless, many such transmembrane segments have been claimed to show signs of sidedness. Helical plots, either of the “cylindrical” type or the “helical wheel” type (Schulz and Schirmer, 19791, have often been drawn, and a seeming clustering of polar residues on one side has been interpreted as resulting from a selection for sidedness, perhaps stabilizing dimeric associations in the membrane (Segrest and Jackson, 1977; Argos et al., 1982b). Two examples are shown in Fig. 5. An analysis of the statistics behind these claims, however, shows that many of the purportedly amphipathic helices do not in fact deviate significantly in their degree of sidedness from what can be expected from chance alone (Flinta et al., 1983).

168

GUNNAR VON HEIJNE

FIG.5 . Helical wheel plots of two transmembrane segments from the phage MI3 coat protein (left) and the bacteriorhodopsin C-helix (right). Polar and charged residues are shown within circles. Cutting planes derived according to Argos’s method (i.e., planes that divide the helix into two maximally dissimilar halves) are indicated by broken lines. The probability for a “3 - 0” distribution (the MI3 helix) in a sample of randomly generated 12-residue helices is 0.54, and the probability for a distribution that is at least as biased as the one on the right (bacteriorhodopsin helix) is 0.09.

A better measure of helix amphiphilicity should be the so-called “helical hydrophobic moment” (Eisenberg et al., 1982) in which one defines a hydrophobicity vector for each residue in the helix, with a length proportional to the hydrophobicity of the residue and pointing from the center of the helical wheel to the position of the residue on the circumference. The net helical moment is then obtained as the vector sum of all individual hydrophobicity vectors in the segment. Using the helical moment thus defined, a better test for sidedness can be designed (Flinta et al., 1983; Pownall et al., 1983). All things considered, there are very good reasons for suggesting that a typical transmembrane segment is a hydrophobic a-helix, that is approximately 21 residues long and that has a markedly basic end on the cytoplasmic side. In most cases, the membrane-spanning segment seems to be under selection for overall hydrophobicity but not much else, and, indeed, the gross amino acid composition of a sample of transmembrane segments has been quite successfully reproduced in an ‘‘evolutionary game”-simulation with absence of charged residues and overall hydrophobicity as the only selection criteria (von Heijne, 198 lb). A notable exception is provided by the histocompatibility antigens, (610, Table IU), in which the transmembrane segment is highly conserved. In some cases, there are indications that particular serine residues in the transmembrane segment are covalently bound to fatty acids via an ester linkage (Schmidt, 1982). What the functional significance of this is is not clear. Con-

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

169

ceivably the lipid chains could provide extra stabilization to the protein-membrane interaction; they could be active in the complex transport and sorting activities that determine the final cellular destination of the protein; or they could function in membrane-fusion reactions.

V.

PROTEIN EXPORT: RULES OF THE GAME

Clearly, much of the secret behind the export process resides in the signal sequence and the transmembrane segment. By combining these two elements, it should be easy to construct proteins that will be inserted into a membrane in any conceivable orientation (Blobel, 1980). A nice illustration is provided by work on the influenza hemagglutinin (Gething and Sambrook, 1982), in which the wild-type gene codes for a membrane protein with both a signal sequence and a transmembrane segment. Deleting the transmembrane segment produces a protein that is efficiently secreted into the medium; deleting the signal sequences produces a cytoplasmic protein. Even though the simple logic inherent in the concepts of signal (or “start”) sequences and transmembrane (or ‘‘stop”) sequences is immediately clear, it has proved exceedingly difficult to really pinpoint the structural and physicochemical basis for the functioning of these sequences. A number of characteristic features of start and stop signals have been discussed in this article, and some interpretations of their possible functional relevance have been attempted, but there is still a long way to go before the export process as a whole can be said to be sufficiently well understood. In this section, I will try to piece together some of the elements entering into the problem, namely, the signal sequence data, the roles of the signal recognition particle and other proteins implicated in the export process, the structure of transmembrane segments, and some information regarding the behavior of the exported chain proper during export, with a view toward a critical evaluation of the different models of the export process that have been put forward since 1975.

A. Models of Protein Export Generally speaking, two different kinds of models have been considered in the literature. In its original formulation (Blobel and Dobberstein, 1975), the “signal hypothesis” postulated the existence of a protein pore or “channel” guiding the nascent chain through the membrane and shielding it from contact with the lipid membrane core. The channel was further thought to have a signal recognition site and a ribosome binding site at its cytoplasmic end. Thus, a signal sequence emerging from a cytoplasmic ribosome would guide the translation

170

GUNNAR VON HEIJNE

complex to the signal recognition site, the ribosome would bind to “channel subunits” in the membrane and assemble them into a complete channel, whereupon the growing chain would exit through the proteinaceous pore thus formed. In opposition to this model, a number of authors have envisioned processes whereby the nascent chain could make it directly through the lipid interior of the membrane. In one widely cited model (DiRienzo et al., 1978), the positively charged N-terminus of the signal sequence binds to the negatively charged inner membrane surface; the hydrophobic signal sequence core partitions into the membrane with the N-terminus remaining on the cytoplasmic side (hence the name “loop model”); and the growing chain is pulled through the membrane by folding on the outside. In a similar vein, Engelman and Steitz (1981) have put forward the so-called helical hairpin hypothesis. Here, the signal sequence and the first part of the mature chain form a hydrophobic “hairpin” structure composed of two interacting a-helices. This hairpin then penetrates the membrane pretty much as in the “loop model,” and export of the remaining parts of the chain is hypothesized to be driven by a gradient of increasing hydrophobicity toward the C-terminus, i.e., the more C-terminal parts push the chain through the membrane by virtue of their greater attraction for the hydrophobic membrane interior. A third variation on this theme is the “direct transfer model” (von Heijne and Blomberg, 1979), which is similar to the loop model except that a ribosomebinding protein is postulated to provide a firm ribosome-membrane bond, making it possible for the growing chain to be pushed through the membrane as it is elongated. Finally, Wickner’s (1980) “membrane trigger hypothesis” does away with cotranslational export altogether and postulates that the chain, including the signal sequence, folds up posttranslationally in the cytoplasm in such a way that, upon exposure to a membrane, it undergoes some kind of conformational rearrangement that carries it through to the other side. These, then, are the major contestants that have entered the arena to date. As I now show, they all have their weak spots and none is in complete agreement with all the experimental data.

B. “Something Is Rotten in the State of Denmark’’’ First and foremost, cotranslational export has been amply demonstrated in the great majority of systems, although in a handful of cases posttranslational insertion of membrane proteins with small extracytoplasmic domains has been shown to be possible (Ito et al., 1980; Goodman et al., 1981; see also Randall, 1983). ‘Shakespeare, W. (1602). from “Hamlet,” Act I, Scene IV

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

171

Thus, the membrane trigger hypothesis is of no more than limited applicability. [Proteins not covered by the signal hypothesis (e.g., mitochondria1 and peroxisomal proteins) do, however, use poorly understood posttranslational routes in negotiating the various membranes barring the way to their final destinations (Kreil, 1981)]. Recently it has become clear that the export machinery is very complex, at least in eukaryotic cells, involving not only the signal recognition particle already referred to, but also a fairly large ribosome-binding protein complex in the membrane [the ribophorins (Kreibich el a(., 1978)], and a membrane-bound “docking protein” or “signal recognition particle receptor” (Meyer et al., 1982; Gilmore and Blobel, 1983) which is required to release the translational block imposed by the signal recognition particle. Clearly, the original signal hypothesis was correct in stressing the importance of having a protein machinery in the membrane. Similarly, the direct transfer model relies on the existence of a ribosome-binding protein, whereas the loop model and helical hairpin model do not. What is not clear, however, is the extent to which the nascent chain is exposed to a lipid environment during its passage through the membrane. A perhaps naive argument in favor of lipid exposure would be that this makes it easier to understand how a hydrophobic transmembrane segment becomes embedded in the membrane; if it were exported through a porteinaceous channel it could hardly be expected to “sense” the presence of the intramembranous lipid and would not be able to halt export. On the other hand, export directly through a lipid barrier is only possible with some kind of force or energy source for driving polar and charged residues through to the outside. The direct transfer model tries to meet this demand by suggesting that the free energy of interaction between the ribosome and the ribosome-binding site on the membrane is sufficiently large to drive the equilibrium distribution of all possible states (i.e., the state with the ribosome bound to the membrane, plus all states with an unbound ribosome and the nascent chain free to move back and forth through the membrane) toward the bound state. Only if a strongly hydrophobic segment, followed on the C-terminal side by a strongly hydrophilic one, appears in the nascent chain would the ribosome-ribosome-binding site interaction be overcome, resulting in a trans-membrane protein. This model is the only one that has been developed into a quantitative prediction scheme. Using the buried transfer free energy scale, free energy profiles for the passage of the nascent chain through the membrane can be calculated from a knowledge of the amino acid sequence. One example is shown in Fig. 6. This is the curve obtained for the phage M I 3 coat protein, an unusual molecule that can be posttranslationally inserted into the E . coli membrane (Wickner, 1980). As is clear from the figure, this protein

172

GUNNAR VON HEIJNE 0

I

I

I

I

I

I

K)

20

I

2 .

-100

E

-

7

1

u

a

-200

- 300 0

30 Step number

I

40

FIG.6. Free energy profile for the transfer of the phage M I 3 coat protein across a membrane. Step i refers to the situation when residue i has just entered the membrane, and step zero refers to the situation with only the signal sequence bound to the membrane. The minimum corresponds to the equilibrium state with the protein partly extruded through the membrane. Scale in kJ/mol (1 kl = 0.24 kcal). From von Heijne and Blomberg (1979) (with permission).

encounters only a moderately high energy barrier early on in the insertion process, followed by a downhill gradient leading to a state of minimum free energy with a transmembrane segment located close to the C-terminus. Thus, this particular protein may not need the elaborate export machinery to find its proper place in the membrane and can, in fact, be incorporated with the correct transmembrane orientation into artificial liposomes (Wickner, 1983). A similar curve calculated for the E . coli lipoprotein suggests that it, too, should be able to insert posttranslationally into membranes (von Heijne, 1980), a prediction that has now been verified (Inouye et al., 1982). Further, bee venom melittin was predicted to end up as a transmembrane protein even though it is known to be secreted from the venom gland. Translation of microinjected melittin mRNA in Xenopus oocytes, however, produces vesicle-associated molecules, possibly inserted across the membrane as predicted (Lane et al., 1981). In spite of these successful predictions, it is clear by now that the model cannot correctly predict the behavior of all experimentally investigated proteins. One notable failure is the case of some bacterial fusion proteins, where signal sequences or longer segments from normally exported proteins have been fused to the gene for the cytoplasmic protein P-galactosidase, the product nevertheless remaining in the cytoplasm (Moreno er al., 1980; Kadonaga el al., 1984; see article by Bankaitis et al., this volume). So far, only one successful fusion turning a cytoplasmic protein into an exported one has been reported (Yost et al., 1983). This difficulty, implying rather strict constraints on the sequences of exported proteins, was not anticipated by the “direct transfer model.” In the case that the growing nascent chain is actually in direct contact with

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

173

membrane lipids during transit, however, some free energy calculation of the kind employed in the direct transfer model should be applicable. Indeed, estimates of the total hydrophobicity of the signal sequence cores show that a looptype insertion into a membrane is thermodynamically plausible (von Heijne, 1980; Engelman and Steitz, 1981; Jahnig, 1983). On the other hand, such calculations have also shown that most exported proteins lack the hydrophobic gradient required by the helical hairpin hypothesis (von Heijne and Blomberg, 1979; von Heijne, 1980). Finally, it may be asked whether the hydrophobic transmembrane segments really act as true stop signals or only as anchors. The vast majority are located not far (less than 40 residues) from the C-terminus of the protein, and this makes it likely that they are more or less embedded in the membrane when translation stops, since the ribosome itself covers some 40 residues of the chain. Thus, they may never have to function as stop signals at all (Zilberstein et al., 1981). In a handful of cases, though, the transmembrane segment is located far from the C-terminus. The E. coli serine chemoreceptor has a large cytoplasmic domain some 315 residues long, and approximately 200 residues are on the extracytoplasmic side of the membrane (Boyd et al., 1983). The Sindbis and Semliki Forest viruses also provide examples of transmembrane segments that probably act as true stop signals (Garoff et al., 1980; Rice and Strauss, 1981). Nevertheless, transmembrane segments more than 40 residues distant from the C-terminus seem to be selected against in most instances. Summing up this section, it is clear that no one model has come out clearly ahead of its competitors. The loop, helical hairpin, and membrane trigger hypotheses all have been discredited as generally valid models by experimental data, and the direct transfer model has not been able to live up to all of its predictive aspirations. The signal hypothesis, finally, leaves so many important questions regarding the actual workings of the export machinery unanswered that it is more of a research paradigm than a specific, quantifiable model.

VI.

CONCLUSION

As I hope this review has made clear, a richer picture of the biogenesis of exported and membrane-bound proteins is slowly developing. Much inspiration and guidance have come from the signal hypothesis, and this may be part of the reason for our relatively advanced understanding of the early steps in the process, specifically those directly involving the signal sequence. The ensuing steps are less well characterized. Almost nothing is known about the constraints imposed by the export process on the sequence of the exported protein, except that they exist. How and when transmembrane segments function as stop signals is another question for the future-the basic design of such segments is obvious enough, but how wide are the tolerable variations? And why

174

GUNNAR VON HEIJNE

are examples with more than 40 residues in between the transmembrane segment and the C-terminus so scarce? Finally, the most important question in the whole field-one that experimenters have been carefully avoiding since 1975 and the question which in itself holds the answers to most of the earlier ones-is still just as much a matter of conjecture as it was in the early days of the game: What sort of environment does the nascent chain encounter on its way through the membrane? Only when this is known can we hope to really understand protein export.

REFERENCES Anilionis, A., Wunner, W. H., and Curtis, P.J. (1981). Structure of the glycoprotein gene in rabies virus. Nature (London) 294, 275-278. Argos, P., Pedersen, K., Marks, M. D., and Larkins, B. A.(1982a). A structural model for maize zein proteins. J. Biol. Chem. 257, 9984-9990. Argos, P., Rao, J . K. M., and Hargrave, P. A. (1982b). Structural prediction of membrane-bound proteins. Eur. J. Biochem. 128, 565-575. Austen, B. M. (1979). Predicted secondary structures of amino-terminal extension sequences of secreted proteins. FEES Lett. 103, 308-313. Austen, B. M., and Ridd, D. H. (1981). The signal peptide and its role in membrane penetration. Biochem. Soc. Symp. 46, 235-258. Bedouelle, H., and Hofnung, M. (1981). Functional implications of secondary structure analysis of wild type and mutant bacterial signal peptides. In “Membrane Transport and Neuroreceptors,” pp. 399-403. Liss, New York. Bedouelle, H., Bassford, P. J., Fowler, A. V., Zabin, I., Beckwith, J., and Hofnung, M. (1980). Mutations which alter the function of the signal sequence of the maltose binding protein of Escherichia coli. Nature (London) 285, 78-8 1. Blobel, G. (1980). lntracellular protein topogenesis. Proc. Natl. Acad. Sci. U.S.A. 77, 1496-1500. Blobel, G., and Dobberstein, B. (1975). Transfer of proteins across membranes. J . Cell Biol. 67, 835-851. Blobel, G . , and Sabatini, D. (1971). Ribosome-membrane interaction in eukaryotic cells. Eiomembranes 2, 193-195. Boyd, A., Kendall, K., and Simon, M. I. (1983). Structure of the serine chemoreceptor in Exherichia coli. Nature (London) 301, 623-626. Bull, H. B., and Breese, K.(1974). Surface tension of amino acid solutions: A hydrophobicity scale of amino acid residues. Arch. Biochem. Biophys. 161, 665-670. Capaldi, R. A., and Green, D. E. (1972). Membrane proteins and membrane structure. FEES Lett. 25, 205-209. Capaldi, R. A., Marshall, F. A,, and Staples, S. J. (1983). Structure of intrinsic membrane proteins and their amino acid sequences. Comments Mol. Cell. Biophys. I, 365-381. Charton, M., and Charton, B. I. (1982). The structural dependence of amino acid hydrophobicity parameters. J. Theor. Biol. 99, 629-644. Cheng, H.-L., Blattner, F. R., Fitzmaurice, L., Mushinski, J. F., and Tucker, P. W. (1982). Structure of genes for membrane and secreted murine IgD heavy chains. Nature (London) 296, 410-41 5 . Chothia, C. (1974). Hydrophobic bonding and accessible surface area in proteins. Nature (London) 248, 338-339.

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

175

Daniels, C. J., Bole, D. G.,Quay, S. C., and Oxender, D. L. (1981). Role for membrane potential in the secretion of protein into the periplasm of Escherichia coli. Proc. Narl. Acad. Sci. U.S.A. 78, 5396-5400. Date, T., Goodman, J . M., and Wickner, W. (1980). Procoat, the precursor of M13 coat protein. requires an electrochemical potential for membrane insertion. Proc. Narl. Acad. Sci. U.S.A. 77, 4669-4673. DiRienzo, J . M., Nakamura, K., and Inouye, M. (1978). The outer membrane proteins of Gramnegative bacteria: Biosynthesis, assembly, and functions. Annu. Rev. Biochem. 47, 48 1-532. Eisenberg, D., Weiss, R. M., and TenviUiger, T. C. (1982). The helical hydrophobic moment: a measure of the amphiphilicity of a helix. Nature (London) 299, 371-374. Em, S. D., and Bassford, P. J. (1982). Localization of outer membrane and periplasmic proteins in Escherichia coli strains harboring export-specific suppressor mutations. J. Biol. Chem. 257, 5852-5860. E m , S . D., and Silhavy, T. J. (1982). Molecular components of the signal sequence that function in the initiation of protein export. J. Cell Biol. 95, 689-696. E m , S. D., Hedgpeth, J., Clement, J.-M., Silhavy, T. J., and Hofnung, M. (1980). Sequence analysis of mutations that prevent export of A receptor, an Escherichia coli outer membrane protein. Nature (London) 285, 82-85, Em, S. D., Hanley-Way, S., and Silhavy, T. J. (1981).Suppressor mutations that restore export of a protein with a defective signal sequence. Cell 23, 79-88. Engelman, D. M., and Steitz, T. A. (1981). The spontaneous insertion of proteins into and across membranes: The helical hairpin hypothesis. Cell 23, 41 1-422. Engelman, D. M . , and Zaccai, G . (1980). Bacteriorhodopsin is an inside-out protein. Proc. Narl. Acad. Sci. U.S.A. 77, 5894-5898. Finney, J . L., Gellatly, B. J., Golton, I. C . , and Goodfellow, J . (1980). Solvent effects and polar interactions in the structural stability and dynamics of globular proteins. Biophys. J. 32, 17-33. Flinta. C . , von Heijne. G., and Johansson, J. (1983). “Helical sidedness” and the distribution of polar residues in trans-membrane helices. J. Mol. Biol. 168, 193-196. Garnier, J., Gaye, P., Mercier, J.-C., and Robson, B . (1980). Structural properties of signal peptides and their membrane insertion. Biochimie 62, 23 1-239. Garoff, H . , Frischauf, A.-M., Simons, K., Lehrach, H., and Delius, H. (1980). Nucleotide sequence of cDNA coding for Semliki Forest virus membrane glycoproteins. Narure (London) 288, 236-241. Gething, M-J., and Sambrook, J . (1982). Construction of influenza haemagglutinin genes that code for intracellular and secreted forms of the protein. Nature (London) 300, 598-603. Gilmore, R., and Blobel, G . (1983). Transient involvement of signal recognition particle and its receptor in the microsomal membrane prior to protein translocation. Cell 35, 677-685. Goodman, J. M., Watts, C . , and Wickner, W . (1981). Membrane assembly: Posttranslational insertion of MI3 coat protein into E. coli membranes and its proteolytic conversion to coat protein in vitro. Cell 24, 437-441, Green, N., Shinnick, T. M . , Witte, O., Ponticelli, A , , Sutcliffe, J . G.,and Lemer, R. A . (1981). Sequence-specific antibodies show that maturation of Moloney leukemia virus envelope polyprotein involves removal of a COOH-terminal peptide. Pror. Nail. Acad. Sci. U.S.A. 78,60236027. Henderson, R., and Unwin, P. N. T. (1975). Three-dimensional model of purple membrane obtained by electron microscopy. Nature (London) 257, 28-32. Hitzeman, R . A., Leung, D. W., Perry, L. J., Kohr, W. J., Levine, H. L., and Goeddel, D. V. (1983). Secretion of human interferons by yeast. Science 219, 620-625. Hopp, T. P., and Woods, K. R. (1981). Prediction of protein antigenic determinants from amino acid sequences. Proc. Nut/. Acad. Sci. U.S.A. 78, 3824-3828.

176

GUNNAR VON HEIJNE

Hortin, G., and Boime, I. (1980). Inhibition of preprotein processing in ascites tumor lysates by incorporation of a leucine analog. Proc. Natl. Acad. Sci. U.S.A. 77, 1356-1360. Hortin, G., and Boime, I. (1981). Misclevage at the presequence of rat preprolactin synthesized in pituitary cells incubated iwth a threonine analog. Cell 24, 453-461. Inouye, S., Soberon, X., Franceschini, T.,Nakamura, K., Itakura, K., and Inouye, M. (1982). Role of positive charge on the amino-terminal region of the signal peptide in protein secretion across the membrane. Proc. Natl. Acad. Sci. U.S.A. 79, 3438-3441. Ito, K., Date, T., and Wickner, W. (1980). Synthesis, assembly into the cytoplasmic membrane, and proteolytic processing of the precursor of coliphage M13 coat protein. J. B i d . Chem. 255, 21 23-2 130. Jihnig, F. (1983). Thermodynamics and kinetics of protein incorporation into membranes. Proc. Natl. Acad. Sci. V.S.A. 80, 3691-3695. Janin, J. (1979). Surface and inside volumes in globular proteins. Nature (London) 277, 491-492. Jones, D. D. (1975). Amino acid properties and side-chain orientation in proteins: A cross c o d a tional approach. J . Theor. Biol. 50, 167-183. Kadonaga, J. T., Gautier, A. E., Straw, D. R., Charles, A. W., Edge, M. D., and Knowles, J. R. (1984). The role of the p-lactamase signal sequence in the secretion of proteins by Escherichia coli. J . Biol. Chem. 259, 2149-2154. Kauzmann, W . (1959). Some factors in the interpretation of protein denaturation. Adv. Protein Chem. 14, 1-63. Koshland, D., Sauer, R. T., and Botstein, D. (1982). Diverse effects of mutations in the signal sequence on the secretion of p-lactamase in Salmonella ryphimurium. Cell 30, 903-914. Kreibich, G . , Czako-Graham, M., Grebenau, R., Mok, W., Rodriguez-Boulan, E., and Sabatini, D. D. (1978). Characterization of the ribosomal binding site in rat liver rough microsomes: Ribophorins I and 11, two integral membrane proteins related to ribosome binding. J . Supramol. Struct. 8, 279-302. Kreil, G. (1981). Transfer of proteins across membranes. Annu. Rev. Biochem. 50, 317-348. Kumamoto, C. A,, Oliver, D. B., and Beckwith, J. (1984). Signal sequence mutations disrupt feedback between secretion of an exported protein and its synthesis in E. coli. Nature (London) 308, 863-864. Kvist, S., Bregegere, F., Rask, L., Cami, B., Garoff, H., Daniel, F., Wiman, K., Larhammar, D., Abastado, J. P., Gachelin, G., Peterson, P. A., Dobberstein, B., and Kourilsky, P. (1981). cDNA clone coding for part of a mouse H-2d major histocompatibility antigen. Proc. Narl. Acad. Sci. U.S.A. 78, 2772-2776. Kyte, J., and Doolittle, R. F. (1982). A simple method for displaying the hydropathic character of a protein. J. Mol. Biol. 157, 105-132. Lane, C. D., Champion, J., Haiml, L., and Kreil, G. (1981). The sequestration, processing, and retention of honey-bee promelittin made in amphibian oocytes. Eur. J. Biochem. 113,273-281. Lesk, A. M., and Chothia, C. (1980). Solvent accessibility, protein surfaces, and protein folding. Biophys. J . 32, 35-47. Lin, J. J. C., Kanazawa, H., Ozols, J., and Wu, H. C. (1978). An Escherichia coli mutant with an amino acid alteration within the signal sequence of outer membrane prolipoprotein. Proc. Narf. Acad. Sci. U.S.A. 75, 4891-4895. McNicholas, J., Steinmetz, M., Hunkapiller, T., Jones, P., and Hood, L. (1982). DNA sequence of the gene encoding the W, la polypeptide of the BALB/c mouse. Science 218, 1229-1232. Manavalan, P., and Ponnuswamy, P. K. (1978). Hydrophobic character of amino acid residues in globular proteins. Nature (London) 275, 673-674. Meyer, D. I., Krause, E., and Dobberstein, B. (1982). Secretory protein translocation across membranes-the role of the ‘docking protein.’ Nature (London) 297, 647-650.

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

177

Michaelis, S . , and Beckwith, J. (1982). Mechanism of incorporation of cell envelope proteins in Escherichiu coli. Annu. Rev. Microbiol. 36, 435-465. Milstein, C., Brownlee, G., Harrison, T., and Mathews. M. (1972). A possible precursor of immunoglobulin light chains. Nature (London) New Biol. 239, 117-120. Min Jou, W., Verhoeyen, M., Devos, R., Saman, E., Fang, R., Huylebroeck, D., Fiers, W., Threlfall, G . , Barber, C., Carey, N., and Emtage, S. (1980). Complete structure of the hemagglutinin gene from the human influenza A/Victoria/3/75 (H3N2) strain as determined from cloned DNA. Cell 19, 683-696. Moore, K. W., Sher, B. T., Sun, Y. H., Eakle, K. A., and Hood, L. (1982). DNA sequence of a gene encoding a BALBlc mouse Ld transplantation antigen. Science 215, 679-682. Moreno, F., Fowler, A. V., Hall, M., Silhavy, T. J., Zabin, I., and Schwartz, M. (1980). A signal sequence is not sufficient to lead P-galactosidase out of the cytoplasm. Nufure (London) 286, 356-359. Nozaki, Y.,and Tanford, C. (1971). The solubility of amino acids and two glycine peptides in aqueous ethanol and dioxane solutions. J . Biol. Chem. 246, 221 1-2217. Pa&, J.-M., and Lazdunski, C. (1982). Membrane potential (A+) depolarizing agents inhibit maturation. FEES Lefr. 149, 5 1-54. Persson, H., Jomvall, H., and Zabielski, J. (1980). Multiple mRNA species for the precursor to an adenovirus-encoded glycoprotein: Identification and structure of the sequence. Proc. Nud. Acud. Sri. U.S.A. 77, 6349-6353. Pincus, M. R., and Klausner, R. D. (1982). Prediction of the three-dimensional structure of the leader sequence of pre-K light chain, a hexadecapeptide. Proc. Nurl. Acud. Sci. U.S.A. 79, 3413-34 17. Ploegh, H. L., Om, H. T.. and Strominger, J. L. (1981). Major histocompatibility antigens: The human (HLA-A, -B, -C) and murine (H-2K, H-2D) class I molecules. Cell 24, 287-299. Pownall, H. I., Knapp, R. D., Gotto, A. M., and Massey, J. B. (1983). Helical amphipathic moment: Application to plasma lipoproteins. FEES Left. 159, 17-23. Randall, L. L. (1983). Translocation of domains of nascent periplasmic proteins across the cytoplasmic membrane is independent of elongation. Cell 33, 23 1-240. Reyes, A. A,, Schold, M., Itakura, K.,and Wallace, B. (1982). Isolation of a cDNA clone for the murine transplantation antigen H-2Kb. Proc. Nutl. Acud. Sci. U.S.A. 79, 3270-3274. Rice, C. M., and Straws, I. H. (1981). Nucleotide sequence of the 26s mRNA of Sindbis virus and deduced sequence of the encoded virus structural proteins. Proc. Nufl. Acud. Sci. U.S.A. 78, 2062- 2066. Robson, B., and Osguthorpe, D. J. (1979). Refined models for computer simulation of protein folding. J . Mol. Biol. 132, 19-51. Rogers, J., Early, P., Carter, C., Calame, K., Bond, M., Hood, L., and Wall, R. (1980). Two mRNAs with different 3’ ends encode membrane-bound and secreted forms of immunoglobulin chain. Cell 20, 303-312. Rose, J . K., Welch, W. I . , Sefton, B. M., Esch, F. S., and Ling, N. C. (1980). Vesicularstomatitis virus glycoprotein is anchored in the viral membrane by a hydrophobic domain near the COOH terminus. Proc. Nufl. Acud. Sci. U.S.A. 77, 3884-3888. Rosenblatt, M., Beaudette, N. V., and Fasman, G. D. (1980). Conformational studies of the synthetic precursor-specific region of preproparathyroid hormone. P roc. Nut!. Acud. Sci. U.S.A. 77, 3983-3987. Russel, M., and Model, P. (1981). A mutation downstream from the signal peptidase cleavage site affects cleavage but not membrane insertion of phage coat protein. Proc. Nail. Acud. Sci. U.S.A. 78, 1717-1721. Schmidt, M. F. G. (1982). Acylation of proteins-a new type of modification of membrane glycoproteins. Trends Biochem. Sci. 7, 322-324.

178

GUNNAR VON HEIJNE

Schulz, G . E., and Schirmer, R. H. (1979). “Principles of Protein Structure.” Springer-Verlag, Berlin and New York. Segrest, J. P., and Feldmann, R. J . (1974). Membrane proteins: Amino acid sequence and membrane penetration. J. Mol. Biol. 87, 853-858. Segrest, J. P., and Jackson, R. L. (1977). Molecular properties of membrane proteins. In “Membrane Proteins and their Interaction with Lipids” (R. Capaldi, ed.), pp. 21-45. Dekker, New York. Shaw, M. W., Lamb, R. A., Erickson, B. W., Briedis, D. J . , and Choppin, P. W. (1982). Complete nucleotide sequence of the neuraminidase gene of influenza B virus. Proc. Null. Acud. Sci. U.S.A. 79, 6817-6821. Singer, S . J . , and Nicolson, G. L. (1972). The fluid mosaic model of the structure of cell membranes. Science 175, 720-731. Smith, W. P., Tai, P.-C., and Davis, B. D. (1978). Interaction of secreted nascent chains with surrounding membrane in Bacillus subtilis. Proc. Nurl. Acud. Sci. U.S.A. 75, 5922-5925. Strauss, A. W., and Boime, I. (1982). Compartmentation of newly synthesized proteins. CRC Crir. Rev. Biochem. 12, 205-235. Talmadge, K., and Gilbert, W. (1982). Cellular location affects protein stability in Escherichiu coli. Proc. Nutl. Acad. Sci. U.S.A. 79, 1830-1833. Tanford, C. (1979). Interfacial free energy and the hydrophobic effect. Proc. Nail. Acud. Sci. U.S.A. 76, 4175-4176. Tyler, B. M., Cowman, A. F., Gerondakis, S. D., Adams, J. M., and Bernard, 0. (1982). mRNA for surface immunoglobulin y chains encodes a highly conserved transmembrane sequence and a 28-residue intracellular domain. Proc. Null. Acad. Sci. U.S.A. 79, 2008-2012. Van Rompuy, L., Min Jou, W., Huylebroeck, D., and Fiers, W. (1982). Complete nucleotide sequence of a human influenza neuraminidase gene of subtype N2 (A/Victoria/3/75). J. Mol. Biol. 161, 1-11. Vlasuk, G. P.,Inouye, S., Ito, H., Itakura, K., and Inouye, M. (1983). Effects of the complete removal of basic amino acid residues from the signal peptide on secretion of lipoprotein in Escherichiu coli. J . Biol. Chem. 258, 7141-7148. Von Heijne, G. (1980). Trans-membrane translocation of proteins: A detailed physico-chemical analysis. Eur. J . Biochem. 103, 431-438. Von Heijne, G. (1981a). On the hydrophobic nature of signal sequences. Eur. J. Biochem. 116,419422. Von Heijne, G. (1981b). Membrane proteins: The amino acid composition of membrane-penetrating segments. Eur. J . Biochem. 120, 275-278. Von Heijne, G . (1983). Patterns of amino acids near signal sequence cleavage sites. Eur. J . Biochem. 133, 17-21. Von Heijne, G. (1984a). How signal sequences maintain cleavage specificity. J. Mol. Biol. 173, 243-25 1. Von Heijne, G. (1984b). Analysis of the distribution of charged residues in the N-terminal region of signal sequences: implications for protein export in prokaryotic and eukaryotic cells. EMBO J . 3, 2315-2318. Von Heijne, G., and Blomberg, C. (1979). Trans-membrane translocation of proteins: The direct transfer model. Eur. J. Biochem. 97, 175-181. Walter, P., and Blobel, G. (1982). Signal recognition particle contains a 7 s RNA essential for protein translocation across the endoplasmic reticulum. Nurure (London) 299, 69 1-698. Watson, R. J., Weis, J. H., Salstrom, J. S., and Enquist, L. W. (1982). Herpes simplex virus type-1 glycoprotein D gene: Nucleotide sequence and expression in Escherichiu coli. Science 218, 38 1-384. Wickner, W. (1980). Assembly of proteins into membranes. Science 210, 861-868.

4. PROTEIN TRANSFER: STRUCTURE AND THERMODYNAMICS

179

Wickner, W. (1983). MI3 coat protein as a model of membrane assembly. Trends Biochem. Sci. 8, 90-94. Wolfenden, R. V., Cullis. P. M., and Southgate, C. C. F. (1979). Water, protein folding, and the genetic code. Srience 206, 575-577. Yelverton, E., Norton, S., Obijeski, J. F., and Goeddel, D. V. (1983). Rabies virus glycoprotein analogs: Biosynthesis in Escherichiu roli. Science 219, 614-619. Yost, C. S., Hedgpeth, J., and Lingappa, V. (1983). Stop transfer sequence confers predictable orientation to a previously secreted protein in cell-free systems. Cell 34, 759-766. Zilberstein, A , , Snider, M. D., and Lodish, H. F. (1981). Synthesis and assembly of the vesicular stomatitis virus glycoprotein. Cold Spring Harbor Symp. Quanr. B i d . 46, 785-795.

This Page Intentionally Left Blank

CURRENT TOPICS IN MEMBRANES AND TRANSPORT. VOLUME 24

Chapter 5

Mechanisms and Functional Role of Glycosylation in Membrane Protein Synthesis SHARON S. KRAG Deparmenr of Biochemistry The Johns Hopkins Universiry School of Hygiene and Public Health Baltimore, Maryland

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Involvement of Oligosaccharide Lipid Intermediates . . . . . . . . . . . . A. Lipid Carrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B . Saccharide Moieties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Synthetic Pathway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Role of the Carbohydrate Moiety.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. General Approaches and Methodologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B . Protein Solubility and Structure.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Protein Turnover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Enzymatic Function . ......... ............................ E. Interactive Functions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ......... F. Compartmentalization of Proteins . . . . . . . . . . . . IV. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.

181

192 201 217 226 226 228 229 230 230 231 232 233

INTRODUCTION

Many membrane-associated and secreted proteins of eukaryotes are glycoproteins, with oligosaccharides attached in 0-glycosidic linkage to serine, threonine, hydroxyllysine, and hydroxylproline or in N-glycosidic linkage to asparagine (Kornfeld and Kornfeld, 1980). Oligosaccharides linked to serine and threonine residues contain N-acetylgalactosamine (GalNAc) as the linkage sugar, 181

Copyright 0 19x5 by Acddemic Prers. In' All rights VI reproduction in m y lomi re\erved ISBN 0-12-157324 7

182

SHARON S.KRAG

galactose (Gal), N-acetylglucosamine (GlcNAc), sialic acid (SA), and fucose (Fuc). A single protein may have oligosaccharides 0-glycosidically linked to numerous serine and threonine residues. For example, an erythrocyte membrane glycoprotein, glycophorin A, has 12 to 15 tri- and tetrasaccharide units of two types attached to it (Adamany el al., 1983). Collagens and basement membranes of animal cells and the glycoproteins of plant cell walls also contain oligosaccharides attached via 0-glycosidic linkages. In collagens, numerous hydroxyllysine residues are modified by Gal and glucose(G1c)-Gal disaccharides; in basement membranes, arabinose and Gal are attached to hydroxylproline residues (Kornfeld and Kornfeld, 1980). The oligosaccharide structures found linked to asparagine residues (Am) on membrane proteins contain SA, Gal, GlcNAc, Fuc, and mannose (Man). The sugar residue covalently attached to the asparagine in these glycoproteins is GlcNAc. The absence of GalNAc and the presence of Man are quite characteristic of this type of structure, although mannose is found attached to serine or threonine residues via an 0-glycosidic bond in glycoproteins of yeast and fungi. There are two major types of N-linked oligosaccharides, termed complex and high-mannose. Two representative structures of the complex oligosaccharides are shown in I and 11: SA-Gal-GlcNAc-Man

\

klcNAc-Man-GlcNAc-GlcNAc- Asn: I / SA- Gal- GlcNAc-Man tFUC I

SA- Gal- GlcNAc \

SA- Gal- G i C N A Y a i ~ G ~ c N A cManGlcNAc-GlcNAc- Am< I

fFUC

SA- Gal- GlcNAc II

These complex oligosaccharides can have more branches than the biantennary structure (I) or triantennary structure (11) (Gleeson and Schachter, 1983). They sometimes contain Fuc or a GlcNAc residue which bisects the terminal branching sequences of SA-Gal-GlcNAc (Narasimhan, 1982). The high-mannose oligosaccharides contain only Man and GlcNAc, in structures similar to those often found on secretory proteins. A representative structure for these high-mannose chains is shown in 111.

5. GLYCOSYLATION OF MEMBRANE PROTEINS

183

Man

\

Man

/

\

Man

Man- GlcNAc- GlcNAc- A m ( / Man / Man Ill

Few if any membrane proteins have been reported to contain only high-mannose oligosaccharides, although 3-hydroxy-3-methylglutaryl-CoA reductase of the endoplasmic reticulum appears to contain primarily high-mannose structures (Liscum et al., 1983). A third type of oligosaccharide side chain attached to asparagine residues is called a hybrid structure and has the representative structure shown in IV. Man- Man &an Man/

\

iGlcNAc-Man- GlcNAc- GlcNAc- A m < /

SA- Gal- GlcNAC

IV

There are generally a limited number, from one to six, of N-glycosidically linked oligosaccharides on a protein. A single protein may contain oligosaccharides linked by both 0- and N-glycosidic bonds. For example, the lowdensity lipoprotein receptor contains one complex oligosaccharide linked to an Asn residue and six to nine oligosaccharides linked to Ser/Thr residues (Cummings et al., 1983). One protein may also contain more than one type of Nlinked structure. As an example, the envelope glycoprotein E2 of Sindbis virus contains one high-mannose and one complex oligosaccharide attached to asparagine residues (Burke and Keegstra, 1979). The oligosaccharides at any one glycosylation site in a protein may be heterogeneous: more than seven different high-mannose and hybrid oligosaccharides have been found attached to a single asparagine in ovalbumin (Yamashita er al., 1978). Alternatively, the oligosaccharides located at one site in a protein may be limited and constant, but may differ from the oligosaccharides at another site on the same protein. For example, rhe H-2Kk antigen has two glycosylation sites; the same spectrum of complex oligosaccharides was always found at each site, but the two sites have different spectra of oligosaccharides (Swiedler et al., 1983). The biosynthesis of 0-glycosidically linked oligosaccharide moieties occurs by the sequential action of glycosyltransferases, adding individual sugars directly

184

SHARON S. KRAG

onto protein, after the translation of the protein is complete (Beyer and Hill, 1982; Hanover et al., 1982; Cummings et al., 1983). The final structure of these oligosaccharides presumably depends on the specificities of the individual transferases and on the interactions between these enzymes and their protein or glycoprotein acceptors. These membrane-associated enzymes are thought to be located facing the Golgi lumen and to utilize sugar nucleotide substrates which are transported into this organelle (see Section II,D,3). Glycoproteins with oligosaccharides attached to Asn residues are synthesized by a series of reactions which can be divided into four major steps. The first step is the synthesis of the phosphorylated isoprenoid lipid intermediate which is involved in the synthesis of both the complex and high-mannose types of side chains. The synthesis of the lipid involves both cytoplasmic and membraneassociated, endoplasmic reticular enzymes. The second step is the assembly of an oligosaccharide containing GlcNAc, Man, and Glc on a pyrophosphorylated lipid intermediate; this assembly involves membrane-associated enzymes of the rough endoplasmic reticulum utilizing both sugar nucleotides and monoglycosylated phosphorylated isoprenoid lipids as sugar donors. The third step is the transfer of this preassembled oligosaccharide unit to an Asn on protein, in some cases while that protein is still being translated in the rough endoplasmic reticulum. The final step of N-linked glycoprotein biosynthesis involves modification of the transferred oligosaccharide while it is on the protein. It is this step which distinguishes the complex and high-mannose oligosaccharides. For high-mannose oligosaccharides, trimming of the Glc and some of the Man residues occurs in the rough endoplasmic reticulum and the Golgi. More extensive trimming and stepwise addition of GlcNAc, Gal, SA, and Fuc occurs in the Golgi for complex oligosaccharides (Kornfeld, 1982). The fact that the biosynthesis of both types of Asn-linked structures have these common steps involving the phosphorylated polyisoprenoid lipid explains why, although diverse in their final structures, most oligosaccharide side chains attached to Asn residues contain a core sequence of (Man),(GlcNAc),. All the mechanisms which regulate whether the final structure at a particular glycosylation site will become high-mannose or complex are not clear, although accessibility of the site within the protein to the processing enzymes appears to be one important determinant (Hsieh et al., 1983). This review will concentrate on the first major steps in the synthesis of Asnlinked glycoproteins, that is, the synthesis of the oligosaccharide lipid intermediate, both the lipid moiety and the carbohydrate portion, and the transfer of the preassembled oligosaccharide to the protein. Reviews (Beyer and Hill, 1982; Komfeld, 1982) of the final trimming and modification steps have recently appeared.

5. GLYCOSYLATION OF MEMBRANE PROTEINS

185

II. INVOLVEMENT OF OLIGOSACCHARIDE LIPID INTERMEDIATES A. Lipid Carrier 1. IDENTITY The lipid moiety of the glycosylation intermediates in bacterial glycan synthesis is known to be a phosphorylated polyisoprenol lipid (undecaprenyl phosphate). Since dolichol, a family of lipids characterized by having 16-22 isoprene units (C80-Cllo) with the a-isoprene unit saturated, is the abundant polyisoprenol lipid in mammalian tissues (Pennock et af., 1960), even the earliest report of glycosylation intermediates in mammalian tissues (Behrens and Leloir, 1970) implicated dolichyl phosphate as the lipid carrier. In their report and numerous subsequent reports (reviewed by Waechter and Lennarz, 1976; Parodi and Leloir, 1979; Hubbard and Ivatt, 1981; Staneloni and Leloir, 1982), the chemical and chromatographic properties of glycosylated endogenous lipids were compared with those of glycosylated dolichyl phosphate and were found to be similar. These comparisons were facilitated by the availability of authentic, chemically synthesized mannosylphosphoryldolichol (Warren and Jeanloz, 1973). There are three distinctive properties of dolichol which are often exploited. First, the saturation of the a-isoprene unit prevents degradation of the glycosylated, phosphorylated polyisoprenol during treatment with hot phenol (Murazumi et al., 1979) and during hydrogenolysis (Evans and Hemming, 1973; Tkacz et al., 1974), although this latter technique appears unreliable (Kean, 1977a). Second, the phosphoglycosidic bond is more labile to mild acid hydrolysis when the a-isoprene unit is unsaturated. Finally, derivatives of dolichol, which contain 16-22 isoprene units, can easily be separated by paper, thin-layer, or gel permeation chromatography from the bacterial polyisoprenyl lipids, which contain 11 isoprene units. In two cases, mannosylated phosphorylpolyisoprenyl lipids have been purified in chemical amounts from tissue (10 kg of pig liver and 6.5 kg of bovine liver) and characterized (Evans and Hemming, 1973; Baynes et al., 1973). In both cases the ratio of organic phosphorus to acid-labile, reducing sugar in the purified lipid was approximately 1 : 1. The ultraviolet spectrum of the purified mannosylated lipid was featureless in the 300- to 400-nm range, a finding indicating that the lipid did not contain a conjugated system. The infrared spectrum was consistent with the spectrum of dolichol, including the characteristic pattern for a terminal saturated isoprene residue. Finally, the mass spectral analysis was

186

SHARON S. KRAG

consistent with a polyisoprenoid, but in neither study was the molecular ion obtained. Therefore, the precise number of isoprene units was not established. Before considering the metabolism of dolichol, it should be pointed out that lipid moieties other than dolichol may be involved in protein glycosylation in higher eukaryotic tissues. First, retinyl phosphate can serve as an acceptor of mannosyl residues (DeLuca et al., 1973; Martin and Thorne, 1974a) in rat liver and intestine, and glycosylated derivatives of retinol have been implicated in glycoprotein biosynthesis (see review by DeLuca, 1982). Second, shorter homologs of dolichol (molecules with a saturated a-isoprene unit) have been isolated from pig liver (Mankowski ef al., 1976), and a mannosylated, short-chain polyisoprenyl phosphate derivative has been characterized from protozoa (Quesada-Allue and Parodi, 1983). Third, two fully unsaturated derivatives have been described. An N-acetylglucosaminylmonophosphoryl long-chain polyisoprenyl lipid with all unsaturated isoprene units was obtained after incubation of hen oviduct membranes with sugar nucleotide (Hayes and Lucas, 1980). A fully unsaturated, short-chain polyprenol has been isolated from bovine pituitary (Radominska-Pyrek ef al., 1979). The involvement of these lipids in the glycosylation of eukaryotic proteins remains to be established. However, in view of the numerous glycosylated lipids synthesized, the diversity of proteins and glycans to be glycosylated, the various subcellular compartments involved, and the as yet undefined regulatory mechanisms in operation, lipids other than dolichyl phosphate may serve as lipid carriers. 2. METABOLISM Dolichol is found in many tissues: liver, spleen, kidney, brain, pancreas, intestine, oviduct, pituitary, and skeletal muscle (Martin and Thorne, 1974b; Keller and Adair, 1977). The lipid can be synthesized de novo in each of these tissues; alternatively, it may be stored in the liver and subsequently dispensed to other tissues. In this latter case, either the dolichol could be made de novo by the liver, or polyisoprenols might be obtained from the diet and stored and/or modified in this organ. There have been two recent studies concerning the uptake of dietary polyprenols and storage of the lipid in liver. Keller et al. (1982) and Chojnacki and Dallner (1983) concluded that dietary uptake by rats of dolichol and long-chain (Cg0or Cg5)polyprenols (all unsaturated isoprene units) was insignificant (0.060.18%). However, Chojnacki and Dallner (1983) found significant (2%) uptake of both unsaturated and a-saturated polyprenols of C,, chain length and observed subsequent saturation and phosphorylation of the polyprenols by the liver microsomes; elongation of the polyprenols was not detected. The significance of dietary contribution to body polyprenol stores seems unresolved.

187

5. GLYCOSYLATION OF MEMBRANE PROTEINS

It is clear that liver tissue can synthesize dolichol de n o w from mevalonic acid (Buttenvorth et ul., 1966; Martin and Thome, 1974b). Dolichol is a poly-cisisoprenol; the incorporation of labeled mevalonate into dolichol is significantly less than into cholesterol and ubiquinone, both of which are trans derivatives. The precise reaction sequence from mevalonate to dolichol is still obscure; the structures of some of the early intermediates are shown in Fig. 1. Isopentenyl pyrophosphate and farnesyl pyrophosphate seem to be involved, since they have been used as substrates for polyisoprenoid synthesis in in vitro reactions using various liver fractions (Daleo et ul., 1977; Wellner and Lucas, 1979; Adair and Keller, 1982; Wong and Lennarz, 1982a). While these four studies report the synthesis of 1- 10 pmol polyisoprenoid hr - I mg protein - I , they differ with respect to reaction conditions, the identity of the active subcellular fractions (see below), and the products formed. The products of incubation of liver fractions with isopentenyl pyrophosphate include dolichol (Wong and Lennarz, 1982a), dolichyl phosphate (Daleo et uf., 1977), dehydropolyprenyl pyrophosphate

0

Acetate

II

CH,-C-0-

f Mevalonate

b

t

Isopentenyl

COOI

CH,-

I

-C"-

CH,OH

OH

F"

CH,- C-C€i,-

pyrophosphate

0

P Dolichols

0

7H3 (CH,-C=CB-CCH,),-

7%

II

A-

P Farnesyl pyrophosphate

0

II

C~-O-P-O-P-OI 0-

II

0 II

O-P-O-P-O-

A-

y

I

0-

3

CH,-C=CH-C&(C~-C=CH-C&)l,-CI&-

7% C-C&I

C&OH

H FIG. 1. Structures of some of the known intermediates in dolichol synthesis. Multiple steps exist between each of these intermediates and have yet to be completely delineated.

188

SHARON S. KRAG

(Wellner and Lucas, 1979), and polyprenyl monophosphate (Adair and Keller, 1982). The identification of the final product of synthesis is crucial to an understanding of the further metabolism of dolichol (see Fig. 2 and discussion below). If the liver is the prime dolichol storage/synthesis organ of the body, then transport of dolichol throughout the organism would be required. When radioactive dolichol was injected intravenously into a rat, the lipid rapidly associated with high-density lipoprotein (HDL) (Keenan et al., 1977), a finding suggesting that HDL might be responsible for transport. The radioactivity disappeared from the circulation with a half-life of approximately 9 hours. Radioactivity appeared subsequently in all tissues, with the majority being recovered in the liver (Keenan et al., 1977). An analysis of the steady-state level of dolichol in the plasma of chickens indicated a very low level (0.5 pg/g) as compared to liver (182 pg/g) or oviduct (about 20 pg/g) (Keller and Adair, 1977). Nonliver tissues appear able to synthesize dolichol from early precursors. For example, Wong and Lennarz (1982b) described the synthesis of dolichol from acetate by slices of three different chicken embryo organs: liver, heart, and brain. Synthesis of dolichol from acetate was also observed in murine testes, preputial glands (Potter and Kandutsch, 1983), and spleens (Potter and Kandutsch, 1982). The lipid moiety of glycosylated dolichyl derivatives was labeled by incubating radioactive acetate with cultured aortic smooth muscle cells (Mills and Adamany, 1978) and thyroid slices (Spiro et al., 1976a). Incubations of chicken oviduct preparations with farnesyl pyrophosphate and isopentenyl pyrophosphate produced long-chain unsaturated polyprenyl phosphate (Grange and Adair,

Dehydrodolichyl pyrophosphate

phosphatase

Dehydrodolichyl phosphate

J

reduc tase Dolichyl pyrophosphate

phosphatase

Glycosylated forms

reduc tase

Dolichyl phosphate

Dolichol esterase

11

synthetase

Dolichyl ester

FIG. 2. Known interconversions of the various forms of dolichol in animal tissue. The compounds which are underlined have been reported to be products of de novo synthesis (see text).

5. GLYCOSYLATION OF MEMBRANE PROTEINS

189

1977). Finally, the synthesis of C,, and C,, isoprenoid acids, some containing cis double bonds, was recently reported using mevalonate and homogenates of bovine retina (Fliesler and Schroepfer, 1983). No matter how a cell obtains its dolichol, the lipid is found in a number of enzymatically interconvertible forms (see Fig. 2). The form which is a substrate for glycosylation reactions is dolichyl phosphate. Dolichol can be phosphorylated by a CTP-dependent kinase described in brain (Burton et af.,1979), plasmacytoma, Chinese hamster ovary cells, and liver (Allen et al., 1978; Rip and Carroll, 1980). Dolichol is also found in liver tissue as the ester, with up to 63% of the total being esterified (Butterworth and Hemming, 1968); a transesterification reaction implicating phosphatidylcholine as an acyl donor to dolichol has been described in rat liver microsomes by Keenan and Kruczek ( 1976). The amount of esterified product produced during an incubation of labeled dolichol with membranes was stimulated by phosphatidylcholine (eight-fold by varying the concentration from 0 to 7 p,mol/ml) but not by the cofactors needed to generate fatty acyl-CoA derivatives (ATP, CoA, Mg2+). The conversion of labeled dolichol to dolichyl ester was not affected by addition of palmityl-CoA to the reaction mixture. A dolichyl esterase activity was observed in brain membranes by Scher and Waechter (1981); its pH optimum (pH 7.5) appears to distinguish it from lysosomal esterases. Additional metabolic conversions of dolichol involve removal of phosphate. For the case of dolichyl phosphate, phosphatases have been reported in lymphocytes (Wedgewood and Strominger, 1980), brain (Burton er af., 1981b), and liver (Appelkvist et af., 1981; Belcopitow and Boscoboinik, 1982; Rip et af., 1981). The phosphatase which degraded exogenous dolichyl phosphate in these tissues had a pH optimum of 6.5-7, was inhibited by millimolar concentrations of fluoride, and did not require metal ions; the enzymes in the various studies appeared to differ in their sensitivity to the presence of phosphate. The enzyme in brain degraded dolichyl phosphate generated in situ by incubating preparations with CTP (Burton et af., 1981b). One obvious problem in characterizing this enzyme in crude preparations is specificity; all four groups approached this problem by doing inhibition studies with a variety of potential substrates. In addition, Burton er af. (1981b) demonstrated that the enzyme which degrades dolichyl phosphate in brain is less thermolabile than a p-nitrophenyl phosphatecleaving enzyme. Also, the phosphatidic acid phosphatase of brain was found to be less sensitive to the presence of inorganic phosphate than was the dolichyl phosphate phosphatase. These studies suggest that dolichyl phosphate is not cleaved by all cellular phosphatases, but it remains to be determined if there is a specific dolichyl (polyprenyl) phosphatase. Dolichyl pyrophosphatases have been observed in rat liver (Appelkvist et al., 1981; Belcopitow and Boscoboinik, 1982; Kato et af., 1980) and lymphocytes

190

SHARON S.KRAG

(Wedgewood and Strominger, 1980). This activity had a pH optimum of 8, was not inhibited by fluoride, but was inhibited by bacitracin. Dolichyl pyrophosphate is generated after the transfer of an oligosaccharide unit from dolichyl pyrophosphate to protein and may also arise from de novo synthesis (see Fig. 2). Interestingly, the reported properties of this pyrophosphatase activity in rat liver were similar for long-chain (Kato et al., 1980) and short-chain (Appelkvist et al., 198 1) saturated polyprenyl pyrophosphates. 3. SUBCELLULAR LOCATION In addition to knowing which dolichol metabolites and enzymes are present in a cell, it is of interest to know their subcellular localization. An early report suggested that in pig liver dolichol was fairly evenly distributed in four fractions: cytoplasmic, mitochondrial, microsomal, and nuclearhnbroken cells (Burgos and Morton, 1962). A subsequent study by Butterworth and Hemming (1968) indicated that while total dolichol was evenly distributed, the majority of free dolichol in pig liver was located in the mitochondria1fraction, while most of the esterified dolichol was found in the nuclearhnbroken cell fraction. Recent studies using rat liver (Wong et al., 1982; Adair and Keller, 1982; Rip and Carroll, 1982; Rip et al., 1983) found an enrichment of total dolichol in the mitochondrial/lysosomal fraction, with this fraction accounting for 45-60% of the total amount of dolichol. Wong et al. (1982) further fractionated the mitochondrial/lysosomal fraction and found that the greatest amount and enrichment of dolichol was in the fractions which colocalized with a lysosomal marker, p-Nacetylhexosaminidase. Rip and Carroll (1982) and Rip et al. (1983) found that although only 11% of the total dolichol was in the microsomal fraction of rat liver, it was enriched 14-fold in the Golgi over other submicrosomal fractions. Wong and Lennarz (1982a) also examined the localization of newly synthesized polyprenols, after incubation of rat liver slices with [3H]mevalonolactone. Interestingly, labeled polyprenols appeared at early times (up to 30 minutes) in both the mitochondrial/lysosomal fraction and the microsomal fraction; radioactivity in the latter fraction remained constant for 1 hour while it continued to increase in the mitochondrial/lysosomal fraction during the incubation (2 hours). Since their methodology did not separate unsaturated (free or phosphorylated) and a-saturated polyprenols and therefore the precise molecular species were not determined in these experiments, it is not clear whether the suggestion of translocation of newly synthesized molecules from microsomes to lysosomes, based on the incorporation kinetics, is valid. Localization studies of dolichyl phosphate have also been done. These studies are technically more difficult because of the much lower steady-state level of this intermediate. For example, in testicular glands, although the de novo synthesis of

5. GLYCOSYLATION OF MEMBRANE PROTEINS

191

dolichyl phosphate could be measured, no chemical amounts of this lipid could be detected (Potter and Kandutsch, 1983). Adair and Keller (1982) and Rip et al. (1983) were able to detect dolichyl phosphate in rat liver; it was 5-10% of the amount of total dolichol in this tissue and was not localized to any particular subcellular fraction; the relative specific activity only varied twofold among the nuclear, microsomal, submicrosomal , mitochondrial , lysosomal, and cytoplasmic fractions. Likewise, Spiro et al. (1976a) found that dolichyl phosphate oligosaccharide, labeled de novo in thyroid slices by mevalonic acid, was not in the cytoplasmic nor lumenal fractions of the cell but was widely distributed in the various membranous subfractions of the cell. Finally, the subcellular locations of some of the various dolichol metabolizing enzymes have been examined. Dolichyl phosphate phosphatase was enriched in the plasma membrane, although the majority of the activity was distributed among the nuclear, microsomal, and cytoplasmic fractions (Rip et al, 1981); of the submicrosomal fractions, the activity was enriched in the Golgi fraction (Rip et al., 1983). Appelkvist et al. (1981) had found both short-chain, a-saturated polyprenyl phosphate phosphatase and prenyl pyrophosphate phosphatase enriched in the plasma and lysosomal membranes, but also present in microsomal and outer mitochondrial membranes. Rupar et al. (1982) found dolichol kinase in both the microsomal and plasma membrane fractions, and the dolichyl ester synthetase enriched in the cytoplasmic fraction. Although no clear picture has emerged from the localization studies of either the lipids or the enzymes, two things are clear. First, the possibility of intracellular translocation of dolichol or its metabolites, although interesting, is a potential problem in any localization study. Depending on the mechanism of this translocation, subcellular localization of these lipids may be variable. Second, there are difficulties in working with membrane-associated enzymes and lipophilic substrates. For example, the specific activity of these enzymes in the different subcellular fractions varies with the concentrations of detergent in the assay. The amount of activity detected may also depend on the level of endogenous lipid present. Thus, the precise determination of the subcellular distribution of these enzymes awaits quantitation of the enzymes as proteins rather than measurement of the enzymatic activities. In summary, although it is clear that dolichyl phosphate is involved in glycosylation reactions, a great deal remains to be established concerning the uptake, synthesis, metabolism, and translocation of this molecule. Basic questions must be answered before we can begin to contemplate regulatory mechanisms. For example, what is the product of de novo synthesis? At what step is the polyprenol reductase involved in de novo synthesis and/or dietary uptake? Is there organismal and/or intracellular transport of dolichyl derivatives? Where in the cell are the enzymes and lipids located?

192

SHARON S.KRAG

B. Saccharide Moieties 1. MANNOSYL-, GLUCOSYL-, AND

N-ACETYLGLUCOSAMINYLPHOSPHORYL LIPIDS The carbohydrate portions of the lipid intermediates involved in protein glycosylation range from a single unit, the monosaccharides mannose, glucose, and N-acetylglucosamine, to oligomers of these three sugars of up to 14 residues. The syntheses of labeled monoglycosyl lipids (see Table I) have been reviewed numerous times (Waechter and Lennarz, 1976; Parodi and Leloir, 1979; Hubbard and Ivatt, 198 1) in systems including Mycobacceriurn srnegrnacis (Schultz and Elbein, 1974), Volvox (Bause et al., 1983), cotton (Forsee and Elbein, 1975), yeast, and avian and mammalian cultured cells and tissues. The common structure found for the mannose-containing molecule is mannosylphosphoryl lipid: the phosphodiester bond involves the hydroxyl group of the isoprenyl lipid arid the C, anomeric carbon of mannose. The labeled product of incubations using calf pancreas microsomes was found to be the p isomer (Tkacz et al., 1974; Tkacz and Herscovics, 1975). Mannosylphosphoryl lipid is easily detected after incubations of membrane fractions with GDP-[ *4C]mannose, especially if the reactions have been supplemented with exogenous polyisoprenyl phosphate. Radiolabeling of this lipid in vivo is often more difficult, (Krag, 1979; Chapman et al., 1979b), presumably because of a small, metabolically active pool. Evidence that mannosylphosphoryl lipid is a glycosylation intermediate comes from three lines of research. First, the transfer of mannose from exogenous mannosylphosphoryl lipid (MPL) to endogenous acceptors occurs under conditions which minimize the formation of GDP-mannose; GDP-mannose, another mannosyl donor, could be formed during certain incubations by simple reversal of the MPL synthase reaction (Lucas et al., 1975; Chambers et al., 1977). TABLE I VARIOUS MONOSACCHARIDE AND DISACCHARIDE PHOSPHORYLATED LIPIDSWHICHHAVEBEEN DESCRIBED I N EUKARYOTIC CELLS Sugar component Mannose Glucose A'-Acetylglucosamine Glucuronic acid Xylose Galactose Glucuronic acid-N-acetylglucosamine Chitobiose

Reference Tanner (1969); Caccam et a / . (1969); Zatz and Barondes (1969); Rosso et al. (1977) Behrens and Leloir (1970) Tetas et ul. (1970) Cummings and Roth (1982) Waechter er al. (1974) Peterson et al. (1976) Turco and Heath (1977) Leloir e/ a / . (1973)

5. GLYCOSYLATION OF MEMBRANE PROTEINS

193

Second, mannosylation of endogenous acceptors by GDP-mannose and MPL made from GDP-mannose is altered in the presence of the antibiotic amphomycin, an inhibitor of MPL synthesis (Kang et al., 1978). Finally, two mutants, one from a lymphoma cell line (Chapman et al., 1979b) and one from a CHO cell line (Stoll et al., 1982), have been described which lack MPL synthase activity and display alterations in glycosylation. The synthesis of glucosylphosphoryl lipid was first observed on incubation of UDP-[ ''C]glucose with rat liver microsomes (Behrens and Leloir, 1970). The general structure seems similar to that of MPL. On the basis of its rate of degradation in alkali, the lipid made by microsomes of both liver and pancreas is thought to be the p anomer (Behrens and Leloir, 1970; Herscovics et al., 1977a). Glucose from exogenous, labeled glucosylphosphoryl lipid is incorporated into endogenous acceptors during incubations with membrane preparations, a finding indicating that this lipid is also an intermediate in glycosylation reactions (Behens et af.,1971; Parodi et al., 1972; Waechter and Scher, 1978). N-Acetylglucosamine is also found covalently linked to a phosphorylisoprenyl lipid; this monosaccharide lipid is unique in that it contains a pyrophosphate linkage, synthesized by the transfer of N-acetylglucosaminyl phosphate from UDP-N-acetylglucosamine to phosphorylated lipid (Tetas et al., 1970; Ghalambor et al., 1974; Behrens et al., 1971). This reaction was inhibited by uridine monophosphate (Palamarczyk and Hemming, 1975; Zatta et al., 1976). The synthesis of this lipid was also inhibited by the antibiotic tunicamycin; this lipophilic, glycose- and uridine-containing drug is a tight-binding inhibitor which can be viewed as a bisubstrate analog (Keller et al., 1979) resembling both the sugar nucleotide and the polyisoprenoid substrate. The important role of the N-acetylglucosaminyl lipid will be discussed below. AND DISACCHARIDE 2. OTHERMONOSACCHARIDE PHOSPHORYL LIPIDS

Sugars other than glucose, mannose, and N-acetylglucosamine (see Table I) have been found attached to polyisoprenoids by a glycosidic-phosphoryl linkage, a linkage characterized by lability to treatment with mild acid. Glucuronosyl-N-acetylglucosaminylpyrophosphoryllipid has been detected in SV40-transformed human lung fibroblasts after incubation with labeled glucosamine and phosphate (Turco and Heath, 1977); this lipid was also detected after incubations of UDP-glucuronic acid and UDP-N-acetylglucosamine with microsomes prepared from rat lung (Turco and Heath, 1977) or rat fibrosarcoma (Hopwood and Dorfman, 1977). The glycosidic bond of the disaccharide in the lipid labeled by the transformed lung fibroblasts was a 1,4 linkage, as determined by periodate and mild alkaline hydrolysis. These two sugars in a 1,4 linkage constitute the repeating disaccharide unit of the glycosaminoglycan portion of heparan sulfate

194

SHARON S. KRAG

and heparin (Roden, 1980). A glucuronylphosphoryl lipid has also been detected after incubation of labeled UDP-glucuronic acid with homogenates of adult and phenobarbital-treated embryonic chicken liver (Cummings and Roth, 1982). Some of these lipids may be involved in the synthesis of complex glycans other than asparagine-linked glycoproteins, such as glycosaminoglycans and proteoglycans. This is suggested by both the sugar composition of the saccharide lipids and the fact that tunicamycin does inhibit the synthesis of at least one type of proteoglycan, corneal keratan sulfate (Hart and Lennarz, 1978). However, in this case tunicamycin may not only be inhibiting the synthesis of the putative mono- or disaccharide lipid intermediates of the glycosaminoglycan chains, but also may be blocking the synthesis of the N-glycosidic linkage region. In corneal tissue, the structure of the linkage region indicates it is derived from a biantennary complex oligosaccharide, the synthesis of which would be inhibited by tunicamycin (Nilsson et al., 1983). Consistent with the idea that tunicamycin does not affect the synthesis of the disaccharide portion of these molecules is that no inhibition was seen in the synthesis of hyaluronic acid, and the inhibition of heparan sulfate and chondroitin sulfate could be explained by an inhibition of protein synthesis (Hartand Lennarz, 1978). Also, neither the mono- nor disaccharide lipids containing glucuronic acid are able to transfer radioactivity to endogenous macromolecules when added exogenously to in vitro reactions. Thus, the role of these lipids in glycan synthesis remains unresolved. A xylosylphosphorylpolyisoprenol(Table I) has also been characterized; this pentose-containing lipid was synthesized during incubations of UDP-xylose with membranes prepared from hen oviduct (Waechter et al., 1974). Labeled monosaccharide was shown to be transferred during incubations of microsomes with exogenous xylosylphosphoryl lipid to two types of endogenous acceptors. One acceptor molecule (10% transfer of label) was soluble in mixtures of chloroform, methanol, and water, a finding indicative of an amphipathic molecule; an oligosaccharide with xylose on the nonreducing end was released from this molecule after mild acid hydrolysis. The other acceptor (0.5% transfer of label) was sensitive to Pronase digestion but not to digestion with testicular hyaluronidase. Although xylose is a common constituent of proteoglycans, its addition to those molecules is not thought to involve lipid intermediates, based on studies with purified xylosyl transferases (Roden, 1980). Xylose has been reported in a limited number of glycoproteins (Kornfeld and Kornfeld, 1980). Finally, synthesis of galactosylphosphoryl lipid (Table I) has been reported after incubation of sugar nucleotides with crude homogenates (Parodi and Leloir, 1979). In only two of these reports was the labeled monosaccharide identified as galactose, a crucial control, since homogenates can convert UDP-galactose to UDP-glucose. In the first report (Zatta et al., 1975), rat liver microsomes incubated with UDP-[ 14C]galactoseyielded a labeled lipid which comigrated with mannosylphosphoryl lipid during thin-layer chromatography in mixtures of chloroform, methanol, and water. Following mild acid hydrolysis of the washed lipid

5. GLYCOSYLATION OF MEMBRANE PROTEINS

195

extract, the radioactivity comigrated with galactose on paper chromatography. Since sugar nucleotides are also labile to mild acid hydrolysis and persist in a washed extract, this report would have been much more convincing if the analysis had been done on lipid purified from the extract. In a study of galactosylphosphorylretinol (Peterson et a / ., 1976), purified lipid was used; acidreleased label was found to comigrate with unlabeled galactose in four chromatographic systems. Therefore, it seems likely that both mannose (Rosso et al., 1977) and galactose can be found esterified to retinyl phosphate. 3. CHITOBIOSYLPYROPHOSPHORYL LIPID:SYNTHESIS

As mentioned earlier, N-acetylglucosaminylpyrophosphoryllipid was synthesized during incubations of membranes with UDP-N-acetylglucosamine. In addition, the formation of chitobiosylpyrophosphoryl lipid (Table I) was also observed (Leloir et af., 1973). (Chitobiose is the trivial name for a disaccharide of N-acetylglucosamine with a 1,4 linkage.) The chitobiosyl moiety appears to be linked to the phosphate group of the lipid in an a! anomeric configuration, based on a comparison with the mobilities of synthetic a! and p derivatives in three different solvent systems (Herscovics et a / ., 1978). In many reports using membranes from a variety of sources and labeled UDPN-acetylglucosaminc, the synthesis of both the mono- and disaccharide lipids was reported, although the proportion of each vaned depending on the metal ion (Kean, 1983) and the concentration of substrate (Keller et a l ., 1979) and membranes (Ghalambor et al., 1974) in the assay. Both products are also produced during incubations using enzyme preparations obtained by treating membranes with sufficient detergent (Heifetz and Elbein, 1977b) or sonic oscillations (Villemez and Carlo, 1980) to solubilize the activity. Chitobiosylpyrophosphoryl lipid was synthesized when labeled exogenous N-acetylglucosaminylpyrophosphoryl lipid was mixed with unlabeled UDP-N-acetylglucosamine, a finding indicating that the monosaccharide lipid was a precursor of the disaccharide lipid (Herscovics et a / ., 1978; Chen and Lennarz, 1977). Also, chitobiosylpyrophosphoryl lipid was made only when exogenous N-acetylglucosaminylpyrophosphoryl lipid was incubated with sugar nucleotide, not when exogenous lipid was incubated alone, a finding indicating that the sugar donor for the disaccharide lipid is the sugar nucleotide. The synthesis of chitobiosylpyrophosphoryl lipid from UDP-N-acetylglucosamine probably involves two enzymatic activities, although the definitive experiment awaits purification of the enzyme(s). The formation of N-acetylglucosaminylpyrophosphoryl lipid involves the transfer of a sugar phosphate and is inhibited by UMP (Harford and Waechter, 1979b; Zatta et al., 1976); the second reaction involves the transfer of a sugar and is inhibited by UDP (Zatta et al., 1976). As mentioned, the antibiotic tunicamycin inhibits the synthesis of Nacetylglucosaminylpyrophosphoryl lipid; it also inhibits the synthesis of the

196

SHARON S.KRAG

chitobiosyl derivative, if one looks at the coupled reaction starting from nucleotide sugar. However, when exogenous, labeled lipids were added to membranes and the conversion of mono- to disaccharide lipid was monitored in the presence of sugar nucleotide, no effect of tunicamycin was seen (Lehle and Tanner, 1976). This strongly suggests that there are two enzymatic activities, assuming that the potency of tunicamycin is not affected by the presence of exogenous saccharide lipids.

4. CHITOBIOSYLPYROPHOSPHORYL LIPID:FATES There are four possible fates of the chitobiosylpyrophosphoryl lipid pool (Fig. 3). First, it may serve as a substrate for conversion to mannosylchitobiosylpyrophosphoryl lipid. This conversion is catalyzed by a membrane-bound enzyme using the substrate GDP-a-mannose; the mannose is incorporated into the lipid as the p anomer (Levy et af., 1974; Heifetz and Elbein, 1977a; Chen and Lennarz, 1976). Second, the disaccharide from the lipid can be transferred directly to protein, a reaction which has been detected in membranes prepared from hen oviduct (Chen and Lennarz, 1977), brain gray matter (Harford and Waechter, 1979a), and yeast (Lehle and Tanner, 1978a). The transfer of N-acetylglucosamine and chitobiose to both polyisoprenoid derivatives and protein was detected in nondisrupted rat spleen lymphocytes (Hoflack et af., 1982). In this study, data from a pulse-chase experiment using UDP-N-acetylglucosamine suggested direct transfer of label from the sugar-nucleotideto the protein. Yet, the transfer of label during a pulse was inhibited by tunicamycin, a finding indicating the saccharide lipid was an 0

0

A-

6- 0 -Lipid

I1

II

Mannose-Chitobiose -0- P- 0-P-

GDP -mannose

0 0 II II Chitobiose-0-P- 0 - P - 0-Lipid

A- A-

UDP-glucose Chitobiosyl-Asn-Protein

0 0-P-

I1

0-P-0-Lipid

t Chitobiose

FIG. 3. Potential fates of the chitobiosylpyrophosphoryl lipid pool in eukaryotic cells.

5. GLYCOSYLATION OF MEMBRANE PROTEINS

197

intermediate in this system. The radioactivity found in the mono- and disaccharide lipid disappeared during the chase, but most appeared as free chitobiose rather than protein or oligosaccharide lipid. This result indicated a third possible fate of chitobiosylpyrophosphoryl lipid: hydrolysis. Of most interest was that some of the label (N-acetylglucosamine and chitobiose) recovered from the protein after hydrazinolysis had been converted during an incubation with unlabeled GDP-mannose to a form which could be retained on a concanavalin ASepharose column. This result suggested conversion of the disaccharide attached to the protein to a species containing at least some a-linked mannosyl residues. A fourth potential fate of the chitobiosylpyrophosphoryl lipid was suggested in 1976, by Zatta, Zakim, and Vessey when they reported the synthesis of a glucosylated, N-acetylglucosaminyl lipid. No ratio of glucose to N-acetylglucosamine in this novel lipid was determined, so its precise structure is unclear. A determination of which of these various possibilities occurs in vivo awaits further study. Analysis of the yeast mutant algl-l will be useful in this regard (Huffaker and Robbins, 1982). At the nonpermissive temperature, a f g l - l cells were unable to synthesize mild acid-labile, mannosyl-containing lipids because of a defect of the P-mannosyl transferase responsible for the addition of the first mannosyl residue. Interestingly, at the nonpermissive temperature these cells synthesized mannosyl compounds precipitable by trichloroacetic acid at 6 1% the rate of wild-type cells. It remains to be determined if the residual mannosylation results from transfer of chitobiose to protein and subsequent elongation of the glycose chains on the protein. Both wild-type cells and the algl-1 mutant cells also synthesized at the nonpermissive temperature uncharacterized N-acetylglucosaminyl lipids which might correspond to glucosyl or glucuronyl derivatives. The most widely accepted fate of chitobiosylpyrophosphoryl lipid is its conversion to the mannosylated trisaccharide lipid. This trisaccharide lipid can serve as a direct donor of its glycosyl moiety to protein (Chen and Lennarz, 1976). Alternatively, this trisaccharide lipid can be further elongated by the addition of mannose units to oligosaccharide lipids. This was first shown to occur in membranes prepared from hen oviduct (Lucas et al., 1975) and rat liver (Levy et a l . , 1974). Details of this elongation reaction will be presented in the next section.

5. OLtGOSACCHARIDE LIPIDSCONTAINING ONLY AND MANNOSE N-ACETYLGLUCOSAMINE Oligosaccharide lipids containing only N-acetyiglucosamine and mannose have been shown to be substrates for the transfer of the oligosaccharide to protein using purified, labeled oligosaccharide lipids and unlabeled microsomes prepared from either rat liver (Behrens et a l . , 1973), niurine myeloma (Hsu et al., 1974), or hen oviduct (Lucas et al., 1975; Pless and Lennarz, 1975). The oligosaccharide released from lipid by mild acid hydrolysis in these studies

198

SHARON S.KRAG

appeared to have five to seven mannose residues, the majority in a linkages, and a chitobiosyl core at the phosphoryl end of the oligosaccharide. This oligosaccharide was sensitive to hydrolysis by endo-p-N-acetylglucosaminidase H (Tarentino et al., 1978; Chen et al., 1975), the reaction yielding an oligosaccharide and N-acetylglucosamine. Treatment of the intact oligosaccharide with exoglycosidases, first a-mannosidase and then P-mannosidase, resulted in the production of chitobiose. When the order of enzymes was reversed, no hydrolysis occurred (Chen et al., 1975; Lucas et al., 1975). The data from both the endo- and exoglycosidase treatments are consistent with the structure of M,S shown in V.

\

M a n L GlcNAc- GlcNAc- P- P- Lipid

/

V

Another oligosaccharide containing only mannose and N-acetylglucosamine has been detected as a minor species in Chinese hamster ovary cells (Li and Kornfeld, 1979), both uninfected and infected with vesicular stomatitis virus. A variety of techniques including treatment with exoglycosidases and endoglycosidases, acetolysis, methylation, and periodate oxidation followed by NaBH, reduction (Smith periodate degradation) were performed on material isolated from cells incubated with [2-3H]mannose. The structure proposed for MSReSfrom these studies is shown in VI. Man a \

196

M y - GlcNAc-GlcNAc-P-P-Lipid

Man

a 1,2 a 1,2 Man -Man

VI

5. GLYCOSYLATION OF MEMBRANE PROTEINS

199

An important structural difference between VI (MsReS) and V, which was labeled during in vitro reactions (M,S), is the lack of susceptibility to endoglycosidase H, a finding which suggests that the a 1,6-mannose is not blocked in VI (Kobata, 1978). Although the oligosaccharide used for these structural studies appeared homogeneous after descending paper chromatography, it was either a mixture of oligosaccharides or the mannoses in the molecule were not uniformly labeled. The former possibility was suggested by the heterogeneity of acetolysis products (acetolysis preferentially cleaves (Y 1,6 linkages). The latter possibility was suggested by the lower than expected ratio (3.5 to 1) of radioactivity in the products of digestion with a-mannosidase, namely, mannose and trisaccharide. The fate of this oligosaccharide (MsReS) was examined during a 10-minute incubation of cells with [3H]glucosamine followed by a 10-minute incubation in unlabeled medium (Li and Kornfeld, 1979). The label in the oligosaccharide lipid containing the MsRes structure disappeared completely during the chase period, while the amount of the label in larger oligosaccharide lipids increased. The possibilities that the MsRe5oligosaccharide was also transferred to protein and/or hydrolyzed from the lipid were not examined directly.

6. OLIGOSACCHARIDE LIPIDSCONTAINING N-ACETYLGLUCOSAMINE, MANNOSE,AND GLUCOSE Oligosaccharides containing all three sugars (N-acetylglucosamine, mannose, and glucose) have been detected attached to isoprenyl lipids by an acid-labile, phosphoglycosidic bond. Their presence has been reported in thyroid slices (Spiro et al., 1976a), slices from hen oviduct, thymus, kidney, and liver (Spiro er al., 1976b), and pancreas microsomes (Herscovics et al., 1977a). Cell-free incubations of microsomes from cultured fibroblasts (Robbins et al., 1977b), hen oviduct (Chen and Lennarz, 1978), and brain white matter (Scher et al., 1977) with mixtures of UDP-N-acetylglucosamine, GDP-mannose, and UDP-glucose resulted in a lipid-linked oligosaccharide containing all three sugars. Molecules containing six to eight or more sugars linked to a large polyisoprenyl lipid such as dolichyl phosphate have unusual solubility properties: they are insoluble in water and in ch1oroform:rnethanol 2: 1 (vol/vol) mixtures, but are soluble in mixtures of ch1oroform:methanol:water 10:10:3 (Behrens et al., 1973). When the ratio of gluc0se:mannose:N-acetylglucosamine in unlabeled material extracted from calf pancreas microsomes (Herscovics et al., 1977b) and calf thyroid tissue (Spiro et al., 1976a) was determined, values ranging from 1.4:7.0:2 to 1.25:6.0:2 were obtained. Purification by DEAEcellulose chromatography of the phosphooligosaccharide lipid from possible glycosphingolipid contamination yielded ratios of 1.4:8.0:2.0 for the oligosaccharide lipid isolated from thyroid and ratios ranging from 0.8:5.0:2.0 to 2:8.2:2

200

SHARON S. KRAG

for the molecule isolated from pancreas. After mild acid hydrolysis and purification of the resultant oligosaccharide by chromatography on BioGel P (acrylamide) columns, the ratio of the component sugars (g1ucose:mannose:N-acetylglucosamine) was 1.5: 10.9:2.0 for the molecule from thyroid tissue and 7.5:8.8:2.0 for a similar preparation from cultured Chinese hamster ovary cells (Li et al., 1978). The structure of labeled oligosaccharide released by mild acid hydrolysis of the lipid was analyzed in detail in two systems. Labeled material was obtained either by incubating thyroid slices with radioactive glucose (Spiro et al., 1976c) or CHO cells with [2-3H]mannose (Li er al., 1978). Both studies were performed on a mixture of oligosaccharides containing nonuniformly labeled material, as indicated by the data obtained after acetolysis of the oligosaccharide. It will be informative to reexamine the structural features of these molecules, after separation by high-pressure liquid chromatography. The structures of the oligosaccharide composed of mannose, glucose, and Nacetylglucosamine produced in these two studies had common features. First, there were a l , 6 branches containing two and three mannoses linked to a Man,% ManKGlcNAc-GlcNAc core. Second, there were also mannose residues in the structures which were resistant to digestion by a-mannosidase, apparently blocked by glucose residues. There was, however, one major difference. Two large fragments resulted from acetolysis of the oligosaccharides labeled by CHO cells; they both contained glucose, mannose, and N-acetylglucosamine. In the oligosaccharides from thyroid, one of the large acetolysis products appeared to contain only mannose and glucose, suggesting the presence of glucose on branches linked a l , 6 to the oligosaccharide. The structure proposed by Li er al. (1978) for the CHO oligosaccharide on the basis of their studies was identical with structures found on secreted glycoproteins except that it contained glucose on the a l , 3 branch (Kornfeld and Kornfeld, 1980). Their structure is shown in VII. Man-

ff 1,2

Man Man 0

Man

(Glc),-Man

1,y

012

-Man

l t 2 Man

(y1s2

\,I,, Man- GlcNAc- GlcNAc- P- P- Lipid

10

Man

1,s

5. GLVCOSVLATIONOF MEMBRANE PROTEINS

201

C. Synthetic Pathway The first steps in the assembly of the oligosaccharide of the saccharide lipids which contain N-acetylglucosamine, mannose, and glucose were detailed in the last section and are as follows: UDP-GlcNAc + Lipid-P $ GlcNAc-P-P-Lipid + UMP GlcNAc-P-P-Lipid + UDP-GlcNAc 8 ( G I C N A C ) ~ - P - P -+L ~UDP ~~~ ( C I C N A C ) ~ - P - P - L+~ PGDP ~~ GDP-Man F? Man (GlcNAc),-P-P-Lipid

+

This section will deal with the elongation reactions which convert the trisaccharide to larger oligosaccharides, the structures of the oligosaccharides which are transferred to proteins, and the transferase reaction itself. 1. REACTIONSCONVERTING TRISACCHARIDE LIPIDTO OLIGOSACCHARIDE LIPID The general approach used in defining the steps of the pathway entails (1) incubation of membrane preparations with labeled sugar nucleotides (in vitro) or cells with labeled sugars (in vivo);( 2 ) extraction of labeled oligosaccharide lipids from the mixture; (3) hydrolysis of oligosaccharide lipids to free oligosaccharides; (4)separation of the oligosaccharides by gel filtration or paper chromatography, and (5) structural analyses of the oligosaccharides using both a battery of exo- and endoglycosidases and chemical treatments such as acetolysis (Tai et al., 1975), Smith periodate oxidation (Spiro, 1966), and methylation (Hakomori, 1964). The basic assumption in these studies is that the oligosaccharides analyzed are the primary intermediates. However, it is possible that the oligosaccharides obtained in quantities sufficient for analysis are those which accumulate because they are elongated or transferred more slowly. Intermediates that have a rapid rate of turnover may not be detected in such studies. For example, mannosylphosphoryldolichol is rarely detected in cells following incubation with labeled mannose (Krag, 1979). Most oligosaccharides analyzed to date exhibit an uneven distribution of label in the various glycosyl moieties. This probably happens because both membrane preparations and cells contain unlabeled pools of endogenous saccharide lipids. Also, at least in the case of mannose, there are two glycosyl donors, sugar nucleotide and mannosylphosphoryldolichol. Nonuniformity of labeling does not necessarily affect the identification of the oligosaccharide, but it may mask isomers of oligosaccharides which copurify . Also, without uniform labeling, precursor-product relationships are obscured. The variations in the sizes of endogenous saccharide lipid pools may also affect which oligosaccharides are labeled sufficiently to permit analyses.

202

SHARON S. KRAG

The transfer of a mannosyl residue in an a l , 3 linkage to the trisaccharide lipid was demonstrated by experiments of Vijay et al. (1980). In these studies membranes prepared from lactating bovine mammary tissues were incubated with various labeled sugar nucleotides. These membranes contained unlabeled N acetylglucosaminylpyrophosphoryldolichol.This was determined by analyzing the distribution of radioactivity in glucosamine and glucosaminitol in saccharides reduced with NaBH,, labeled in the presence of UDP-N-[3H]acetylglucosamine, and finally subjected to strong acid hydrolysis. In the case of the trisaccharide, ManK(GlcNAc),, the ratio of labeled glucosamine to glucosaminitol was 1:0.77. Interestingly, the label in the two glucosaminyl and the two mannosyl residues in the tetrasaccharide Man"ManA(GlcNAc), was uniformly distributed; the distribution of label between the mannosyl residues was determined by analysis of the products of digestion by a-mannosidase. These constrasting results suggest that the trisaccharide- and tetrasaccharide-lipids arose from different pools. The linkage between the mannoses in the tetrasaccharide was found to be a 1,3 based on (1) release of mannose and a trisaccharide after a-mannosidase treatment; (2) cleavage of the molecule by endoglycosidase D (Muramatsu, 1978) to a trisaccharide and N-acetylglucosamine; (3) resistance to degradation during acetolysis (which preferentially cleaves 1,6 linkages); (4) lack of hydrolysis of the molecule by endoglycosidase H (Kobata, 1978); and ( 5 ) isolation of 2,4,6trimethylmannose after methylation analysis. In these studies, the donor of the mannosyl residues was not identified. An al,3-mannosyl transferase was recently solubilized by nonionic detergents from calf pancreas microsomes (Herscovics et al., 1983). In this case, a synthetic trisaccharide-P-P-lipid rather than an endogenous one was used as substrate. The tetrasaccharide released from the lipid was characterized by its chromatographic properties (gel filtration and HPLC) and sensitivity to glycosidases. The transfer of a 1,6-mannosyl residue to the trisaccharide lipid has not been detected either in vitro or in vivo. However, a tetrasaccharide lipid with the structure ManzManA(GlcNAc),-P-P-lipid was isolated from porcine liver by Jensen et al. (1980). A pentasaccharide lipid of the structure shown in VIII was obtained by mixing the unlabeled tetrasaccharide lipid with GDP-[ l4C]manMall

\.

196

M ~ P ( G ~ c N A c P) ~P-Lipid -

/.

193

[14C]Man

Vlll

203

5. GLYCOSYLATION OF MEMBRANE PROTEINS

nose, MgCl,, and a solubilized microsomal preparation from rabbit liver (Jensen et al., 1980). The reactions were performed in the absence of exogenous dolichyl phosphate; under these conditions no [ ~4Clmannosylphosphoryldolichol was detected, indicating that the al ,3-mannosyl transferase utilized GDP-mannose as substrate. Rabbit liver microsomes also contained a 1,2-mannosyl transferases (Schutzbach et ul., 1980) which transferred (14C]rnannose from GDP-[ 14C]mannose to an endogenous acceptor lipid, forming a heptasaccharide-lipid of the structure shown in IX.

\.

Man

196

Man- (GlcNAc X- P-P- Lipid /

Ql2

a1 2

[ "C 1 Man A ['T] Man A Man IX

This heptasaccharide is the MJRebstructure mentioned earlier, which is characterized by resistance to endoglycosidase H. The in vitro reactions described thus far resulted in the labeling of the mannosyl residues on the a I ,3 branch. The mannose linked al,6 in the MgReastructure was labeled preferentially (70%of the label released as mannose after acetolysis) during incubations of particulate preparations of porcine aorta with GDP-[I4C]mannose (Chambers et al., 1977). This reaction was not inhibited by EDTA (Spencer and Elbein, 1980) or by amphomycin (Kang ef al., 1978), both of which inhibit the synthesis of mannosylphosphoryldolichol. GDP-mannose seems to be the donor for all the mannosyl residues in MsKeb. This finding was corroborated in studies using membranes from the Thy- 1 murine lymphoma cell line (Rearick et al., 1981b). Thy-I cells do not synthesize mannosylphosphoryldolichol (Chapman et ul. , 1980), yet the M5ReS structure was obtained when membranes were incubated with GDP-I 14C]mannose. An unlabeled pentasaccharide lipid was present in these membranes since the labeled structure X was found. ~

\.

Man

196

Man- (GlcNAch- P- P- Lipid I

[IiC]Man-

(Y

1,2

Q 1,2 [ LIC]Man ___ Man

X

204

SHARON S. KRAG

After labeling wild-type Chinese hamster ovary cells with [2-3H]mannose, two oligosaccharide lipids with saccharide moieties of less than seven units were isolated (Li and Kornfeld, 1979). The first was M a n G M a n l ( G l c N A c ) , ; the distribution of label in the mannose residues was 49%:51%. In contrast, the labeling pattern in the other oligosaccharide lipid structure, a heptasaccharide lipid (XI),was uneven (% denotes percentage of total label at each position),

indicating the presence of unlabeled oligosaccharide lipids in these cells. Preferential labeling of the a1,6 residue of this heptasaccharide lipid was also observed in Thy-1- lymphoma cells (Chapman el al., 1980). In addition to the tetra and heptasaccharide lipids labeled in Chinese hamster ovary (CHO) cells, small amounts of three others were detected. After performing structural analyses on all five oligosaccharides, Chapman er al. (1979b) proposed an order of mannosyl addition leading to the MgRes in CHO cells; the order is designated by the Roman numerals shown in XII. This pathway is not

\.

Man

'I1

1,6

ManP(G1cNAck-P-P-Lipid a 1,2 Man V

~

ff

Man

Iv

~

1,2

p 1 , 3

Man I1 XI1

consistent with all of the experiments listed above, for example, the finding of M a n u M a n l ( G l c N A c ) , tetrasaccharide lipid in porcine liver. Whether there are different pathways operating in different cells or more than one pathway in the same cell remains to be determined. It is clear that the heptasaccharide is a transient species. When CHO cells were given a 2.5-minute pulse of [2-3H]mannose, during which time the M5ReSwas labeled, followed by a 2.5-minute chase, all of the radioactivity in the MgRes

205

5. GLYCOSYLATION OF MEMBRANE PROTEINS

species disappeared (Hubbard and Robbins, 1980). What are the metabolic fates of this intermediate'? First, the n/15ReS species can be glucosylated. This was shown most clearly in the Thy-I - lymphoma cells (Chapman er al., 1979b). Oligosaccharides of the structure shown in XI11 were labeled during incubations with either [2-3H]manMan Man- (GlcNAc A- P- P- Lipid /

*

(Glc),,,-Glc-Man

Man

Man Xlll

nose or [3H]galactose. (The latter appears as [ 3H]glucose in oligosaccharides via epimerization of UDP-galactose to UDP-glucose.) Second, the M5ReScan be elongated by further mannosylation. 114C]M,Res was prepared in calf brain membranes by incubating in the presence of amphomycin and GDP-[ ''C]mannose; membranes were then washed free of labeled sugar nucleotide and incubated with unlabeled GDP-mannose, amphomycin, and enough exogenous dolichyl phosphate to reverse the drug effect and allow the synthesis of mannosylphosphoryldolichol. The oligosaccharide obtained was larger than (Man), and was sensitive to endoglycosidase H (Banerjee er al., 1981). The endoglycosidase H-sensitive, mannosylated oligosaccharide lipid can be glucosylated to form oligosaccharide lipids containing one, two, or three glucosyl moieties (Hubbard and Ivatt, 1981). The structure of the oligosaccharide lipid containing nine mannoses, two N-acetylglucosamines, and three glucoses (G,M,, Li et af., 1978) is as shown in XIV. The donor of the five Q

1,2

Man Man

Man

Q

@

Q 1,2 Man 1

1,2

Q

(GlcNAch- P- P- Lipid

1,2 XIV

206

SHARON S. KRAG

circled mannoses is GDP-mannose, and the donor of the other four mannosyl residues is mannosylphosphoryldolichol. Early studies showed that glucosylphosphoryl lipid (GPL) could serve as a glucosyl donor for synthesis of glucosylated oligosaccharide lipids (Behrens and Leloir, 1970; Herscovics et al., 1977a; Parodi et al., 1972). In studies by Staneloni et al. (1980) and Murphy and Spiro (1981), labeled GPL was incubated with membranes prepared from either rat liver or calf thyroid in the presence of EDTA, which inhibits UDP-glucose-GPL interconversion. In both studies three glucosylated molecules were detected. In the case of the thyroid membranes, the molecules corresponded to (Glc) ,-,(Man),(GlcNAc),-P-Plipid. The label was evenly distributed among the glucosyl residues of the three molecules, indicating that they were all derived from the labeled GPL. The donor of the glucosyl moieties to the glucosylated M5ReSis not yet clear. One proposed pathway from M5ReSto G,M, is based on a structural analysis of the molecules detected after labeling CHO cells with [2-,H]mannose (Chapman et al., 1979b). The steps are indicated by Roman numerals in structure XV.

Man 1,2 . I , +

ff

Man Man

n

(Glc)3-

Iv

Man

012 Man

ff

1,2

~

\d,6

Man- (GlcNAch-P-P-Lipid

/.i

1,s

Man

xv No order was designated for addition of the ninth mannose versus addition of glucose since a nonglucosylated oligosaccharide with nine mannose residues (M,) was not detected. Oligosaccharides with structures consistent with this pathway, including M,, have also been detected following incubation of [3H]mannosylphosphoryldolichol and membranes from Thy- 1 - lymphoma cells (Rearick et al., 1981b) and of GDP-[14C]mannose and membranes from bovine lactating mammary tissue (Vijay et al., 1980). In general, more isomers were detected as products of these in vitro incubations than were seen in the in vivo experiments. For example, a structure consistent with reversing the order of steps I1 and 111 [see (XV)] was detected. Vijay and Perdew (1983) found less heterogeneity of labeled structures when the assay mixtures were supplemented with dolichyl phosphate, although the mechanism of this effect is unclear. By adding dolichyl phosphate,

5. GLYCOSYLATION OF MEMBRANE PROTEINS

207

one is simultaneously changing the ratio and the specific activity of the mannosyl donors and the levels of the endogenous, unlabeled oligosaccharide lipids. Structures have been detected which suggest that during the synthesis of oligosaccharide lipids, mannosyl residues are added to the mannose which is linked a 1,6 to the P-mannosyl residue before mannosyl residues are added to the mannose which is linked a1,3 to the P-mannosyl residue (Vijay and Perdew, 1983; Rearick et al., 198I b; Chambers et al., 1977). Addition of these mannoses to the (Y 1,6 branch occurred using MPL (Rearick, et al., 198Ib; Chambers et al., 1977). These results suggest that not all of the larger oligosaccharide lipids arise from MsReS. For example, the endoglycosidase H-sensitive structure shown in XVI was detected.

[ 3H]Man

LY ~

1,3

Man \

Man- (GlcNAch- P- P- Lipid /a 1,s

Man XVI

The addition of glucosyl residues to endogenous (Man),(GlcNAc),-P-P-lipid in vitro (Murphy and Spiro, 1981; Vijay and Perdew, 1982) resulted in three oligosaccharides differing by one glucosyl residue. The three oligosaccharides obtained after incubating membranes from lactating bovine mammary tissue with both labeled UDP-glucose and UDP-N-acetylglucosamine were released by mild acid hydrolysis, separated by paper and gel filtration chromatography, reduced with NaBH,, and converted to component monosaccharides by strong acid hydrolysis. Two of the labeled products, glucosamine and glucosaminitol, were then separated by paper chromatography. If the three oligosaccharides were homogeneous and synthesized from the same pool of mannosylated, glucosaminylated oligosaccharide lipid, then one would predict that the ratios of [3H]glucosamine to [3H]glucosaminitol from the three would be equal. However, ratios ranged from 1:0.81 to 1:0.6, suggesting multiple isomers and pathways. Different tissues and cells appeared to contain large pools of different endogenous oligosaccharide lipids (Table 11). This affects which oligosaccharide lipids are detected after radioactive labeling. Other factors shown to affect the extent to which different oligosaccharide lipids were labeled are the ratios of sugar nucleotides present in in vitro reactions (Liu et al., 1979; Robbins et al., 1977b), the levels of mannose and glucose

208

SHARON S.KRAG

TABLE I1 ENDOGENOUS OLIGOSACCHARIDE LIPIDS IN EUKARYOTIC CELLS

Tissue or cell

Endogenous oligosaccharide lipid

Bovine mammary glands

GlcNAc-P-P-Lipid

Porcine liver

Man=ManL(GlcNAc)2-P-P-lipid

Rabbit liver Thy- 1 - lymphoma

M",a1.6 ManJ(GlcNAc)2-P-P-lipid Man/CTL.3

Porcine aorta Thy- 1 - lymphoma Wild-type CHO cells

Man~Man~Man~Man~(Gl~NAc)~-P-P-lipid

present in in vivo experiments (Krag, 1979), and the method of stopping the in vivo incubations (Hubbard and Robbins, 1979). To reiterate, since the oligosaccharide lipids are intermediates, the ones most easily isolated are those metabolized the slowest and/or present in the largest amounts. Which oligosaccharide lipids one labels depends on the nature and size of the pools of unlabeled oligosaccharide lipids and on the relative pool sizes of the two mannosyl donors in the tissue or cell.

2. OLIGOSACCHARIDE STRUCTURES TRANSFERRED TO PROTEIN Which of the oligosaccharides is initially transferred from lipid to protein? To answer this question, the structures of the oligosaccharides from both oligosaccharide lipid and glycoprotein have been compared. These analyses are complicated by the fact that both oligosaccharides, the one from the lipid and the one from the protein, are transient intermediates. Also, since mixtures of protein acceptors are often employed in these studies, it is difficult to evaluate whether different proteins are glycosylated with different oligosaccharides. The oligosaccharide moiety present on many proteins at times soon after addition differs from that found on the mature protein. Newly synthesized glycoproteins were susceptible to P-N-acetylglucosaminidase H or C,, (Kobata, 1978), while mature proteins were not. This was first shown for G protein of vesicular stomatitis virus (Robbins et al., 1977a; Tabas et al., 1978; Hunt et al., 1978), PE, of Sindbis virus (Robbins et al., 1977a), and IgG (Tabas et al., 1978), and has subsequently been reported for a variety of proteins. Subsequent to transfer from the lipid intermediate, the protein-bound oligosaccharides may be acted upon by glucosidases, mannosidases, and glycosyl transferases, the combination of which produce oligosaccharide structures resistant to endoglycosidase H. After a very short (1.3-minute) pulse of secondary chick embryo fibroblast

5. GLYCOSYLATION OF MEMBRANE PROTEINS

209

cells with [2-,H]mannose, four labeled oligosaccharides were obtained from delipidated protein (Hubbard and Robbins, 1979); three (approximately 95% of the radioactivity) co-eluted on gel filtration chromatography with glucosylated, mannosylated structures previously characterized as G ,-,M9. The fourth oligosaccharide co-eluted with a mannosylated structure, M,. Approximately 30% of the label appeared identical to G,M,. Mixtures of oligosaccharides were also obtained from protein on varying the pulse up to 60 minutes. A mixture of oligosaccharides was also detected on the protein after a 10-minute pulse of [2-3H]mannose followed by a chase with nonradioactive glucose for periods of 10-120 minutes. Structures corresponding to G,M, or G,M, were not present after 30 minutes of chase, although approximately 10-20% of the radioactivity co-eluted with G,M, even after 120 minutes of chase (Hubbard and Robbins, 1979). From these kinetic studies, it is impossible to determine whether oligosaccharides other than G,M, were initially transferred from the lipid or arose by transfer of one oligosaccharide that was then modified to another. The modification of larger oligosaccharide to smaller ones by hydrolases has already been mentioned. In addition, it is possible that oligosaccharides can be elongated while on the protein; the transfer of glucose directly from UDP-glucose to proteins susceptible to endoglycosidase H has been reported in thyroid microsomes (Ronin and Caseti, 1981) and thyroid slices (Parodi et al., 1983). A number of results suggest that although the G,M, is transferred from lipid to protein, it is not the only oligosaccharide that can be transferred. Four mutant cell lines have been characterized which do not synthesize G3M, oligosaccharide lipid, but do glycosylate proteins. Two of these, Thy-1 - lymphoma cells (Chapman et al., 1980) and B4-2-1 CHO cells (Stoll et al., 1982), do not synthesize mannosylphosphoryldolichol. The largest lipid intermediate they synthesize is a truncated oligosaccharide lipid which can be glucosylated and is resistant to endoglycosidase H. All oligosaccharides found on the proteins are also resistant to endoglycosidase H. The oligosaccharides which were initially transferred ( 10-minute pulse) by the Thy- I cells were G,M, (major), G,M,, and G IM, (Kornfeld et al., 1979); the labeled oligosaccharides attached to lipid were MSReS,G,M,, and G,M, (Chapman et al., 1979a). In parallel experiments using lymphoma cells, labeled G,M, and G,M, structures were found on lipids while at 10 minutes only labeled G,M, and M, oligosaccharides were attached to protein. One might argue that the Thy-I - mutant uses a pathway not normally operating in the parental cells because the initial oligosaccharides detected on most proteins, including those of lymphoma and CHO cells, are sensitive to endoglycosidase H. However, the results of Yamashita et al. (1983) suggest that transfer of oligosaccharides resistant to endoglycosidase H may occur normally. In this study, structural analyses were performed on most of the glycopeptides of hen ovomucoid; they found an oligosaccharide with the structure shown in XVII. This structure could not ~

210

SHARON S.KRAG

\

A a n L (GlcNAch-Asn< I

Man XVll

have arisen from an endoglycosidase H-sensitive oligosaccharide by a-mannosidase action since the removal of the last two mannosyl residues linked to the a 1,6-mannose requires prior transfer of an N-acetylglucosaminyl residue to the al,3-mannose (Kornfeld et al., 1978). A similar structure suggesting normal transfer of oligosaccharides resistant to endoglycosidase H has recently been reported in the linkage region for keratan sulfate (Nilsson et al., 1983). Two other cell lines synthesize endoglycosidase H-sensitive, lipid-linked oligosaccharides which are smaller than G3M,; these cells mannosylate their proteins. One of these cell lines, the phytohemagglutinin-resistant,concanavalin A-resistant CHO cell termed Lecl.Lec6 (Stanley, 1984), has two defects: (1) it lacks the ability to add an a-mannosyl residue to a M, oligosaccharide lipid (Hunt, 1980a,b) and (2) it lacks N-acetylglucosaminyl transferase I activity (Stanley et al., 1975). The other cell line, the concanavalin A-resistant CHO cell B211, is unable to glucosylate its oligosaccharide lipids and proteins in vitro or in vivo, although it does synthesize glucosylphosphoryl lipid (Krag, 1979, and unpublished observations). The key point is that in these two cell lines, as in Thy-1 - and B4-2-1, the failure to synthesize G,M, did not lead to cessation of glycosylation; rather, the synthesis of an oligosaccharide other than G3M, resulted in the transfer of the altered oligosaccharide to protein acceptors. The results on the four cell lines mentioned above indicate that G3M, is not the only oligosaccharide that can be transferred from lipid to protein. Additional support for this idea may be obtained from studies of a lymphoma cell line that has recently been characterized as deficient in glucosidase 11, the enzyme which normally cleaves the last two glucoses from the protein-bound oligosaccharide (Reitman et al., 1982). Membranes prepared from the mutant have 0.3% of the parental enzymatic activity in reactions using [3H]G,M, structures as substrates. These cells accumulated glucosylated oligosaccharides on their proteins; however, a variety of nonglucosylated oligosaccharides were also present on the proteins. Also, oligosaccharide structures resistant to endoglycosidase H were found on mature proteins. Oligosaccharides which remain glucosylated on the a1,3 branch are not processed to endoglycosidase H-resistant forms because removal of the last mannosyl residues linked to the a 1,6-mannose of the oligosaccharides depends on the addition of an N-acetylglucosaminyl residue to the al,3-linked

5. GLYCOSYLATION OF MEMBRANE PROTEINS

21 1

mannose (Narasimhan et a l . , 1977; Kornfeld et al., 1978; Harpaz and Schachter, 1980); fucosylation of an oligosaccharide also depends on the addition of that Nacetylglucosaminyl residue (Wilson et al., 1976). Little alteration was seen in the incorporation of glucosamine and mannose into proteins of the glucosidasedeficient mutant as compared with wild-type cells (Trowbridge et al., 1978b). Thus, it is possible that the defect seen in vitro is not expressed in vivo. Alternatively, it may be that nonglucosylated oligosaccharides are transferred, circumventing the block in the mutant. Further characterization of this mutant, complemented with studies using glucosidase inhibitors, such as 1 -deoxynojirimycin (Saunier et al., 1982), may distinguish between these alternatives. Studies with two parasites, Crithidia fasciculata and Trypanosoma cruzi, suggest that in these organisms oligosaccharides transferred from lipid to protein do not contain glucose (Parodi and Quesada-Allue, 1982; Parodi and Cazzulo, 1982; Parodi et al., 1981). In Trypanosoma, labeled glucosylphosphoryldolichol, the glucosyl donor, was not detected; the largest and major oligosaccharide found attached to lipid had nine mannosyl residues and two N-acetylglucosaminyl residues, but contained no glucose. In Crirhidia, the only labeled oligosaccharide attached to lipid had seven mannosyl and two N-acetylglucosaminyl residues. Interestingly, in both cases some of the oligosaccharides found attached to protein at early times had already been modified by the addition of either glucose or galactose. In vitro experiments examining transfer of oligosaccharides from labeled oligosaccharide lipids to protein acceptors also indicate that a variety of structures can be transferred. Many such experiments were cited in preceding sections, illustrating the transfer of many different oligosaccharides. The effect of glucose on the efficiency of transfer has been studied. Microsomes from thyroid (Murphy and Spiro, 1981) and rat liver (Staneloni et al., 1980) transferred G,M, from lipid to protein but did not utilize G,M, lipid as an oligosaccharide donor. On the other hand, microsomes prepared from NIL cells (Turco and Robbins, 1979) transferred both G,M, and G,M, from their respective lipids to protein, although G,M, appeared to be transferred preferentially. Comparisons between glucosylated and nonglucosylated oligosaccharide lipids as donors in in vitro reactions have been done in two ways. First, the ability of an oligosaccharide lipid to be a donor to endogenous protein in microsomes prepared from thyroid was examined before and after treatment of the oligosaccharide lipid with a rnembrane-bound glucosidase (Spiro et al., 1979a). Transfer of the treated oligosaccharide was 5% of that of the untreated one. The glucosidase used in these experiments was from thyroid; the product of digestion with this enzyme did not appear to be completely sensitive to further digestion by a-mannosidase (Spiro et al., 1979b), indicating that all of the glucosyl moieties may not have been removed. Second, the various species were differentially labeled, by incubating micro-

212

SHARON S. KRAG

somes with GDP-[ 14C]mannose or GDP-mannose in the presence or absence of UDP-glucose or UDP-[3H]glucose. The labeled molecules were mixed and added to unlabeled membranes, which contained the transferase and endogenous acceptors; incubations were done with or without exogenous peptide acceptors. Using membranes prepared from NIL cells and endogenous protein acceptors (Turco et al., 1977), the glucosylated oligosaccharide lipid appeared to be a better donor than the nonglucosylated molecule, in terms of both rate and extent of transfer; similar experiments done with microsomes from thyroid and exogenous protein acceptors indicated no differences in the rate or extent of transfer of glucosylated and nonglucosylated oligosaccharides from the lipids (Ronin et al., 198la). Using detergent-solubilized preparations of membranes prepared from yeast and monitoring transfer to endogenous protein, one group found the extent of transfer of the two types of oligosaccharides was the same, but the rate of transfer of the glucosylated oligosaccharide was 15-fold higher (Trimble et al., 1980). Another study comparing the rate of transfer of glucosylated and a number of nonglucosylated oligosaccharides to exogenous protein acceptors by detergent-solubilized preparations of membranes from yeast (Sharma et al., 1981) found the rates to differ 50-fold; the structures G,M,, Man%GlcNAc),, and (GlcNAc), were transferred efficiently, while very little transfer of M, and GlcNAc occurred. The differences in rate lessened and the relative efficiencies changed when examining transfer to endogenous protein acceptors that were solubilized with the transferase. No experiments comparing oligosaccharide lipid substrates were done using the transferase activity (about 30%) which could not be solubilized from the membranes. Clearly, there are differences in efficiency of transfer depending on the experimental system; these differences are usually attributed to expected variation in the interaction of a variety of lipid-linked substrates with a single transferase. They could also reflect variable levels of protein acceptors in the preparations. Finally the oligosaccharide transferase has yet to be purified to homogeneity; it is possible that the differences are due to multiple transferases, all present in the same membrane or solubilized preparations. 3. OLIGOSACCHARIDE TRANSFERASE The oligosaccharide transferase has been solubilized by both nonionic detergents such as NP-40 and by mixtures of salt and deoxycholate from microsomes prepared from yeast (Trimble et al., 1980; Sharma et al., 1981), hen oviduct (Aubert et al., 1982; Das and Heath, 1980), and porcine thyroid (Ronin, 1980). The solubilized activity was assayed using both exogenous oligosaccharide lipids and peptide acceptors, although both endogenous lipids and protein acceptors were also present in the solubilized extracts (Trimble et d., 1980; Sharma et al., 1981). In addition, activities competing for the substrates and

5. GLYCOSYLATION OF MEMBRANE PROTEINS

213

products may also be solubilized, as was the glucosidase in the thyroid system (Ronin, 1980). All the solubilized preparations have optimal activity in the presence of manganese ions. The pH optimum of the activity varies from 6.5 to 8.5, depending on the source. In two cases, the solubilized transferase has been purified approximately 2000-fold using affinity chromatography; columns were made with a-lactalbumin (Das and Heath, 1980) or a tetradecapeptide with a potential glycosylation site (Aubert et a/., 1982). In the latter case, the purified fraction was analyzed by polyacrylamide gel electrophoresis in sodium dodecyl sulfate and was found to contain three bands as detected by staining with Coomassie blue. However, there was no indication of which, if any, of the bands was the transferase. The pH optimum and ion specificity of these partially purified transferases did not differ from that of the membrane-associated activity. No studies comparing different oligosaccharide lipid substrates were reported. The subcellular location of the oligosaccharide transferase has yet to be determined by a method which is independent of endogenous lipid substrates, protein acceptors, or both. When labeled glucosylated oligosaccharide lipids were added to separated rough (containing ribosomes) and smooth microsomes and incorporation of label into endogenous proteins measured, both fractions had equal specific activities (Vargas and Carminatti, 1977). However, the level of glycosylation in these experiments may reflect the level of endogenous proteins rather than the activity of the transferase itself. Experiments utilizing submicrosomal fractions of hen oviduct, GDP-[ 14C]mannoseand exogenous protein acceptor indicated 10-30 times the specific activity of the transferase in rough microsomes as compared with that in smooth microsomes (Czichi and Lennarz, 1977). In these experiments one must assume that oligosaccharide lipid intermediates formed in both submicrosomal compartments were the same and had the same specific activities. Three enzymes, the ones which transfer mannose, glucose, and N-acetylglucosamine I-phosphate to dolichyl phosphate, are equally active in rough and smooth microsomes (Ravoet et al., 1981). 1 have discussed the oligosaccharide lipid substrate of the oligosaccharide transferase, but what is known about the protein-peptide substrate? It appears that the tripeptide Asn-X-Ser(Thr) is a necessary requirement for glycosylation. Analyses of 25 glycoproteins showed the sequence Asn-X-Ser(Thr) was invariant at the asparaginyl-carbohydrate linkage (Marshall, 1974). The synthetic peptide Asn-Ala-Thr was a substrate for the in vitro transfer of mannosyloligosaccharide using thyroid microsomes and labeled oligosaccharide lipid (Ronin et al., 1981b); blocking the amino terminus of the synthetic tripeptide Asn-X-Ser(Thr) with an acetyl group substantially increased its acceptor activity in a number of systems (Hart et al., 1979; Ronin et al., 1981b; Bause and Legler, 1981). Blocking the amino terminus with benzoyl and octanoyl groups resulted in a still greater increase of the acceptor activity in membrane preparations of hen

214

SHARON S. KRAG

oviduct (Welply et al., 1983). An increase in the acceptor activity of a hydrophobic heptapeptide, in this case a dinitrophenyl amino-terminal derivative, was observed when its acceptor activity was compared to that of an unmodified peptide in a microsomal system; no difference between these two peptides was observed when a solubilized preparation of the transferase was used (Ronin et al., 1978). Finally, the acceptor activity of the tripeptide also increased when the carboxy terminus was blocked (Hart et al., 1979). Interpretation of these comparisons among peptides with differing modifications is complicated by the questions of the substrate’s membrane permeability and solubility and by the alterations in the activity of the transferase itself when in the membrane, when in the presence of detergent, and when solubilized by detergent. Peptides with almost any residue as X appear to function as acceptors; for example, X can be a cysteine involved in a disulfide bond (Bause et al., 1982). However, X cannot be proline (Ronin et al., 1978; Bause, 1983a). The necessity of a hydroxy (or -SH) amino acid in the next position of this particular sequence was shown by comparing the acceptor activity of five different peptides Thr-AsnGly-Y-Ser-Val in an incubation with bovine microsomes and labeled chitobioseP-P-lipid (Bause and Legler, 1981). When Y was Thr, Ser, or Cys, the peptide was active; when Y was Val or O-methylthreonine, the peptide was not. Peptides containing derivatives of the asparagine residue were not acceptors (Welply et al., 1983). These studies not only suggest elements of the peptide acceptor which are important in interacting with the enzyme, but they also provide information which will lead to the synthesis of inhibitors of the transferase. One such inhibitor, an epoxy-peptide derivative, has been described (Bause, 1983b). Inhibitors, both irreversible and reversible ones, will be valuable for the purification, localization, and characterization of the transferase. Not all proteins which contain the Asn-X-Thr(Ser) sequence are glycosylated. These sequences in cytoplasmic proteins and the cytoplasmic domains of transmembrane proteins are perhaps not glycosylated because they are not in the same subcellular compartment as the transferase. However, not all these sequences in secreted proteins, which are located in the same compartment, are glycosylated. What other features of the peptide/protein acceptors are known to be necessary for activity? Bause et al. (1982) synthesized peptides containing the Asn-Gly-Thr triplet sequence and two cysteine residues at variable positions on either site of the triplet. They compared the kinetics of transfer using these peptides in the oxidized (cyclic) or reduced forms as exogenous acceptors in reactions containing All the cyclic forms microsomes of bovine liver and [ 14C]chitobio~yl-P-P-lipid. but one were poorer substrates (in general had higher K,’s) than the linear peptides, suggesting that restricting the conformation of the peptide results in loss of acceptor activity. The one peptide which had the same activity in the linear versus cyclic form was not one surrounded by cysteine residues but rather

5. GLYCOSYLATION OF MEMBRANE PROTEINS

215

the Cys-Thr-Asn-Cys-Thr-Ser-Val. Thus, the mobility of the Thr residue appears important, supporting the idea that in the glycosylation site there is a hydrogen bond formed between the oxygen of the threonine and the P-amide of the asparagine (Marshall, 1974; Bause and Legler, 1981). This hydrogen bond would also be disrupted if X were replaced by proline (Bause, 1983b; Ronin et al., 1978) or if the Thr were replaced by valine (Bause and Legler, 1981); as mentioned above, both situations are known to destroy acceptor activity. Other potential determinants for peptide acceptor activity are composition and chain length. The effect of composition is illustrated by comparison between the linear peptides Cys-Asn-Gly-Thr-Cys-Gly and Tyr-Asn-Gly-Thr-Ser-Val; the former peptide has 20% the acceptor activity of the latter (Bause et al., 1982). By comparing peptides ranging from six to nine amino acids, these workers suggested a correlation might exist between the number of amino acids in the peptide and acceptor activity (Bause et al.. 1982); however, the correlation was low, suggesting instead that compositional differences are more important than length. No correlation between acceptor activity and chain length was seen in an earlier study using hen oviduct membranes comparing the acceptor activity of alactalbumin and three of its peptide fragments (Struck et al., 1978). Denatured a-lactalbumin ( 123 amino acids) and a heptapeptide containing the glycosylation site were two to three times more active as acceptors than a peptide of 90 amino acids which also contained the glycosylation site. The ability of this triplet sequence to be glycosylated when present in a protein has been studied in an in vitro system of hen oviduct membranes and oligosaccharide lipid (Pless and Lennarz, 1977) or sugar nucleotide (Kronquist and Lennarz, 1978). When three known glycoproteins (ovalbumin, a-lactalbumin, and ribonuclease) were added as their nonglycosylated or deglycosylated forms to these reactions, no addition of carbohydrate occurred. However, when the proteins were denatured by procedures of sulfitolysis, S-carboxymethylation or S-aminoethylation, they served as glycosyl acceptors. When denatured under these conditions, some proteins not normally glycosylated, but containing the crucial tripeptide sequence, became glycosylated. Other denatured proteins served as acceptors only after cleavage to peptides using cyanogen bromide. These results strongly suggest that in many mature glycoproteins, the glycosylation site is not accessible to the membrane-associated transferase. These observations and the finding of glycosylated nascent chains of ovalbumin (Kiely et al., 1976; Glabe et al., 1980), immunoglobulin heavy chain (Bergman and Kuehl, 1977), and the G protein of vesicular stomatitis virus (Rothman and Lodish, 1977; Rothman et al., 1978) have led to the hypothesis that addition of the oligosaccharide from the lipid to the protein is a cotranslational event, occurring when the tripeptide in the nascent chain is exposed in a denatured form to the transferase. In the case of G protein, posttranslational glycosylation is not thought to occur (Rothman et al., 1978). Translation of VSV

216

SHARON S.KRAG

mRNA using a wheat germ extract and dog pancreatic microsomes resulted in the glycosylation of G protein. In microsomes pretreated with detergent, under- and nonglycosylated forms of G protein were synthesized. An additional incubation (approximately 2.5 times as long as the original translation incubation) resulted in no conversion of these underglycosylated molecules to fully glycosylated ones, a finding suggesting that only nascent chains are acceptors. However, the integrity of the glycosylation system of the treated membranes was not monitored after the original incubation. In the cases of murine immunoglobulin heavy chain (Schubert, 1970) and rat a,-acid glycoprotein (Jamieson, 1977), the addition of the oligosaccharide containing N-acetylglucosaminyl and mannosyl residues is thought to occur after synthesis of the protein is complete. Incubation of crude membranes or microsomal preparations with GDP-[ I4C]mannose resulted in the incorporation of [ 14C]mannoseinto endogenous proteins which migrate during electrophoresis in polyacrylamide gels as discrete bands (DeRosa and Lucas, 1982; Harford and Waechter 1979a; Hanover and Lennarz, 1980). The labeled material produced by hen oviduct membranes did not appear to be ovalbumin or nascent chains of ovalbumin (Hanover and Lennarz, 1980). The mobility of the labeled material was not changed by pretreatment with base, which would discharge nascent chains from tRNA (DeRosa and Lucas, 1982). In most cases the identity of these endogenous proteins was unknown, so a comparison of the apparent molecular weights of the labeled material and authentic material was not possible. In one case, however, this comparison could be made; incubation of membranes prepared from Sindbis-infected chicken embryo fibroblasts with GDP-[ I4C]mannose resulted in labeled proteins comigrating with the Sindbis envelope proteins (Krag and Robbins, 1977). It is possible that in the cases suggesting posttranslational modification, only nascent chains very near completion are being glycosylated. Alternatively, some proteins may be glycosylated after their translation is complete. Proteins may not attain their final conformations immediately; various posttranslational events may affect a protein’s structure in such a way that a glycosylation site is exposed. A review which discusses in more depth the question of cotranslational versus posttranslational glycosylation has been published (Bergman and Kuehl, 1982). As I mentioned, the oligosaccharide moieties which are transferred to the protein undergo additional modifications of their structure. The enzymology of this series of modifications, termed processing, has been reviewed extensively elsewhere (Kornfeld, 1982; Schachter et al., 1983; Schachter and Roseman, 1980). The mechanisms which determine the nature of a protein’s final oligosaccharide structure remain to be elucidated. A number of possibilities exist, such as (1) the oligosaccharide structure which is transferred, (2) the intracellular route of the protein, (3) the processing enzymes of the particular cell, (4) features of the polypeptide sequences, and (5) the positioning of the oligosaccharide in relation to the tertiary structure of the protein.

5. GLYCOSYLATION OF MEMBRANE PROTEINS

217

D. Regulation Our understanding of the regulatory mechanisms affecting glycosylation of asparagine residues in proteins is still limited. Thus far, changes in the pathway can be categorized as those which affect the level of activity of the overall pathway and those which affect a single enzymatic activity. Systems which are currently under study, the work from many of which will be detailed below, include cultured cells, which are responsive to nutritional factors or are altered by mutations or inhibitors, oviduct of estrogen-treated chicks, Dictyostefium discoidurn (Ivatt et al., 1981), and developmental systems such as chick embryo (Tavares and Hemming, 1982), brain, and sea urchins. 1. LEVELOF DOLICHYL PHOSPHATE

The level of dolichyl phosphate in the membrane may regulate the level of glycosylation of mannosylated proteins. The syntheses of the biosynthetic intermediates such as mannosylphosphoryl lipid, glucosylphosphoryl lipid, and N acetylglucosaminylpyrophosphoryl lipid can be increased by the addition of dolichyl phosphate to in vitro reactions (see earlier sections of this article). As detailed in Section II,A,2, dolichyl phosphate can be generated ( I ) from dolichol via a kinase, (2) from de novo synthesis, or (3) from dolichyl pyrophosphate via a phosphatase; dolichyl phosphate is the lipid carrier for a number of intermediates. Therefore, the level of dolichyl phosphate could be altered by a number of mechanisms. A low level of dolichyl phosphate may reflect decreased utilization of lipid-linked intermediate; for example, the mutant B211 which synthesizes but fails to use glucosylphosphoryl lipid for glucosylation of oligosaccharide lipid or protein has a low level of dolichyl phosphate (Krag, 1979). The possibility of regulation of glycosylation via the level of dolichyl phosphate has been examined in three developmental systems. In early studies, the amount of dolichyl phosphate in the membranes was estimated by its ability to serve as the endogenous acceptor of I14C]mannosefrom GDP-[ 14C]mannose. The synthesis of MPL in membranes from actively myelinating brains of young piglets was threefold higher than in membranes from adult animals (Harford et al., 1977). The enhancement of this synthesis by exogenous dolichyl phosphate was greater using membranes from adults than membranes from younger animals, although even at the highest concentration of exogenous lipid, MPL synthesis was higher using membranes from the actively myelinating animals. Similarly, MPL synthesis in the absence of exogenous dolichyl phosphate increased in membranes prepared from chicks treated for increasing periods with diethylstilbesterol (Lucas and Levin, 1977). Both these observations suggested that the level of dolichyl phosphate varied in the membranes of developing brain and

218

SHARON S.KRAG

hormone-treated chicks. In the latter case, increased incorporation of mannose into protein in in vitro reactions was also detected as a function of hormone treatment; an increase in MPL synthesis was seen during 6 days of hormone treatment whereas mannosylated protein synthesis increased during the entire 10 days of treatment. In the case of estrogen-induced differentiation, the change in the level of dolichol kinase was examined as a potential mechanism for the increase in dolichyl phosphate (Burton et al., 1981a). A 2.5-fold enhancement of dolichol kinase activity, independent of the level of exogenous dolichol present in the assay, was observed during the first 6 days of the 10-day hormone treatment. This increase in enzymatic activity paralleled the observed increase in the level of endogenous dolichyl phosphate (Lucas and Levine, 1977) in that both increased during the first 6 days and neither parameter changed from 6 to 10 days of treatment. A potential mechanism for regulating the dolichol kinase was reported by Gandhi and Keenan (1983). The kinase assayed in rat brain and Tetrahymena was inhibited by trifluoroperazine and chlorpromazine, known inhibitors of calmodulin action. After preincubating membranes with antiserum (to precipitate any endogenous calmodulin) and EGTA (to chelate Ca2+), the kinase activity was stimulated only by a mixture of Ca2+ and calmodulin, but not by either compound alone. A third system which has been studied in relation to changes in dolichyl phosphate levels is sea urchin embryos (Carson and Lennarz, 1981). In this system, the rate of asparagine-linked glycoprotein synthesis, assayed by either [3H]mannoseor [3H]glucosamine incorporation in vivo, increased 20- to 30-fold during the first 50 hours after fertilization. During the first 25 hours, the rate of dolichyl phosphate synthesis, measured by incorporation of either [I4C]acetate or [32Pi]into both free and saccharide-linked dolichyl phosphate, increased fiveto eightfold (Rossignol et a l . , 1981). The possibility that this increased synthesis of dolichyl phosphate was due to an increase in the activity of dolichol kinase was examined in both in vitro and in vivo studies. During the first 25 hours after fertilization, the specific activity of the dolichol kinase in the membranes increased twofold when assayed in the presence of exogenous dolichol (Rossignol et al., 1981). The contribution of the kinase to regulating the level of dolichyl phosphate in vivo was estimated by determining the rate of dolichyl phosphate synthesis using 32Piin the presence of an inhibitor of de novo synthesis, compactin (Carson and Lennarz, 1981; Rossingol et al., 1981, 1983). After a 20-hour preincubation of embryos (20hour postfertilization) in compactin, the 32Pi incorporation was 25% that of untreated embryos; after only 3.5 hours of preincubation of embryos (20-hour postfertilization) in the drug, the 32Piincorporation was equal to that of untreated embryos. In both cases, the [ 14C]acetateincorporation into dolichol and dolichyl

5. GLYCOSYLATION OF MEMBRANE PROTEINS

219

phosphate was inhibited (80-95%) by the drug. It appears that in these embryos (20-hour postfertilization), there is a pool of dolichol which can be converted to dolichyl phosphate by dolichol kinase activity. This pool was depleted during the 20-hour incubation with compactin. The size of this endogenous pool of dolichol during the entire period of increased glycosylation and increased dolichyl phosphate synthesis in these embryos must be determined in order to use this labeling procedure to access the role of the dolichol kinase in the increase of dolichyl phosphate in this system. Tissue slices of hen oviduct and bovine pancreas incubated in the presence of 20 pg/ml dolichyl phosphate showed an overall twofold increase in the glycosylation (mannose and glucosamine) of protein, without an increase in the incorporation of amino acids (Carson et al., 1981). No increase in the glycosylation of ovalbumin was seen in oviduct slices, but five times more mannose was incorporated into ribonuclease after dolichyl phosphate-supplementation of the pancreas slices. No increase in incorporation of mannose into MPL or oligosaccharide lipid was detected in these in vivo incubations. Membranes prepared from the tissue slices supplemented by dolichyl phosphate displayed enchanced (twofold) incorporation of glucosamine from UDP-[3H]glucosamine into endogenous acceptors in the presence of UDP-Glc and GDP-Man. OF PROTEIN SYNTHESIS 2. CESSATION

Another mechanism involved in regulating the entire synthetic pathway of lipidlinked saccharides has been uncovered in studies using protein synthesis inhibitors such as cycloheximide and puromycin. Incubations of CHO (Hubbard and Robbins, 1980), MDCK (Schmitt and Elbein, 1979), and LM (Grant and Lennarz, 1983) cells with these inhibitors rapidly inhibits synthesis of oligosaccharide lipids; incorporation of mannose into oligosaccharide lipid is reduced to 15% of the untreated level, while glucosamine incorporation is reduced to 40% of untreated levels. Treatment of LM cells with dolichyl phosphate did not reverse the effect (Grant and Lennarz, 1983), thus ruling out inhibition due to trapping dolichyl phosphate in nontransferred intermediates. The in vivo synthesis of MPL, N-acetylglucosaminyl- and chitobiosyl-P-P-lipid, and Man-GlcNAc-GlcNAc-PP-lipid are not affected by cycloheximide. Adding cycloheximide to microsomes of untreated cells had no effect on any of the in vitro reactions, including oligosaccharide lipid synthesis. Microsomes prepared from drug-treated cultures synthesized MPL, (GlcNAc), ,,-P-P-lipid, and Man-(GlcNAc),-P-P-lipid. In treated cells, no new oligosaccharide lipids appeared to accumulate; instead there appeared to be reduced synthesis of the intermediates detected in untreated cells. The turnover (t,,J for the major labeled oligosaccharide lipid in CHO cells and LM cells increased from 5-15 minutes to 60 minutes in the presence of cycloheximide. There were no reports of effects on the turnover time of shorter

220

SHARON S. KRAG

saccharide intermediates. In LM cells, the synthesis of oligosaccharide lipid resumed after 4 hours of continual drug treatment; no resumption of synthesis was detected after an 8-hour treatment of MDCK cells. It appears that cessation of protein synthesis by addition of cycloheximide inhibits a step between Man (GlcNAc),-P-P-lipid and G,M,-P-P-lipid, while resulting in no accumulation of the intermediate before the block. One suggested mechanism explaining this effect is that GTP, which might accumulate transiently in the absence of protein synthesis, inhibits oligosaccharide lipid synthesis (Grant and Lennarz, 1983). Addition of the al,3-mannose to a tetrasaccharide lipid is inhibited by GTP (Jensen et al., 1980). It is not clear why the tetrasaccharide lipid does not accumulate. In view of the observations by Paiement and Bergeron (1983) that GTP promotes fusion of endoplasmic membranes stripped of ribosomes, one could speculate that increased GTP levels enhance translocation of saccharide lipids, such as the tetrasaccharide lipid, from the cytoplasmic to lumenal face of the membrane. In this orientation, they would not be substrates for elongation by the mannosyl transferases that utilize GDP-mannose; GDP-mannose appears not to be transported into the lumen of the reticulum (Hanover and Lennarz, 1982). The saccharide lipids in the lumen might then be rapidly degraded, cleaved by hydrolases which convert them to phosphooligosaccharides; these hydrolases appear to prefer nonglucosylated substrates (Hoflack et al., 1981), thus sparing the G,M,P-P-lipid. A potential role for cytoplasmic GTP in regulating the membranebound glycosyl transferases raises the question of orientation of these enzymes in the membrane. AND ORIENTATION 3. SUBMICROSOMAL LOCATION

The level of activity of the glycosylation pathway may be regulated by the localization and orientation of the enzymes and substrates in the various cellular compartments. The oligosaccharide lipid transferase and the glycosyl transferases involved in the synthesis of the oligosaccharide lipid intermediates appear to be present in the endoplasmic reticulum. This cellular compartment is thought to be impermeable to sugar nucleotides (Hanover and Lennarz, 1982), in contrast to Golgi membranes which transport UDP-galactose, GDP-fucose, UDP-N-acetylglucosamine, and CMP-sialic acid (Sommers and Hirschberg 1982; Perez and Hirschberg, 1983; Fleischer, 1983). The synthesis of both the carrier lipid and the oligosaccharide moiety involves substrates synthesized in the cytoplasm: isopentenyl pyrophosphate and sugar nucleotides, respectively. The endogenous proteins glycosylated in reactions with GDP-[ 14C]mannose and sealed hen oviduct membranes are located within the lumen of the reticulum (see Table III), as judged by their resistance to cleavage by P-N-acetylglucosaminidase H (used as an impermeant probe). Cleavage by this enzyme was obtained when sufficient

221

5. GLYCOSYLATION OF MEMBRANE PROTEINS

TABLE III ENZYME, GLYCOPROTEIN, A N D GLYCOLIPID ORIENTATIONS I N THE ROUGH ENDOPLASMIC RETICULUM o b EUKARYOTIC CELLS Molecules Endogenous glycosylated proteins G3M9-P-P-lipid Chitobiosyl-P-P-lipid Polyprenyl transferase Dolichyl kinase Mannosylphosphoryl lipid synthase Chitobiosylpyrophosphoryl lipid synthase N-Acet ylglucosaminyl- 1-P-lipid transferase Glucosylphosphoryl lipid

Location in the reticulum Lumenal face Lumenal face Lumenal face Cytoplasmic face Cytoplasmic face Cytoplasmic face Cytoplasmic face Ambiguous Ambiguous

detergent was added to allow maximal expression of the activity of two presumably lumenal enzymes, P-glucuronidase and mannose 6-phosphatase (Hanover and Lennarz, 1980). So far, the orientation of only two of the lipid intermediates has been examined (see Table 111). The major mannose-labeled oligosaccharide lipid, G,M,-PP-lipid, of CHO cells infected by vesicular stomatitis virus (VSV) was found to be localized primarily within the reticular lumen. About 12-25% of this oligosaccharide lipid bound concanavalin A, when this impermeant reagent was incubated with labeled, sealed microsomes (Snider and Robbins, 1982). In the presence of detergent about 65-75% of the labeled G,M,-P-P-lipid in the microsomes bound the lectin; the concentration of detergent used to render the microsomes permeable to concanavalin A did not solublize the G,M,-P-P-lipid. The binding observed without detergent was attributed to unsealed microsomes (about 20%) in the preparation; the percentage of unsealed microsomes was estimated from the amount of mannose-labeled G protein of VSV cleaved to small peptides by trypsin. A second intermediate, chitobiosylpyrophosphoryl lipid, was found to face the lumen (90%),judged by its inability to accept galactosyl residues when sealed microsomes were incubated with UDP-galactose and the impermeant galactosyl transferase (Hanover and Lennarz, 1978). Chitobiosyl-P-P-lipid was labeled by incubating membranes with UDP-N-acetyl-[''C]glucosamine; then the membranes were subjected to centrifugation, resuspended, and determined to be 8590% sealed by the latency of mannose 6-phosphatase and P-glucuronidase activities. Quantitative conversion of the [ 14CJchitobiosyllipid to Gal-GlcNAcGlcNAc-P-P-lipid was obtained with galactosyltransferase after treating the sealed microsomes with low concentrations of detergents. The labeled chitobiosyl lipid in the washed microsomal pellet was stable for an hour in a buffered

222

SHARON S.KRAG

sucrose/NaCl solution containing 10 mM MnCl,. Addition of unlabeled GDPmannose for an hour resulted in loss of all the radioactivity in the chitobiosyl lipid fraction, with a concomitant increase in labeling of larger oligosaccharide lipids and protein. While these two saccharide lipid intermediates, G,M,-P-P-lipid and chitobiosyl-P-P-lipid, have been found to face the lumen, four of the enzymes involved in the pathway appear accessible to the cytoplasmic side of the reticulum (Table 111). Polyprenyl transferase, which converts labeled isopentenyl pyrophosphate and farnesyl pyrophosphate to dolichyl phosphate (Fig. l), and dolichol kinase were both inactivated in intact rat liver microsomes by trypsin treatment (90 and 40%, respectively) and treatment with the impermeant inhibitor mercury dextran (50% inhibition of each) (Adair and Cafmeyer, 1983). Mannosylphosphoryl lipid synthase in rat liver microsomes was inactivated (90%) by protease treatment in the absence of detergent; (1 mg/ml Pronase or 2 mg/ml trypsin, 30 minutes, 30°C; Snider et af., 1980); glucose 6-phosphatase was inactivated (75%) only in the presence of detergent. This result suggests that portions of the mannosylphosphoryl lipid synthase necessary for activity are facing the cytoplasmic side; alternatively, a component required for transfer of the GDP-mannose to a lumenally oriented transferase is protease sensitive. The synthase could not be localized in hen oviduct microsomes because it appeared insensitive (maximum of 25% inactivation) to protease treatment (50 pg/ml protease, 10 minutes, 0°C; Hanover and Lennarz, 1982). Note the marked differences in the conditions of protease treatment in these two studies. MPL synthase activity from yeast has been solubilized, partially purified, and incorporated into liposomes of phosphatidylcholine (PC) (Haselbeck and Tanner, 1982). The enzyme in the liposomes was active in the synthesis of MPL using extraliposomal GDP-[ 14C]mannoseonly when dolichyl phosphate was incorporated with the PC into the liposomes. When the liposomes also contained unlabeled GDP, label was detected within the liposomes which comigrated with GDP-mannose on paper chromatography, suggesting the enzyme was active in the reverse direction using substrates within the water space of the liposomes. This result, assuming GDP-mannose and GDP are impermeable to the liposomes, implies either that the enzyme was randomly oriented in the liposomes or that the active site of the enzyme can move from outside to inside; in addition, either the MPL can traverse the membrane or is carried by the synthase. The enzyme which transfers GlcNAc from UDP-GlcNAc to N-acetylglucosaminyl-P-P-lipid forming chitobiosyl-P-P-lipid appears to have its active site facing the cytoplasm (Table 111). Both impermeable derivatives of N-ethylmaleimide (Lennarz, 1979) and Pronase (1 mg/ml, 30 minutes, 30°C) inactivated (90%) this enzyme in intact microsomes. The activity could not be localized using trypsin since the enzyme is insensitive to treatment with trypsin (2 mg/ml, 30 minutes, 30°C) in intact or disrupted microsomes. Since the product

5. GLYCOSYLATION OF MEMBRANE PROTEINS

223

of this enzyme appears to face the reticular lumen (see above), this enzyme may facilitate translocation of the product as part of its enzymatic activity (Hanover and Lennarz, 1982). However, this hypothesis still leaves the question of how the GDP-mannose (or GDP) needed as the mannosyl donor for mannosylated oligosaccharide lipids enters the lumen. The orientation of glucosylphosphory1 lipid (GPL) synthase in closed microsomally derived vesicles is ambiguous (Table 111). The enzyme was inactivated (95%) in the absence and presence of detergent by Pronase (1 mg/ml, 30 minutes, 30°C), partially inactivated (67%) with or without detergent by trypsin (2 mg/ml, 30 minutes, 30"C), and 40% inactivated in intact microsomal vesicles and 70% inactivated in detergent-treated microsomes using protease (50 pg/ml, 10 minutes, 0°C). The orientation of the enzyme which synthesizes N-acetylglucosaminylpyrophosphoryl lipid is also unclear. This enzyme was inactivated no more than 30% by protease treatment with or without detergent and was stimulated by treatment with N-ethylmaleimide (Hanover and Lennarz, 1981). Finally, the activities of steps beyond chitobiosyl lipid have not been assayed independently of the early synthetic activities; all the assays have utilized endogenous rather than exogenous substrates. Therefore, their orientation with respect to the membrane is unknown.

4. LEVELOF GLUCOSEIN LABELING MEDIUM Finally, incubations of certain cultured cells in medium lacking glucose produces changes in the overall glycosylation pathway. These changes have been studied in two ways. First comparisons have been made between the mobility of known proteins on SDS-polyacrylamide gels after labeling cells in the presence or absence of glucose. There are now at least six cases in which proteins synthesized during glucose starvation were heterogeneous; in each case, some of the proteins labeled under starvation conditions migrated normally, some migrated as an unglycosylated form, and some as intermediate forms. These cases include the envelope proteins of Sindbis virus (Sefton, 1977) and Semliki Forest virus (Kaluza, 1975), the immunoglobulin K light chain of murine myeloma cells (Stark and Heath, 1979), G protein of vesicular stomatitis virus grown in BHK cells (Turco, 1980; Turco and Pickard, 1982), a,-acid glycoprotein of rat hepatoma cells (Baumann and Jahreis, 1983), and a subunit of glycoprotein hormones in Chang human liver cells (Morrow et al., 1983). Rat a,-acid glycoprotein normally contains six asparagine-linked oligosaccharides, all of which are sensitive to endoglycosidase H treatment. The intermediate forms of a,-acid glycoprotein seen during glucose starvation corresponded to proteins having zero to six glycans attached; each intermediate appeared to contain both oligosaccharides sensitive and resistant to treatment with endoglycosidase H. Mature G protein from VSV-infected BHK cells contains two general types of

224

SHARON S. KRAG

glycopeptides: those which do not bind to a concanavalin A-Sepharose column (presumably oligosaccharides with triantennary structures; see Introduction) and those which do bind and are eluted with a-methylmannoside (presumably oligosaccharide with biantennary structures). The mixture of [3H]mannose-labeled G proteins (one of which comigrated with the normal G protein) labeled during glucose starvation appeared to have only biantennary structures. The second method of analysis used was comparison of the oligosaccharides released from the oligosaccharide lipids and glycopeptides of cells labeled with mannose in the presence or absence of glucose. In the case of CHO cells (Rearick ef al., 1981a) and virally infected BHK cells (Turco, 1980; Turco and Pickard, 1982), the G,M, oligosaccharide lipid was not labeled during glucose starvation. Instead, a range of oligosaccharide lipids from M,-P-P-lipid to Man-(GlcNAc),P-P-lipid were labeled in the virally infected BHK cells; primarily the M,Res-P-Plipid (see Section II,C, 1) was labeled by the CHO cells. In NIL 8 cells, the G,M,P-P-lipid was still detectable during glucose starvation, but its level in relation to M,Res-P-P-lipid and (Man),-(GlcNAc),-P-P-lipid was greatly reduced (Gershman and Robbins, 1981). The oligosaccharide structures found attached to glycopeptides from the proteins labeled during short incubations with [2-3H]mannosewere G1c2-MSReS in CHO cells and a mixture of Glc,-M,Res and (Man),-,-(GlcNAc), in virally infected BHK cells. Over 90% of the glycopeptides in both cases were resistant to digestion by endoglycosidase H. Thus, incubation of most cultured cells without glucose resulted in the synthesis of shorter oligosaccharides on the lipid intermediates, some of which were transferred to protein; as a result, the proteins had altered mobilities. Not all cultured cells displayed altered glycosylation during glucose starvation; those not affected included SV40-transformed BALB/c 3T3 cells (although uninfected cells were) (Gershman and Robbins, 1981), uninfected BHK cells, 3T3 (mouse strain unspecified) cells, and mouse lymphoma cells (Rearick ef al., 1981a). The effect of glucose starvation seen in cells such as BALB/c 3T3, CHO, and NIL 8 at low culture densities was not found when the cultures were near or at confluence (Gershman and Robbins, 1981). In some cell lines, such as NIL 8, the effect of glucose starvation was only transient; 40 minutes into the starvation period the ratio of oligosaccharide lipids synthesized in a pulse experiment was normal (Gershman and Robbins, 1981). The effects of glucose starvation were reversed by the addition of glucose, mannose, or butyrate, partially reversed by galactose, but not reversed by pyruvate, glycerol, glutamine, glycine, inositol, fructose, N-acetylglucosamine, or ribose (Stark and Heath, 1979; Turco, 1980; Gershman and Robbins, 1981; Morrow ef al., 1983). The mechanism by which glucose starvation caused these changes in glycosylation is not yet clear. One change that occurs when one performs labeling experiments during glucose starvation is a change in the specific activity of sugar

225

5. GLYCOSYLATION OF MEMBRANE PROTEINS

nucleotides (Kim and Conrad, 1976); the specific activity of UDP-GlcNAc increased exponentially as the level of glucose in the medium decreased. Also, the level of mannosylphosphoryl lipid decreased as a function of time under starvation conditions (Rearick et a / . , 1981a). Since MPL is the donor of only a subset of the mannose residues in the G,M,-P-P-lipid [circled in (XVIII) below], a

Ly

(Glc) -Man

~

1,2

Ly

Man

~

1,2

1

Man XVlll

decrease in its level could account for the altered pattern of lipid intermediates detected. The decrease in MPL does not seem to be the result of an increase in the GDP:GDP-mannose ratio, which would affect the equilibrium of the freely reversible MPL synthase reaction (Gershman and Robbins, 1981). If the effects of glucose starvation are solely due to a reduction in the level of mannosylphosphoryl lipid, then at its extreme cells starved for glucose should have similar phenotypes to the Thy- 1 - lymphoma cells (Chapman ef al., 1980) and the B4-2-1 CHO cells (Stoll et al., 1982); these mutants are unable to synthesize MPL. However, in at least one aspect this prediction does not hold: mature G protein of vesicular stomatitis virus grown in the Thy-1 cells has its normal oligosaccharide structures (Kornfeld et al., 1979) while under glucose starvation conditions the mature G protein does not have all its normal structures (Turco and Pickard, 1982). ~

5 . REGULATION OF INDIVIDUAL ENZYMES

Not only does MPL serve as a donor of mannosyl residues, but it has been postulated to be a positive regulator of glycoprotein biosynthesis in chick retina (Kean, 1982). The addition of mannosylphosphoryldolichol to incubations of UDP-N-[3H]acetylglucosamine, dolichyl phosphate, and homogenates of chick retina increased the V,,, of glucosaminyl lipid formation 10-fold. It also increased the K , for UDP-GlcNAc two- to fourfold; the result was an overall

226

SHARON S.KRAG

stimulation of the rate of the reaction. It is not yet clear whether this activation by MPL is unique or can also be demonstrated by glucosylphosphoryl lipid. There are two examples of enzymes in the glycosylation pathway that are activated by specific phospholipids. The enzyme activity in microsomes of rat lung which synthesizes GlcNAc-P-P-lipid was inhibited by treatment of the microsomes with phospholipase A,. Activity was completely restored by the addition of phosphatidylglycerol in Triton X- 100; other phosphatides such as phosphatidylcholine, phosphatidylinositol, and cardiolipin partially restored activity but phosphatidylserine, phosphatidylethanolamine, and the nonionic detergent Triton X-100 alone did not (Plouhar and Bretthauer, 1982). Interestingly, phosphatidylglycerol is present at very low levels in most animal tissues but is found in significant amounts in lung (Mason and Williams, 1980). The second exam le is the enzyme which transfers mannose from GDPmannose to M a n d M a n L ( G l c N A c ) , - P - P - l i p i d resulting in a pentasaccharide with an al,3-rnannosyl linkage. Following solubilization from rabbit liver microsomes, this activity was optimally reconstituted in the presence of phosphatidylethanolamine with unsaturated acyl chains (Jensen and Schutzbach, 1982). In this section I have given examples of regulatory mechanisms affecting the level of all or certain oligosaccharide lipid intermediates. There also appear to be regulatory mechanisms affecting the activity of the oligosaccharide transferase. The transferase activity in the membranes of estrogen-treated chicks was found to increase three- to fourfold during hormone treatment. The transferase activity was assayed using an exogenous acceptor, carboxymethylated a-lactalbumin, and exogenous oligosaccharide lipid, so the observed increase in activity was not due to known changes in the level of dolichyl phosphate or potential changes in the nature of endogenous protein acceptors (Singh and Lucas, 1981).

111.

ROLE OF THE CARBOHYDRATE MOIETY

A. General Approaches and Methodologies There are three general approaches used to determine the role of protein-bound carbohydrate. First, the properties of glycosylated and nonglycosylated forms of an isolated protein are examined. The methods used to obtain nonglycosylated forms of N-linked glycoproteins include both chemical treatments of the glycosylated protein, such as anhydrous hydrogen fluoride (Mort and Lamport, 1977) or trifluoromethanesulfonic acid (Edge et al., 19811, and enzymatic treatments with a battery of both exoglycosidases (including neuraminidase, P-galactosidase , P-N-acetylglucosaminidase, and a-mannosidase) and endoglycosidases (such as P-N-acetylglucosaminidasesH and D) (Muramatsu, 1978; Kobata, 1978). The major concern with regard to these methods is obtaining maximum

5. GLYCOSYLATION OF MEMBRANE PROTEINS

227

removal of carbohydrate with minimal disruption of the protein. Alternatively, one can isolate the protein from a cell which either does not glycosylate the protein or has been treated with a drug which prevents glycosylation. Second, one can compare the function and compartmentalization of a protein in an untreated cell to that in a cell treated with an inhibitor of glycosylation. Using this approach, one has to assume that the inhibitor is specific. Even if the inhibitor is specific for a single step in glycosylation, these studies can often define only whether the carbohydrate has an overall role in the function of a protein. Determining the precise role of the carbohydrate is often complicated because glycosylation is such an early step in the synthesis of a glycoprotein; many glycoproteins are destined for further posttranslational modifications (0linked glycosylation, acylation, sulfation, proteolysis) which may be necessary for their function and which may depend on prior glycosylation. A comprehensive review of a variety of inhibitors and studies using those inhibitors has recently been published by Schwartz and Datema (1982). One of the most widely used inhibitors is tunicamycin, a nucleoside analog which was isolated from Streptomyces lysosuperifiicus by Takatsuki et al. (1971). As mentioned in Section 11, this antibiotic inhibits the transfer of N-acetylglucosamine 1phosphate from UDP-N-acetylglucosamine to dolichyl phosphate (Tkacz and Lampen, 1975; Takatsuki et a l . , 1975); since this step is first in the lipid-linked pathway, tunicamycin blocks all synthesis of sugar-P-P-lipids and thus blocks any further modification or transport which depends on glycosylation of the protein. Certain proteins, such as apoprotein B (Siuta-Mangano et a l . , 1982) and proopiomelanocortin (Budarf and Herbert, 1982), are correctly assembled, cleaved, and transported in the presence of tunicamycin. Other proteins are not, such as certain immunoglobulins (Sidman, 1981), the insulin receptor (Reed et a l . , 1981), murine mammary tumor viral glycoproteins (Firestone, 1983), and the sodium channel (Waechter et a l . , 1983). Third, one can compare the function of a glycoprotein in a normal cell with that in a mutant cell defective in glycosylation. Using this technique, one faces the same problems as with inhibitors; cells may harbor more than one mutation, and even if a single lesion is demonstrated, defects in glycosylation a priori have pleiotropic effects. A variety of mutants can be obtained. For example, glycosylation mutants have been isolated from cultured cell lines and eukaryotic protists employing either selection (Stanley, 1980; Briles, 1982; Hyman and Trowbridge, 1977; Criscuolo and Krag, 1982) or screening (Raetz et d., 1982; Robbins et a l . , 1981; Snider et a l . , 1982). Mutants have been isolated from animal viruses which contain and encode envelope glycoproteins (Ruta et a l . , 1979). Finally, the effect of naturally occurring mutations on the function of certain glycoproteins can be studied. For example, the necessity for carbohydrate on the fourth component of murine complement was suggested by correlating the presence of carbo-

228

SHARON S.KRAG

hydrate with the hemolytic activity observed in four mouse strains (Karp et al., 1982).

B. Protein Solubility and Structure Some proteins are less soluble in their nonglycosylated form and tend to aggregate. The G protein of vesicular stomatitis virus (VSV) and PE, and E, of Sindbis virus are normally glycosylated and transported to the plasma membrane in infected cells; however, when infected cells were treated with tunicamycin, these proteins were only detected inside the cell and not on the cell surface. These proteins were insoluble in nonionic detergents which extract the normally glycosylated proteins. This effect of the carbohydrate moieties on protein solubility differed, depending on the amino acid sequence and the temperature. In some strains of VSV, nonglycosylated G proteins aggregated at both 37 and 30"C, while in other strains, aggregation was seen at 37 but not at 30°C (Gibson et al., 1978, 1979). This relationship of solubility, temperature, and the presence of an oligosaccharide moiety was also studied using polyethylene glycol precipitation (Lawson et al., 1983). Ribonuclease A (nonglycosylated) was less soluble than ribonuclease B (one N-glycan containing only mannose and N-acetylglucosamine) at temperatures between 0 and 30"C, but at higher temperatures the B form was slightly less soluble. Inhibition of the glycosylation of the soluble, secreted acid phosphatase of yeast by tunicamycin resulted in synthesis of inactive, nonglycosylated, membrane-associated forms of the enzyme (Mizunaga and Noguchi, 1982). These forms had mobilities on SDS-polyacrylamide gels identical to products formed by treating native enzyme with hydrogen fluoride and P-N-acetylglucosaminidase H. These deglycosylated and nonglycosylated proteins were insoluble in water, high salt, and detergent, conditions which solubilized the glycosylated protein. Two monomeric, soluble glycoproteins from yeast, invertase and carboxypeptidase Y, have been examined. In the case of invertase, the presence of the carbohydrate moiety stabilized the enzyme to denaturation by mild acid, heat, or repeated cycles of freezing and thawing (Chu et al., 1978). Removal of the oligosaccharides of carboxypeptidase Y by digestion with endoglycosidase H doubled the rate of inactivation by treatment with sodium dodecyl sulfate as compared to the glycosylated enzyme (Chu and Maley, 1982). Assuming both forms bound the same amount of detergent, this result suggests the carbohydrate may help maintain the active conformation of this enzyme. On the other hand, deglycosylated human choriogonadotropin was more resistant to thermal denaturation (temperatures above 50°C) than was the glycosylated form (Manjunath and Sairem, 1983).

5. GLYCOSYLATION OF MEMBRANE PROTEINS

229

The presence and size of carbohydrate may influence the folding of a protein to its active conformation. When the extent of renaturation of glycosylated and deglycosylated invertase was compared after treatment with guanidine hydrochloride, the former renatured twice as efficiently as the latter (Chu et al., 1978). Observations suggesting this role for the carbohydrate moiety have been made using the mutant cell lines Thy-1 and B4-2-1, neither of which synthesize mannosylphosphoryl lipid and therefore make and transfer truncated oligosaccharides (Chapman et ul., 1980; Stoll et ul., 1982). At least one glycoprotein, G protein of vesicular stomatitis virus, has been shown in Thy-I - cells to have normal, mature oligosaccharide structures (complex oligosaccharide) despite the transfer of truncated oligosaccharide (see Section 11). However, two proteins containing complex oligosaccharides (the Thy antigen in the lymphoma mutant and the mannose 6-phosphate receptor in the CHO mutant) are affected by the transfer of the truncated species, although they presumably contain normal, mature oligosaccharides. The Thy antigen is not active on the cell surface and is rapidly degraded (tllZof 4.5 hours rather than 30 hours) (Trowbridge et af., 1978b); the mannose 6-phosphate receptor reaches the cell surface, but displays altered binding properties (Robbins and Myerowitz, 1981). Two studies have suggested a role for carbohydrate in the assembly of oligomeric structures. The a subunit of acetylcholine receptor, synthesized by a mouse cell line (BC3H- I), was immunoprecipitated by incubating cell extracts with a-bungarotoxin followed by antiserum to the toxin. Only assembled acetylcholine receptor binds a-bungarotoxin. However, nonglycosylated u subunits, synthesized in the presence of tunicamycin, were not precipitated by the toxinantitoxin treatment, although the subunit was detected by antiserum specific for that subunit (Merlie et al., 1982). Second, maintenance of the normal large multimers of factor VlII (von Willebrand factor) depends on the presence of galactose residues on the final mature complex structure (Gralnick et al., 1983). The large multimers and the activity of the factor both decreased as the release of galactose increased during P-galactosidase treatment. ~

C. Protein Turnover There are numerous studies which suggest that the presence of carbohydrate on a protein may decrease its susceptibility to proteases (reviewed by Olden et al., 1982). These studies include both comparisons of loss of activity during treatments with proteases in vitro and comparisons of turnover rates in vivo after treatment of cells with inhibitors of glycosylation. The reason for the increased susceptibility of nonglycosylated protein probably varies with the protein; factors such as decreased solubility, altered conformation, and improper compartmentalization may lead to increased protease sensitivity. Examples of proteins known

230

SHARON S. KRAG

to have an increased susceptibility to proteases in their non- or deglycosylated form include fibronectin in chicken embryo fibroblasts (Bernard et al., 1982; Hynes and Yamada, 1982), acetylcholine receptor in cultured chick embryo muscle cells, (Prives and Bar-Sagi, 1983), alkaline phosphatase in murine cells (Firestone and Heath, 1981), and carboxypeptidase Y of yeast cells (Chu and Maley, 1982).

D. Enzymatic Function Many enzymes are equally active in their glycosylated and nonglycosylated forms, indicating that in these cases the carbohydrate has no role in enzyme activity. Ribonucleases A and B have similar activities (Tarentino et al., 1974). Acid phosphatase in yeast synthesized in the presence of tunicamycin was inactive; however, in this system it is difficult to determine whether the enzyme was inactive because it was aggregated, because it was membrane-associated, or because it was nonglycosylated (Mizunaga and Noguchi, 1982). There may be one example where the presence of the correct oligosaccharide is necessary for enzymatic activity. Immunoprecipitable radioactivity and enzymatic activity of the lysosomal enzyme a-L-iduronidase were compared using enzyme secreted from the mutant B4-2-1 and parental cells during metabolic labeling. Specific activity of enzyme from the mutant was one-tenth of that from the parent (Stoll et al., 1982; Robbins, unpublished data). This reduction in enzyme activity did not result from proteolytic cleavage, since the secreted enzyme showed the same mobility on SDS-polyacrylamide gel electrophoresis as the biosynthetic precursor (Robbins and Myerowitz, 1981).

E. Interactive Functions There are a number of systems in which carbohydrates are thought to play a role in interactions among cells, including mixed lymphocyte reactions (Hart, 1982), adhesion in intestinal epithelial cells (Sasak et al., 1983), lymphocyte circulation through blood, lymph, and tissues (Stoolman and Rosen, 1983) and neural cell adhesion (Cunningham et al., 1983). Recently, a number of studies have been done evaluating the role of the carbohydrate in the interaction of human chorionic gonadotropin with its membrane-bound receptor resulting in the activation of adenylate cyclase and steroid production (Goverman et al., 1982; A m et al., 1982; Chen et al., 1982; Manjunath and Sairam, 1982; Kalyan et al., 1982; Sairam and Manjunath, 1983; Kalyan and Bahl, 1983). This hormone consists of two glycosylated, nonidentical subunits. The carbohydrate moieties of the subunits were altered by a variety of treatments, including exo- and endoglycosidases, anhydrous hydrogen

5. GLYCOSYLATION OF MEMBRANE PROTEINS

231

fluoride, periodate, or trifluoromethanesulfonic acid, and then tested for binding and biological activity. The results of all these studies are similar. Modification of the carbohydrate moiety on the two subunits of the hormone obliterates binding to liver cell membranes but does not affect binding to membrane-associated receptor in gonadal tissue. However, no production of cAMP or steroid occurs when deglycosylated hormones bind to the receptor, even though the adenylate cyclase in that membrane can be stimulated by guanine nucleotides or catecholamines (Thotakura and Bahl, 1982). Intermediate production of cAMP and steroids occurs with one glycosylated and one deglycosylated subunit. This same result was obtained when a related oligomeric glycoprotein hormone, ovine lutropin, was studied (Manjunath et al., 1982). After deglycosylation with anhydrous hydrogen fluoride, the hormone reacted with gonadal receptor and antiserum, but cAMP production did not result from hormone binding to the receptor. Two recent studies have suggested a role for the carbohydrate moiety in interactions in the immune system. Nose and Wigzell(l983) found that carbohydrate-depleted mouse IgG2b antibodies could bind antigen normally, but the carbohydrate-depleted antibody-antigen complexes lost the following activities: to activate complement, to bind to macrophage Fc receptors, to induce antibodydependent cellular cytotoxicity , and to be eliminated rapidly from the circulation. Cowing and Chapdelaine (1983) presented results suggesting that sialylation differences between Ia molecules was the basis of the different T-cell mitogenic potential of accessory cells and B cells.

F. Compartmentalizationof Proteins Compartmentalization of enzymes and proteins often involves carbohydrate recognition systems (Neufeld and Ashwell, 1980). These include the asialoglycoprotein receptor of liver, which binds circulating glycoproteins lacking sialic acid prior to receptor-mediated endocytosis, and the mannose 6-phosphate receptor of fibroblasts, which directs newly synthesized hydrolases to the lysosome. The mannose 6-phosphate receptor system is affected in some glycosylation mutants (Neufeld and Robbins, 1982). As discussed in Section lll,C, the lack of synthesis of mannosylphosphoryldolichol altered the function of the mannose 6phosphate receptor in mutant CHO cells (Robbins and Myerowitz, 1981). These cells and the Thy-1 - lymphoma cells with a similar lesion (Chapman et al., 1980) produce lysosomal enzymes containing truncated oligosaccharides that bind to the normal mannose 6-phosphate receptor. This implies that the enzyme responsible for phosphorylation of the mannosyl residues (Reitman and Kornfeld, 1981; Hasilik et al., 1981) recognizes the truncated oligosaccharides.

232

SHARON S.KRAG

A different phenotype is displayed by a CHO cell mutant B2 11, which does not glucosylate its oligosaccharides and proteins (Krag, 1979). In this cell line, the mannose 6-phosphate receptor and the enzyme which phosphorylates the mannose residues on lysosomal enzymes are both normally active. However, lysosomal enzymes are not phosphorylated in this mutant and are not segregated to the lysosomes; instead, they contain complex oligosaccharides and are secreted (Krag and Robbins, 1982). It is not clear why the phosphotransferase fails to recognize the lysosomal enzymes of B211, but it is clear that normal compartmentalization of lysosomal hydrolases can be affected by alterations in glycosylation.

IV. SUMMARY The synthesis of asparagine-linked glycoproteins involves a complex series of reactions catalyzed by soluble and membrane-associated enzymes in at least three cellular compartments. In this article I have concentrated on four areas. First, I considered the synthesis, storage, and interconversions of the polyisoprenol lipids which are involved in glycosylation reactions. Dolichyl phosphate is clearly involved in glycosylation; it remains to be determined whether other polyisoprenoids are as well. Future research in this area will answer questions such as what is the relative contribution of de n o w synthesis by each cell type to the uptake, synthesis, storage, and dissemination of dolichyl metabolites by the liver; how are the interconversions of dolichyl metabolites to dolichyl phosphate regulated and where in the cell do the interconversions occur; do different enzymes which utilize dolichyl phosphate derivatives display specificity for isoprenol chain length which could modulate enzymatic activity and therefore regulate glycosylation. Second, I reviewed the steps involved in the assembly of oligosaccharides on dolichyl phosphate and considered which of these various structures could be transferred to protein. Are there only one or two oligosaccharide structures transferred to all glycoproteins? Is there only one o1igosaccharide:protein transferase? These questions will be answered by purification of the oligosaccharide:protein transferase(s) and by an examination of the glycosylation of a variety of individual proteins (both normal and modified through genetic engineering) in different mutants and in the presence of different inhibitors. Third, I considered what is currently known about the regulation of glycosylation, and it is clear that many regulatory mechanisms are involved at both the enzymatic and cellular level. Studies on these regulatory mechanisms will depend on the isolation of regulatory mutants and will necessitate purification of the proteins and enzymes involved; purification of many of these proteins will be made feasible by DNA recombination techniques.

5. GLYCOSYLATION

OF MEMBRANE PROTEINS

233

Finally, I reviewed current studies on the functional role of the glycose moiety on proteins. Future studies will no doubt expand the list of examples in which it can be demonstrated that the presence of oligosaccharide (and in some cases, the precise structure of the oligosaccharide) on a protein affects its activity, solubility, stability, or ability to interact with another molecule or cell. It is clear, however, from the examples we already have that the functional significance may vary from glycoprotein to glycoprotein and from cell type to cell type. ACKNOWLEDGMENTS

I thank Barbara Criscuolo and Jeff Robinson for their helpful comments on this review, and Michele DuVal and Dawn McKinney for typing it. In addition. 1 would like to thank Adina K. Student, Elizabeth F. Neufeld, John J . Scocca, Pamela Stanley, and Philip A. Knauf for their helpful, critical comments. Finally. I especially thank April R. Robbins for her thorough, careful, critical evaluation of this manuscript and her continual encouragement during its preparation. I acknowledge the support of NIH Grants CA2042 1 and CA00640. REFERENCES Adair, W. L., Jr., and Cafnieyer, N. (1983). Topography of dolichyl phosphate synthesis in rat liver microsomes. Transbilayer arrangement of dolichol kinase and long-chain prenyltransferase. Eiochim. Eiophys. Acta 751, 21-26. Adair, W. L., Jr., and Keller, R. K. (1982). Dolichol metabolism in rat liver. Determination of the subcellular distribution of dolichyl phosphate and its site and rate of de n o w biosynthesis. J. Eiol. Chem. 257, 8990-8996. Adamany, A. M., Blumenfeld, 0. O., Sabo, B., and McCreary, J. (1983). Acarbohydrate structural variant of MM glycoprotein (glycophorin A). J. B i d . Chem. 258, 11537-1 1545. Allen, C. M., Jr., Kalin, J. R., Sack, J . , and Verizzo. D. (1978). CTP-dependent dolichol phosphorylation by mammalian cell homogenates. Eiochemistty 17, 5020-5026. A m , S., Shimohigashi, Y.. Carayon, P., Chen, M.-C., and Nisula, B. (1982). Sialic acid residues of the cx-subunit are required for the thymotropic activity of HCG. Eiochem. Eiophys. Res. Commun. 109, 146-151. Appelkvist. E. L., Chojnacki, T., and Dallner, G . (1981). Dolichol (C55) mono- and pyrophosphat a m of the rat liver. Eiosci. Rep. 1, 619-626. Aubert, J . P., Chiroutre, M., Kerchaert, J . P., Helbecque, N., and Louchereux-Lefebore, M. H. (1982). Purification by affinity chromatography of the solubilized oligosaccharyl transferase from hen oviducts using a privileged secondary structure adopting peptide. Eiochem. Eiophys. Res. Commun. 104, 1550-1559. Banerjee, D. K., Scher, M. G., and Waechter, C. J . (1981). Amphomycin: Effect of the lipopeptide antibiotic on the glycosylation and extraction of dolichyl monophosphate in calf brain membranes. Biochemistry 20, I561- 1568. Baumann, H., and Jahreis, G. P. (1983). Glucose starvation leads in rat hepatoma cells to partially N-glycosylated glycoproteins including a I -acid glycoproteins. Identification by endoglycolytic digestions in polyacrylamide gels. J. Eiol. Chem. 258, 3942-3949. Bause, E. (1983a). Active-site-directed inhibition of asparagine N-glycosyltransferases with epoxypeptide derivatives. Eiochem. J. 209, 323-330. Bause, E. (1983b). Structural requirements of N-glycosylation of proteins. Studies with proline peptides as conformational probes. Biochem. J. 209, 33 1-336.

234

SHARON S. KRAG

Bause, E . , and Legler, G. (1981). The role of the hydroxy amino acid in the triplet sequence of AsnXaa-Thr (Ser) for the N-glycosylation step during glycoprotein biosynthesis. Biochem. J. 195, 639-644. Bause, E., Hettkamp, H., and Legler, G. (1982). Conformational aspects of N-glycosylation of proteins. Studies with linear and cyclic peptides as probes. Biochem. J. 203, 761-768. Bause, E., Muller, T., and Jaenicke, L. (1983). Synthesis and characterization of lipid-linked mannosyl oligosaccharides in Vulvux curteri f. nuguriensis. Arch. Biochem. Biophys. 220,200207. Baynes, 1. W., Hsu, A.-F., and Heath, E. C. (1973). The role of mannosyl-phosphoryl dihydropolyisoprenol in the synthesis of mammalian glycoproteins. J. Biol. Chem. 248, 56935704. Behrens. N. H., and Leloir, L. H. (1970). Dolichol monophosphate glucose: An intermediate in glucose transfer in liver. Proc. Nurl. Acud. Sci. U.S.A. 66, 153-159. Behrens, N. H., Parodi, A. J., Leloir, L. F., and Krisman, C. R. (1971). The role of dolichol monophosphate in sugar transfer. Arch. Biochem. Biophys. 143, 375-383. Behrens, N. H., Carminatti, H., Staneloni, R. J., Leloir, L. F., and Cantarella, A. I. (1973). Formation of lipid-bound oligosaccahrides containing mannose: Their role in glycoprotein synthesis. Proc. Nutl. Acad. Sci. U.S.A. 70, 3390-3394. Belcopitow, E., and Boscoboinik, D. (1982). Dolichyl-phosphate phosphatase and dolichyl-diphosphate phosphatase in rat liver microsomes. Eur. J. Biochem. 125, 167-173. Bergman, L. W., and Kuehl, W. M. (1977). Addition of glucosamine and mannose to nascent immunoglobulin heavy chains. Biochemistry 16, 4490-4497. Bergman, L. W., and Kuehl, W. M. (1982). The variable temporal relationship between translation and glycosylation and its effects on the efficiency of glycosylation. In “The Glycoconjugates,” (M. I. Horowitz, ed.), Vol. 3, pp. 81-98. Academic Press, New York. Bernard, B. A., Yamada, K. M., and Olden, K. (1982). Carbohydrates selectively protect a specific domain of fibronectin against proteases. J. Biol. Chem. 257, 8549-8554. Beyer, T. A., and Hill, R. L. (1982). Glycosylation pathways in the biosynthesis of nonreducing terminal sequences in oligosaccharides of glycoproteins. In “The Glycoconjugates’ ’ (M. I. Horowitz, ed.), Vol. 3, pp. 25-45. Academic Press, New York. Briles, E. B. (1982). Lectin-resistant cell surface variants of eukaryotic cells. Int. Rev. Cytol. 75, 101- 165. Budad, M. L., and Herbert, E. (1982). Effect of tunicamycin on the synthesis, processing, and secretion of pro-opiomelanocortin peptides in mouse pituitary cells. J. B i d . Chem. 257, 1012810135. Burgos, J., and Morton, R. A. (1962). The intracellular distribution of dolichol in pig liver. Biochern. J. 82, 454-456. Burke, D., and Keegstra, K. (1979). Carbohydrate structure of Sindbis virus glycoprotein E2 from virus grown in hamster and chicken cells. J. Virol. 29, 546-554. Burton, W. A., Scher, M. G . , and Waechter, C. J. (1979). Enzymatic phosphorylation of dolichol in central nervous tissue. J. Biul. Chem. 254, 7129-7136. Burton, W. A., Lucas, J. J., and Waechter, C. J. (1981a). Enhanced chick oviduct dolichol kinase activity during estrogen-induced differentiation. J. Biol. Chem. 256, 632-635. Burton, W. A., Scher, M. G., and Waechter, C. J. (1981b). Enzymatic dephosphorylation of endogenous and exogenous dolichyl monophosphate by calf brain membranes. Arch. Biochem. Biophys. 208, 409-417. Butterworth, P. H. W., and Hemming, F. W. (1968). Intracellular distribution of the free and esterified forms of dolichol in pig liver. Arch. Biochem. Biophys. 128, 503-508. Butterworth, P. H. W., Draper, H. H., Hemming, F. W., and Morton, R. A. (1966). In vivo incorporation of [2-L4C]mevalonateinto dolichol of rabbit and pig liver. Arch. Biochern. Biophys. 113, 646-653.

5. GLYCOSYLATJON OF MEMBRANE PROTEINS

235

Caccam. J. F., Jackson, J. J . , and Eylar, E. H. (1969). The biosynthesis of mannose-containing glycoproteins: A possible lipid intermediate. Biochem. Biophvs. Res. Commun. 35, 505-5 1 I , Carson, D. D., and Lennarz, W. J . (1981). Relationships of dolichol synthesis to glycoprotein synthesis during embryonic development. J . B i d . Chem. 256, 4679-4686. Carson, D. D., Earles, B. J.. and Lennarz, W. J. (1981). Enhancement of protein glycosylation in tissue slices by dolichylphosphate. J . Biol. Chem. 256, 11552. Chambers, J., and Elbein, A. D. (1975). Biosynthesis and characterization of lipid-linked sugars and glycoproteins in aorta. J . Biol. Chem. 250, 6904-6915. Chambers, J . , Forsee, W. T., and Elbein, A. D. (1977). Enzymatic transfer of mannose from mannosyl-phosphoryl-polyprenolto lipid-linked oligosaccharides by pig aorta. J . Biol. Chem. 252, 2498-2506. Chapman, A . , Trowbridge, 1. S . , Hyman, R., and Kornfeld, S . (1979a). Structureofthe lipid-linked oligosaccharides that accumulate in class E Thy- I-negative mutant lymphomas. Cell 17, 509515.

Chapman, A . , Li, E . , and Kornfeld, S . (1979b). The biosynthesis of the major lipid-linked oligosaccharide of Chinese hamster ovary cells occurs by the ordered addition of mannose residues. J . Biol. Chem. 254, 10243-10249. Chapman, A . , Fujimoto, K., and Kornfeld, S . (1980). The primary glycosylation defect in class E Thy-1 - mutant mouse lymphoma cells is an inability to synthesize dolichol-P-mannose.J . B i d . Chem. 255, 4441-4446. Chen, M.-C., Shimohigashi. Y . , Dufau, M. L., and Catt, K. F. (1982). Characterization and biological properties of chemically deglycosylated human chorionic gonadotropin. Role of carbohydrate moieties in adenylate cyclase activation. J . B i d . Chem. 257, 14446- 14452. Chen, W. W., and Lennarz, W. J. (1976). Participation of a trisaccharide-lipid in glycosylation of oviduct membrane glycoproteins. J . Biol. Chem. 251, 7802-7809. Chen, W. W., and Lennarz, W. J . (1977). Metabolism of lipid-linked N-acetylglucosamine intermediates. J . B i d . Chem. 252, 3473-3479. Chen, W. W., and Lennarz, W. J. (1978). Enzymatic synthesis of a glucose-containing oligosaccharide-lipid involved in glycosylation of proteins. J . B i d . Chem. 253, 5774-5779. Chen, W. W . , Lennarz, W. J., Tarentino, A. L.. and Maley. F. (1975). A lipid-linked oligosaccharide intermediate in glycoprotein synthesis in oviduct. Structural studies on the oligosaccharide chain. J . Biol. Chem. 250, 7006-7013. Chojnacki, T., and Dallner, G . (1983). The uptake of dietary polyprenols and their modification to active dolichols by the rat liver. J . Blol. Chem. 258, 916-922. Chu, F. K., and Maley. F. (1982). Stabilization of the structure and activity of yeast carboxypeptidase Y by its high-mannose oligosaccharide chains. Arch. Biochem. Biophys. 214, 134-1 39. Chu, F. K . , Trinible, R. B.. and Maley. F. (1978). The effect of carbohydrate depletion on the properties of yeast external invertase. J . Biol. Chem. 253, 8691-8693. Cowing, C., and Chapdelaine, J. M. (1983). T cells discriminate between la antigens expressed on allogeneic accessory cells and B cells: A potential function for carbohydrate side chains on la molecules. Proc. Nail. Acad. Sci. U . S . A . 80, 6000-6004. Criscuolo, B. A , , and Krag, S . S . (1982). Selection of tunicaniycin-resistant Chinese hamster ovary cells with increased N-acetylglucosaminyl transferase activity. J . Cell B i d . 94, 586-591 , Cumniings, R. D., and Roth, S. (1982). The discovery o f a lipid-linked glucuronide and its synthesis by chicken liver. J . Biol. Chem. 257, 1755-1764. Cummings, R. D.. Kornfeld, S . , Schneider. W. J.. Hohgood, K. K., Tolleshaug, H., Brown, M. S., and Goldstein, J. L. (1983). Biosynthesis of N- and 0-linked oligosaccharides of the low density lipoprotein receptor. J . Biol. Chem. 258, 15261- 15273. Cunningham. B. A . , Hoffman, S . , Rutishauser, U . , Hemperly. J . J . , and Edelman, G. M. (1983). Molecular topography of the neural cell adhesion moleculc N-CAM: Surface orientation and location of sialic acid-rich binding regions. Proc. N u / . Acod. Sci. U . S . A . 80, 31 16-3120.

236

SHARON S. KRAG

Czichi, U., and Lennarz, W. J. (1977).Localization of the enzyme system for glycosylation of proteins via the lipid-linked pathway in rough endoplasmic reticulum. J . Biol. Chem. 252, 7901-7904. Daleo, G. R., Hopp, H. E., Romero, P. A , , and Pont Lezica, R. (1977).Biosynthesis of dolichol phosphate by subcellular fractions from liver. FEBS Lett. 81, 41 1-414. Das, R. C., and Heath, E. C. (1980).Dolichyl-diphosphoryloligosaccharide-protein oligosaccharyltransferase: Solubilization, purification, and properties. Proc. Narl. Acad. Sci. U.S.A. 77, 381 1-3815. DeLuca, L. M. (1982).Studies on mannosyl carrier functions of retinol and retinoic acid in epithelia and mesenchymal tissue. J. Am. Acad. Dermarol. 6, 61 1-618. DeLuca, L. M., Maestri, N . , Rosso, G., and Wolf, G. (1973).Retinol glycolipids. J . Biol. Chem. 248, 641-648. DeRosa, P. A., and Lucas, J. J. (1982).Estrogen-induced changes in chick oviduct membrane glycoproteins. J . Biol. Chem. 257, 1017-1024. Edge, A. S.,Faltynek, C. R., Hof, L., Reichert, L. E., Jr., and Weber, P. (1981).Deglycosylation of glycoproteins by trifluoromethanesulfonic acid. Anal. Biochem 118, I3 1 - 137. Evans, P. J., and Hemming, F. W. (1973).The unambiguous characterization of dolichol phosphate mannose as a product of mannosyl transferase in pig liver endoplasmic reticulum. FEBS Lett. 31, 335-338. Firestone, G. L. (1983). The role of G protein glycosylation in the compartmentalization and processing of mouse mammary tumor virus glycoproteins in mouse mammary tumor virusinfected rat hepatoma cells. J . Biol. Chem. 258, 6155-6161. Firestone, G. L., and Heath, E. C. (1981).Role of protein glycosylation in the CAMP-mediated induction of alkaline phosphatatase in mouse L-cells. J . B i d . Chem. 256, 1404-1411. Fleischer, B. (1983).Mechanism of glycosylation in the Golgi apparatus. J. Histochem. Cytochem. 31, 1033-1040. Fliesler, S . J., and Schroepfer, G. J., Jr. (1983).Metabolism of mevalonic acid in cell-free homogenates of bovine retinas. Formation of novel isoprenoid acids. J . Biol. Chem. 258, 1506215070. Forsee, W. T., and Elbein, A. D. (1975).Glycoprotein biosynthesis in plants. Demonstration of lipid-linked oligosaccharides of mannose and N-acetylglucosamine. J. Biol. Chem. 250, 92839293. Forsee, W. T., and Elbein, A. D. (1976). Biosynthesis of lipid-linked oligosaccharides and glycoprotein in aorta: Stimulation by acceptor lipids isolated from liver. Proc. Nafl. Acad. Sci. U.S.A. 73, 2574-2578. Forsee, W. T., Griffin, J. A . , and Schutzbach, J. S. (1977).Mannosyl transfer from GDP-mannose to oligosaccharide-lipids. Biochem. Biophys. Res. Commun. 75, 799-805. Gandhi, C. R., and Kennan, R. W. (1983).The role of calmodulin in the regulation of dolichol kinase. J. Biol. Chem. 258, 7639-7643. Gershman, H., and Robbins, P. W. (1981).Transitory effects of glucose starvation on the synthesis of dolichol-linked oligosaccharides in mammalian cells. J . Biol. Chem. 256, 7774-7780. Ghalambor, M. A , , Warren, C. D., and Jeanloz, R. W. (1974). Biosynthesis of an Pl-2acetamido-2-deoxy-D-glucosyl Pz-polyisoprenyl pyrophosphate by calf pancreas microsomes . Biochem. J. 148, 245-251. Gibson, R., Leavitt, R., Komfeld, S., and Schlesinger, S. (1978). Synthesis and infectivity of vesicular stomatitis virus containing nonglycosylated G protein. Cell 13, 671-679. Gibson, R., Schlesinger, S., and Kornfeld, S. (1979).The nonglycosylated glycoprotein of vesicular stomatitis virus is temperature-sensitive and undergoes intracellular aggregation at elevated temperatures. J . Biol. Chem. 254, 3600-3607. Glabe, G. G., Hanover, J . A , , and Lennarz, W. J. (1980).Glycosylation of ovalbumin nascent

5. GLYCOSYLATION OF MEMBRANE PROTEINS

237

chains. The spatial relationship between translation and glycosylation. J . B i d . Chem. 255, 9236-9242.

Gleeson, P. A,. and Schachter, H. (1983). Control of glycoprotein synthesis. J . B i d . Chrm. 258, 6 162-6 173.

Goverman. J . M . , Parsons, T. F.. and Pierce, J . G. (1982). Enzymatic deglycosylation of the subunits of chorionic gonadotropin. Effects on formation of tertiary structure and biological activity. J. B i d . Chem. 257, 15059-15064. Gralnick, H. R., Williams, S. B., and Rick, M. E. (1983). Role of carbohydrate in multimeric structure of factor Vlllivon Willebrand factor protein. Proc. Nad. Acad. Sci. U.S.A. 80, 27712774.

Grange, D. K . , and Adair, W. L., Jr. (1977). Studies on the biosynthesis of dolichyl phosphate: Evidence for the in v i m formation of 2.3-dehydrodolichyl phosphate. Biochem. Biophys. Res. Commun. 79, 734-740. Grant, S. R., and Lennarz, W. J. (1983). Relationship between oligosaccharides-lipid synthesis and protein synthesis in mouse LM cells. Eur. J . Biochem. 134, 575-583. Hakomori, S . (1964). A rapid permethylation of glycolipid, and polysaccharide catalyzed by methylsulfinylcarbanion in dimethyl sulfoxide. J . Biochem. (Tokyo) 55, 205-208. Hanover, J . A,, and Lennarz, W. J. (1979). The topological orientation of N’Nl-diacetylchitobiosylpyrophosphoryldolichol in artificial and natural membranes. J . B i d . Chem. 254, 9237-9246. Hanover, J. A,, and Lennarz, W. J. (1980). N-linked glycoprotein assembly. Evidence that oligosaccharide attachment occurs within the lumen of the endoplasmic reticulum. J . Biol. Chem. 255, 3600- 3604.

Hanover, J. A,, and Lennarz, W. J. (1981). Transmembrane assembly of membrane and secretory glycoproteins. Arch. Biochem. Biophys. 211, 1-19. Hanover, J. A,, and Lennarz, W. J. (1982). Transmembrane assembly of N-linked glycoproteins. Studies on the topology of saccharide-lipid synthesis. J . B i d . Chem. 257, 2787-2794. Hanover, J. A,, Eking, I., Mintz, G. R., and Lennarz, W. J . (1982). Temporal aspects ofthe N- and 0-glycosylation of human chorionic gonadotropin. J . Biol. Chem. 257, 10172- 10177. Harford, J . , and Ashwell, G . (1982). The hepatic receptor for asialoglycoproteins. In “The Glycoconjugates” (M. I. Horowitz, ed.), Vol. 4, pp. 27-55. Academic Press, New York. Harford, J . B., and Waechter, C . J. (1979a). Lipid-mediated glycosylation of a white matter membrane glycoprotein with an apparent molecular weight of 24,000. J . Neurochem. 32, 17071715.

Harford, J. B . , and Waechter, C. J. (1979b). Transfer of N , I-diacetylchitobiose from dolichyl diphosphate into a gray matter membrane glycoprotein. Arch. Biochem. Biophys. 197, 424435.

Harford, J. B., Waechter, C. J., and Earl, F. L. (1977). Effect of exogenous dolichyl monophosphate on a developmental change in mannosylphosphoryldolichol biosynthesis. Biochem. Biophys. Res. Commun. 76, 1036-1043. Harpaz, N., and Schachter, H. (1980). Control of glycoprotein synthesis. Processing of asparaginelinked oligosaccharides by one or more rat liver Golgi a-D-mannosidases dependent on the prior I. J . action of UDP-N-acetylglucosamine:a-D-mannoside-~2-N-acetylglucosaminyltransferase Biol. Chem. 255, 4894-4902. Hart, G. W. (1982). The role of asparagine-linked oligosaccharides in cellular recognition by thymic lymphocytes. Effects of tunicamycin on the mixed lymphocyte reaction. J . Biol. Chem. 257, I5 1-1 58. Hart, G. W., and Lennarz, W. J . (1978). Effects of tunicamycin on the biosynthesis ofglycosaminoglycans by embryonic chick cornea. J . B i d . Chem. 253, 579555801. Hart, G. W.. Brew, K . , Grant, C. A,, Bradshaw, R. A,, and Lennarz, W. J. (1979). Primary

238

SHARON S.KRAG

structural requirements for the enzymatic formation of the N-glycosidic bond in glycoproteins. Studies with natural and synthetic peptides. J. Eiol. Chem. 254, 9747-9753. Haselbeck, A,, and Tanner, W. (1982). Dolichyl phosphate-mediated mannosyl transfer through liposomal membranes. Proc. Narl. Acud. Sci. U.S.A. 79, 1520-1524. Hasilik, A , , Waheed, A., and von Figura, K. (1981). Enzymatic phosphorylation of lysosomal enzymes in the presence of UDP-N-acetylglucosamine. Absence of the activity in I-cell fibroblasts. Eiochem. Biophys. Res. Commun. 98, 761-767. Hayes, G. R., and Lucas, J. J. (1980). A novel N-acetylglucosaminyl polyprene in hen oviduct. J. Eiol. Chem. 255, 7536-7539. Heifetz, A,, and Elbein, A. D. (1977a). Biosynthesis of Man-GlcNAc-GlcNAc-pyrophosphorylpolyprenol by a solubilized enzyme from aorta. Eiochem. Biophys. Res. Commun. 75, 20-28. Heifetz, A., and Elbein, A. D. (1977b). Solubilization and properties of mannose and N-acetylglucosamine transferases involved in formation of polyprenyl-sugar intermediates. J. Eiol. Chem. 252, 3057-3063. Herscovics, A,, Warren, C. D., and Jeanloz, R. W. (1975). Anomeric configuration of the dolichyl D-mannosyl phosphate formed in calf pancreas microsomes. J. Eiol. Chem. 205, 8079-8084. Herscovics, A,, Bugge, B., and Jeanloz, R. W. (1977a). Glucosyltransferase activity in calf pancreas microsomes. Formation of dolichyl D-[ ~4C]glucosy1phosphate and I4C-labelled lipidlinked oligosaccharides from UDP-D-[14C]glucose. J. B i d . Chem. 252, 2271-2277. Herscovics, A,, Goluvtchenko, A. M., Warren, C. D., Bugge, B., and Jeanloz, R. W. (1977b). Mannosyltransferase activity in calf pancreas microsomes. Formation of 14C-labeled lipidlinked oligosaccharides from GDP-D-[ “Tlmannose and pancreatic dolichylp-D-[ I4C]mannopyranosyl phosphate. J. Eiol. Chem. 252, 224-234. Herscovics, A,, Warren, C . D., Bugge, B., and Jeanloz, R. W. (1978). Biosynthesis of PI-Di-Nacetyl-a-chitobiosyl P2-dolichyl pyrophosphate in calf pancreas microsomes. J. Eiol. Chem. 253, 160-165. Herscovics, A,, Warren, C. D., Bugge, B., and Jeanloz, R. W. (1980). Biosynthesis of dolichyl pyrophosphate trisaccharide from synthetic dolichyl pyrophosphate di-N-acetylchitobiose and GDP-[I4 Clmannose in calf pancreas microsomes. FEES Left. 120, 271-274. Herscovics, A., Warren, C . D., and leanloz, R. W. (1983). Solubilization of an a-(I,3)-D-mannosyl transferase from pancreas which utilizes synthetic dolichyl pyrophosphate trisaccharide as substrate. FEBS Leu. 156, 298-302. Hoflack, B., Cacan, R., and Verbert, A. (1981). Dolichol pathway in lymphocytes from rat spleen. Influence of the glucosylation on the cleavage of dolichyl diphosphate oligosaccharides into phosphooligosaccharides. Eur. J. Biochem. 117, 285-290. Hoflack, B., Debeine, P., Cacan, R., Montreuil, J . , and Verbert, A. (1982). Dolichol-dependent synthesis of chitobiosyl proteins and their further mannosylation. A second route for glycosylation of proteins in rat-spleen lymphocytes? Eur. J. Eiochem. 124, 527-531. Hopwood, J. J., and Dorfman, A. (1977). Isolation of lipid glucuronic acid and N-acetylglucosamine derivatives from a rat fibrosarcoma. Eiochern. Biophys. Res. Commun. 75, 472-479. Hsieh, P., Rosner, M. R., and Robbins, P. W. (1983). Selective cleavage by endo-e-N-acetylglucosaminidase H at individual glycosylation sites of Sindbis virion envelope glycoproteins. J. Eiol. Chem. 258, 2555-2561. Hsu, A.-F., Baynes. J. W., and Heath, E. C . (1974). The role of a dolichol-oligosaccharideas an intermediate in glycoprotein biosynthesis. Proc. Nurl. Acud. Sci. U.S.A. 71, 2391-2395. Hubbard, S. C., and Ivatt, R. J. (1981). Synthesis and processing of asparagine-linked oligosaccharides. Annu. Rev. Eiochem. 50, 555-583. Hubbard, S . C., and Robbins, P. W. (1979). Synthesis and processing of protein-linked oligosaccharides in vivo. J. B i d . Chem. 254, 4568-4576. Hubbard, S . C., and Robbins, P. W. (1980). Synthesis of the N-linked oligosaccharides of glycopro-

5. GLYCOSYLATION OF MEMBRANE PROTEINS

239

teins. Assembly of the lipid-linked precursor oligosaccharide and its relation to protein synthesis in uiuo. J . Biol. Chem. 255, 11782-1 1793. Huffaker, T. C . , and Robbins, P. W. (1982). Temperature-sensitive yeast mutants deficient in asparagine-linked glycosylation. J . Biol. Chem. 257, 3203-32 10. Hunt, L. A. (1980a). CHO cells selected for phytohemagglutinin and conA resistance are defective in both early and late stages of protein glycosylation. Cell 21, 407-415. Hunt, L. A. (l980b). Altered synthesis and processing of oligosaccharides of vesicular stomatitis virus glycoprotein in different lectin-resistant Chinese hamster ovary cell lines. J . Virul. 35, 362-370. Hunt, L. A , , Etchison, J. R., and Summers, D. F. (1978). Oligosaccharide chains are trimmed during synthesis of the envelope glycoprotein of vesicular stomatitis virus. Proc. Natl. Acad. Sci. U.S.A. 75, 754-758. Hyman. R., and Trowbridge, I. (1977). Analysis of lymphocyte surface antigen expression by the use of variant cell lines. Cold Spring Harbor Symp. Qitunr. Biol. 41, 407-415. Hynes, R. O., and Yamada, K. M. (1982). Fibronectins: Multifunctional modular glycoproteins. J . Cell Biol. 95, 369-377. Ivatt, R. J., Das, 0. P., Henderson, E. J., and Robbins, P. W. (1981). Developmental regulation of glycoprotein biosynthesis in Dictyostelium. J . Suprarnol. Strucr. 17, 359-368. Jamieson, J. C. (1977). Studies on the site of addition of sialic acid and glucosamine to rat a,-acid glycoprotein. Can. J . Biochem. 55, 408-414. Jensen, I. W . , and Schutzbach, J. S. (1982). The biosynthesis of oligosaccharide-lipids. Activation of mannosyl transferase I1 by specific phospholipids. J . Biol. Chem. 257, 9025-9029. Jensen, J. W., Springfield, J. D., and Schutzbach, J. S . (1980). The biosynthesis ofoligosaccharidelipids. Isolation of an oligosaccharide-P-P-lipidacceptor. J . Biol. Chem. 255, 1 1268-1 1272. Kaluza, G . (1975). Effect of impaired glycosylation on the biosynthesis of Semliki forest virus glycoproteins. J . Virul. 16, 602-612. Kalyan, N. K., and Bahl, 0. P. (1983). Role of carbohydrate in human chorionic gonadotropin. Effect of deglycosylation on the subunit interaction and on its in uirru and in uiuo biological properties. J . Biul. Chem. 258, 67-74. Kalyan, N. K., Lippes, H. A., and Bahl, 0. P. (1982). Role of carbohydrate in human chorionic gonadotropin. Effect of periodate oxidation and reduction on its in v i m and in uiuo biological properties. J . Biol. Chem. 257, 12624-12631. Kang. M . S., Spencer, J. P., and Elbein, A. D. (1978). Amphomycin inhibition of mannose and GlcNAc incorporation into lipid-linked saccharides. J . Biol. Chem. 253, 8860-8866. Karp, D. R., Atkinson, J. P., and Shreffler, D. C. (1982). Genetic variation in glycosylation of the fourth component of murine complement. Association with hemolytic activity. J . Biul. Chem. 257, 7330-7335. Kato, S . , Tsuji, M . , Nakaniski, Y., and Sueuki, S. (1980). Enzymatic dephosphorylation of dolichyl pyrophosphate-the bacitracin-sensitive, rate-limiting step for dolichyl mannosyl phosphate synthesis in rat liver microsomes. Biochem. Biuphys. Res. Commun. 95, 770-776. Kean, E. L. (l977a). Concerning the catalytic hydrogenation of polyprenol phosphate-mannose synthesized by the retina. J . Biol. Chem. 252, 5619-5621. Kean. E. L. (1977b). GDP-mannose-polyprenyl phosphate mannosyltransferases of the retina. J . Biol. Chem. 252, 5622-5629. Kean, E. L. (1982). Activation by dolichol phosphate-niannose of the biosynthesis of N-acetylglucosaminylpyrophosphoryl polyprenols by the retina. J . Biol. Chem. 257, 7952-7954. Kean, E. L. ( 1983). Influence of metal ions on the biosynthesis on N-acetylglucosaminyl polyprenols by the retina. Biochim. Biuphys. Artu 750, 268-273. Keenan, R . W . , and Kruczek, M. E. (1976). The esterification of dolichol by rat liver microsomes. Biochemisrry 15, 1586-1591.

240

SHARON S. KRAG

Keenan, R. W., Fischer, J. B., and Kruczek, M. E. (1977). The tissue and subcellular distribution of [3H]dolichol in the rat. Arch. Biochem. Biophys. 179, 634-642. Keller, R. K., and Adair, W. L. (1977). Microdetermination of dolichol in tissues. Biochim. Biophys. Acta 489, 330-336. Keller, R. K., Boon, D. Y.,and Crum, F. C. (1979). N-Acetylglucosamine-I-phosphatetransferase from hen oviduct: Solubilization, characterization and inhibition by tunicamycin. Biochemistry 18, 3946-3952. Keller, R. K., Jehle, E., and Adair, W. L. (1982). The origin of dolichol in the liver of the rat. Determination of the dietary contribution. J. Biol. Chem. 257, 8985-8989. Kiely, M. L., McKnight, G. S., and Schimke, R. T. (1976). Studies on the attachment of carbohydrate to ovalbumin nascent chains in hen oviduct. J. Biol. Chem. 251, 5490-5495. Kim, J. I., and Conrad, H. E. (1976). Kinetics of mucopolysaccharide and glycoprotein synthesis by chick embryo chondrocytes. J . Biol. Chem. 251, 6210-6217. Kobata, A. (1978). Endo-P-N-acetylglucosaminidasesCI and C,, from Clostridium perfringens. In “Methods in Enzymology” (V. Ginsburg, ed.), Vol. 50, pp. 567-574. Academic Press, New York. Kornfeld, R., and Kornfeld, S. (1980). Structure of glycoproteins and their oligosaccharide units. In “The Biochemistry of Glycoproteins and Proteoglycans” (W. I. Lennarz, ed.), pp. 1-34. Academic Press, New York. Kornfeld, S. (1982). Oligosaccharide processing during glycoprotein biosynthesis. In “The Glycoconjugates” (M. I. Horowitz, ed.), Vol. 3, pp. 3-23. Academic Press, New York. Kornfeld, S., Li, E., and Tabas, I. (1978). The synthesis of complex-type oligosaccharides. 11. Characterization of the processing intermediates in the synthesis of the complex oligosaccharide units of the vesicular stomatitis virus G protein. J. Biol. Chem. 253, 7771-7778. Kornfeld, S., Gregory, W., and Chapman, A. (1979). Class E Thy-I negative mouse lymphoma cells utilize an alternate pathway of oligosaccharide processing to synthesize complex-type oligosaccharides. J. Biol. Chem. 254, 11649-1 1654. Krag, S. S . (1979). A concanavalin A-resistant Chinese hamster ovary cell line is deficient in the synthesis of [3H]glucosyl oligosaccharide-lipid. J . Biol. Chem. 254, 9167-9177. Krag, S. S., and Robbins, P. W. (1977). Sindbis envelope proteins as endogenous acceptors in reactions of guanosine diphosphate [ 14C]mannosewith preparations of infected chicken embryo fibroblasts. J. Biol. Chem. 252, 2621-2629. Krag, S. S., and Robbins, A. R. (1982). A Chinese hamster ovary cell mutant deficient in glucosylation of lipid-linked oligosaccharides synthesizes lysosomal enzymes of altered structure and function. J. Biol. Chem. 257, 8424-843 1. Kronquist, K. E., and Lennarz, W. J. (1978). Enzymatic conversion of proteins to glycoproteins by lipid-linked saccharides: A study of potential exogeneous acceptor proteins. J. Suprarnol. Struct. 8, 51-65. Lawson, E. Q., Hedlund, B. E., Ericson, M. E., Mood, D. A,, Litman, G. W., and Middaugh, R. (1983). Effect of Carbohydrate on Protein Solubility. Arch. Biochem. Biophys. 220, 572-575. Leavitt, R., Schlesinger, S . , and Komfeld, S. (1977). Impaired intracellular migration and altered solubility of nonglycosylated glycoproteins of vesicular stomatitis virus and Sindbis virus. J. Biol. Chem. 252, 9018-9023. Lehle, L., and Tanner, W. (1976). The specific site of tunicamycin inhibition in the formation of dolichol-bound N-acetylglucosamine derivatives. FEBS Lett. 71, 167- 170. Lehle, L., and Tanner, W. (1978a). Glycosyl transfer from dolichyl phosphate sugars to endogenous and exogenous glycoprotein acceptors in yeast. Eur. J. Biochem. 83, 563-570. Lehle, L., and Tanner, W. (1978b). Biosynthesis and characterization of large dolichyl disphosphate-linked oligosaccharides in Succharornyces cerevisiae. Biochim. Biophys. Acra 539,2 18229.

5. GLYCOSYLATION OF MEMBRANE PROTEINS

24 1

Leloir, L. F., Staneloni, R. J., Carminatti, H., and Behrens. N. H. (1973). The biosynthesis of a NN‘-diacetylchitobiose containing lipid by liver microsonies. A probable dolichol pyrophosphate derivative. Biochem. Biophys. Res. Commun. 52, 1285- 1292. Lennarz. W. J . (1979). The enzymatic conversion of membrane and secretory proteins to glycoproteins. Miami Winrer Symp. 16, 407-423. Levy. J. A., Carminatti, H., Cantarella, A. J . , Behrens, N. H., Leloir, L. F . , and Tabora, E. ( 1974). Mannose transfer to lipid linked di-N-acetylchitobiose. Biochem Biophys. Res. Commun. 60, I 18- 125. Li, E.. and Kornfeld, S. (1979). Biosynthesis of lipid-linked oligosaccharides. Isolation and structure of a second lipid-linked oligosaccharide in Chinese hamster ovary cells. J . Biol. Chem. 254, 2754-2758. Li, E., Tabas, I . , and Kornfeld, S . (1978). The synthesis of complex-type oligosaccharides. 1. Structure of the lipid-linked oligosaccharide precursor of the complex-type oligosaccharides of the vesicular stomatitis virus G protein. J . B i d . Chem. 253, 7762-7770. Liscum, L., Cummings, R. D., Anderson, R. G . W., DeMartino, G. N., Goldstein, J. L., and reductase: A transmembrane glycoproBrown, M. S . ( 1 983). 3-Hydroxy-3-methylglutaryl-CoA tein of the endoplasmic reticulum with N-linked “high-mannose” oligosaccharides. P roc. Nut/. Acad. Sci. U . S . A . 80, 7165-7169. Liu, T., Stetson, B., Turco, S. J., Hubbard, S. C . , and Robbins, P. W. (1979). Arrangement of glucose residues in the lipid-linked oligosaccharide precursor of asparaginyl oligosaccharides. J . Biol. Chem. 254, 4554-4559. Lucas. J . J., and Levin, E. (1977). Increase in the lipid intermediate pathway of protein glycosylation during hen oviduct differentiation. J. Biol. Chem. 252, 4330-4336. Lucas, J. J., Waechter, C. J., and Lennarz, W. J. (1975). Membrane glycoproteins 11. The participation of lipid-linked oligosaccharide in membrane glycoprotein synthesis. J . Biol. Chem. 250, 1992-2002. Manjunath, P., and Sairam, M. R. (1982). Biochemical, biological, and immunological properties of chemically deglycosylated human choriogonadotropin. J . Biol. Chem. 257, 7109-7 I 15. Manjunath, P., and Sairam, M. R. (1983). Enhanced thermal stability of chemically deglycosylated human choriogonadotropin. 1. Biol. Chem. 258, 3554-3558. Manjunath, P., Sairam, M. R., and Schiller, P. W. (1982). Chemical deglycosylation of ovine pituitary lutropin. Biochem. J . 207, 11-19. Mankowski, T., Jankowski, W . , Chojnacki, T . , and Franke, P. (1976). Css-dolichol: Occurrence in pig liver and preparation by hydrogenation of plant undecaprenol. Biochemistry 15,2 125-2130. Marshall, R. D. (1974). The nature and metabolism of the carbohydrate-peptide linkage of glycoproteins. Biochem. SOC. Symp. 40, 17-26. Martin, H. G., and Thorne, K. J. I . (1974a). The involvement of endogeneous dolichol in the formation of lipid-linked precursors of glycoproteins in rat liver. Biochem. J . 138, 281289. Martin, H. G . , and Thorne, K. J. I. (1974b). Synthesis of radioactive dolichol from [4S-3H]mevalonate in the regenerating rat liver. Biochem. J . 138, 277-280. Mason, R. J., and Williams, M. C. (1980). Phospholipid composition and ultrastructure of A549 cells and other cultured pulmonary epithelial cells of presumed type I1 cell origin. Biochim Biophys. A C ~ U617, 36-50. Merlie, J. P., Sebbane, R.. Tzartos, S . , and Lindstrom, J. (1982). Inhibition of glycosylation with tunicamycin blocks assembly of newly synthesized acetylcholine receptor subunits in muscle cells. J . Biol. Chem. 257, 2694-2701. Mills, I. T., and Adamany, A. M. (1978). Impairment of dolichyl saccharide synthesis and dolicholmediated glycoprotein assembly in the aortic smooth muscle cell in culture by inhibitors of cholesterol biosynthesis. J . Biol. Chem. 253, 5270-5273.

242

SHARON S.KRAG

Mizunaga, T., and Noguchi, T. (1982). The role of core-oligosaccharide in formation of an active acid phosphatase and its secretion by yeast protoplasts. 1. Biochem. 91, 191-200. Morrow, J. S., Weintraub, B. D., and Rosen, S. W. (1983). Butyrate regulates glycosylation of the glycoprotein hormone alpha subunit secreted by glucose-starved human liver cells. Biochem Biophys. Res. Commun. 112, 115-125. Mort, A. J., and Lamport, D. T. A. (1977). Anhydrous hydrogen fluoride deglycosylates glycoproteins. Anal. Biochem. 82, 289-309. Muramatsu, T. (1978). Endo-P-N-acetylglucosaminidaseD from Diplococcus pneumoniae. In “Methods in Enzymology” (V. Ginsburg, ed.), Vol. 50, pp. 555-559. Academic Press, New York. Murazumi, N., Yamamori, S., Araki, Y., and Ito, E. (1979). Anomeric configuration of N-acetylglucosaminyl phosphorylundecaprenols formed in Bacillus cereus membranes. J . B i d . Chem. 254, 11791-11793. Murphy, L. A,, and Spiro, R. G . (1981). Transfer of glucose to oligosaccharide-lipid intermediates by thyroid microsomal enzymes and its relationship to the N-glycosylation of proteins. J . B i d . Chem. 256, 7487-7494. Narasirnhan, S . (1982). Control of Glycoprotein Synthesis. J. Biol. Chem. 257, 10235-10242. Narasimhan, S., Stanley, P., and Schachter, H. (1977). Control of glycoprotein synthesis. Lectinresistant mutant containing only one of two distinct N-acetylglucosaminyltransferase activities present in wild-type Chinese hamster ovary cells. J. B i d . Chem. 252, 3926-3933. Neufeld, E. F., and Ashwell, G. (1980). Carbohydrate recognition systems for receptor-mediated pinocytosis. In “The Biochemistry of Glycoproteins and Proteoglycans” (W. J. Lennarz, ed.), pp. 241-266. Plenum, New York. Neufeld, E. F., and Robbins, A. R. (1982). Pleiotropic mutations of lysosomal function in human patients and Chinese hamster ovary cells. Prog. Clin. Biol. Res. 103B, 176-185. Nilsson, B., Nakazawa, K., Hassell, J. R., Newsome, D. A,, and Hascall, V. (1983). Structure of oligosacchaiides and the linkage region between keratan sulfate and the core protein on proteoglycans from monkey cornea. J . B i d . Chem. 258, 6056-6063. Nose, M., and Wigzell, H. (1983). Biological significance of carbohydrate chains on monoclonal antibodies. Proc. Narl. Acud. Sci. U.S.A. 80, 6632-6636. Olden, K., Parent, J. B., and White, S. S . (1982). Carbohydrate moieties of glycoproteins. A reevaluation of their function. Biochim. Biophys. Acru 650, 209-232. Paiement, J., and Bergeron, J. J. M. (1983). Localization of GTP-stimulated core glycosylation to fused microsomes. J . Cell Biol. 96, 1791-1796. Palamarczyk, G., and Hemming, F. W. (1975). The formation of mono-N-acetylhexosamine derivatives of dolichol diphosphate by pig liver microsomal fractions. Biochem. J . 148, 245-251. Pan, Y.-T., and Elbein, A. D. (1982). The formation of lipid-linked oligosaccharides in MadinDarby canine kidney cells. Changes in oligosaccharide profiles induced by glucosamine. J . Biol. Chem. 257, 2795-2801. Parodi, A. J., and Cazzulo, J . J. (1982). Protein glycosylation in Trypanosoma cruzi. 11. Partial characterization of protein-bound oligosaccharides labelled in vivo. J. Biol. Chem. 257, 76417645. Parodi, A. J., and Leloir, L. F. (1979). The role of lipid intermediates in the glycosylation of proteins in eukaryotic cells. Biochim. Biophys. Acru 559, 1-37. Parodi, A. J., and Quesada-Allue, L. A. (1982). Protein glycosylation in Trypanosoma cruzi. 1. Characterization of dolichol-bound monosaccharides and oligosaccharides synthesized in vivo. J . Biol. Chem. 257, 7637-7640. Parodi, A. J., Behrens, N. H., Leloir, L. F., and Carminatti, H. (1972). The role of polyprenolbound saccharides as intermediates in glycoprotein synthesis in liver. Proc. Narl. Acud. Sci. U.S.A. 69, 3268-3272.

5. GLYCOSYLATION OF MEMBRANE PROTEINS

243

Parodi, A. J., Quesada-Allue, L. A,, and Cazzulo, J. J. (1981). Pathway of protein glycosylation in the trypanosomatic Crirhidiu fuscirulu/u. Proc. Nut/. Acud. Sci. U . S . A . 78, 6201-6205. Parodi, A. J . , Mendelzon, D. M., and Lederkremer, G . Z. (1983). Transient glucosylation of protein-bound Man9GlcNAc2, Man8GlcNAc2, and Man7GlcNAc2in calf thyroid cells. A possible recognition signal in the processing of glycoproteins. J . B i d . Chem. 258, 8260-8265. Pennock, J. F.. Hemming, F. W., and Morton, R. A. (1960). Dolichol: A naturally occurring isoprenoid alcohol. Nurure (London) 186, 470-472. Perez, M., and Hirschberg, C. ( 1983). Translocation of uridine diphospho-N-acetylglucosamineinto rat liver golgi-derived vesicles. J . Cell Biol. 97, 442a. Peterson, P. A., Rask, L., Helting, T., Ostberg. L., and Fernstedt, Y. (1976). Formation and properties of retinylphosphate galactose. J . B i d . Chem. 251, 4986-4995. Pless, D. D., and Lennarz, W. J . (1975). A lipid-linked oligosaccharide intermediate in glycoprotein synthesis. Characterization of [Man-14C]glycoproteinslabelled from [Man-14C]oligosaccharide-lipid and GDP-["T]man. J . B i d . Chem. 250, 7014-7019. Pless, D. D., and Lennarz, W. J. (1977). Enzymatic conversion of proteins to glycoproteins. Proc. Narl. Acud. Sci. U.S.A. 74, 134-138. Plouhar, P. L., and Bretthauer, R. K . (1982). A phospholipid requirement for dolichol pyrophosphate N-acetylglucosamine synthesis in phospholipase A2-treated rat lung niicrosomes. J . Biol. Chem. 257, 8907-89 I I . Potter, J . E. R., and Kandutsch, A. A. (1982). Increased synthesis and concentration of dolichyl phosphate in mouse spleens during phenylhydrazine-induced erythropoiesis. Biochem. Biophys. Res. Commun. 106, 691-696. Potter, J . E. R., and Kandutsch. A. A. (1983). Different tissue concentrations and rates of synthesis of dolichol, dolichyl acyl esters, and dolichyl phosphate in mouse testes and preputial glands. Biochem. Biophys. Res. Commun. 110, 512-518. Prakash, C., and Vijay, I . K . (1982). characterization of intermediates up to lipid-linked heptasaccharide implicated in the biosynthesis of Succhuromyces cerevisie mannoproteins. Biochemistry 21, 4810-4818. Prives, J., and Ban-Sagi, D. (1983). Effect of tunicamycin, an inhibitor of protein glycosylation, on the biological properties of acetylcholine receptor in cultured muscle cells. J . B i d . Chem. 258, 1775-1780. Quesda-Allue, L. A,, and Parodi, A. J. (1983). Novcl mannose carrier in the trypanosomatid Crithidiu fusiculuru behaving as a short a-saturated polyprenylphosphate. Biochem. J . 212, 123- 128. Radominska-Pyrek, A., Chojnacki, T., and St. Pyrek, J . (1979). Fully unsaturated decaprenol from bovine pituitary glands. Biochem. Biophys. Res. Commun. 86, 395-401. Raetz, C. R. M., Wermuth. M. M., Mclntyre. T. M.. Esko, J . D., and Wing, D. C. (1982). Somatic cell cloning in polyester stacks. Proc. Nutl. Acud. Sri. U.S.A. 79, 3223-3227. Ravoet, A.-M., Amar-Costesec, A,, Godelaine, D., and Beaufay, H. (1981). Quantitative assay and subcellular distribution of enzymes acting on dolichyl phosphate in rat liver. J . Cell B i d . 91, 679-688. Rearick, J. I . . Chapman, A , , and Kornfeld, S . (1981a). Glucose starvation alters lipid-linked oligosaccharide biosynthesis in Chinese hamster ovary cells. J . Biol. Chem. 256, 62556261, Rearick, J . I . , Fujimoto, K., and Kornfeld, S . (1981b). Identification of the mannosyl donors involved in the synthesis of lipid-linked oligosaccharides. J . Biol. Chem. 256, 3762-3769. Reed, B. C., Ronnett, G. V . . and Lane. M. D. (1981). Role of glycosylation and protein synthesis in insulin receptor metabolism by 3T3-LI mouse adipocytes. Proc. Nutl. Acud. Sci. U.S.A. 78, 2908-29 12. Reitman, M. L., and Kornfeld, S . (1981). UDP-N-Acetylglucosamine: glycoprotein N-acetyl-

244

SHARON S. KRAG

glucosamine- 1-phosphotransferase. Proposed enzyme for the phosphorylation of the high mannose oligosaccharide units of lysosomal enzymes. J. Biol. Chem. 256, 4275-4281. Reitman, M. L., Trowbridge, I. S., and Komfeld, S . (1982). A lectin-resistant mouse lymphoma cell line is deficient in glucosidase 11, a glycoprotein-processing enzyme. J . Eiol. Chem. 257, 10357- 10363. Richards, J. B., and Hemming, F. W. (1972). The transfer of mannose from guanosine diphosphate mannose to dolichol phosphate and protein by pig liver endoplasmic reticulum. Biochem. J. 130, 77-93. Rip, J. W., and Carroll, K. K. (1980). Properties of a dolichol phosphokinase activity associated with rat liver microsomes. Can J . Biochem. 58, 1051-1056. Rip, J. W., and Carroll, K. K. (1982). The submicrosomal distribution of dolichol and dolichol phosphokinase activity in rat liver. Can. J. Biochem. 60, 36-41. Rip, J. W., Rupar, C. A., Chaundhary, N., and Carroll, K. K. (1981). Localization of a dolichyl phosphate phosphatase in plasma membranes of rat liver. J . Eiol. Chem. 256, 1929-1934. Rip, J. W., Chaudbary, N., and Carroll, K. K. (1983). The submicrosomal distribution of dolichyl phosphate and dolichyl phosphate phosphatase in rat liver. J . Eiol. Chem. 258, 14926-14930. Robbins, A. R., and Myerowitz, R. (1981). The mannose 6-phosphate receptor of Chinese hamster ovary cells. Compartmentalization of acid hydrolases in mutants with altered receptor. J . Biol. Chem. 256, 10623-10627. Robbins, P. W., Hubbard, S. C., Turco, S. J . , and Wirth, D. F. (1977a). Proposal for a common oligosaccharide intermediate in the synthesis of membrane glycoproteins. Cell 12, 893-900. Robbins, P. W., Krag, S . S., and Liu, T. (1977b). Effects of UDP-glucose addition on the synthesis of mannosyl lipid-linked oligosaccharides by cell-free fibroblast preparations. J. Eiol. Chem. 252, 1780-1785. Robbins, A. R., Myerowitz, R., Youle, R. J., Murray, G. J., and Neville, D. M., Jr. (1981). The mannose 6-phosphate receptor of Chinese hamster ovary cells. Isolation of mutants with altered receptors. J . B i d . Chem. 256, 10618-10622. Roden, L. (1980). Structure and metabolism of connective tissue proteoglycans. In “The Biochemistry of Glycoproteins and Proteoglycans” (W. J. Lennarz, ed.), pp. 267-371. Plenum, New York. Ronin, C. (1980). Solubilization of oligosaccharide transferase and glucosidase activities from thyroid rough microsomes. FEES Lett. 113, 340-344. Ronin, C., and Caseti, C. (1981). Transfer of glucose in the biosynthesis of thyroid glycoproteins. 11. Possibility of a direct transfer of glucose from UDP-glucose to proteins. Biochim. Biophys. Acta 674, 58-64. Ronin, C., Bouchilloux, S., Granier, C., and van Rietschoten, J. (1978). Enzymatic N-glycosylation of synthetic asn-x-thr containing peptides. FEES Lett. 96, 179- 182. Ronin, C . , Caseti, C., and Bouchilloux, S. (1981a). Transfer of glucose in the biosynthesis of thyroid glycoproteins. I. Inhibition of glucose transfer to oligosaccharide lipids by GDP-mannose. Biochim. Eiophys. Acta 674, 48-57. Ronin, C., Granier, C., Caseti, C., Bouchilloux, S . , and van Rietschoten, J. (1981b). Synthetic substrates for thyroid oligosaccharide transferase. Effect of peptide chain length and modifications in the -am-xaa-thr region. Eur. J. Biochem. 118, 159-164. Rossignol, D. P., Lennarz, W. J., and Waechter, C. I. (1981). Induction of phosphorylation of dolichol during embryonic development of the sea urchin. 1.Biol. Chem. 256, 10538-10542. Rossignol, D. P., Scher, M., Waechter, C. J., and Lennarz, W. J. (1983). Metabolic interconversion of dolichol and dolichyl phosphate during development of the sea urchin embryo. J. Eiol. Chem., in press. Rosso, G. C., Masushige, S., Quill, H., and Wolf, 0.(1977). Transfer of mannose from mannosyl retinyl phosphate to protein. Proc. Narl. Acad. Sci. U.S.A. 74, 3762-3766.

5. GLYCOSYLATION OF MEMBRANE PROTEINS

245

Rothman, J. E., and Lodish. H. F. (1977). Synchronised trdnsniembrane insertion and glycosylation of a nascent membrane protein. Nurure (London) 269, 775-780. Rothman, J. E., Katz, F. N., and Lodish, H. F. (1978). Glycosylation of a membrane protein is restricted to the growing polypeptide chain but is not necessary for insertion as a transmembrane protein. Cell 15, 1447-1454. Rupar. C. A , , Rip, J. W., Chaundhary, N . , and Carroll, K. K. (1982). The subcellular localization of enzymes of dolichol metabolism in rat liver. J . B i d . Chem. 257, 3090-3094. Ruta, M., Murray. M. J., Webb, M. C., and Kabat, D. (1979). A niurine leukemia virus mutant with a temperature-sensitive defect in membrane glycoprotein synthesis. Cell 16, 77-88. Sairam, M. R.,and Manjunath, P. (19R3). Hormonal antagonistic properties of chemically deglycosylated human choriogonadotropin. J . B i d . Chem. 258, 445-449. Sasak, W., Quaroni, A,, and Herscovics, A . (1983). Changes in cell-surface fucose-containing glycopeptides and adhesion of cultured intestinal epithelial cells as a function of cell density. Biochem. J . 211, 75-80. Saunier, B., Kilker, R. D., Tkacz, J. S . , Quaroni, A . . and Herscovics, A. (1982). Inhibition of Nlinked complex oligosaccharide formation by I -deoxynojirimycin, an inhibitor of processing glucosidases. J . B i d . Chem. 257, 14155-14161. Schachter, H., and Roseman, S. (1980). Mammalian glycosyltransferases. Their role in the synthesis and function of complex carbohydrates and glycolipids. I n "The Biochemistry of Glycoproteins and Proteoglycans (W. J. Lennarz, ed.), pp. 85-160. Plenum, New York. Schachter, H.. Nardsimhan, S . , Gleeson, P., and Vella, G. (1983). Control of branching during the biosynthesis of asparagine-linked oligosaccharides. Can. J . Biochem. Cell B i d . 61, 10491066. Scher, M. G., and Waechter. C. J. (1981). Lipolytic cleavage of dolichyl oleate catalyzed by calf brain membranes. Biochem. Biophys. Res. Commun. 99, 675-681. Scher, M. G., Jochen, A., and Waechter, C. J. (1977). Biosynthesis of glucosylated derivatives of dolichol: Possible intermediates in the assembly of white matter glycoproteins. Biochemisrry 16, 5037- 5044. Schmitt, J. W., and Elbein, A. D. (1979). Inhibition of protein synthesis also inhibits synthesis of lipid-linked oligosaccharides. J . B i d . Chem. 254, I229 I- 12294. Schubert, D. (1970). Immunoglobulin biosynthesis. IV. Carbohydrate attachment to immunoglobulin subunits. J . Mol. Biol. 51, 287-301. Schultz, J., and Elbein, A. D. (1974). Biosynthesis of mannosyl- and glucosyl-phosphoryl polyprenols in Mycobacrerium smegmaris. Evidence for oligosaccharide-phosphoryl-polyprenols. Arch. Biochem. Biophys. 160, 31 1-322. Schutzbach, J. S.,Springfield, J. D., and Jensen, J. W. (1980). The biosynthesis of oligosaccharidelipids. Formation of an a-l,2-mannosyl-mannose linkage. J . B i d . Chem. 255, 41704175. Schwarz, R. T.,and Datema, R. (1982). The lipid pathway of protein glycosylation and its inhibitors: The biological significance of protein-bound carbohydrates. Adv. Carbohydr. Chem. Biochem. 40, 287-379. Sefton, B. M. (1977). Immediate glycosylation of Sindbis virus membrane proteins. Cell 10, 659668, Sharma, C . , Lehle, L., and Tanner, W. (1981). N-Glycosylation of yeast proteins. Characterization of the solubilized oligosaccharyl transferase. Eur. J . Biochem. 116, 101-108. Sharma, C. B., Lehle, L., and Tanner, W. (1982). Solubilization and characterization of the initial enzymes of the dolichol pathway from yeast. Eur. 1. Biochem. 126, 319-325. Sidman, C. (I98 1). Differing requirements for glycosylation in the secretion of related glycoproteins is determined neither by the producing cell nor by the relative number of oligosaccharide units. J . B i d . Chem. 256, 9374-9376.

246

SHARON S.KRAG

Singh, B. N., and Lucas, J. J. (1981). Increased transfer of oligosaccharide from oligosaccharide pyrophosphoryl dolichol to protein acceptors upon estrogen-induced chick oviduct differentiation. J. Biol. Chem. 256, 12018-12022. Siuta-Mangano, P., Janero, D. R., and Lane, M. D. (1982). Association and assembly of triglyceride and phospholipid with glycosylated and unglycosylated apoproteins of very low density lipoprotein in the intact liver cell. J. Biol. Chem. 257, 11463-11467. Snider, M. D., and Robbins, P. W. (1982). Transmembrane organization of protein glycosylation. Mature oligosaccharide-lipid is located on the lumenal side of microsomes from Chinese hamster ovary cells. J. Biol. Chem. 257, 6796-6801. Snider, M. D., Sultzman, L. A,, and Robbins, P. W. (1980). Transmembrane location of oligosaccharide-lipid synthesis in microsomal vesicles. Cell 21, 385-392. Snider, M. D., Huffaker, T. C., Couto, J. R., and Robbins, P. W. (1982). Genetic and biochemical studies of asparagine-linked oligosaccharide assembly. Philos. Trans. R. Soc. London Ser. B 300, 207-223. Sommers, L. W., and Hirschberg, C. B. (1982). Transport of sugar nucleotides into rat liver golgi. A new golgi marker activity. J. Biol. Chem. 257, 10811-10817. Spencer, J. P., and Elbein, A. D. (1980). Transfer of rnannose from GDP-mannose to lipid-linked oligosaccharide by soluble mannosyl transferase. Proc. Natl. Acad. Sci. U.S.A. 77, 25242527. Spiro, R. G. (1966). Characterization of carbohydrate units of glycoproteins. In “Methods in Enzymology” (E. F. Neufield and V. Ginsburg, eds.), Vol. 8, pp. 26-52. Academic Press, New York. Spiro, M. J., Spiro, R. G., and Bhoyroo, V. D. (1976a). Lipid-saccharide intermediates in glycoprotein biosynthesis. I. Formation of an oligosaccharide-lipid by thyroid slices and evaluation of its role in protein glycosylation. J . Biol. Chem. 251, 6400-6408. Spiro, M. J., Spiro, R. 0 . .and Bhoyroo, V. D. (1976b). Lipid-saccharide intermediates in glycoprotein biosynthesis. 111. Comparison of oligosaccharide-lipids formed by slices from several tissues. J. Biol. Chem. 251, 6420-6425. Spiro, R. G., Spiro, M. J., and Bhoyroo, V. D. (1976~).Lipid-saccharide intermediates in glycoprotein biosynthesis. 11. Studies on the structure of an oligosaccharide-lipid from thyroid. J . Biol. Chem. 251, 6409-6419. Spiro, M. J., Spiro, R. G., and Bhoyroo, V. D. (1979a). Glycosylation of proteins by oligosaccharide-lipids. Studies on a thyroid enzyme involved in oligosaccharide transfer and the role of glucose in this reaction. J. Biol. Chem. 254, 7668-7674. Spiro, M. J., Spiro, R. G., and Bhoyroo, V. D. (1979b). Processing of carbohydrate units of glycoproteins. Characterization of a thyroid glucosidase. J. Biol. Chem. 254, 76597667. Staneloni, R. J., and Leloir, L. F. (1982). The biosynthetic pathway of the asparagine-linked oligosaccharides of glycoproteins. Crit. Rev. Biochem. 12, 289-326. Staneloni, R. J., Ugalda, R. A,, and Leloir, L. F. (1980). Addition of glucose to dolichyl diphosphate oligosaccharide and transfer to protein. Eur. J . Biochem. 105, 275-278. Stanley, P. (1980). Surface carbohydrate alterations of mutant mammalian cells selected for resistance to plant lectins. In “The Biochemistry of Glycoproteins and Proteoglycans” (W. J. Lennarz, ed.), pp. 161-189. Plenum, New York. Stanley, P. (1984). Glycosylation mutants of animal cells. Annu. Rev. Genet. 18, 525-552. Stanley, P., Caillibot, V., and Siminovitch, L. (1975). Selection and characterization of eight phenotypically distinct lines of lectin-resistant Chinese hamster ovary cells. Cell 6, 121- 128. Stark, N. J., and Heath, E. C. (1979). Glucose-dependent glycosylation of secretory glycoprotein in mouse myeloma cells. Arch. Biochem. Biophys. 192, 599-609. Stoll, J., Robbins, A. R., and Krag, S. S. (1982). Mutant of Chinese hamster ovary cells with altered

5. GLYCOSYLATION

OF MEMBRANE PROTEINS

247

mannose 6-phosphate receptor activity is unable to synthesize mannosylphosphoryldolichol. Proc. Narl. Acad. Sci. U . S . A . 79, 2296-2300.

Stoolman, L. M., and Rosen, S. D. (1083). Possible role for cell-surface carbohydrate-binding molecules in lymphocyte recirculation. J . Cell Eiol. 96, 722-729. Struck, D. K., and Lennarz. W. J . (1977). Evidence for the participation of saccharide-lipids in the synthesis of the oligosaccharide chain of ovalbumin. J. B i d . Chern. 252, 1007-1013. Struck, D. K . , Lennarz, W. J . , and Brew, K . (1978). Primary structural requirements for the enzymatic formation of the N-glycosidic bond in glycoproteins, studies with a-lactalbumin. J . Eiol. Chem. 253, 5786-5794. Swiedler, S . J., Hart, G. W., Tarentino, A. L., Plummer, T. H., Jr., and Freed, J . H. (1983). Stable oligosaccharide microheterogeneity at individual glycosylation sites of the H-2Kk membrane glycoprotein. J. Eiol. Chem. 258, I I5 15- I 1523. Tabas, I., Schlesinger, S., and Kornfeld, S . (1978). Processing of high mannose oligosaccharides to form complex type oligosaccharides on the newly synthesized polypeptides of the vesicular stomatitis virus G protein and the IgG heavy chains. J . Eiol. Chem. 253, 716-722. Tai, T . , Yamashita, K., Ogata-Arakdwa, M., Koide. N., Muramatsu, T . , Iwashita, S . , h u e , Y., and Kobata, A. (1975). Structural studies of two ovalbumin glycopeptides in relation to the endo-P-N-acetylglucosaminidasespecificity. J. B i d . Chem. 250, 8569-8575. Takatsuki, A., Arima, K., and Tamura, G . (1971). Tunicamycin, a new antibiotic. I. Isolation and characterization of tunicamycin. J . Anribiot. 24, 2 15-223. Takatsuki, A,, Kohno, K., and Tamura, G. (1975). Inhibition of biosynthesis of polyisoprenol sugars in chick embryo microsomes by tunicamycin. Agric. Biol. Chem. 39, 2089-2091. Tanner, W. (1969). A lipid intermediate in mannan biosynthesis in yeast. Eiochem. Eiophys. Res. Cummun. 35, 144-150. Tarentino, A. L., Plummer, T. H., Jr., and Maley, F. (1974). The release of intact oligosaccharides from specific glycoproteins by endo-P-N-acetylglucosaminidaseH. J . Eiol. Chem. 249, 8 I8824. Tarentino, A. L., Trimble, R. B., and Maley, F. (1978). Endo-P-N-acetylglucosaminidasefrom Strepromyces plicarus. In “Methods in Enzymology” (V. Ginsburg, ed.), Vol. 50, pp. 574580. Academic Press, New York. Tavares. 1. A , . and Hemming, F. W. (1982). N-Glycosylation of proteins by homogenate preparations of developing chick embryos. Eur. J. Eiochem. 122, 591-600. Tetas, M., Chao, M., and Molnar, J . (1970). Incorporation of carbohydrates into endogenous acceptors of liver microsomal fractions. Arch. Eiochem. Biophys. 138, 135- 146. Thotakura. N. R., and Bahl, 0. P. (1982). Role of carbohydrate in human chorionic gonadotropin: Deglycosylation uncouples hormone-receptor complex and adenylate cyclase system. Eiochem. f3iophys. Res. Commun. 108, 339-405. Tkacz, J. S . , and Herscovics, A. (1975). Ozonolytic cleavage of authentic and pancreatic dolichylmannosylpyranosyl phosphates: Determination of sugar configuration in the fragments with (Y and p mannosidases. Eiochem. Eiophys. Res. Commun. 64, 1009-1017. Tkacz, J . S., and Lampen, J. B. (1975). Tunicamycin inhibition of polyisoprenyl N-acetylglucosaminyl pyrophosphate formation in calf-liver microsomes. Eiochem. Eiophys. Res. Comtniin. 65, 248-257. Tkacz, J . S . , Herscovics, A,, Warren, C. D., and leanloz, R . W. (1974). Mannosyltransferase activity in calf pancreas microsomes. Formation from guanosine diphosphate-D-[ 14C]mannose of a 14C-labelledmannolipid with properties of dolichyl mannopyranosyl phosphate. J . Eiol. Chem. 249, 6372-638 I . Trimble, R . B., Byrd, J . C., and Maley, F. (1980). Effect ofglucosylation of lipid intermediates on ohgosaccharide transfer in solubilized microsomes from Saccharomyces cerevisiue. J . Eiol. Chem. 255. 11892-1 1895.

248

SHARON S.KRAG

Trowbridge, I. S . , Hyman, R., Ferson, T., and Mazauskas, C. (1978a). Expression of Thy-I glycoprotein on lectin-resistant lymphoma cell lines. Eur. J. Irnmunol. 8, 7 16-723. Trowbridge, I. S., Hyman, R., and Mazauskas, C. (1978b). The synthesis and properties of T25 glycoprotein in Thy-1 negative mutant lymphoma cells. Cell 14, 21-32. Turco, S. J. (1980). Modification of oligosaccharide-lipid synthesis and protein glycosylation in glucose-deprived cells. Arch. Eiochem. Biophys. 205, 330-339. Turco, S. J., and Heath, E. C. ( 1977). Glucuronosyl-N-acetylglucosaminylpyrophosphoryldolichol. Formation in SV40-transformed human lung fibroblasts and biosynthesis in rat lung microsomal preparations. J. Eiol. Chem. 252, 2918-2928. Turco, S. J., and Pickard. J. L. (1982). Altered G-protein glycosylation in vesicular stomatitis virusinfected glucose-deprived baby hamster kidney cells. J. B i d . Chem. 257, 8674-8679. Turco, S. J., and Robbins, P. W. (1979). The initial stages of processing of protein-bound oligosaccharide in v i m . J. Eiol. Chem. 254, 4560-4567. Turco, S. J., Stetson, B., and Robbins, P. W. (1977). Comparative rates of transfer of lipid-linked oligosaccharides to endogenous glycoprotein acceptors in vifro. Proc. Nafl. Acad. Sci. U.S.A. 74, 4411-4414. Vargas, V. I . , and Carminatti, H. (1977). Glycosylation of endogenous protein(s) of the rough and smooth microsomes by a lipid sugar intermediate. Mol. Cell. Eiochem. 16, 171-176. Vijay, I. K., and Perdew, G. H. (1982). Biosynthesis of mammary glycoproteins. Structural characterization of lipid-linked glucosyloligosaccharides.Eur. J. Biochem. 126, 167- 172. Vijay, 1. K., and Perdew, G. H. (1983). Characterization of lipid-linked octa-, nona-, and decasaccharides formed during in virro synthesis of mammary glycoproteins. Arch. Eiochem. Biophys. 220, 605-614. Vijay, I. K., Perdew, G. H., and Lewis, D. E. (1980). Biosynthesis of mammary glycoproteins. Partial characterization of the sequence for the assembly of lipid-linked saccharides. J. Biol. Chern. 255, 11210-1 1220. Villemez, C. L., and Carlo, P. L. (1980). Properties of a soluble polyprenol phosphate. UDP-D-Nacetyl-glucosamine N-acetylglucosamine- I -phosphate transferase. J. Biol. Chem. 255, 8 1748178. Waechter, C. J., and Harford, J. B. (1977). Evidence for the enzymatic transfer of N-acetylglucosamine from UDP-N-acetylglucosamine into dolichol derivatives and glycoproteins by calf brain membranes. Arch. Biochem. Biophys. 181, 185-198. Waechter, C. J., and Lennarz, W. J. (1976). The role of polyisoprenol-linked sugars in glycoprotein biosynthesis. Annu. Rev. Eiochem. 45, 95-1 12. Waechter, C. J., and Scher, M. G. (1978). Glucosylphosphoryldolichol:Role as a glucosyl donor in the biosynthesis of an oligosaccharide lipid intermediate by calf brain membranes. Arch. Eiochem. Biophys. 188, 385-393. Waechter, C. J., Lucas, J. J., and Lennarz, W. J. (1974). Evidence for xylosyl lipids as intermediates in xylosyl transfers in hen oviduct membranes. Biochem. Biophys. Res. Comrnun. 56,343350. Waechter, C. J., Schmidt, J. W., and Catterall, W. A. (1983). Glycosylation is required for maintenance of functional sodium channels in neuroblastoma cells. J. Biol. Chem. 258, 51 175123. Warren, C. D., and Jeanloz, R. W. (1973). The characterization of glycolipids derived from longchain polyprenols: Chemical synthesis of P-D-mannopyranosyl dolichyl phosphate. FEBS Left. 31, 332-334. Wedgewood, J. F., and Strominger, J. L. (1980). Enzymatic activities in cultured human lymphocytes that dephosphorylate dolichyl pyrophosphate and dolichyl phosphate. J. Biol. Chem. 255, 1120- 1123.

5. GLYCOSYLATION OF MEMBRANE PROTEINS

249

Wellner, R . B., and Lucas. I. J . (1979). Evidence for a compound with the properties of 2.3dehydrodolichyl pyrophosphate. FEBS Lerr. 104, 379-383. Welply, J . K., Shenbaganiurthi. P.. Lennarz. W. J . , and Naider. F. (1983). Substrate recognition by oligosaccharyltrdnsferase: Studies on glycosylation of modified asn-x-thriser tripeptides. J . B i d . Chem. 258, 11856- I 1863. Wilson, J . R . , Williams, D., and Schachter, H. (1976). The control of glycoprotein synthesis: N acetylglucosamine linkage to a mannose residue as a signal for the attachment of 1.-fucose to the asparagine-linked N-acetylglucosamine residue of glycopeptide from a,-acid glycoprotein, Biochem. Biuphys. Res. Commun. 72, 909-916. Wong, T. K . , and Lennarz, W. J . (1982a). The site of biosynthesis and intracellular deposition of dolichol in rat liver. J. B i d . Cheni. 257, 6619-6624. Wong, T. K., and Lennarz,W. J . (1982b). Biosynthesis ofdolichol and cholesterol during embryonic development of the chicken. Biochim. Biuphys. Acru 710, 32-38. Wong, T. K . , Decker, G . L., and Lennarz, W. J. (1982). Localization of dolichol in the lysosomal fraction of rat liver. J . Biol. Chem. 257, 6614-6618. Yamashita, K . , Tachibana, Y., and Kobata, A . (1978). The structures of the galactose-containing sugar chains of ovalbumin. J . B i d . Chem. 253, 3862-3869. Yamashita, K., Karnerling, J . P., and Kobata, A . (1983). Structural studies of the sugar chains of hen ovomucoid. Evidence indicating that they are formed mainly by the alternate biosynthetic pathway of asparagine-linked sugar chains. J . B i d . Chem. 258, 3099-3106. Zatta, P., Zakim, D., and Vessey, D. A. (1975). The transfer of galactose from UDP-galactose to endogenous lipid acceptors in liver microsomes. Biuchim. Biophys. Acfu 392, 361-365. Zatta, P., Zakim, D., and Vessey, D. A. (1976). Incorporation of N-acetylglucosdmine into lipid linked oligosaccharides. Biochem. Biophys. Res. Cummun. 70, 1014- 1019. Zatz, M . , and Barondes, S . H. (1969). Incorporation of mannose into mouse brain lipid. Biuchem. Biophys. Res. Cummun. 36, 51 1-517.

This Page Intentionally Left Blank

CURRENT TOPICS IN MEMBRANES AND TRANSPORT. VOLUME 24

Chapter 6 Protein Sorting in the Secretory Pathway ENRIQUE RODRIGUEZ-BOULAN,* DAVID E . MISEK, DORA VEGA D E SALAS,* PEDRO J . I . SALAS,* AND ENZO BARDt *Department of Cell Biology and Anatomy Cornell University Medical College New York. New York tDepartmenr of Pathology State University of New York Downstare Medical Center Brooklyn, New York

I.

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.....................

11. Molecular Sorting: Definitions and Factors Invol

Site of Synthesis.. . . . . .................................... Sorting Signals or “Zip Codes,” Addressing Signals . . . . . . . . . . . . . . . . . . . . . . Signal Recognition or Decoding Mechanisms.. . . . . . . . . . . . . . . . . . . . . . . . . . . . Cotranslational and Posttranslational Mechanisms for Translocation across Membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Specific Fusion-Fission Interactions between Intracellular Vesicular Compartments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. Restrictions on Lateral Mobility in the Plane of the Bilayer . . . . . . . . . . . . . . . . . C. Lipid Sorting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ill. Molecular Sorting in the Secretory Pathway. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. The Secretory Pathway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B . Sorting in the Endoplasmic Reticulum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Sorting in the Colgi Apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Sorting of Lysosomal Proteins.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Sorting of Plasma Membrane Glycoproteins .......................... F. Molecular Sorting, Endocytosis, and Menib Recycling. . . . . . . . . . . . . . . . . . IV. Model Systems for the Study of Molecular Sorting in Eukaryotic Cells. . . . . . . . . . . . A. Reconstitution Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ....... B . Genetic Systems . . . . .......................................... C. Other Systems.. . . . . . . . . . . . . . . . . . . . . . . .................. D. Recombinant DNA Technology Applied to rotein Sorting. . . . . . . V. Summary and Perspectives. . . . . . . . . . . . . . . . . . .................... .............. References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. B. C. D.

252 252 253 253 256 256 251 251 258 259 259 259 260 262 262 266 268 268 210 212 211 280 280

25 1 Copynghi 0 IYXS by Academic Press, Inc. . . All rights of reproduction in any form reserved.

ISBN 0-12-lS3324-7

252

ENRIQUE RODRIGUEZ-BOULAN ET AL.

1.

INTRODUCTION

From primary and secondary biosynthetic sites in the cytosol and mitochondria1 matrix, respectively, proteins and lipids are distributed to more than 30 final destinations in membranes or membrane-bound spaces, where they carry out their programmed function. Work in recent years has started to elucidate how the cell manages to perform this considerable sorting task. Thirty years after Porter et al. (1945) discovered the endoplasmic reticulum (ER)-the main avenue for the transport of proteins and lipids in eukaryotic cells (Jamieson and Palade, 1971; Palade, 1975)-the signal hypothesis (Blobel and Sabatini, 1971; Blobel and Dobberstein, 1975a,b) provided the first molecular explanation for the sorting of secretory proteins into the ER lumen. The rate of progress in the last 8 years has been exponential. Palade’s suggestion (1975) that the secretory pathway could be utilized to transport molecules destined to reside in the various segments of the pathway itself or in branches of it (such as membrane glycoproteins of the ER, lysosomes, and plasma membrane) has been widely confirmed. Considerable information has been gathered on the intracellular processing of these proteins, particularly on their glycosylation, on the role of signal sequences in protein translocation, and on how hydrolases are imported into lysosomes. The transport of proteins into mitochondria and chloroplasts is now much better understood. In spite of these advances, our knowledge of the precise mechanisms determining the final location of cellular proteins and lipids is still very scant. The initial hope of finding just a few very general mechanisms has been replaced by more eclectic expectations, as new exceptions to “rules’’ are described. The development of new model systems, our recently acquired ability to modify genes and reintroduce and express them in eukaryotic cells, and more powerful and sensitive immunolocalization techniques are expected to enable us to provide answers to some of the questions. In this article we will update the information available on the sorting of molecules in the secretory pathway. Rather than extensively reviewing each subject, we refer the reader to some excellent reviews recently published whenever needed. We have concentrated on a few areas which, we feel, will provide important information in the near future.

II. MOLECULAR SORTING: DEFINITIONS AND FACTORS INVOLVED Molecular sorting is defined in this article, in its most general sense, as the sum of the mechanisms that determine the distribution of a given molecule from its site of synthesis to its site of function in the cell. Because most of the work in this regard has been centered on protein sorting, we deal mostly with this aspect.

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

253

The final site of residence of a protein in a eukaryotic cell is determined by a combination of various factors, acting in concert: site of synthesis; sorting signals or “zip codes;” signal recognition or decoding mechanisms; cotranslational or posttranslational mechanisms for translocation across membranes; specific fusion-fission interactions between intracellular vesicular compartments; and restrictions to the lateral mobility in the plane of the bilayer. Very little is known about how the differences in lipid compositions of the various cell membranes are generated or about what mechanisms are resposible for their characteristic asymmetric distribution in different halves of the bilayer.

A. Site of Synthesis Except for a small percentage of mitochondria1 and chloroplast polypeptides which are coded by local DNA and synthesized within these organelles, all other cell proteins have a major site of synthesis: the cytosolic polysomes, either free in the cytoplasmic matrix or bound to the membrane of the endoplasmic reticulum. Two principal sites of synthesis also exist for the other major component of membranes, the lipids; these sites are the mitochondria (minor site, mostly producer of cardiolipin) and the ER (major producer of phospholipids and cholesterol) (van Golde et al., 1974; Jelsema and Morre, 1978; Op den Kamp, 1979; Bell and Coleman, 1980). Addition of carbohydrates to glycoproteins and glycolipids also occurs in two major stages: core glycosylation in the ER followed by trimming of this core and addition of branch carbohydrates in the ER and Golgi apparatus (Parodi and Leloir, 1979; Lennarz, 1980; Hubbard and Ivatt, 1981).

6. Sorting Signals or “Zip Codes,” Addressing Signals From the two major sites of synthesis, proteins, glycoproteins, and lipids are distributed to more than 30 membranes or membrane-bound organellar spaces, each of them with a very characteristic composition. Progress in recent years, in particular the confirmation of the “signal hypothesis” (Blobel and Sabatini, 1971; Blobel and Dobberstein, 1975a,b), has substantiated the belief that specific structural features in polypeptides, encoded in their mRNA, guide their subcellular distribution. Furthermore, proteins, lipids, and carbohydrates have very defined orientations with respect to the plane of the bilayer, a characteristic which implies that the sorting mechanisms must work in close collaboration with systems to translocate the molecules across the bilayer. At the moment of synthesis, the cell makes an initial decision on the destination of the proteins based on the possession, by some of them, of specific structural features or “primary addressing signals” (Table I). Secretory proteins and transmembrane or luminal glycoproteins destined to go to the various seg-

254

ENRIQUE RODRIGUEZ-BOULAN ET AL.

TABLE I MECHANISMS FOR ADDRESSING PROTEINSTO ORGANELLES

Protein destination Cotranslational insertion Secretory proteins Integral membrane glycoproteins of ER, Golgi, lysosomes, peroxisomes Plasma membrane Integral proteins on cytoplasmic side of ER Luminal proteins of ER, Golgi Lysosornal hydrolases Posttranslational insertion Peripheral and integral proteins on cytoplasmic side of all organelles Mitochondria1 proteins (inner or outer membrane, 2 spaces) Chloroplast proteins (inner, outer, thylakoid membranes, 3 spaces) Luminal peroxisomal proteins

Primary addressing signals

Signal sequences

Secondary addressing signals

Ref. a

Probably none

1-5

Signal sequences ?

?

1-8

Signal sequences Insertion signals

? ?

1-5,9 3,9,14

Signal sequences ?

?

1-5

Signal sequences

Mannose 6-P

10,11

Binding sites for integral or peripheral proteins, insertion signals Extra peptides, precursor polyproteins or internal sequences Extra peptides or internal sequences

Probably none

4,9,12,14

Internal sequences

?

13,14

?

13,14

?

2,4,6,8

a Key to references: (1) Blobel er al. (1979); (2) Blobel (1980); (3) Kreil(1981); (4) Sabatini er al. (1982); ( 5 ) Lusis and Swank (1980);(6) Novikoff (1976); (7) Kreibich et al. (1978a-c); (8) Lazarow (1980); (9) Lodish et al. (1981); (10) Strawser and Touster (1980); (11) Sly (1982); (12) Branton e? al. (1981); (13) Chua and Schmidt (1979); (14) Poyton (1983).

ments of the secretory pathway (ER, Golgi, plasma membrane), or to branches of it (lysosomes), are cotrunslutionally translocated through the membrane of the rough ER. The structural information in these proteins, or primary addressing signal, is the signal sequence, which is a stretch of 15-30 hydrophobic amino acids, is usually located at the amino terminal end, and is cleaved cotranslationally by a signal peptidase. As predicted by Blobel and Sabatini (1971), the first evidence for its existence came from in vitro translation experiments by Milstein et ul. (1972), who showed that immunoglobulin light chains exhibited extra amino acids at the amino-terminal end when synthesized from mRNA in a reticulocyte lysate in the absence, but not in the presence, of microsomal mem-

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

255

branes. Blobel and Dobberstein (1975a,b) obtained the crucial experimental evidence that led to the formulation of the signal hypothesis and the concept of cotranslational segregation. They found ( 1 ) synthesis of a large precursor in a cell-free translation system programmed with Ig light chain mRNA, and (2) segregation of normal-size Ig light chains in the microsomal lumen when membranes were added before the start, but not after the completion, of protein synthesis. Subsequently, the existence of precursors was demonstrated for a wide variety of secretory (Blobel et al., 1979) and membrane-bound (Lodish et al., 1981; Sabatini et al., 1982) polypeptides. Sequence information for many of them has been obtained by protein sequencing or, more frequently in recent years, from sequencing cDNA or cloned genes. A discussion of the structure of the signal sequences and their role in protein translocation can be found in the first four articles of this volume. Much of the direct evidence for this role comes from work in prokaryotic systems (reviewed by Davis and Tai, 1980; Emr et a/., 1980; Kreil, 1981; see also articles by Duffaud et al. and Bankaitis el al., this volume). Cotranslationally operating hydrophobic sequences which are permanent features of the protein participate in the insertion of cytochrome P-450 and in the translocation across the ER membrane of influenza virus neuraminidase (BarNun et al., 1980; Blok er d . , 1982). Transient amino-terminal sequences are apparently also responsible for the transfer of several proteins into mitochondria and chloroplasts (Chua and Schmidt, 1979; Poyton, 1983; and article by Reid, this volume). The function of this second type of primary addressing signals is different, however, from that of the signal sequences, since these proteins are synthesized by free polysomes and translocated posttranslationally. Other primary addressing signals operating via posttranslational translocation mechanisms are involved in the distribution of peroxisomal and nuclear matrix proteins, which are also synthesized in free cytoplasmic polysomes (Table I). Free polysomes also participate in the synthesis of proteins associated with the cytoplasmic domain of several organelles, such as cytochrome b, and NADHcytochrome b, reductase (Borgese and Gaetani, 1980; Rachubinski et al., 1980). Proteins that remain in the cytosol after synthesis by free polysomes presumably lack organellar addressing signals. In addition to the primary addressing signals mentioned above, other features in the polypeptide chain are thought to be responsible for the particular distribution of proteins within organelles or chains of interconnected organelles (such as those in the secretory pathway). These secondary “sorting signals,” or “addressing signals,” may also be added co- or posttranslationally to the protein but, of course, the type or the degree of the modification is ultimately determined by structural information in the peptide chain. For example, only proteins which expose in the lumen of the ER Asn-X-Ser(Thr) groups will be core glycosylated;

256

ENRIQUE RODRIGUEZ-BOULAN ET AL.

additional information is required to determine whether that core will be processed to simple or complex residues or to terminal mannose 6-phosphate groups that will target the protein to the lysosomes (Hubbard and Ivatt, 1981; Sly, 1982; Pollack and Atkinson, 1983). The mannose 6-phosphate group of lysosomal proteins is the only well-characterized “secondary addressing signal. ” Like secretory proteins, lysosomal hydrolases are synthesized as larger precursors, inserted into the ER lumen via transient signal sequences, and glycosylated by transfer of a mannose-rich core oligosaccharide (see Sly, 1982). This core is processed differently, however, in lysosomal proteins: the action of two enzymes in the Golgi apparatus results in the exposure of several mannose 6-phosphate residues, responsible for the targeting to their final destination (see Section IILD).

C. Signal Recognition or Decoding Mechanisms The sorting signals must be read and interpreted by specific decoding mechanisms. Advances have been made in recent years in identifying the decoding mechanism for the signal sequence (see Section IV,A,l). A receptor for the mannose 6-P groups of lysosomal proteins-located on the cell surface and, most crucially, in the Golgi apparatus-appears to play a role in the transport to lysosomes (Sly, 1982; see Section 111,D). No information is available yet on the decoding systems for the signals of other organellar proteins.

D. Cotranslational and Posttranslational Mechanisms for Translocation across Membranes Proteins which have hydrophilic segments exposed on the noncytoplasmic or “ecto” domain of membranes require special translocation systems. It is now clear that proteins are translocated across membranes using either co- or posttranslational mechanisms but very little is known about the events taking place at the molecular level. Cotranslational translocation is the preferred procedure for exported proteins and transmembrane glycoproteins; it appears to be a very useful method to prevent the accumulation of potentially dangerous secretory toxins in the cytosol. As mentioned, posttranslational translocation has been shown to occur for mitochondria1 and chloroplast proteins and peroxisomal en1977; Goldman and Blobel, 1978; Chua and Schmidt, zymes (Dobberstein et d., 1979; Poyton, 1983). An interesting example of posttranslational translocation is the entrance of bacterial and plant toxins into cells (Pappenheimer, 1978). These toxins (such as diphtheria and tetanus toxin and ricin) possess a domain that recognizes specific

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

257

receptors in the plasma membrane and mediates translocation across the bilayer and a second domain responsible for the toxicity. Membrane penetration is triggered by a slightly acidic pH; it has been shown that these toxins enter the cell via receptor-mediated endocytosis and reach the cytoplasm through the membrane of prelysosomal vesicles (or “endosomes”) with acid pH contents (Donovan et a / . , 1981; Boquet and Duflot, 1982). Similar mechanisms appear to be responsible for the fusogenic properties of various viral envelope glycoproteins (see White et a / . , 1983) and the penetration of enveloped RNA viruses into cells (Marsh et al., 1983).

E. Specific Fusion-Fission interactions between lntracellular Vesicular Compartments Work during recent years has clearly demonstrated that intracellular compartments communicate with each other through a vast and apparently complex traffic of vesicles. The best-studied process is the endocytosis of macromolecules, a property which, like secretion, appears to be constitutive to all cells (see Steinman el al., 1983, for a review). It has been shown that some receptors are interiorized after binding to their respective ligands, dissociated under the effect of the low pH of the “endosomal” compartments, and recycled back to the cell surface. This mechanism implies that a given protein may be found, albeit at different concentrations, in various cellular compartments. The same concept applies to the segregation, during biogenesis, of components destined to the different segments of the secretory pathway (rough and smooth ER, Golgi, plasma membrane). Specific fusion-fission interactions between the different cellular compartments are responsible for conferring selectivity to this movement of material, but no information is available on the molecular details underlying these interactions. Evidence from work with fusogenic viral glycoproteins indicates that restricted amino acid sequences may be responsible for the fusing capacity and that single amino acid substitutions may block it (White et a / . , 1983). Further work is needed to determine the generality of this mechanism in the fusion between cellular vesicles.

F. Restrictions on Lateral Mobility in the Plane of the Bilayer Interactions with peripheral or transmenibrane proteins which are specific to a particular membrane and which do not participate in recycling may be an important mechanism to stabilize the composition of a given membrane. All proteins belonging to such a membrane would share “retention signals” (specific se-

258

ENRIQUE RODRIGUEZ-BOULAN ET AL.

quences or conformational features) that recognize the primary organellar structural protein which, in order for the system to work, must possess the ability to interact with other similar molecules. Organelles connected by flow of membrane material (such as the different segments of the secretory pathway) would be, according to this view, composed of one or more primary structural proteins (probably conferring the shape of the organelle), associated secondary proteins (carrying out the specific functions of the organelle), and proteins in transit to other organelles. A beautiful example of the interactions described in this paragraph is provided by the work of Branton and colleagues, describing the relationship of peripheral proteins in red blood cells, such as ankyryn and spectrin, with integral membrane proteins and the cytoskeleton (Branton et al., 1981). Another possible example is the ribophorins I and 11, which have been postulated to carry the ribosome binding sites and perhaps to be responsible for the typical shape of the rough ER (Kreibich et al., 1978a-c). Restrictions to the lateral mobility in the plane of the bilayer may also be caused by interaction with extracellular molecules, such as collagen, fibronectin, or laminin, or by the presence of uninterrupted discontinuities in the bilayer, such as the occluding junctions that separate the apical and basolateral regions of the epithelial cell plasmalemma (see Section IV,C,2).

G. Lipid Sorting Most of the terminal steps in phospholipid and cholesterol synthesis are carried out by enzymes in the ER (van Golde et al., 1974; Jelsema and Morre, 1978; Op de Kamp, 1979; Bell and Coleman, 1980). However, cholesterol is almost absent from ER membranes (cholestero1:phospholipid ratio, 1:lo), has intermediate concentrations in the Golgi apparatus, and is a major component of plasma membranes (cholestero1:phospholipidratio, 1:1). The factors that govern these differential lipid distributions in specific subcellular locations are completely unknown. Equally unknown are the mechanisms responsible for the asymmetric localization of phospholipids in the plasma membrane: aminophosphatides in the inner half, choline phosphatides and glycolipids in the outer half (Bretscher, 1973; Verkleij et al., 1973; Renooij et al., 1974; Rothman and Lenard, 1977). Other important questions are as follows: Do all or only some of the cell membranes have the necessary machinery to generate lipid asymmetry? Are the intracellular pathways for lipids the same or do they differ to a certain extent from those followed by the proteins? What is the biological role of lipid asymmetry? The observation that fluorescent lipid analogs adopt specific surface and intracellular distributions, depending on the lipid, and are transported and metabolized within the cell (Pagan0 and Longmuir, 1983) opens the possibility of providing answers to some of these questions.

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

111.

259

MOLECULAR SORTING IN THE SECRETORY PATHWAY

A. The Secretory Pathway It is now well documented that the secretory pathway not only functions as a protein export system but is also a main molecular distribution system in the cell. Originally described in glandular epithelial cells, it is found, although modified, in every eukaryotic cell type (Palade, 19751, including yeast. Secretory proteins undergo the following biosynthetic and processing steps (Jamieson and Palade, 1971, 1977; Palade, 1975): (1) synthesis of a precursor (or preprotein) by polysomes attached to the outer surface of the rough ER; (2) segregation into the ER cistemae, mediated by a signal sequence; (3) intracellular transport via transitional smooth ER vesicles to the Golgi complex; (4) concentration and storage in secretory granules in cells with “regulated” secretion; this step is absent in cells with continuous or “nonregulated” secretion; and (5) exocytosis.

B. Sorting in the Endoplasmic Reticulum There is now a large body of evidence indicating that the secretory pathway participates in the synthesis, processing, and distribution of lysosomal hydrolases and plasma membrane glycoproteins. Although it is believed that the same is true for integral membrane proteins of ER, Golgi, lysosomes, and peroxisomes, no experimental proof is available because of the unavailability of purified markers. The ER appears to possess a population of intrinsic transmembrane glycoproteins (Rodriguez-Boulan et al., 1978a,b), with their carbohydrate residues exposed in the luminal surface (Hirano et al., 1972). Lectins specific for mannose or glucose residues (such as concanavalin A), but not lectins with affinity for branch carbohydrates (such as ricin or wheat germ agglutinin) bind to the luminal side of ER membranes (Hirano et al., 1972; Rodriguez-Boulan et a/., 1978a). Only two integral ER glycoproteins have been characterized in some detail: ribophorins I and 11, believed to be the ribosome attachment sites (Kreibich et al., 1978a-c). The membrane glycoproteins of viruses that bud from the RER, such as the rotaviruses (see Section IV,C,3), or of mutants of surface-budding viruses with defects in the exit of the glycoproteins from the ER (Lafay, 1974; Knipe et al., 1977a; Lohmeyer and Klenk, 1979; Rodriguez-Boulan el al., 1984), should provide a useful system to study the biogenesis of ER integral transmembrane proteins. From a biogenetic point of view, it is important to elucidate what factors determine that some proteins remain in the ER after synthesis, while others move ahead to the Golgi, lysosomes, or plasma membrane. Is it the possession of a

260

ENRIQUE RODRIGUEZ-BOULANET AL.

“retention signal” (see Section II,F) in ER proteins designed to interact with other ER-specific proteins or is it the lack of “addressing signals” designed to interact with specific mechanisms that mediate the transport away from the ER? Different secretory or membrane proteins are transported from ER to the Golgi at very different rates, a finding which has been interpreted as meaning that one or more specific receptors, presumably carbohydrates, are involved in this process (Strous and Lodish, 1980; Lodish et al., 1983; Lodish and Kong, 1984; Fitting and Kabat, 1982; Ledford and Davis, 1983). An alternative explanation of these experiments would be that secretory proteins differ in the strength of their interactions with ER components, a characteristic which results in different retention times in this organelle. It is known that some ER proteins can be detected, albeit at reduced concentrations, in the Golgi apparatus (Ehrenreich et al., 1973; It0 and Palade, 1978), possibly in the proximal or cis cisternae. It is possible that multiple mechanisms coexist and determine transport between ER and Golgi: Retention signals in ER proteins may prevent their diffusion away from the ER; those that escape and are carried to the Golgi together with proteins destined to other organelles are recycled back to the ER because of the lack of “addressing signals’’ that determine the compartmentalization into specific post-Golgi vesicles. A discussion of some recent molecular data on the movement of proteins between the ER and Golgi, provided by recombinant DNA work, can be found in Section IV,E.

C. Sorting in the Golgi Apparatus The role of the Golgi apparatus as a major sorting center in the cell is well recognized (Farquhar and Palade, 1981; Rothman, 1981; Farquhar, 1983). The Golgi apparatus is ideally placed at the crossroads of the exocytic and endocytic pathways to carry out this function. The structure and function of this organelle have been extensively reviewed (Whaley and Dauwalder, 1979; Tartakoff, 1980; Farquhar and Palade, 1981). It consists of a stack of 4-10 smooth-surfaced cisternae, usually with dilated rims, associated vesicles (some of them coated), and vacuoles. Morphologically, the Golgi complex sometimes displays a clear polarity, with the proximal or cis cisternae facing the endoplasmic reticulum and the distal or trans cisternae facing either the nucleus or the plasmalemma (because of a general spiral or coiled organization). Membrane thickness increases progressively from proximal to distal cisternae. Biochemical polarity is evident in the differential enzyme distribution, detected by cytochemical and immunocytological procedures [thiamine pyrophosphatase (TPPase), nucleoside diphosphatase (NDPase), and galactosyltransferase are markers of the distal cisternae (Novikoff et al., 1971; Novikoff and Novikoff, 1977; Roth and Berger, 1982)], and in the asymmetric distribution of lectin-binding sites [Ricinus

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

261

communis agglutinin, with affinity for galactose residues, binds preferentially to the distal cisternae in some cell types (Griffiths et al., 1982; Tartakoff and Vassali, 1983)l. Major enzymatic activities associated with the Golgi stack are glycosidases involved in the trimming of carbohydrate residues from core oligosaccharides, transferases for addition of N-acetylglucosamine, galactose, fucose, and sialic acid to glycoproteins (Parodi and Leloir, 1979; Schachter and Roseman, 1980; Lennarz, 1980), transferases for the incorporation of galactose and sialic acid to glycolipids (cerebrosides and gangliosides), sulfotransferases, and proteases involved in the processing of proproteins (Young, 1973; Steiner et af., 1970; reviewed in Farquhar and Palade, 1981). The two enzymes responsible for the addition of mannose 6-phosphate groups to lysosomal hydrolases, an N-acetylglucosaminyltransferase and a phosphodiesterase, also appear to be localized in the Golgi complex (Varki and Kornfeld, 1980; Reitman and Kornfeld, 1981). More recently, four initial Golgi enzymes of the glycosylation pathway (mannosidase I, N-acetylglucosaminetransferase I, mannosidase 11, and N-acetylglucosaminetransferase 11) were resolved from two late enzymes (galactosyltransferase and sialyltransferase) by sucrose-gradient centrifugation (Dunphy and Rothman, 1983). These typical enzymatic activities of the Golgi apparatus have been very useful in assessing the passage of various classes of proteins through the organelle. As a consequence of the addition of terminal carbohydrates, glycoproteins become resistant to the action of endoglycosidase H; this provides a simple test for the passage of some glycoproteins through the Golgi complex (Robbins et af., 1977). It is now clear that secretory proteins, plasma membrane glycoproteins, and lysosomal hydrolases undergo biochemical modifications typical of passage through the Golgi apparatus (Farquhar and Palade, 1981; Rothman, 1981; Strawser and Touster, 1980; Sly, 1982). Coated vesicles have been postulated to participate in the post-Golgi transport of plasma membrane proteins (Rothman and Fine, 1980; Rothman et uf., 1980) and lysosomal hydrolases (Friend and Farquhar, 1967). A special cellular structure associated with the ER and the Golgi, or GERL, has been proposed on the basis of cytochemical evidence as an organelle involved in the biogenesis of lysosomal hydrolases (Novikoff, 1976). Although no structural or biogenetic information is available on intrinsic Golgi proteins, it may be assumed that transmembrane glycoproteins in the organelle are inserted into the ER membrane via signal sequences and then transported to the Golgi apparatus, as shown for plasma membrane glycoproteins (see Section III,D and E). At least three Golgi proteins, including the classic marker galactosyltransferase, have already been purified (Louvard et af., 1982; Roth and Berger, 1982) and antibodies against them have been raised and used for immunolocalization studies. It has been shown, using antibodies against galac-

262

ENRIQUE RODRIGUEZ-BOULAN ET AL.

tosyltransferase, that this typical Golgi enzyme is synthesized in the ER and transported to the Golgi apparatus in approximately 20 minutes, where it stays for an average of about 20 hours (Strous et al., 1983; Strous and Berger, 1982).

D. Sorting of Lysosomal Proteins Lysosomal hydrolases are synthesized as larger precursors and inserted via signal sequences into the ER lumen (Erickson and Blobel, 1979; Hasilik and Neufeld, 1980a; Rosenfeld et al., 1982). Like secretory proteins, they are cotranslationally glycosylated by transfer of a GlcNAc,-Man,-Glc, oligosaccharide from dolichol phosphate. The glucose residues are trimmed by specific glycosidases in the ER, and N-acetylglucosamine 1-phosphate is added to the 6 position of several mannose residues of the high-mannose oligosaccharides (Tabas and Kornfeld, 1980; Hasilik and Neufeld, 1980b). Subsequently, a phosphodiesterase removes, probably in the Golgi, the N-acetylglucosamineresidues and exposes the phosphate groups on the mannose residues (Varki and Kornfeld, 1980; Hasilik et al., 1981). These mannose 6-phosphate groups are recognized by specific receptors, presumably in the Golgi apparatus or GERL region, that mediate the transfer to the lysosomes (see Sly, 1982, for a recent review). Recently, Brown and Farquhar (1984) have localized these receptors with monospecific antibodies and immunoelectron microscopy to cisternae and coated vesicles in the cis Golgi apparatus. Mannose 6-P groups also mediate an alternative pathway of the hydrolases to the lysosomes: the uptake from the medium via receptor-mediated endocytosis. The relationship between the receptors that mediate this secretion-recapture pathway (Neufeld et al., 1977) and the intracellular receptors for mannose 6-P is not yet totally understood.

E. Sorting of Plasma Membrane Glycoproteins The last decade has seen an explosive increase in information on the structure and biogenesis of plasma membrane proteins. The availability of DNA cloning and sequencing techniques is accelerating this process even further. Table I1 summarizes structural and biogenetic data for some of the best-studied plasma membrane glycoproteins. It is now well established that, like secretory proteins, plasma membrane glycoproteins are (1) produced by polysomes bound to the ER and inserted cotranslationally into its membrane via signal sequences (Rothman and Lodish, 1977; Garoff et al., 1978; Lingappa et al., 1978; Dobberstein et al., 1979; Krangel et al., 1979); (2) cotranslationally glycosylated by transfer in block from dolichol phosphate of a core oligosaccharide, GlcNAc,-Man,-Glc,, which is later trimmed and further processed by ER and Golgi enzymes to the simple or

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

263

complex residues found in mature glycoproteins (Hunt et a l . , 1978; Leblond and Bennet, 1977; Lennarz, 1980; Parodi and Leloir, 1979; Robbins et al., 1977; Schachter and P.oseman, 1980; Tabas et al., 1978; Hubbard and Ivatt, 1981); (3) subject to additional processing, such as proteolytic cleavage, sulfation, and covalent linking of lipids (Lazarowitz and Choppin, 1975; Nakamura and Compans, 1977; Schmidt and Schlessinger, 1979). A major difference from secretory proteins is the possession of additional hydrophobic sequences responsible for the anchoring to the lipid bilayer (Garoff and Soderlund, 1978; Kehry er al., 1980; Rose er al., 1980; Gething and Sambrook, 1982; Sveda et al., 1982). Because of the large amounts produced by the infected cells and other experimental advantages (see Section IV,C, l ) , the membrane glycoproteins of three enveloped viruses-vesicular stomatitis (VSV) G protein, Semliki Forest virus El, E2, and E3 envelope proteins, and the influenza hemagglutinin (HA)-are the best-Characterized members of the group (Lenard, 1978; Simons and Garoff, 1980). Morphological details of the intracellular pathway are available, in fact, only for them (Bergmann et al., 1981; J . Green et al., 1981; Wehland et ul., 1982; Bergeron et al., 1982; Rindler et al., 1982; Rodriguez-Boulan et al., 1984). These studies have confirmed previous biochemical evidence: the glycoproteins are detected initially in the ER, 10 minutes later in the Golgi apparatus, and 30-40 minutes later in the cell surface. Although these major steps are clear, some confusion remains regarding the nature of the intermediate carrier vesicles. Evidence obtained by cell fractionation experiments suggests the participation of coated vesicles in the transport of G protein between ER and Golgi, and between Golgi and plasma membrane (Rothman and Fine, 1980; Rothman et al., 1980). A morphological demonstration, however, was not available. In fact, in immunoelectron microscopy experiments using a temperaturesensitive mutant of influenza, with a reversible block in the HA exit from the ER that was used to synchronize its migration, the vesicles that became labeled at intermediate times between Golgi and the cell surface were smooth and larger than coated vesicles (Rodriguez-Boulan et al., 1984). In spite of the considerable amount of information on the structure and processing of plasma membrane glycoproteins, very little is known about how their sorting is carried out. Different glycoproteins exhibit different orientations in the plasma membrane; the most frequent is amino-terminal end in the external domain, carboxy-terminal end in the cytoplasmic domain (Table 11). Orientation, though, is not even related to finer degrees of sorting since, in epithelial cells, two proteins with the same orientation (HA and VSV G protein) are distributed to different surfaces (apical and basolateral, respectively), whereas proteins with opposite orientation (HA and neuraminidase of influenza virus, and sucraseisomaltase) go to the same surface (apical) (Rodriguez-Boulan and Sabatini, 1978; Rodriguez-Boulan, 1983; Srinivas et al., 1983; see Table 11). In contrast to their role in lysosomal proteins, carbohydrates do not seem to be part of the

264

ENRIQUE RODRIGUEZ-BOULAN ET AL.

TABLE 11. BIOCENETIC DATA Orientation

Glycoprotein

Molecular weight (kDa)

N

N-ter

Viral glycoproteins vsv G 61 3 Ext Influenza HA 74 3 Ext NA -50 4 Cyt SVF El 51 3 Ext 62 3 Ext P62 Histocompatibility antigens 44 2 Ext Mouse H-2 Human HLA 41 2 Ext (heavy chain) Immunoglobulins Membrane IgM, 70 2 Ext IgG (heavy chain) Erythrocyte glycoproteins Glycophorin 30 ? Ext Band 3 93 2 Cyt Plasma membrane proteins of polarized cells 150 2 Cyt Sucrase-isomal tase Aminopeptidase 130 2 Cyt Receptors 82 '? Ext IgA 5 Ext Acetylcholine 250

C-ter

Signal sequence

Localization

Cleav.

Length (amino acids)

N-ter

Yes

16

N-ter N-ter

Yes No

16-18 -30

N-ter N-ter

Yes No

14 12

N-ter N-ter

Yes Yes

15-20 20-24

N-ter

Yes

18

N-ter Internal

Yes No

?

N-ter N-ter

No No?

-20 33-42

N-ter N-ter

Yes Yes

18 17-24

?

Key to references: (1) Mudd (1974); Lingappa er al. (1978); Robbins er al. (1977); Tabas er al. (1978); Hunt etal. (1978); Rose et al. (1980); Lodish et al. (1981); (2) Ward and Dopheide (1979); Waterfield et al. (1979); Porter er al. (1979); Hiti er al. (1981); Gething et al. (1980); Sleigh er al. (1980); Min Jou et al. (1980); Davies et al. (1980); Wilson et al. (1981); Winter er al. (1981); (3) Colman er al. (1983); Fields et al. (1981); Blok eta[. (1982); Varghese et al. (1983); (4) Garoff and Soderlund (1978); Garoff et al. (1980); Simons and Garoff (1980); (5) Dobberstein er al. (1979);

sorting signals of plasma membrane glycoproteins, since they reach the cell surface-even the correct surface domain (apical or basolateral) in epithelial cells-in the presence of tunicamycin, an inhibitor of ASN-linked glycosylation (Gibson et al., 1978; Roth et al., 1979; Strous and Lodish, 1980; R. F. Green et al., 1981). Ionophores, such as monensin, have been shown to inhibit the transport to the cell surface (Tartakoff and Vasalli, 1977; Johnson and Schlesinger,

265

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

PLASMA MEMBRANE GLYCOPROTHNS

ON

Anchor pieces

Localization

n

Carbohydrates

Length (amino acids)

Percentage Type

Proteolytic fragments

Migration half-time (niin)

Surface localization in epithelia

Ref."

I

C-ter

20

10

S,C

No

50

Basal

I

1

C-ter N-ter

24-30 -30

19 20

S,C S,C

HAI, H A 2 No

60

I

> 2 hours

Apical Apical

2 3

1 1

C-ter C-ter

27 32

7 9

S,C S,C

No E2, E3

50 50

Basal Basal

4 4

I I

C-ter C-ter

24 10

8

C C

No No

60 60

Basal Basal?

5 6

I

C-ter

26

15

S.C

No

'?

7

1

C-ter -

12

60

13

10

0,C C

No No

7

3

7

8 9

I I

N-ter N-ter

20 33-42

S.C S,C

S.1

45

No

-

Apical Apical

10 II

C-ter

23 18-30

-

15.69 kd

-

NO

30-60 -3 hours

b-Apical ?

12 13

1 -4

-

-

4

15

13-35 -

3

40

Pease el al. (1982); Ploegh er al. (1981); Evans er a / . (1982); (6) Parham el a / . (1977); Ploegh el a / . (1979, 1981); Krangel er nl. (1979); Owen e/ a / . (1981); (7) Rogers er al. (1980); (8) Tomite and Marchesi (1975); Furthmayr (1977); Jokinen et a / . (1979); (9) Steck (1978); Braell and Lodish (1982); (10) Semenza (1976); Brunner PI a / . (1979); Frank era/. (1978); Hauri er al. (1979. 1982); (11) Lojda (1972); Kenny and Maroux (1982); (12) Kraehenbuhl and Kuhn (1978): Solari and Kraehenbuhl (1984); Mostov e t a / . (1984); (13) Noda era/. (1982, 1983a,b); Claudio e r a / . (1983).

1980; Kaariainen ef uf., 1980; Tartakoff, 1983), presumably by interfering with translocation within the Golgi (Griffiths et al., 1983). Low temperature (20°C) blocks a later transport step, presumably at the level of post-Golgi vesicles ( M a t h and Simons, 1983). Mutants of enveloped viruses in which membrane glycoproteins fail to reach the cell surface do not fit into a consistent pattern so far. Mutations of a vaccinia virus-coded surface glycoprotein which result in

266

ENRIQUE RODRIGUEZ-BOULAN ET AL.

extra amino acids in the cytoplasmic domain are retained in the ER (Shida and Matsumoto, 1983) but deletion of the cytoplasmic domain of G protein (Rose and Bergman, 1982), HA (Gething and Sambrook, 1981; Sveda et al., 1981), or Semliki Forest virus glycoproteins (Garoff et af., 1983) by recombinant DNA technology does not affect their expression via the appropriate vectors (see Section IV,D). On the other hand, mutant cell lines which fail to produce p2microglobulin do not express HLA on the cell surface (Ploegh et al., 1979). Since P,-microglobulin interacts with the ectoplasmic domain of HLA in the ER, this observation suggests a role for this domain in the sorting of HLA. Obviously, interpretation of any data as discussed above must take into account possible changes in conformation in the protein tertiary or even quaternary structure (since all integral membrane proteins seem to be in the state of, at least, dimers; see Table 11). Recent data on recombinant DNA work related to the sorting of viral envelope glycoproteins are discussed in Section IV,D.

F. Molecular Sorting, Endocytosis, and Membrane Recycling Whereas some proteins (such as secretory proteins) undergo only one round of sorting during their biogenesis, it is clear that other proteins are subject to many rounds of recycling through various cellular compartments (and therefore, to many sorting events) during their lifetime. The best-known example are some plasma membrane glycoproteins, in particular various receptors to hormones, growth factors, and viruses, which are removed from the cell surface to be quickly incorporated into intracellular vesicles, called ‘‘endosomes” or “receptosomes” (Helenius et al., 1983; Pastan and Willingham, 1983), and recycled back to the plasmalemma. The physiological role of recycling through intracellular compartments, in addition to allowing the endocytosis of various types of molecules or particles, appears to be in many cases the dissociation of ligand from receptor (e.g., low-density lipoprotein, epidermal growth factor, asialoglycoprotein, and a,-macroglobulin receptors) or of an essential factor from its carrier protein (such as iron from transferrin), which allows the reutilization of the receptor or carrier protein for subsequent rounds of endocytosis or transport (Anderson and Kaplan, 1983). Dissociation is triggered by the existence of a low pH (5-6) in the endosomal compartment (Tycko and Maxfield, 1982), possibly generated by a special proton pump present in endosomes, coated vesicles, lysosomes, and even in the Golgi apparatus (Schneider, 1981; Forgac et al., 1983; Marsh et al., 1983; Glickman et al., 1983). In addition to this “receptor-mediated endocytosis” (Goldstein et al., 1979), cells are able to interiorize large amounts of fluid and dissolved solutes by fluidphase pinocytosis. The magnitude of the movement of membrane material during

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

267

these endocytic processes is very large: fibroblasts interiorize 0.9% and macrophages 3.1% of their cell surface every minute (Steinman e? al., 1976). Most of the membrane returns to the cell surface, while the endocytosed fluid and solutes are delivered to the lysosomes, indicating the existence of sorting mechanisms operating at this level. Very little information is available, however, on how specific the movement of membrane material is (i.e., whether only some plasma membrane proteins are subject to recycling) and on what the forces responsible for this process are. Bretscher (1981) has shown that certain major cell surface proteins, such as theta and H63 antigens, appear not to be endocytosed and recycled. Most of the information available on the routes of membrane internalization come from studies on receptor-mediated endocytosis. Upon ligand binding, receptors cluster in coated pits (or may be preclustered there) (Goldstein ef al., 1979; Anderson and Kaplan, 1983; see article by Wiley, this volume), are internalized in coated vesicles, and appear within seconds in uncoated vesicles (endocytic vesicles, “receptosomes,” or “endosomes”) (Wall et a f . , 1980; Helenius et al., 1983; Pastan and Willingham, 1983; Steinman et al., 1983). These vesicles display saltatory movement in the cytoplasm and fuse in 2-20 minutes with lysosomes (Pastan and Willingham, 1983) in a process inhibited by incubation at temperatures below 20°C (Dunn et al., 1980). The routes of membrane recycling vary according to the cell type. While many ligands and fluidphase proteins are delivered to the lysosomes, passage through the Golgi apparatus has been described (Herzog and Farquhar, 1977; Herzog and Miller, 1979; Farguhar, 1982; Pastan and Willingham, 1983). In epithelial and endothelial cells, mechanisms exist to transfer proteins from one to the other side of the monolayer, bypassing the lysosomes. Examples of this process are the transfer of maternal IgG via Fc receptors from the luminal to the basal medium in newborn rat intestine and the transport of IgA via a membrane receptor that is cleaved and released as a complex with the immunoglobulin (secretory component) in various glandular epithelia (Rodewald, 1973, 1980; Kraehenbuhl and Kuhn, 1978; Mostov et al., 1984). From the viewpoint of this article, it is important to stress that many sorting steps take place during this movement of membrane material: sorting of plasma membrane proteins that enter into coated vesicles and endosomes from those that stay in the cell surface, sorting of receptor from ligand in the endosomes, sorting of endosomal membrane proteins from receptors that return to the cell surface, sorting of endosomal fluid phase proteins to the lysosomes, sorting of proteins transported across the monolayer from those delivered to the lysosomes in epithelial cells. Very little is known about the nature of the signals that mediate these sorting events. The best-known sorting signal for endocytosed ligands is also one of the best-characterized signals of exocytic processes: the mannose 6-P group of lysosomal hydrolases (Sly, 1982). No common feature of recycling

268

ENRIQUE RODRIGUEZ-BOULAN ET AL.

receptors has been found so far; carbohydrates do not appear to be involved (Brown et al., 1983). Regarding the nature of the structures that recognize the sorting information in the proteins, a large amount of circumstantial evidence has established an important role for clathrin-coated vesicles (Anderson and Kaplan, 1983; Brown et al., 1983). Geuze et al. (1983) have provided immunoelectron microscopic evidence indicating that, after dissociation from the ligand in the endosomes, receptors are segregated into a population of thin tubular vesicles designated by the acronym CURL (compartment of uncoupling of receptor and ligand). This mechanism provides a very simple possible explanation of how the bulk of the endosomal volume containing dissociated ligand destined to the lysosomes is segregated from the receptors, which return in small-volume vesicles to the cell surface. Another important unsolved question is the extent to which the exocytic and endocytic pathways overlap. Do plasma membrane proteins being delivered from Golgi to the cell surface for the first time travel through the endosomal compartment? Recent data from our laboratory (Salas, unpublished results) indicate that anti-influenza hemagglutinin (HA) antibody incorporated into endosomal vesicles of MDCK cells fails to react with newly synthesized HA migrating to the cell surface, suggesting that the two pathways do not overlap.

IV. MODEL SYSTEMS FOR THE STUDY OF MOLECULAR SORTING IN EUKARYOTIC CELLS A. Reconstitution Systems A main advantage of reconstitution systems is that they permit direct studies of particular sorting steps. Improvements in cell fractionation, protein separation, and immunoprecipitation procedures in the last decade have made them possible. They are currently used to study the translocation of proteins into the ER or into the nonsecretory organelles (mitochondria, chloroplasts, nucleus) and the movement of proteins between Golgi subcompartments. Theoretically, they could be applied to the study of other sorting steps in the secretory pathway; the main limitation is imposed by the availability of pure subcellular fractions and recognizable biochemical modifications of the molecule under study in the target organelle.

SYSTEMS 1. In Vitro TRANSLATION-TRANSLOCATION The development of an in vitro translation system that translocated immunoglobulin light chains across dog pancreas microsomal membranes (Blobel and Dobberstein, 1975a,b) constituted a major step forward in the elucidation of the

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

269

mechanisms responsible for protein translocation across the ER membrane. Further refined to allow synchronization of protein synthesis (Rothman and Lodish, 1977), it provided the first direct proof that the amino-terminal hydrophobic sequences of secretory proteins and integral membrane glycoproteins were somehow responsible for the translocation process. Other important findings with this system were the demonstration that core glycosylation and signal sequence cleavage were cotranslational (Rothman and Lodish, 1977), that some translocatable proteins that lacked cleavable signal sequences had internal segments responsible for vectorial discharge (Lingappa et ul., 1978; Braell and Lodish, 1982), and that lysosomal hydrolases were synthesized as precursors and segregated into the ER lumen via signal sequences (Erickson and Blobel, 1979). In recent years, a series of elegant experiments have demonstrated the existence of a ribonucleoprotein complex [the signal recognition particle (SRP)] (Walter and Blobel, 1980, 1982) that arrests protein synthesis through an interaction with the ribosome and the signal sequence (Walter et al., 1981; Walter and Blobel, 1981a,b) and mediates translocation through the ER membrane via attachment to an ER membrane protein (the signal recognition receptor or “docking protein”) (Meyer and Dobberstein, 1980a,b; Gilmore et ul., 1982a,b; Meyer et al., 1982). The effect of a hydrophobic sequence spliced into secretory or cytosolic proteins was studied by Yost et d.(1983), using a cell-free transcription-linked translation system and recombinant DNA technology. They utilized a hybrid gene coding for a chimeric secretory protein composed of 182 amino acids of bacterial pre-P-lactamase toward the N-terminus, and 142 amino acids of chimpanzee a-globin toward the C-terminus. When its mRNA, transcribed in vitro, was translated in a wheat germ system in the presence of dog pancreas microsoma1 membranes, it directed the synthesis of a protein that was translocated into the microsomal lumen and had its signal sequence removed. Introduction of the anchor sequence of the membrane-bound form of IgM IJ. chain between the lactamase and the globin resulted in the synthesis of a protein that was also correctly translocated and had its signal peptide removed but remained membrane attached, with the lactamase portion facing the lumen and the globin portion exposed to the cytoplasm, a finding showing that the IgM sequence conserved its anchor function in spite of the long cytoplasmic domain. When the IgM anchor sequence was placed, instead, at the N-terminal end of a cytosolic fusion protein containing mainly globin sequences, the resulting protein was not membrane bound, indicating that the extra hydrophobic segment failed to function both as a signal sequence and as an anchor piece. In vitro translocation systems have also been used to study the posttranslational transfer of proteins into chloroplasts (Dobberstein et al., 1977; Chua and Schmidt, 1979), mitochondria (see Chua and Schmidt, 1979; Poyton, 1983; and article by Reid, this volume, for reviews), and peroxisomes (Goldman and Blobel, 1978).

270

ENRIQUE RODRIGUEZ-BOULAN ET AL.

2 . In Vitro SYSTEMS TO STUDY VESICULAR TRANSPORT BETWEEN ORGANELLES Fries and Rothman have introduced a cell-free system to study the in vitro transport of a plasma membrane glycoprotein (the envelope glycoprotein G of vesicular stomatitis virus) to the Golgi apparatus (Fries and Rothman, 1980; Rothman and Fries, 1981). Donor membranes were extracts of a mutant Chinese hamster ovary (CHO) cell line infected with VSV. Because of the lack of UDP-N-acetylglucosaminyltransferaseI, the G protein in these extracts has carbohydrate residues sensitive to the action of endoglycosidase H. Processing to the typical, endo H-resistant form was observed after addition of exogenous Golgi membranes. Although originally designed to study protein transfer from ER to Golgi, more controlled recent experiments suggest that the G protein is indeed being transported between two different sets of Golgi membranes (Fries and Rothman, 1981). The process requires ATP and cytosolic factors. In a similar line are the experiments by Altstiel and Branton (1983) involving the fusion in vitro of highly purified brain coated vesicles with purified kidney lysosomes. Fusion, followed by the activation of the nonfluorescent compound 6-carboxydiacetylfluorescein to the fluorescent 6-carboxyfluorescein by lysosomal hydrolases, was found to require free calcium and the stripping of the vesicular coat.

B. Genetic Systems 1. SECRETION MUTANTS IN YEAST Schekman and his collaborators have developed a model system involving the use of mutants of Saccharomyces cerevisiae with temperature-sensitive defects in secretory steps, with the aim of identifying the full range of cell functions required for intracellular protein transport (see Schekman, 1982, for a review). The yeast cell surface consists of three layers: a cell wall (consisting of mannoproteins and polysaccharides), a periplasm, and a plasmalemma. Except for small-molecular-weight toxins secreted in the medium, most of the secreted enzymes accumulate in the periplasm. These are exocytosed at the same surface point where new material is added to the surface for bud growth (Tkacz and Lampen, 1972; Field and Schekman, 1980); presumably the same exocytic vesicles are used for surface and secreted molecules (Novick and Schekman, 1983). A strilung feature of yeast is the low level of secretory organelles detected by electron microscopy, which correlates with a fast intracellular transport and small precursor pools of secreted proteins (Novick and Schekman, 1979; Novick et al., 1981). Based on the observation that ts secretory mutants are denser than normal cells

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

271

when incubated at the nonpermissive temperature (37°C) more than 200 sec mutants were isolated; these fell into two major classes: A and B, with 23 and 4 loci, respectively (Novick ef al., 1980; Ferro-Novick et al., 1984a,b). In all mutants, secretion and surface growth stopped at 37°C. Group A sec mutants continued to synthesize and accumulate active secretory enzymes; interestingly, accumulation resulted in the exaggeration of the specific organelle behind the block (ER, Golgi-like, vesicles), where acid phosphatase (a secretory enzyme) could be detected by immunocytochemistry. Class B sec mutants produce enzymatically inactive forms of invertase and acid phosphatase; in two of them (sec53 and sec59) immunorreactive forms of invertase are found embedded in the ER membrane, presumably as a consequence of a defect in function required for the completion of translocation (Ferro-Novick et al., 1984a,b). Studies of double sec mutants demonstrated the following sequence of mutant functions: B-A (ER)-A (Golgi bodies)-A (vesicles), in which the mutants at the left are epistatic to the other mutants. Glycosylation of yeast glycoproteins apparently proceeds in two compartmentalized steps, similar to those observed in mammalian cells: addition of an Nlinked GlcNAc,-Man, core in the ER and of an outer chain of up to 150 mannose residues, presumably in the Golgi apparatus (Esmon et al., 1980; Lehle et al., 1979). Transport to the cell surface of several glycoproteins with asparaginelinked carbohydrates is not inhibited by tunicamycin (Novick and Schekman, 1983). Mutants with a defect in the exit from the ER accumulate glycoproteins with only the core oligosaccharide. Transport of carboxypeptidase Y to the yeast vacuole, a lysosome analog, is, unlike its transport in mammalian cells, independent of carbohydrates (Hasilik and Tanner, 1978; Onishi et al., 1979). It is affected in class B and class A Golgi or pre-Golgi mutants but not in mutants that accumulate (post-Golgi) smooth vesicles (Schekman, 1982). 2. LOW-DENSITY LIPOPROTEIN RECEPTORMUTANTS Familial hypercholesterolemia is a rather frequent disease in humans and is caused by defects in the function of the low-density lipoprotein (LDL) receptor of fibroblasts and other cell types (Brown and Goldstein, 1979). This receptor system has provided very important insights on the mechanisms of receptormediated endocytosis and the role of coated pits in the movement of membrane material. After binding of LDL at 4°C the receptors appear randomly distributed on the cell surface; a few minutes after warming to 37"C, the receptors cluster in coated pits and are interiorized within coated vesicles. Under the influence of low pH, the LDLs dissociate from the receptors, presumably in the endosomes (Helenius et al.. 1983; Anderson and Kaplan, 1983), and are transferred to the lysosomes for degradation. The receptors are recycled back to the cell surface. Two major genetic defects have been described as causes of familial hyper-

272

ENRIQUE RODRIGUEZ-BOULAN ET AL.

cholesterolemia: a receptor with a very decreased binding capacity and a receptor that has a normal binding capacity but is unable to cluster in the coated pits (and, therefore, to be interiorized). 3. GENESOF SECRETORY AND MEMBRANE-BOUND IMMUNOGLOBULINS Lymphocytes at different stages of development express genes for secretory or membrane-bound immunoglobulins. Attachment to the membrane in the bound JgM forms is carried out by hydrophobic extra segments in the carboxy-terminal ends (Vassalli et al., 1979; Kehry et al., 1980; Singer and Williamson, 1980; see Table 11). Both forms are coded by the same genomic arrangement of exons and introns, which generates two different mRNAs upon differential processing of the RNA transcript (Early et al., 1980). Thus, this system provides an interesting natural example of the role of hydrophobic extra segments in membrane attachment. A similar idea has been explored by deleting the transmembrane segment of integral glycoproteins through recombinant DNA technology (see Section IV,D).

4. LYSOSOMAL PROTEINMISLOCATION MUTANTS I-cell disease (mucolipidosis 11) and pseudo-Hurler polydystrophy (mucolipidosis 111) are autosomal recessive lysosomal storage disorders characterized by various skeletal and mental abnormalities and eventually a fatal course (McKusik et al., 1972). A main biochemical feature of these diseases is a high level of lysosomal hydrolases in serum. Cells from these patients show decreased intracellular activities of hydrolases. The biochemical defect appears to be a which prevents the deficiency of UDP-N-acetylglucosaminylphosphotransferase, generation of the phosphomannosyl recognition marker of lysosomal enzymes (Reitman et al., 1981; Hasilik et al., 1981) needed for the transfer from Golgi apparatus to the lysosomes. As a consequence, the enzymes are secreted. Lack of the marker prevents the uptake of lysosomal enzymes from the medium (Hickman and Neufeld, 1972; McKusik et al., 1972). Secretion of lysosomal enzymes can also be observed if the synthesis of carbohydrates is inhibited by tunicamycin (von Figura et al., 1979).

C. Other Systems 1. THE GLYCOPROTEINS OF ENVELOPED RNA VIRUSES A group of plasma membrane glycoproteins, the envelope proteins of enveloped RNA viruses, have proved to be excellent model systems for studies on the biogenesis of plasma membrane glycoproteins (Lenard and Compans, 1974;

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

273

Katz et al., 1977; Lenard, 1978; Simons and Garoff, 1980; Lodish et af., 1981). Several experimental advantages have contributed to their popularity: ( I ) They are produced in large amounts by the infected cells (which allows their easy detection in the cytoplasmic precursor pools). (2) They are processed by cellcoded mechanisms (since the small virus genomes code only for essential viral replicative functions). (3) cDNA complementary to their genomes is relatively easy to obtain and to clone. Therefore, the complete primary structure is known for many viral glycoproteins. Viral envelope glycoproteins have provided an invaluable tool to elucidate the steps and mechanisms involved in the processing of ASN-linked carbohydrates of glycoproteins (see Parodi and Leloir, 1979; Lennarz, 1980, for reviews). Improvements in immunoprecipitation and cell fractionation procedures have allowed the extension of these studies to cellular plasma membrane glycoproteins. 2. POLARIZED BUDDINGOF

VIRUSES FROM

EPITHELIAL CELLS

Epithelial cells infected with enveloped RNA viruses provide an additional advantage for the study of sorting of plasma membrane proteins, since viral glycoproteins may be addressed to one of two possible surface destinations. Enveloped viruses bud with striking polarity from the plasmalemma of polarized epithelial cells: influenza virus, Sendai virus, and Simian virus 5 are assembled from the apical surface of Madin-Darby canine kidney (MDCK) cells, a polarized epithelial line (Misfeldt et al., 1976; Cereijido et al., 1978), whereas VSV is selectively produced from the basolateral surface (Rodriguez-Boulan and Sabatini, 1978; Rodriguez-Boulan, 1983). Polarized budding is preceded, and apparently determined, by the asymmetric insertion of viral glycoproteins into the surface later used for budding (Rodriguez-Boulan and Pendergast, 1980). Influenza hemagglutinin expressed in epithelial cells from a cloned cDNA fragment via SV40 vectors is also asymmetrically distributed (Roth et af., 1983), a finding indicating that nucleocapsid and matrix viral proteins are not necessary for its polarity. Thus, viral glycoproteins share with cellular plasma membrane proteins the sorting signals and biogenetic mechanisms that determine their segregation into different surface domains of epithelial cells. Neither tunicamycin nor the use of lectin-resistant mutants of MDCK cells or of mutants of influenza virus defective in the neuraminidase alters the polarity of virus budding (Roth er al., 1979; R. F. Green et al., 1981; Meiss et al., 1982). Thus, unlike the mechanism for lysosomal hydrolases, carbohydrates are not part of the sorting signals for viral glycoprotein surface segregation in epithelial cells. The intracellular pathways of HA and VSV G protein (respectively an apical and a basolateral glycoprotein) are the same until, at least, the Golgi apparatus, as determined by double-labeling immunoelectron microscopy experiments on cells double-infected with influenza and VSV (Rindler et al., 1982, 1984). After

274

ENRIQUE RODRIGUEZ-BOULAN ET AL.

passage through the Golgi, HA is detected in a population of small vesicles that are about twice the size of coated vesicles and that occupy the apical half of the cytoplasm (Rodriguez-Boulan et al., 1984). G protein appears to be predominantly localized into a different population of vesicles in the lower half of the cytoplasm of MDCK cells (Salas and Rodriguez-Boulan, unpublished results). Monensin inhibits the production of the basolateral virus VSV but only delays the production of the apical virus influenza (Alonso and Compans, 1981; Rodriguez-Boulan et al., 1984). Treatment with cytochalasin D or colchicine does not alter the polarized viral budding, a finding indicating that microtubules or microfilaments (at least those sensitive to the drugs) are not involved in polarized viral budding (Salas et al., 1985). Furthermore, addition of anti-HA antibody only to the basolateral surface of confluent MDCK monolayers do not affect the production of influenza virus or the delivery of HA to the apical surface (Misek et al., 1984). These experiments suggest that apical and basolateral glycoproteins are segregated at the level of the Golgi apparatus, by the incorporation into different sets of vesicles that fuse directly with the respective target domains in the cell surface. There is experimental evidence, however, indicating that some sorting of epithelial plasma membrane proteins may occur at the cell surface. Experiments by M a t h et al. (1983) and Pesonen and Simons (1983) show that, after fusion of VSV to the apical surface of MDCK cells, the G protein is interiorized and quickly redistributed to the basolateral domain. Rodewald (1973) has previously shown that IgG is transported from the apical to the basolateral domain of newborn rat intestinal cells by receptor-mediated endocytosis; the fluid marker horseradish peroxidase, added simultaneously to the apical medium, is transported, however, to the lysosomes (Abrahamson and Rodewald, 1981). Transport of IgA in the opposite direction via a specific receptor by several epithelia is well documented (see Kraehenbuhl and Kuhn, 1978, for a review). This receptor is used only once; available evidence suggests that it is delivered first to the basolateral membrane, where it binds IgA, and then migrates to the apical surface, where it is cleaved by specific proteases and released to the medium complexed with IgA (Solari and Kraehenbuhl, 1984). Mostov et al. (1984) have cloned DNA complementary to IgA receptor mRNA and determined its complete primary structure (its main features are summarized in Table 11); an interesting observation derived from this study is the finding of amino acid sequences homologous to immunoglobulin chains. Hauri et al. (1979) have proposed, on the basis of pulse-chase analysis of subcellular fractions, that sucrase-isomaltase, an apical glycoprotein in intestinal epithelium, is first inserted in the basolateral membrane, retrieved, and then transported to the apical surface. However, since the contamination of basolateral membrane fractions with Golgi membranes, or other post-Golgi vesicles, cannot be discarded, these results must be interpreted with caution. The preceding data taken together strongly suggest

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

275

the existence of mechanisms to relocate molecules from one surface to the other in epithelial cells, mechanisms which may contribute to surface polarity. The role of tight junctions and substrate attachment on epithelial polarity has been investigated. Isolated epithelial cells in suspension culture lose, to a large extent, their surface polarity (Pisam and Ripoche, 1976; Ziomek et al., 1980; Rodriguez-Boulan et al., 1983). Attachment to a substrate, however, results in the recovery of the ability to sustain polarized viral budding by MDCK cells, even in the absence of complete tight junctions (Rodriguez-Boulan et al., 1983). The polarity of thyroid cells in monolayer culture can be changed by polymerization of a collagen gel on top of the cells (Chambard er a / . , 1981). Thus, the interaction with substrate material, which most probably causes a restriction in the mobility of specific surface receptors, reorganizes the epithelial cell surface and contributes to the sorting of the plasma membrane glycoproteins. Complete surface segregation, however, may require the existence of functional tight junctions. VIRUSESTHATBUD FROM INTERNAL MEMBRANES 3. ENVELOPED Two groups of RNA viruses bud from internal membranes: coronaviruses and rotaviruses (Holmes, 1983; Stunnan and Holmes, 1983). Coronaviruses bud into intracytoplasmic vesicles from the RER and Golgi. One member of the group, mouse hepatitis virus (MHV) possess two glycoproteins: E l , which is the matrix protein and is a transmembrane glycoprotein with its N-terminus facing the outside of the virion and the C-terminus associated with the nucleocapsid, and E2, which is the spike protein and forms the peplomers on the virion surface, incorporated late in the assembly. Tunicamycin blocks the glycosylation of E2, which has 8 N-linked carbohydrate moiety, but not of E l , which has only 0-linked oligosaccharides. Glycosylation of E2 occurs during its cotranslational translocation, via a signal sequence, across the ER membrane. E 1 does not present a cleavable signal sequence and apparently is directed toward the membrane by an internal hydrophobic sequence. Glycosylation of E 1 occurs late, in the Golgi, and is blocked by monensin, which produces accumulation of enveloped virions in RER vesicles. It seems that the site of budding is determined by inability of El to exit from the RER. Similar results were obtained with bovine and human coronaviruses, but avian infectious bronchitis virus presents N-linked oligosaccharides in both glycoproteins. Knowledge about rotavirus assembly is far from complete. Different laboratories do not agree on how many of the proteins that have a synthesis directed by the viral genome are structural. Mature virions present a double protein capsid; the outer capsid can be removed by calcium-chelating agents (EDTA or EGTA), thereby rendering the virions unable to infect cells. There is no general agreement among investigators about the detailed composition of the capsids; it is

276

ENRIQUE RODRIGUEZ-BOULAN ET AL.

accepted that the inner capsid contains at least one major and two minor proteins (p116, p96, and p42) and that the major component of the outer capsid is a glycoprotein (gp34), together with a nonglycosylated protein (p84) which is cleaved after the virions are released from the cell into peptides (p62 and p28). Several groups would add two or three proteins to the inner capsid and one or two to the outer capsid. In infected cells the synthesis of viral components takes place in viral inclusions, defined by electron microscopy, called ‘‘viroplasm” which are similar to the “virus factories” in reovirus-infected cells. Single-shelled virions (i.e., virions containing only the inner capsid) bud from the RER membranes into the lumen and possess a lipid bilayer envelope ( “pseudoenvelope”) that has to be eliminated to allow the formation of the outer capsid. The viral genome codes for two glycoproteins: gp34, a component of the outer capsid, and gp25, apparently a nonstructural component. Both glycoproteins are found inserted in the RER membrane, and both possess endoglycosidase H-sensitive carbohydrate moieties. The elimination of the pseudoenvelope needs the synthesis of a glycosylated component; in infected cells treated with tunicamycin, the pseudoenvelopes remain around all the particles in the RER vesicles; and upon lysis of the cells, virions with only the inner capsid are released. It has been postulated that the calcium concentration within the RER lumen is sufficient to allow the assembly of the outer capsid upon removal of the pseudoenvelope, but still, the mechanism of translocation across the RER membrane of the nonglycosylated protein(s) of the outer capsid remains unexplored. INTO CONSTITUTIVE OR REGULATED PATHWAYS 4. SORTING

Kelly and his colleagues have presented evidence that AtT-20, a mouse pituitary tumor line that secretes adrenocorticotropin (ACTH) and 6-endorphin through a regulated pathway, externalizes a plasma membrane glycoprotein and a high-molecular-weight precursor of both hormones via a different constitutive pathway (Gumbiner and Kelly, 1982; Moore et al., 1983a). Different vesicles appear to be mediating both types of secretion. Chloroquine, which raises the pH of intracellular vesicles, blocks the storage of newly synthesized ACTH into secretory granules and instead diverts it to the outside of the cell via the constitutive pathway (Moore et al., 1983b). This is similar to previous observations with lysosomal hydrolases, which are secreted in the presence of chloroquine (Gonzalez-Noriega et al., 1980).

5. RECEPTORSORTINGDURING ENDOCYTOSIS This subject has been extensively reviewed (see, for example, Brown et al., 1983; Anderson and Kaplan, 1983; and articles by Wiley and Schneider et al., this volume).

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

277

D. Recombinant DNA Technology Applied to the Study of Protein Sorting The relatively new technologies of molecular cloning and genetic engineering, in combination with the availability of efficient vectors for the expression of cloned genes in eukaryotic cells, have introduced the possibility of studying (and eventually modifying) at the molecular level the nature of protein-sorting signals. Copy DNA genes need to be ligated to a promoter recognized by eukaryotic polymerase in order to be expressed in eukaryotic cells. A variety of expression vectors carrying viral promoters and other regulatory sequences have been used successfully (see Gluzman, 1982; Rigby, 1982). They were derived from different eukaryotic viruses: SV40, adeno-, papilloma-, and retroviruses. They fall into two main classes: plasmid-type and virus-type vectors. Plasrnid vectors are introduced by transfection, a process usually resulting in transient expression of the gene by a small percentage (< 10%) of the cells. On the other hand, infection with viral vectors results in almost all cells expressing the gene at high levels. The most popular viral vector has been SV40 for obvious reasons: the circular genome is relatively small and its complete sequence is known, the promoters and splicing sites have been accurately mapped, all the structural genes have been identified, and it is easy to purify (Elder et al., 1981). The first report of successful expression of a viral membrane protein using a recombinant vector was published in 1981 by Moriarty et al. (1981). A genomic fragment of hepatitis B virus containing the coding sequence for the viral surface antigen (HBA) was incorporated into a defective SV40 genome and used to transfect African green monkey kidney (AGMK) cells previously infected at the nonpermissive temperature with an SV40 early rs mutant in order to obtain a stock recombinant virus. Infection of AGMK cells with this stock virus resulted in the production of HBA by 45% of the cells as judged by immunofluorescence. HBA was released into the medium in 22-nm particles and filaments identical to those found in sera of human patients. Three different groups introduced cDNA genes from influenza’s HA in defective SV40 genomes (Gething and Sambrook, 1981, 1982; Sveda and Lai, 1981; Sveda et al., 1982; Hartman et al., 1982). By immunofluorescence, HA was detected in a Golgi-like perinuclear area and in the plasma membrane (Sveda and Lai, 1981; Sveda el al., 1982; Hartman et al., 1982), where it conferred on the infected cells the ability to agglutinate erythrocytes (Gething and Sambrook, 1981; Sveda et al., 1982). SDS-PAGE analysis of immunoprecipitates showed the presence of a band with the same mobility as mature HA. Constructs of HA were engineered to study the influence of the signal sequence and of the anchor sequence on the fate of HA. A signal-minus construct coded for a nonglycosylated HA, a result suggesting that this mutant protein was

278

ENRIQUE RODRIGUEZ-BOULAN ET AL.

not translocated across the ER membrane (Gething and Sambrook, 1982). Other constructs were developed lacking segments coding for the C-terminal portion of HA, including the anchor sequence (Sveda et al., 1982; Gething and Sambrook, 1982). These anchor-minus proteins were glycosylated and secreted, although with somewhat lower intracellular transport rates than the complete HA. Some contradictions between the results of these two groups may probably be attributed to differences in the size of the C-terminal deletions. Rose and Bergmann (1982) microinjected into the nuclei of L cells a complete cDNA gene of the VSV G protein ligated to a plasmid vector, pSV2, under the control of the early SV40 promoter (Mulligan and Berg, 1980). The surface and intracellular immunofluorescence pattern resembled the one described above for HA. An anchor-minus mutant G protein, expressed via a plasmid vector carrying late SV40 promoter sequences, pJCll9 (Sprague et al., 1983), introduced by transfection, was secreted from the cells with a half-time of 2-4 hours, much longer than the intracellular transit time of the wild-type G, which reaches the plasma membrane 30 minutes after synthesis (Knipe et al., 1977b; Strous and Lodish, 1980; Bergman et al., 1981). Intending to further define the roles of the anchor sequence and the cytoplasmic domain in the intracellular migration of G, the same investigators engineered a series of mutants with different deletions of the C-terminal segment (Rose and Bergman, 1983). Absence of the cytoplasmic domain or addition of extra amino acids resulted to a failure to exit from the ER or Golgi; deletion of half of the cytoplasmic tail delayed the transport to the surface by about 3 hours. Influenza virus neuraminidase (NA) is inserted in the membrane by its Nterminal hydrophobic region, the rest of the protein facing the external domain (see Table 11; Blok et al., 1982). It possesses two hydrophobic regions: the one at its N-terminus is composed of 29 amino acids and apparently has a dual role of signal peptide and anchor sequence; the other is internal and its function is unknown. Davis et al. (1983) have reported expression of a NA cDNA gene ligated to an SV40 defective genome. Immunofluorescence experiments showed staining of perinuclear structures and the cell surface of infected cells. Kondor-Koch et al. (1982) introduced a full length cDNA of the complete genome of Semliki Forest virus (SFV) in pSV2 and microinjected it into the nuclei of BHK cells. Immunofluorescence experiments detected viral capsid protein and viral protein El distributed diffusely throughout the cytoplasm, while protein p62 (precursor of the viral membrane glycoproteins E2 and E3) presented the same pattern as described above for glycoproteins of other viruses. By blocking further protein synthesis with cycloheximide, they were able to chase the migration of the immunoreactive product from ER-like to Golgi-like structures and to the cell surface. In further experiments using a modified vector, they obtained synthesis and transport of both El and E2 viral proteins to the cell

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

279

surface (Kondor-Koch e t a / ., 1983) and were able to show that El is responsible for the fusogenic activity of the virion at low pH. They also engineered mutants of E2 glycoprotein (Garoff ef a / ., 1983), deleting different amounts (up to twothirds) of the cytoplasmic domain, and showed that all of them migrate to the cell surface apparently with the same kinetics as the wild type. All the results mentioned above confirm that lack of the anchor sequence in a plasma membrane protein results in the eventual secretion of a normally glycosylated smaller protein. These experiments not only demonstrate the membrane attachment role of the anchor sequence, but also strongly indicate that both membrane and secreted forms share the same intracellular migration pathway, at least up to the terminal glycosylation site, the Golgi apparatus. As for the cytoplasmic domain, its role in intracellular migration of plasma membrane glycoproteins is still obscure. Its removal may or may not affect transport, according to the protein. It can be presumed that at least some of the effects described may be related to changes in protein conformation secondary to modifications in the C-terminal end. Zuniga et a f . (1983) have engineered mutations in a genomic clone coding for a class I antigen of the mouse major histocompatibility complex. They obtained two mutants lacking different portions of the cytoplasmic domain. When these mutants were introduced into L cells via calcium phosphate, they produced transplantation antigen located in the cell surface, as determined by radioimmunoassay. The antigen was completely functional, as judged by the ability of the cells displaying it to serve as target in specific cytotoxicity assays. Roth et al. ( 1 983) published the first report of polarized expression of influenza’s HA coded by a cDNA gene in polarized epithelial cells. They infected primary AGMK cells plated at high density with a recombinant SV40 virus; and in cell patches that presented characteristics of polarized epithelium they observed by immunoelectron microscopy HA inserted preferentially in the apical surface. Because neither the matrix protein nor influenza nucleocapsids were present in the infected cells, this experiment clearly shows that the information directing HA to the apical surface is included in its amino acid sequence. Recently, a rhesus monkey kidney cell line, MA-104, has been shown to generate transepithelial electrical resistances of over 100 cm2, to sustain polarized budding of influenza and VSV and to express influenza HA preferentially on the apical surface upon infection with SV40 vectors carrying the HA gene (Gundersen, Roth, Gething, Sambrook, and Rodriguez-Boulan, unpublished observations). Vectors carrying chimeric genes coding for hybrids of apical and basolateral glycoproteins have been generated and are being tested in the MA-104 system for possible changes in the localization of HA that may throw light on the protein domain carrying the information for its apical segregation.

280

ENRIQUE RODRIGUEZ-BOULAN ET AL.

V. SUMMARY AND PERSPECTIVES Work in recent years has started to unravel the complex mechanisms that eukaryotic cells use to sort and distribute their proteins to the site of function. Two of the “sorting signals” that participate in this process have been identified: the signal sequence, which is employed in translocation across the ER membrane of proteins destined for the different segments of the secretory pathway and for exocytosis, and the mannose 6-P residues of lysosomal proteins, which are responsible for their targeting to lysosomes. Substantial progress has been made in the characterization of the receptors for these two sorting signals. However, very little is known about the mechanisms that mediate the localization and concentration of specific proteins and lipids within organelles. Various experimental model systems have become available for their study. The advent of recombinant DNA technology has shortened the time needed for obtaining the primary structure of proteins to a few months. Consequently, there are now several membrane proteins, including some receptors, the primary sequences of which are known, and the list is increasing every month. This information, in combination with new and more efficient methods to reintroduce and express genes into cells, makes it possible to plan strategies to identify the protein domains involved in sorting; these, eventually, could be used to identify the “sorters” themselves. The ultimate knowledge of the mechanisms that the cell uses to distribute its components is not only an important aspect of cell biology but may also bear significant practical relevance for the treatment of some genetic and oncogenic disorders. ACKNOWLEDGMENTS We are grateful to Dr. Schekman and his colleagues for sending us their manuscripts before publication. This work was supported by grants from the National Institutes of Health and from the National Science Foundation. E.R.-B. was a recipient of an IrmaT. Hirschl award. P. J. I. Salas was a recipient of a Fogarty International Fellowship from N.I.H.

REFERENCES Abrahamson, D. R . , and Rodewald, R. (1981). Evidence of the sorting of endocytic vesicle contents during the receptor-mediated transport of IgG across the newborn rat intestine. J . Cell Biol. 91, 270-280. Adams, W. R., and Kraft, M. L. (1967). Electron microscopic study of the intestinal epithelium of mice infected with the agent of epizootic diarrhea of infant mice. Am. J . Parhol. 51, 39-60. Alonso, F. V., and Cornpans, R. W. (1981). Differential effect of monensin on enveloped viruses that form at distinct plasma membrane domains. J . Cell B i d . 89, 700-705. Altstiel, L., and Branton, D. (1983). Fusion of coated vesicles with lysosomes: Measurement with a fluorescence assay. Cell 32, 921-929.

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

28 1

Anderson, G . W., and Kaplan, J . (1983). Receptor-mediated endocytosis. I n “Modern Cell Biology” (B. Satir, ed.), Vol. I , pp. 1-52. Liss, New York. Bar-Nun, S . , Kreibich, G . , Adesnik, M.. Alterman. L., Negishi, M., and Sabatini, D. D. (1980). Synthesis and insertion of cytochronie P-450 into endoplasmic reticulum membranes. Proc. Nail. Acad. Sci. U.S.A. 77, 965-969. Bell, R. M., and Coleman. R. A. (1980). Enzymes of glycerolipid synthesis in eukaryotes. Annu. Rev. Biochem. 49, 459-488. Bergeron, J. J. M.. Kotwal. G. J . , Levine, G., Bilan, P., Rachubinsky, R.. Hamilton, M., Shore, G. C., and Ghosh, H. P. (1982). Intracellular transport of the glycoprotein G of vesicular stomatitis virus as visualized by electron microscope radioautography. J . Cell Biol. 94, 36-41, Bergmann, J . E., Tokuyasu. K. T., and Singer, S . J . (1981). Passage of an integral membrane protein, the vesicular stomatitis virus glycoprotein through the Golgi apparatus en route to the plasma membrane. Proc. Natl. Acad. Sci. U.S.A. 78, 1746-1750. Blobel, G. (1980). Intracellular protein topogenesis. Pro(,.Nurl. Acad. Sci. U.S.A. 77, 1496- 1500. Blobel, 0.. and Dobberstein. B. (1975a). Transfer of proteins across membranes. 1. Presence of proteolytically processed and unprocessed na~centimmunoglobulin light chains on membranebound ribosomes of murine myeloma. J . Cell Biol. 67, 835-851. Blobel, G., and Dobberstein, B. (1975b). Transfer of proteins across membranes. 11. Reconstitution of functional rough microsomes from heterologous components. J . Cell B i d . 67, 852-862. Blobel, G., and Sabatini, D. D. (1971). Ribosome-membrane interaction in eukaryotic cells. In “Biomembranes” (L. A. Manson, ed.), Vol. 2. pp. 193-195. Plenum, New York. Blobel, G . , Walter, P., Chang, C. N., Goldman, 8.. Erickson, A. H., and Lingappa, V. R. (1979). Translocation of proteins across membranes: The signal hypothesis and beyond. Symp. Sue. Exp. B i d . 33, 9-36. Blok, J . , Air, G . M., Laver. W. G., Ward, C. W., Lilley, G . G . , Woods, E. F., Roxburgh. C. M.. and Inglis, A. S . (1982). Studies on the size, chemical composition, and partial sequence of the neuraminidase (NA) from type A influenza viruses show that the N-terminal region of the NA is not processed and serves to anchor the NA in the viral membrane. Virology 119, 109-121. Boquet, P., and Dutlot, E. (1982). Tetanus toxin fragment forms channels in lipid vesicles at low pH. Proc. Nail. Acad. Sci. U.S.A. 79, 7614-7618. Borgese. N . , and Gaetani, S. (1980). Site of synthesis of rat liver NADH cytochrome b5 reductase. an integral membrane protein. FEES Lei?. 112, 216-220. Braell. W. A,, and Lodish, H. F. (1982). The erythrocyte anion transport protein is cotranslationally inserted into microsomes. Cell 28, 23-31. Branton, D., Cohen, C. M., and Tyler, J. (1981). Interactions of cytoskeletal proteins on the human erythrocyte membrane. Cell 24, 24-32. Bretscher, M. S. (1973). Membrane structure: Some general principles. Science 181, 622-629. Bretscher. M. S . (1981). Surface uptake by fibroblasts and its consequences. Cold Spring Harbor Svmp. Quant. Biol. 46, 707-712. Brown, M. S.. and Goldstein. J . L. (1979). Receptor-mediated endocytosis: Insights from the lipoprotein receptor system. Proc. Nurl. Acad. Sci. U . S . A . 76, 3330-3337. Brown, M . S., Anderson. R. G . W., and Goldstein, J . L. (1983). Recycling receptors: The roundtrip itinerary of migrant membrane proteins. Cell 32, 663-667. Brown, W. I . , and Farquhar, M. C. (1984). The mannose-6-phosphate receptor is concentrated in cis Golgi cisternae. Cell 36, 295-307. Wacker, H., O’Neill, B., and Semenza. G . Brunner, J., Hauser, H., Braun, H., Wilson, K. .I., (1979). The mode of association of the enzyme complex sucrase-isomaltase with the intestinal brush border membrane. J . B i d . Cham. 254, 1821-1828. Cereijido, M., Robbins, E. S., Dolan, W. J . , Rotunno. C. A., and Sabatini, D. D. (1978). Polarized

282

ENRIQUE RODRIGUEZ-BOULAN ET AL.

monolayers formed by epithelial cells on a permeable and translucent support. J . Cell Biol. 77, 853-880. Chambard, M., Gabrion, J., and Mauchamp, J. (1981). Influence of collagen gel on the orientation of epithelial cell polarity: Follicle formation from isolated thyroid cells and from preformed monolayers. J . Cell B i d . 91, 157-166. Chua, N.-H., and Schmidt, G . W. (1979). Transport of proteins into mitochondria and chloroplasts. J . Cell Biol. 81, 461-483. Claudio, T., Ballivet, M., Patrick, J., and Heineman, S. (1983). Nucleotide and deduced aminoacid sequences of Torpedo calijornica acetylcholine receptor gamma subunit. Proc. Narl. Acad. Sci. U.S.A. 80, 1111-1115. Colman, P. M., Varghese, I. N., and Laver, W. G . (1983). Structure of the catalytic and antigenic sites of influenza virus neuraminidase. Nature (London) 303, 41-44. Davies, P. J. A., Davies, D. R., Levitzky, A., Maxfield, F. R., Milhaud, P., Willingham, M. C., and Pastan, 1. (1980). Transglutaminase is essential in receptor mediated endocytosis of alpha-2-macroglobulin and polypeptide hormones. Nature (London) 283, 162- 167. Davis, A. R., Bos, J. T., and Nayak, D. P. (1983). Active influenza virus neuraminidase is expressed in monkey cells from cDNA cloned in simian virus 40 vectors. Proc. Natl. Acad. Sci. U.S.A. 80, 3976-3980. Davis, B. P., and Tai, P.-C. (1980). The mechanism of protein secretion across membranes. Nature (London) 283, 433-438. Dobberstein, B., Blobel, G . , and Chua, N-H. (1977). In vitro synthesis and processing of a putative precursor for the small subunit of ribulose- I ,5-biphosphate carboxylase of Chlamidomonas reinhardii. Proc. Natl. Acad. Sci. U.S.A. 76, 343-347. Dobberstein, B., Garoff, H., Warren, G . , and Robinson, P. J. (1979). Cell free synthesis and membrane insertion of mouse H2d histocompatibility antigen and beta 2-microglobulin. Cell 17, 759-769. Donovan, J. J., Simon, M. I., Draper, R. K., and Montal, M. (1981). Diphtheria toxin forms transmembrane channels in planar lipid bilayers. Proc. Narl. Acad. Sci. U.S.A. 78, 172-176. Dunn, W. A., Hubbard, A. L., and Aronson, Jr., N. N. (1980). Low temperature selectively inhibits fusion between pinocytic vesicles and lysosomes during heterophagy of 1251-asialofetuinby the perfused rat liver. J . Biol. Chem. 255, 5971-5978. Dunphy, W. G . , and Rothman, J. E. (1983). Comparmentation of asparagine-linked oligosaccharide processing in the Golgi apparatus. J . Cell Biol. 97, 270-275. Early, P., Rogers, J., Davis, M., Calame, K., Bond, M., Wall, R., and Hood, L. (1980). Two mRNAs can be produced from a single immunoglobulin gene by alternative RNA processing pathways. Cell 20, 313-319. Ehrenreich, J. H., Bergeron, J . J. M., Siekievitz, P.. and Palade, G . E. (1973). Golgi fractions prepared from rat liver homogenates. I. Isolation procedure and morphological characterization. J . Cell Biol. 59, 45-72. Elder, J. T., Spritz, R. A,, and Weissman, S. M. (1981). Simian virus 40 as an eukaryotic cloning vehicle. Annu. Rev. Genet. 15, 295-340. E m , S. D., Hall, M. N., and Silhavy, T. J. (1980). A mechanism of protein localization: The signal hypothesis and bacteria. J . Cell Biol. 86, 701-711. Erickson, A. H., and Blobel, G . (1979). Early events in the biosynthesis of the lysosomal enzyme cathepsin D. J . Biol. Chem. 254, 11771-11774. Esmon, B., Novick, P., and Schekman, R. (1980). Compartmentalized assembly of oligosaccharides on exported glycoproteins in yeast. Cell 25, 451-460. Evans, G. A,, Margulies, D. H., Camerini-Otero, R. D., Ozato, K., and Seidman, J. G . (1982). Structure and expression of a mouse histocompatibility antigen gene, H-2Ld. Proc. Narl. Acad. Sci. U.S.A. 79, 1994-1998.

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

283

Farquhar, M. G . (1982). Membrane recycling in secretory cells: Pathway to the Golgi complex. Ciba Found. Symp. 92, 157-183. Farquhar. M. G . (1983). Multiple pathways of exocytosis, endocytosis, and membrane recycling: Validation of a Golgi route. Fed. Proc., Fed. Am. Soc. Exp. B i d . 42, 2407-2413. Farquhar, M., and Palade, G . (1981). The Golgi apparatus (complex)-(1954-1981)-from artifact to center stage. J . Cell B i d . 91, 77s-103s. Ferro-Novick, S . , Hansen, W., Schauer. I., and Scheckman, R. (1984a). Genes required forcompletion of import of proteins into the endoplasmic reticulum in yeast. J . Cell B i d . 98, 44-53. Ferro-Novick, S . , Novick, P., Field, C., and Schekman. R. (1984b). Yeast secretory mutants that block the formation of active cell surface enzymes. J . Cell Biol. 98, 35-43. Field, C., and Schekman. R. (1980). Localized secretion of acid phosphatase reflects the pattern of cell surface growth in Saccharomycrs cereiisiae. J . Cell B i d . 86, 123-128. Fields, S . , Winter, G . , and Brownlee, G . G. (1981). Structure of the neuraminidase gene in human influenza virus A/PR/8/34. Nature (London) 290, 2 13-217. Fitting, T., and Kabat, D. (1982). Evidence for a glycoprotein "signal" involved in transport between subcellular organelles: Two membrane glycoproteins reach the cell surface at different rates. J . B i d . Chem. 257, 1401 1-14017. Forgac, M . . Cantley, L., Wiedenmann, B., Altstiel. L., and Branton, D. (1983). Clathrin coated vesicles contain an ATP-dependent proton pump. Proc. Nail. Acad. Sci. U . S . A . 80, 13001303. Frank, G., Brunner, J., Hauser, H., Wacker, H., Semenza, G . , and Zuber, H. (1978). The hydrophobic anchor of small intestinal sucrase-isomaltase. FEES Lett. 96, 183-188. Friend, D. S . . and Farquhar, M. G . (1967). Functions of coated vesicles during protein absortion in the rat vas deferens. J . Cell. Biol. 35, 357-376. Fries, D. S . , and Rothman, J . E. (1980). Transport of vesicular stomatitis virus glycoprotein in a cell free extract. Proc. Natl. Acad. Sci. U.S.A. 77, 3870-3874. Fries, E.. and Rothman. J . E. (1981). Transient activity of Golgi-like membranes as donors of vesicular stomatitis viral glycoprotein in vhru. J . Cell B i d . 90, 697-704. Furthmayr, H. (1977). Structural analysis of a membrane glycoprotein: Glycophorin A. J . Supramol. Strucf. 7, 121-134. Garoff, H., and Soderlund, H. (1978). The amphiphilic membrane glycoproteins of Semliki Forest virus are attached to the lipid bilayer by their COOH-terminal ends. J . Mol. Biol. 124, 535-549. Garoff, H.. Simons, K., and Dobberstein, B. (1978). Assembly of the Semliki Forest virus membrane glycoproteins in the membrane of the endoplasmic reticulum in vitro. J . Mol. Biol. 124, 587-600. Garoff, H., Frischauf, A,-M., Simons, K . , Lehrach, H., and Delius. H. (1980). Nucleotide sequence of cDNA coding for Semliki forest virus membrane glycoproteins. Nature (London) 288, 236-241. Garoff, H., Kondor-Koch. C., Petterson, R., and Burke, R. (1983). Expression of Semliki Forest virus proteins from cloned complementary DNA. 11. The membrane-spanning glycoprotein E2 is transported to the cell surface without its normal cytoplasmic domain. J . Cell B i d . 97, 652658. Gething, M . J . , and Sambrook, J . ( 1981 ). Cell-surface expression of intluenza hemagglutinin from a cloned DNA copy of the RNA gene. Nature (London) 292, 620-625. Gething, M. J., and Sambrook, J . (1982). Construction of influenza heniagglutinin genes that code for intracellular and secreted forms of the protein. Nature (London) 300, 598-603. Gething, M. J . , Bye, I . , Skehel, J . , and Waterfield, M. (1980). Cloning and DNA sequence of double stranded copies of hemagglutinin from H2 and H3 strains elucidates antigenic shift and drift in human influenza virus. Nature (London) 287, 301-306. Geuze, H. J., Slot, I. W., Strouss, G . J. A. M . , Lodish, H. F., and Schwartz, A. L. (1983).

284

ENRIQUE RODRIGUEZ-BOULAN ET AL.

Intracellular site of asialoglycoprotein receptor-ligand uncoupling: Double-label immunoelectron microscopy during receptor-mediated endocytosis. Cell32, 277-287. Gibson, R., Leavitt, R., Kornfeld, S . , and Schlesinger, S. (1978). Synthesis and infectivity of vesicular stomatitis virus containing non-glycosylated G protein. Cell 13, 671-679. Gilmore, R., Blobel, G., and Walter, P. (1982a). Protein translocation across the endoplasmic reticulum. 1. Detection in the microsomal membrane of a receptor for the signal recognition particle. J . Cell Biol. 95, 463-469. Gilmore, R., Walter, P., and Blobel, G. (1982b). Protein translocation across the endoplasmic reticulum. 11. Isolation and characterization of the signal recognition particle receptor. J . Cell Biol. 95, 470-477. Glickman, J . , Croen, K., Kelly, S . , and Al-Awqati, Q. (1983). Golgi membranes contain an electrogenic H + pump parallel to a chloride conductance. J . Cell Biol. 97, 1303-1308. Gluzman, Y.,ed. (1982). “Eukaryotic Viral Vectors.” Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Goldman, B. M., and Blobel, G. (1978). Biogenesis of peroxysomes: Intracellular site of synthesis of catalase and uricase. Proc. Narl. Acad. Sci. U.S.A. 75, 5066-5070. Goldstein, J. L., Anderson, R. G. W., and Brown, M. S . (1979). Coated pits, coated vesicles, and receptor-mediated endocytosis. Nature (London) 279, 679-685. Gonzalez-Noriega, A., Grubb, J. H., Talkad, V., and Sly, W. S. (1980). Chloroquine inhibits lysosomal enzyme pinocytosis and enhances lysosomal enzyme secretion by impairing receptor recycling. J . Cell B i d . 85, 839-852. Green, J., Griffiths, G., Louvard, D., Quinn, P., and Warren, G. (1981). Passageof viral membrane proteins through the Golgi complex. J . Mol. Biol. 152, 663-698. Green, R. F., Meiss, H. K., and Rodriguez-Boulan, E. (1981). Glycosylation does not determine segregation of viral glycoproteins in the plasma membrane of epithelial cells. J . Cell Biol. 89, 230-239. Griffiths, G., Brands, R . , Burke, B., Louvard, D., and Warren, G . (1982). Viral membrane proteins acquire galactose in trans Golgi cisternae during intracellular transport. J . Cell Biol. 95, 781792. Griffiths, G . , Quinn, P., and Warren, G. (1983). Dissection of the Golgi complex. I. Monensin inhibits the transport of viral membrane proteins from medial to trans Golgi cisternae in baby hamster kidney cell infected with Semliki Forest virus. J . Cell Biol. 96, 835-850. Gumbiner, B., and Kelly, R. B. (1982). Two distinct intracellular pathways transport secretory and membrane glycoproteins to the surface of pituitary tumor cells. Cell 28, 51-59. Hartman, J. R., Nayak, D. P., and Fareed, G . C. (1982). Human influenza virus hemagglutinin is expressed in monkey cells using simian virus 40 vectors. Proc. Narl. Acad. Sci. U.S.A. 79, 233-237. Hasilik, A., and Neufeld, E. (1980a). Biosynthesis of lysosomal enzymes in fibroblasts-synthesis as precursors of higher molecular weight. J . Biol. Chem. 255, 4937-4945. Hasilik, A., and Neufeld, E. (1980b). Biosynthesis of lysosomal enzymes in fibroblasts-phosphorylation of mannose residues. J . B i d . Chem. 255, 4946-4950. Hasilik, A., and Tanner, W. (1978). Carbohydrate moieties of carboxypeptidase Y and perturbation of its biosynthesis. Eur. J . Biochem. 91, 567-571. Hasilik, A., Waheed, A., and von Figura, K. (1981). Enzymatic phosphorylation of lysosomal enzymes in the presence of UDP-N-acetylglucosamine: Absence of activity in I-cell fibroblasts. Biochem. Biophys. Res. Commun. 98, 761-767. Hauri, H. P., Quaroni, A,, and Isselbacher, K . J. (1979). Biosynthesis of intestinal plasma membrane: Post-translational route and cleavage of sucrase-isomaltase. Proc. Narl. Acad. Sci. U.S.A. 76, 5183-5186. Hauri, H. P., Wacker, H., Rickli, E. E., Bigler-Meier, B., Quaroni, A,, and Semenza, G. (1982).

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

285

Biosynthesis of sucrase isomaltase. Purification and NHz-terminal amino acid sequence of thc rat sucrase isomaltase precursor (pro-sucrase-isomaltase) from fetal intestinal transplants. J . Bid. Chem. 257, 4522-4528. Helenius, A , . Mellman. I . . Wall, D., and Hubbard, A. (1983). Endosomes. Trends Biochem. Sci. 8, 245-250. Herzog, V., and Farquhar, M. G . (1977). Luniinal membrane retrieved after exocytosis reaches most Golgi cisternae in secretory cells. Pror. Natl. Acad. Sci. U . S . A . 74, 5073-5077. Herzog, V., and Miller, F. (1979). Membrane retrieval in epithelial cells of isolated thyroid follicles. Eur. J . Cell B i d . 19, 203-215. Hickman, S., and Neufeld. E. F. (1972). A hypothesis for I-cell disease: Defective hydrolases that do not enter the lysosomes. Biochem. Biophys. Res. Cornmun. 49, 992-999. Hirano, H., Parkhouse, B., Nicolson. G. L. Lennox. E. S . . and Singer, J. S. ( 1972). Distribution of saccharide residues on membrane fragments from a myeloma-cell homogenate: Its implication membrane biogenesis. Proc. Natl. Acud. Sri. U . S . A . 69, 2945-2949. Hiti, A. L., Davis, A. R., and Nayak, D. P. (1981). Complete sequence analysis shows that the hemagglutinins of the HO and H2 subtypes of human influenza are closely related. Virology 111, 1 13- 124. Holmes, I. H. (1983). Rotaviruses. In "The Reoviridae" (W. K . Joklik, ed.), pp. 359-423. Plenum, New York. Hubbard, S. C., and Ivatt, R. J. (1981). Synthesis and processing of asparagine-linked oligosaccharides. Annu. Rev. Biochem. 50, 555-583. Hunt, L. A., Etchison, J. A,, and Summers, D. F. (1978). Oligosaccharide chains are trimmed during synthesis of the envelope glycoprotein of vesicular stomatitis virus. Proc. Nut/. Acad. Sci. U . S . A . 75, 754-758. Ito, A , , and Palade, G. E. (1978). Pmence of NADPH-cytochrome P-450 reductase in rat liver Golgi membranes. J . Cell B i d . 79, 590-597. Jamieson, J. D., and Palade, G . (1971). Synthesis. intracellular transport and discharge of secretory proteins in stimulated exocrine cells. J . Cell Biol. 50, 135-158. Janiieson, J. D., and Palade, 0 . E. (1977). Production of secretory proteins in animal cells. In!. Cell Biol., pp. 308-317. Jelsema, C. L., and Morre, D. J. (1978). Distribution of phospholipid biosynthetic enzymes among cell components of rat liver. J . B i d . Chem. 253, 7960-7971. Johnson, D. C., and Schlesinger, M. J. (1980). Vesicular stomatitis virus and Sindbis virus glycoprotein transport to the cell surface is inhibited by ionophores. Virology 103, 407-424. Jokinen, M., Gahmberg, C. G . , and Anderson, L. A. (1979). Biosynthesis of the major human red cell sialoglycoprotein, glycophorin A, in a continuous cell line. Nature (London)279,604-607. Kaariainen, L., Hashimoto, K . , Saraste, J., Virtanen. I., and Penttinen, K . (1980). Monensin and FCCP inhihit the intracellular transport of alphavirus membrane glycoproteins. J . Cell B i d . 87, 783-79 I . Katz, F. N., Rothman, J . E., Knipe, D. M., and Lodish, H. F. (1977). Membrane assembly: Synthesis and intracellular processing of the vesicular stomatitis glycoprotein. J . Suprumol. Strurt. 7, 353-370. Kehry, M., Ewald, S . , Douglas, R., Sibley, C.. Raschke, W., Fambrough. D., and Hood, L. (1980). The immunoglobulin chains of membrane- bound and secreted IgM molecules differ in their C-terminal segments. Cell 21, 393-406. Kenny, A. J . , and Maroux, S. (1982). Topology of microvillar membrane hydrolases of kidney and intestine. Physiol. Rev. 62, 91-128. Knipe, D. M., Baltimore, D., and Lodish. H. F. (1977a). Maturation of viral proteins in cells infected with temperature-sensitive mutants of vesicular stomatitis virus. J . Virol. 21, 11491158.

286

ENRIQUE RODRIGUEZ-BOULAN ET AL.

Knipe, D. M., Lodish, H. F., and Baltimore, D. (1977b). Localization of two cellular forms of the vesicular stomatitis virus glycoprotein. J. Virol. 21, 1121-1 127. Kondor-Koch, C., Riedel, H., Soderberg, K., and Garoff, H. (1982). Expression of the structural protein of Semliki forest virus from cloned cDNA microinjected into the nucleous of baby hamster kidney cells. Proc. Narl. Acad. Sci. U.S.A. 79, 4525-4529. Kondor-Koch, C., Burke, B., and Garoff, H. (1983). Expression of Semliki Forest virus proteins from cloned complementary DNA. I. Fusion activity of the spike glycoprotein. J. CellBiol. 97, 644-65 I . Kraehenbuhl, J. P., and Kuhn, L. (1978). Transport of immunoglobulins across epithelia. In “Transport of macromolecules in cellular systems” (S. C. Silverstein, ed.), pp. 213-228. Dahlem Konferenzen, Berlin. Krangel, M. S., Om,H. T., and Strominger, I. L. (1979). Assembly and maturation of HLA-A and HLA-B in vivo. Cell 18, 979-991. Kreibich, G . , Czako-Graham, M., Grebenau, R., Mok, W., Rodriguez-Boulan, E., and Sabatini, D. D. (1978a). Characterization of the ribosomal binding site in rat liver rough microsomes: Ribophorins I and 11, two integral membrane proteins related to ribosome binding. J . Supramol. Strucr. 8, 279-302. Kreibich, G., Freinstein, C. M., Pereyra, B. N., Ulrich, B. L., and Sabatini, D. D. (1978b). Proteins of rough microsomal membranes related to ribosome binding. 11. Cross-linking of bound ribosomes to specific membrane proteins exposed at the binding sites. J. Cell Eiol. 77, 488-506. Kreibich, G., Ulrich, B. L., and Sabatini, D. D. (1978~).Proteins of rough microsomal membranes related to ribosome binding. I. Identification of ribophorins I and 11, membrane proteins characteristic of rough microsomes. J . Cell Eiol. 77, 464-487. Kreil, G. (1981). Transfer of proteins across membranes. Annu. Rev. Eiochem. 50, 317-348. Lefay, F. (1974). Envelope proteins of vesicular stomatitis virus: Effects of temperature sensitive mutations in complementation groups I11 and V. J . Virol. 14, 1220-1228. Lazarow, P. (1980). Functions and biosynthesis of lysosomes. Int. Cell Eiol. pp. 633-639. Lazarowitz, S. G., and Choppin, P. W. (1975). Enhancement of the infectivity of influenza A and B viruses by proteolytic cleavage of the hemagglutinin polypeptide. Virology 68, 440-454. Leblond, C. P., and Bennett, G. (1977). Role of the Golgi apparatus in terminal glycosylation. Int. Cell B i d . pp. 326-336. Ledford, B. E., and Davis, D. F. (1983). Kinetics of serum protein secretion by cultured hepatoma cells: Evidence for multiple secretory pathways. J . Eiol. Chem. 258, 3304-3308. Lehle, L., Cohen, R. E., and Ballou, C. E. (1979). Carbohydrate structure of yeast invertase. J . Eiol. Chem. 254, 12209-12218. Lenard, J. (1978). Virus envelopes and plasma membranes. Annu. Rev. Eiophys. Bioeng. 7 , 139165. Lenard, I., and Compans, R. W. (1974). The membrane structure of lipid-containing viruses. Eiochim. Eiophys. Acta 344, 5 1-94. Lennarz, W. J. (1980). In “The Biochemistry of Glycoproteins and Proteoglycans” (W. J. Lennarz, ed.), pp. 35-84. Plenum, New York. Lingappa, V. R., Katz, F. N., Lodish, H. F., and Blobel, G. (1978). A signal sequence for the insertion of a transmembrane glycoprotein. Similarities to the signals of secretory proteins in primary structure and function. J . Biol. Chem. 253, 8667-8670. Lodish, H. F., and Kong, N. (1984). Glucose removal from N-linked oligosaccharides is required for efficient maturation of certain secretory glycoproteins from the rough endoplasmic reticulum to the Golgi complex. J. Cell Eiol. 98, 1720-1729. Lodish, H. F., Braell, W. A., Schwartz, A. L., Strous, G . J. A. M., and Zilberstein, A. (1981). Synthesis and assembly of membrane and organelle proteins. Inr. Rev. Cyrol. Suppl. 12, 247307.

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

287

Lodish, H. F., Kong, N., Snider. M., and Strouss. G . J. A. M. (1983). Hepatoma secretory proteins migrate from the rough endoplasmic reticulum to the Golgi at characteristic rates. Nature (London) 304, 80-83. Lohmeyer, J., and Klenk. H. D. (1979). A mutant of influenza virus with a temperature sensitive defect in the posttranslational processing of the hemagglutinin. Virology 93, 134- 145. Lojda, 2. (1974). Cytochemistry of enterocytes and of other cells in the mucous membrane of the small intestine. Biomemhrunes 4A, 43- 122. Louvard, D., Reggio, H., and Warren, G. (1982). Antibodies to the Golgi complex and rough endoplasmic reticulum. J . Cell B i d . 92, 92-107. Lusis, A. J . , and Swank, R. T. (1980). Regulation of location of intracellular proteins. In “Cell Biology, a Comprehensive Treatise” (D. M. Prescott and L. Goldstein, eds.), Vol. 4, pp. 339391. Academic Press, New York. McKusick, V. A , , Neufeld, E. F., and Kelly, T. T. (1972). Two disorders of lysosomal enzyme localization mucolipidosis I1 (ML 11, I-cell disease, inclusion cell disease and mucolipidosis I11 (MLIII), pseudo-Hurler polydistrophy). In “The Metabolic Basis of Inherited Disease” (J. B. Stanbury, J. B. Wyngaarden. and D. S. Fredrickson, eds.), pp. 1299-1302. McGraw Hill, New York. Marchesi, V. T . , Furthmayr, H., and Tomita. M. (1976). The red cell membrane. Annu. Rev. Biochem. 45, 667-698. Marsh, M., Bolzau, E., and Helenius, A. (1983). Penetration of Semliki Forest Virus from acidic prelysosomal vacuoles. Cell 32, 93 1-940. M a t h , A , , and Simons. K. (1983). Reduced temperature prevents transfer of a membrane glycoprotein to the cell surface but not terminal glycosylation. Cell 34, 233-243. M a t h . K., Bainton, D. F., Pesonen, M., Louvard, D., Genty, N., and Simons, K. (1983). Transepithelial transport of a viral membrane glycoprotein implanted into the apical plasma membrane of Madin-Darby Canine Kidney cells. I. Morphological evidence. J . Cell B i d . 97, 627-637. Maxfield, F. R. (1982). Weak bases and ionophores rapidly and reversibly raise the pH of endocytic vesicles in cultured mouse fibroblasts. J. Cell B i d . 95, 676-681. Meiss. H.K., Green. R.. and Rodriguez-Boulan. E. J. (1982). Lectin resistant mutants of polarized epithelial cells. Mol. Cell. Biol. 2, 1287-1294. Meyer, D. I.. and Dobberstein, B. (1980a). A membrane component essential for vectorial translocation of nascent proteins across the endoplasmic reticulum: Requirement for its extraction and reassociation with the membrane. J . Cell Biol. 87, 498-502. Meyer, D. I . , and Dobberstein, B. (1980b). Identification and characterization of a membrane component essential for the translocation of nascent proteins across the membrane of the endoplasmic reticulum. J . Cell B i d . 87, 503-508. Meyer, D. I., Krause, E., and Dobberstein. B. (1982). Secretory protein translocation across membranes-the role of the “docking protein.” Nature (London) 297, 647-650. Milstein. C., Brownlee, G. G., Harrison, T. M., and Mathews, M. B. (1972). A possible precursor of immunoglobulin light chains. N u m e (London) New Biol. 239, 117-120. W., Verhoeyen, M., Devos, R., Saman, E., Fang, R., Huylebroeck, D., Fiers, W . , Min IOU, Threlfall, G . , Barber, C., Carey, N., and Emtage, S. (1980). Complete structure of the hemagglutinin gene from the human influenza AIVictoriai3175 (H3N2) strain as determined from cloned DNA. Cell 19, 683-696. Misek, D. E., Bard, E., and Rodriguez-Boulan, E. (1984). Biogenesis of epithelial cell polarity. Intracellular sorting and vectorial exocytosis of an apical plasma membrane glycoprotein. Cell. 39, 537-546. Misfeldt, D. S., Hamamoto. S. T . , and Pitelka, D. R. (1976). Transepithelial transport in cell culture. Proc. Nutl. Acud. Sci. U.S.A. 73, 1212-1216. Moore, H.-P. Gumbiner, B.. and Kelly, R. B. (1983a). A subclass of proteins and sulfated mac-

288

ENRIQUE RODRIGUEZ-BOULAN ET AL.

romolecules secreted by AtT-20 (mouse pituitary tumor) cells is sorted with adrenocorticotropin into dense secretory granules. J. Cell Biol. 97, 810-817. Moore, H.-P., Gumbiner, B., and Kelly, R. B. (1983b). Chloroquine diverts ACTH from a regulated to a constitutive secretory pathway in AtT-20 cells. Nature (London) 302, 434-436. Moriarty, A. M., Hoyer, B. H., Shih, J. W. K., Gehrin, J. L., and Hamer, D. H. (1981). Expression of the hepatitis B virus surface antigen gene in cell culture by using a simian virus 40 vector. Proc. Nutl. Acad. Sci. U.S.A. 78, 2606-2610. Mostov, K. E., Friedlander, M., and Blobel, G. (1984). The receptor for transepithelial transport of IgA and IgM contains multiple immunoglobulin-like domains. Nature (London) 308, 37-43. Mudd, I. A. (1974). Glycoprotein fragment associated with vesicular stomatitis virus after proteolytic digestion. Virology 62, 573-577. Mulligan, R. C., and Berg, P. (1980). Expression of a bacterial gene in mammalian cells. Science 209, 1422-1427. Nakamura, K., and Compans, R. W. (1977). The cellular site of sulfation of influenza viral glycoproteins. Virology 79, 381-392. Neufeld, E. F., Sando, G. N., Garvin, A. J . , and Rome, L. H. (1977). The transport of lysosomal enzymes. J. Supramol. Struct. 6, 95-101. Noda, M., Takahashi, H., Tanabe, T., Toyosato, M., Furutani, Y.,Hirose, T., Asai, M., Inayama, S., Miyata, T., and Numa, S. (1982). Primary structure of alpha subunit percursor of Torpedo californica acetylcholine receptor deduced from cDNA sequence. Nature (London) 299, 793797. Noda, M., Furutani, Y.,Takahashi, H., Toyosato, M., Tanabe, T., Shimizu, S., Kikyotani, S., Kayano, T., Hirose, T., Inayama, S . , and Numa, S. (1983a). Cloning and sequence analysis of calf cDNA and human genomic DNA encoding -subunit precursor of muscle acetylcholine receptor. Nature (London) 305, 818-823. Noda, M., Takahashi, H . , Tanabe, T., Toyosato, M., Kikyotani, S., Furutani, Y., Hirose, T., Takashima, H., Inayama, S., Miyata, T . , and Numa, S. (1983b). Structural homology of Torpedo californica acetylcholine receptor subunits. Nature (London) 302, 528-532. Noda, M., Takahashi, H., Tanabe, T . , Toyosato, M., Kikyotani, S., Hiose, T . , Asai, M., Takashima, H., Inayama, S., Miyata, T., and Numa, S. (1983~).Primary structures of beta and delta subunit precursors of Torpedo californica acetylcholine receptor deduced from cDNA sequences. Nature (London) 301, 251-255. Novick, P., and Schekman, R. (1979). Secretion and cell surface growth are blocked in a temperature-sensitive mutant of Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. U S A . 76, 1858-1862. Novick, P., and Schekman, R. (1983). Export of major cell surface proteins is blocked in yeast secretory mutants. J . Cell B i d . 96, 541-547. Novick, P., Field, C., and Schekman, R. (1980). Identification of 23 complementation groups required for post-translational events in the yeast secretory pathway. Cell 21, 205-215. Novick, P.,Ferro, S., and Scheckman, R. (1981). Order of events in the yeast secretory pathway. Cell 25, 461-469. Novikoff, A. B. (1976). The endoplasmic reticulum: A cytochemist’s view (a review). Proc. Natl. Acad. Sci. U.S.A. 73, 2781-2787. Novikoff, A. B., and Novikoff, P. M. (1977). Cytochemical contribution to differentiating GERL from Golgi apparatus. Histochem. J . 9, 525-552. Novikoff, P. M., Novikoff, A. B., Quintana, N., and Hauw, J. J. (1971). Golgi apparatus, GERL, and lysosomes of neurons in rat dorsal root ganglia, studied by thick section and thin section cytochemistry. J . Cell Biol. 50, 859-886. Onishi, H. R., Tkacz, J. S., and Lampen, J. 0. (1979). Glycoprotein nature of yeast alkaline phosphatase. Formation of an active enzyme in the presence of tunicamycin. J. B i d . Chem. 254, 11943-1 1952.

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

289

Op den Kamp, J. A. F. (1979). Lipid asymmetry in membranes. Annu. Rev. Biochem. 48, 47-71. Owen. M. I . , Kissonergis, A . M.. Lodish, H. F., and Crumpton, M. J . (1981). Biosynthesis and maturation of HLA-DR antigens in vivo. J. B i d . Chrm. 256, 888778893, Pagano, R. E., and Longmuir, K. J . (1983). Intracellular translocation and metabolism of fluorescent lipid analogues in cultured mammalian cells. Trends Biochem. Sci. 8, 157- 161. Paladc, G . (1975). Intracellular aspects of the process of protcin secretion. Science 189, 347-358. Pappenheimer, A. M., Jr. (1978). Diphtheria: Molecular biology of an infectious process. Trends Eiochetn. Sci. 3, N220-N224. Parham. P., Alpert, B . N., Orr, H. T., and Strominger, I . L. (1977). Carbohydrate moiety of HLA antigens. Antigenic properties and aminoacid sequences around thc site of glycosylation. J. Biol. Chem. 252, 7555-7567. Parodi, A. I . , and Leloir, L. F. (1979). The role of lipid intermediates in the glycosylation of proteins in the eukaryotic cell. Biochim. Biophys. Acta 559, 1-37. Pastan, I . , and Willingham, M. C. (1983). Receptor mediated endocytosis, coated pits, receptosomes and the Golgi. Trends Biochem. Sci. 8, 250-254. Pearson, G. R., and McNulty. M. S. (1979). Ultrastructural changes in small intestinal epithelium of neonatal pigs infected with pig rotavirus. Arch. Virol. 59, 127-136. Pease. L. R., Nathenson, S . C . ,and Leinward. L. A. (1982). Mapping class I gene sequences in the major histocompatibility complex. Nuture (London) 298, 382-385. Pesonen, M., and Simons, K. (1983). Transepithelial transport of a viral membrane glycoprotein implanted into the apical plasma membrane of Madin-Darby Canine kidney cells. 11. Immunological quantitation. J . Cell B i d . 97, 638-643. Pisam, M.. and Ripoche, P. (1976). Redistribution of surface macromolecules in dissociated epithelial cells. J . Cell B i d . 71, 907-920. Ploegh, H. L., Cannon, L. E., and Strominger, J . L. (1979). Cell free translation of the mRNAs for the heavy and light chains of HLA-A and HLA-B antigens. Proc. Nu//. Acud. Sci. U.S.A. 76, 2273-2277. Ploegh, H. L., Orr, H. T., and Strominger, J. L. (1981). Major histocompatibility antigens: The human (HLA-A. -B, -C) and murine (H-2K. H-2D) class I molecules. Cell 24, 287-299. Pollack, L., and Atkinson, P. H. (1983). Correlation of glycosylation forms with position in aminoacid sequence. J . Cell Biol. 97, 293-300. Porter, A. G., Carber, C . , Carey, N. H., Hallewell, R. A,, Threlfall, G., and Emtage, J . S. (1979). Complete nucleotide sequence of an influenza virus hemagglutinin gene from cloned DNA. Nature (London) 282, 471-477. Porter, K. R., Claude, A,, and Fullam, E. (1945). A study of tissue culture cells by electron . 81, 233-244. microscopy. J. E . K ~Med. Poyton, R. 0. (1983). Memory and membranes: The expression of genetic and spatial memory during the assembly of organelle macrocompartments. In "Modem Cell Biology," Vol. 2, pp. 15-72. Liss, New York. Rachubinski, R. A., Verma, D. P. S., and Bergeron, I. J . M. (1980). Synthesis of rat liver microsomal cytochrome b5 by free ribosomes. J . Cell Eiol. 84, 705-716. Reitman, M.. and Kornfeld, S. (I98 I ) . UDP-N-acetylglucosamine:glycoprotein N-acetylglucosamine- I-phosphotransferase: Proposed enzyme for the phosphorylation of the high mannose oligosaccharide units of lysosomal enzynics. J . Eiol. Chem. 256, 4275-4281. Reitman, M., Varki, A,, and Kornfeld, S. (1981). Fibroblasts from patients with I-cell disease and pseudo-Hurler polydistrophy are deficient in UDP-N-acetylg1ucosamine:glycoproteinN-acetylglucosaminylphosphotransferase activity. J . C'lin. fnvest. 67, 1574- 1579. Renooij, W.. van Golde, L. M. G., Zwaal, R. F. A . , Roelofsen, B . , and van Deenen, L. L. M. (1974). Preferential incorporation of fatty acids at the inside of human erythrocyte membranes. Biochim. Biophvs. Acta 363, 287-292. Ribgy, P. W. J . (1982). Expression of cloned genes in cukaryotic cells using vector systems derived

290

ENRIQUE RODRIGUEZ-BOULAN ET AL.

from viral replicons. In “Genetic Engineering’’ (R. Williamson, ed.), Vol. 3, pp. 83-141. Academic Press, New York. Rindler, M. J., Ivanov, I. E., Rodriguez-Boulan, E., and Sabatini, D. D. (1982). Biogenesis of epithelial cell plasma membranes. Ciba Found. Symp. 92, 184-202. Rindler, M. J . , Ivanov, I. E., Plesken, H., Rodriguez-Boulan, E., and Sabatini, D. D. (1984). Viral glycoproteins destined for apical or basolateral plasma membrane domains traverse the same Golgi apparatus during their intracellular transport in doubly infected Madin-Darby Canine Kidney cells. J. Cell B i d . 98, 1304-1319. Robbins, P. W., Hubbard, C., Turco, S . J., and Wirth, D. F. (1977). Proposal for a common oligosaccharide intermediate in the synthesis of membrane glycoproteins. Cell 12, 893-900. Rodewald, R. (1973). Intestinal transport of antibodies in the newborn rat. J . Cell Eiol. 58, 189211. Rodewald, R. (1980). Distribution of immunoglobulin G receptors in the small intestine of the young rat. J. Cell Eiol. 85, 18-32. Rodriguez-Boulan, E. (1983). Membrane biogenesis, enveloped RNA viruses and epithelial polarity. In “Modern Cell Biology,” Vol. I, pp. 119-170. Liss, New York. Rodriguez-Boulan, E., and Pendergast, M. (1980). Polarized distribution of viral envelope proteins in the plasma membrane of infected epithelial cells. Cell 20, 45-54. Rodriguez-Boulan, E., and Sabatini, D. D. (1978). Asymmetric budding of viruses in epithelial monolayers: A model system for study of epithelial polarity. Proc. Natl. Acad. Sci. U.S.A. 75, 5071 -5075. Rodriguez-Boulan, E., Kreibich, G., and Sabatini, D. D. (1978a). Spatial orientation of glycoproteins in membranes of rat liver rough microsomes. I. Localization of lectin-binding sites in microsomal membranes. J. Cell Biol. 78, 874-893. Rodriguez-Boulan, E., Sabatini, D. D., Pereyra, B. N., and Kreibich, G. (1978b). Spatial orientation of glycoproteins in membranes of rat liver rough microsomes. 11. Transmembrane disposition and characterization of glycoproteins. J. Cell Biol. 78, 894-909. Rodriguez-Boulan, E., Paskiet, K. T., and Sabatini, D. D. (1983). Assembly of enveloped viruses in MDCK cells: Polarized budding from single attached cells and from clusters of cells in suspension. J . Cell Biol. 96, 866-874. Rodriguez-Boulan, E., Paskiet, K. T., Salas, P. J. I., and Bard, E. (1984). Intracellular transport of influenza virus hemagglutinin to the apical surface of Madin-Darby Canine Kidney cells. J . Cell Biol. 98, 308-319. Rogers, J., Early, P., Carter, C., Calame, K., Bond, M., Hood, L., and Wall, R. (1980). Two mRNAs with different 3‘ ends encode membrane-bound and secreted forms of immunoglobulin p chain. Cell 20, 303-312. Rose, J. K., and Bergman, J. E. (1982). Expression from cloned cDNA of cell-surface secreted forms of the glycoprotein of vesicular stomatitis virus in eukaryotic cells. Cell 30, 753-762. Rose, J. K., and Bergman, J. E. (1983). Altered cytoplasmic domains affect intracellular transport of the vesicular stomatitis virus glycoprotein. Cell 34, 513-524. Rose, J. K.,Welch, W. J., Sefton, B. M.,Esch, F. S., and Ling, N. C. (1980). Vesicular stomatitis virus glycoprotein is anchored in the viral membrane by a hydrophobic domain near the COOH terminus. Proc. Natl. Acad. Sci. U.S.A. 77, 3884-3888. Rosenfeld, M. G., Kreibich, G., Popov, D . , Kato, K., and Sabatini, D. D. (1982). Biosynthesis of lysosomal hydrolases: Their synthesis in bound polysomes and the role of co- and post-translational processing in determining their subcellular distribution. J. Cell B i d . 93, 135- 143. Roth, J., and Berger, E. G. (1982). Immunocytochemical localization of galactosyl transferase in HeLa cells: Codistribution with thiamine pyrophosphatase in trans golgi cisternae. J. Cell Eiol. 92, 223-229. Roth, M. G., Fitzpatrick, J., and Compans, R. W. (1979). Polarity of influenza and vesicular

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

291

stomatitis virus maturation in MDCK cells: Lack of requirement for glycosylation of viral glycoproteins. Proc. Natl. Acad. Sci. U.S.A. 76, 6430-6434. Roth, M. G., Compans, R. W., Giusti, L., Davis, A. R., Nayak, D. P., Gething, M. I., and Sambrook, J. (1983). Influenza virus hemagglutinin expression is polarized in cells infected with recombinant SV40 viruses carrying cloned hemagglutinin DNA. Cell 33, 435-443. Rothman, J. E. (1981). The Golgi apparatus: Two organelles in tandem. Science 213, 1212-1219. Rothman, J. E., and Fine, R. E. (1980). Coated vesicles transport newly synthesized membrane glycoproteins from endoplasmic reticulum to plasma membrane in two successive stages. P roc. Natl. Acad. Sci. U.S.A. 11, 780-784. Rothman. J. E., and Fries, E. (1981). Transport of newly synthesized vesicular stomatitis viral glycoprotein to purified Golgi membranes. J . Cell Biol. 89, 162-168. Rothman, J. E., and Lenard, J. (1977). Membrane asymmetry. The nature of membrane asymmetry provides clues to the puzzle of how membranes are assembled. Science 195, 743-753. Rothman, J . E., and Lodish, H. F. (1977). Synchronized transmembrane insertion and glycosylation of a nascent membrane protein. Nature (London) 269, 775-780. Rothman. J. E., Bursztyn-Pettegrew, H., and Fine, R. E. (1980). Transport of the membrane glycoprotein of vesicular stomatitis virus to the cell surface in two stages by clathrin coated vesicles. J. Cell Biol. 86, 162-171. Sabatini, D. D., Kreibich, G., Morinioto, T., and Adesnik, M. (1982). Mechanisms for the incorporation of proteins in membranes and organelles. J . Cell Biol. 92, 1-22. Salas, P. J. I., Vega-Salas, D. E., Cereijido, M.. and Rodriguez-Boulan, E. (1985). Role of cytoskeleton in epithelial cell polarity. Effect of cytochalasin D and Colchicine on the polarized budding of enveloped viruses from Madin-Darby Canine Kidney (MDCK) cells. Submitted. Schachter, H., and Roseman, S. (1980). Mammalian glycosyl-transferases. In “The Biochemistry of Glycoproteins and Proteoglycans” (W. Lennarz, ed.), pp. 85-160. Plenum, New York. Schekman, R. (1982). The secretory pathway in the yeast. Trends Biochem. Sci. 7, 243-246. Schmidt, M. F. G., and Schlesinger, M. (1979). Fatty acid binding to vesicular stomatitis virus glycoprotein: A new type of post-translational modification of the viral glycoprotein. Cell 17, 8 13-8 19. Schneider, D. L. (1981). ATP-dependent acidification of intact and disrupted lysosomes. Evidence for an ATP-driven proton pump. J . B i d . Chem. 256, 3858-3864. Semenza, G. (1976). Small intestinal disaccharidases: their properties and role as sugar translocators across natural and artificial membranes. In “The Enzymes of Biological Membranes” (A. Martonosi, ed.). Plenum, New York. Shida, H., and Matsumoto, S. (1983). Analysis of the hemagglutinin glycoprotein from mutants of vaccinia virus that accumulates on the nuclear envelope. Cell 33, 423-434. Simons, K., and Garoff, H. (1980). The budding mechanisms of enveloped animal viruses. J . Gen. Virol. 50, 1-2 I . Singer, P. A., and Williamson, A. R. (1980). Different species of messenger RNA encode receptor and secretory IgM chains differing at their carboxyl termini. Nature (London) 285, 294-299. Sleigh, M. J . , Both, G . W., Brownlee. G . G., Bender, V. J., and Moss, B. A. (1980). The haemagglutinin gene of influenza A virus: nucleotide sequence analysis of cloned DNA copies. In “Structure and variation in influenza virus” ( W . (3. Laver and G. M. Air, eds.), pp. 69-78. Elsevier, Amsterdam, Sly, W. (1982). The uptake and transport of lysosomal enzymes. In “The Glycoconjugates” (M. I. Horowitz, ed.), Vol. IV, pp. 3-25. Academic Press. New York. Solari, R., and Kraehenbuhl, J. P. (1984). Biosynthesis of the IgA antibody receptor: A model for the transepithelial sorting of a membrane glycoprotein. Cell 36, 61-71. Sprague, J . . Condra, J. H., Arnheiter, H., and Lazzarini, R. A. (1983). Expression of a recombinant DNA gene coding for the vesicular stomatitis virus nucleocapsid protein, J . Virol. 45,773-781,

292

ENRIQUE RODRIGUEZ-BOULAN ET AL.

Srinivas, R. V., Alonso-Caplen, F. V., and Compans, R. W. (1983). Glycosylation and transport of two viral membrane glycoproteins that are anchored by amino-terminal signal peptide sequences. J. Cell Biol. 97, 444a. Steck, T. L. (1978). The band 3 protein of the human red cell membrane: A review. J. Supramol. Struct. 8, 311-324. Steiner, D. F., Clark, J. L., Nolan, C., Rubenstein, A. H., Margoliash, E., Melani, F . , and Oyer, P. E. (1970). The biosynthesis of insulin and some speculations regarding the pathogenesis of human diabetes. In “The Pathogenesis of Diabetes Mellitus” (E. Cerasi and R. Luft, eds.), pp. 123- 132. Almqvist & Wiksell, Stockholm. Steinman, R. M., Brodie, S. E., and Cohn, 2. A. (1976). Membrane flow during pinocytosis. A stereologic analysis. J. Cell Biol. 68, 665-687. Steinman, R. M., Mellman, I. S., Muller, W. A,, and Cohn, 2. A. (1983). Endocytosis and the recycling of plasma membrane. J. Cell B i d . 96, 1-27. Strawser, L. D., and Touster, 0. (1980). The cellular processing of lysosomal enzymes and related proteins. Rev. Physiol. Biochem. Pharmacol. 87, 169-210. Strous, G. J. A. M., and Berger, E. C. (1982). Biosynthesis, intracellular transport and release of the Golgi enzyme galactosyl transferace (lactose synthetase A protein) in HeLa cells. J. Biol. Chem. 257, 7623-7628. Strous, G. J. A. M., and Lodish, H. F. (1980). Intracellular transport of secretory and membrane proteins in hepatoma cells infected by vesicular stomatitis virus. Cell 22, 709-717. Strous, G . J., van Kerkhof, P., Willemsen, R., Geuze, H. J., and Berger, E. C. (1983). Transport and topology of galactosyl transferase in endomembranes of HeLa cells. J. Cell Biol. 97, 723727. Sturman, L. S., and Holmes, K. V. (1983). The molecular biology of corona viruses. Adv. Virus Res. 28, 35-112. Sveda, M. M., and Lai, C. J. (1981). Functional expression in primate cells of cloned cDNA coding for the hemagglutinin surface glycoprotein of influenza virus. Proc. Natl. Acud. Sci. U.S.A. 78, 5488-5492. Sveda, M. M., Markoff, L., and Lai, C. J. (1982). Cell surface expression of the influenza virus hemagglutinin requires the hydrophobic carboxy-terminal sequences. Cell 30, 649-656. Tabas, I., and Kornfeld, S. (1980). Biosynthetic intermediates of beta-glucuronidase contain high mannose oligosaccharides with blocked phosphate residues. J. Biol. Chem. 255, 66336639. Tabas, I . , Schlesinger, S., and Kornfeld, S. (1978). Processing of high mannose oligosaccharides to form complex type oligosaccharides on the newly synthesized polypeptides of the vesicular stomatitis virus G protein and the IgG heavy chain. J. Biol. Chem. 253, 716-722. Tartakoff, A. M. (1980). The Golgi complex: Crossroads for vesicular traffic. Inf. Rev. Exp. Parhol. 22, 228-25 I. Tartakoff, A. M. (1983). Perturbation of vesicular traffic with the carboxylic ionophore monensin. Cell 32, 1026-1028. Tartakoff, A. M., and Vassalli, P. (1977). Plasma cell immunoglobulin secretion. Arrest is accompanied by alterations in the Golgi complex. J. Enp. Med. 146, 1332-1345. Tartakoff, A. M., and Vassalli, P. (1983). Lectin binding sites as markers of Golgi subcompartments: Proximal to distal maturation of oligosaccharides. J. Cell Biol. 97, 1243-1248. Tkacz, J. S., and Lampen, J. 0. (1972). Wall replication in Saccharomyces species: use of fluorescein-conjugated concanavalin A to reveal the site of mannan insertion. J. Gen. Microbiol. 72, 243-247. Tomita, M., and Marchesi, V. T. (1975). Aminoacid sequence and oligosaccharide attachment sites of human erythrocyte glycophorin. Proc. Narl. Acad. Sci. U.S.A. 72, 2964-2968.

6. PROTEIN SORTING IN THE SECRETORY PATHWAY

293

Tycko, B., and Maxfield, F. R. (1982). Rapid acidification of endocytic vesicles containing alpha-2 macroglobulin. Cell 28, 643-65 I . Van Golde, L. M . G . , Raben, I . , Batenburg, J. J., Fleischer. B., Zanibrano, F., and Fleischer, S. (1974). Biosynthesis of lipids in Golgi complex and other subcellular fractions from rat liver. Biochini. Biophvs. Actu 360, 179- 192. Varghese, J . N., Laver, W. G . , and Colnian, P. M. (1983). Structure of the influenza virus glycoprotein antigen neuraminidase at 2.9 8, resolution. Nature (London) 303, 35-40. Varki, A.. and Kornfeld, S. (1980). Identification of a rat liver N-acetylglucosaminyl phosphodiesterase capable of removing “blocking” N-acetylglucosamine residues from phosphorylated high mannose oligosaccharides of lysosomal enzymes. J . Biol. Chem. 255, 8398-8401, Vasalli, P., Tedghi, R., Lisowska-Bernstein, B., Tartakoff, A,, and Jaton, J.-C. (1979). Evidence for hydrophobic region within heavy chains of mouse B lymphocyte membrane-bound IgM. Proc. Nail. Acad. Sci. U.S.A. 76, 5515-5519. Verkleij, A. J., Zwaal, R. F. A., Roelofsen, B., Comfurius, P., Kastelijn, D., and van Deenen, L. L. M. (1973). The asymmetric distribution of phospholipids in the human red cell membrane. A combined study using phospholipases and freeze-etch electron microscopy. Biochim. Biophys. Acta 323, 178-193. Von Figura, K., Rey, M., Printz, R., Voss, B., and Ullrich, K. (1979). Effect of tunicamycin on transport of lysosomal enzymes in cultured skin fibroblasts. Eur. J . Biochem. 101, 103-109. Wall, D. A,, Wilson, G . , and Hubbard, A. L. (1980). The galactose-specific recognition system of mammalian liver: The route of ligand internalization in rat hepatocytes. Cell 21, 79-93. Walter, P.. and Blobel, G. (1980). Purification of a membrane associated protein complex required for protein translocation across the endoplasmic reticulum. Proc. Narl. Acad. Sci. U.S.A. 77, 71 12-7116. Walter, P., and Blobel, G. (1981a). Translocation of proteins across the endoplasmic reticulum. 11. Signal recognition protein (SRP) mediates the selective binding to microsomal membranes of in vitro assembled polysomes synthesizing secretory protein. J . Cell Biol. 91, 55 1-556. Walter, P., and Blobel, G. (1981b). Translocation of proteins across the endoplasmic reticulum. 111. Signal recognition protein (SRP) causes signal sequence dependent and site specific arrest of chain elongation that is released by microsomal membranes. J . Cell Biol. 91, 557-561. Walter, P.,and Blobel, G. (1982). Signal recognition particle contains a 7s RNA essential for protein translocation across the endoplasmic reticulum. Nature (London) 299, 691-698. Walter, P., Ibrahimi, I . , and Blobel, G. (1981). Translocation of proteins across the endoplasmic reticulum. I . Signal recognition protein (SRP) binds to in viiro assembled polysomes synthesizing secretory proteins. J . Cell Biol. 91, 545-550. Ward, C. W . , and Dopheide, T. A. (1979). Primary structure of the Hong Kong (H3) hemagglutinin. Br. Med. Bull. 35, 51-56. Waterfield, M. D., Espelie, K., Elder, K., and Skehel, J. J. (1979). Structure of the hemagglutinin of influenza virus. Br. Med. Bull. 35, 57-73. Wehland, J., Willingham, M . C., Gallo, M. G . , and Pastan, I. (1982). The morphologic pathway of exocytosis of the vesicular stomatitis virus G protein in cultured fibroblasts. Cell 28, 831-841. Whaley, W. G., and Dauwalder, M. (1979). The Golgi apparatus, the plasma membrane and functional integration. Int. Rev. Cyrol. 58, 199-245. White, J., Keelian, M . , and Helenius, A. (1983). Membrane fusion proteins of enveloped viruses. Q.Rev. Biophys. 16, 151-195. Wilson, 1. A , , Skehel, J. J . , and Wiley, D. C. (1981). Structure of the hemagglutinin membrane glycoprotein of influenza virus at 3 8, resolution. Nature (London) 289, 366-373. Winter, G . , Fields, S., and Brownlee, G. 0. (1981). Nucleotide sequence of the hemagglutinin gene of a human influenza virus HI subtype. Nature (London) 292, 72-75.

294

ENRIQUE RODRIGUEZ-BOULAN ET AL.

Yost, S. C., Hedgpeth, J . , and Lingappa, V. R . (1983). A stop transfer sequence confers predictable orientation to a previously secreted protein in cell free systems. Cell 34, 759-766. Young, R. W. (1973). The role of Golgi complex in sulfate metabolism. J. Cell B i d . 57, 175-189. Ziomek, C. A,, Schulman, S., and Edidin, M. (1980). Redistribution of membrane proteins in isolated mouse intestinal cells. J. Cell B i d . 86, 849-857. Zuniga, M. C., Malissen, B., McMillan, M., Brayton, P. R., Clark, S. S., Forman, J., and Hood, L. (1983). Expression and function of transplantation antigens with altered or deleted cytoplasmic domains. Cell 34, 535-544.

CURRENT TOPICS IN MEMBRANES AND TRANSPORT, VOLUME 24

Chapter 7

Transport of Proteins into Mitochondria GRAEME A . REID' Department of Biochemistry. Biocenter Universitv of Basel Basel, Switzerland

I. Introduction: Mitochondrial Biogenesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. An Overview of Mitochondrial Protein Import.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Precursor Polypeptides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Mitochondrial Import Receptors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . c. Translocation and Proctssing . . . . . . . . . . . ....................... D. Assembly of Imported Mitochondrial Proteins . . . . . . . . . . . . . . . . . . . . . . . 111. Are Proteins Transported IV. The Molecular Approach ........................................... A. Isolation and Charac .................................. Protein Import. . . . . . . . . . . . B. Isolation and Charac Mitochondrial Polypeptides . . . . ................................ V. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . .......................

1.

295 297 297 299 30 1 312 313 316 317 320 327 329

INTRODUCTION: MITOCHONDRIAL BlOGENESlS

The eukaryotic cell is physically and biochemically divided into several distinct compartments by the presence of intracellular membranes. The organelles delineated by these membranes perform specialized functions, and this specialization is reflected in their polypeptide compositions: most polypeptides are found exclusively in one particular cellular compartment. How is this highly organized distribution generated? A polypeptide translated in the cytosol must find its way to its ultimate destination, be that in the nucleus, the mitochondrion, the cytosol, or elsewhere. What sort of signals are used to control this traffic? 'Present address: Department of Microbiology, University of Edinburgh, Edinburgh, Scotland.

295

Copyright (ill 1985 by Academic Press. Inc. All righrs of reproduction in any form resewed. ISBN 0-12-153324-7

296

GRAEME A. REID

NUCLEAR DNA FIG. 1 . Mitochondria1 biogenesis. A few polypeptides are encoded on mitochondrial DNA, but the majority are encoded on nuclear chromosomes, synthesized in the cytosol, and transported to their particular destination within the mitochondrion.

In this article we shall consider the particular case of mitochondrial biogenesis. Mitochondria contain a small, usually circular genome which has been the focus of much interest in recent years (Borst and Grivell, 1978; Tzagoloff et al., 1979; Dujon, 1981). The complete nucleotide sequence of mitochondrial DNA from man and some other mammals has been determined (Anderson et al., 1981, 1982; Bibb et al., 1981). The mitochondrial genome encodes only a small number of polypeptides (about a dozen in yeasts and mammals, more in higher plants). The majority of mitochondrial polypeptides (about 90% by mass in yeast) are encoded on nuclear chromosomes, translated in the extramitochondrial cytoplasm, and transported into the mitochondria (Schatz and Mason, 1974; Schatz, 1979; Neupert and Schatz, 1981). We would like to know how such a large group of different proteins is directed specifically to the mitochondrion. Further sorting must also take place since the mitochondrion is delimited by two membranes (inner and outer) enclosing two distinct aqueous compartments (matrix and intermembrane space). An imported protein must find its way to the correct intramitochondrial compartment (Fig. 1). The mitochondrial membranes present barriers to proteins: How are polypeptides transported across these barriers? And once internalized by the mitochondrion, these polypeptides must assume an active conformation, perhaps by interacting specifically with other polypeptides. Relatively little is known about this final assembly step. Research from many laboratories has contributed much to provide a general outline of the

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

297

ways by which mitochondria import proteins; current research is broadly aimed at elucidating the molecular mechanisms involved.

II.

AN OVERVIEW OF MITOCHONDRIAL PROTEIN IMPORT

Nuclear-coded mitochondria1 polypeptides are synthesized in the cytoplasm as precursor forms, distinguishable from their mature counterparts. The transport process is initiated by specific binding to receptors on the mitochondrial surface, from which the precursors are translocated into or across the mitochondrial membranes. Most precursors then undergo covalent modification. The maturation pathway is completed by assembly of imported polypeptides into biologically active proteins.

A. Precursor Polypeptides The majority of imported mitochondrial proteins are synthesized as larger precursors. When yeast spheroplasts are pulse-labeled for a short time and then subjected to immunoprecipitation with antibodies against particular mitochondria1 proteins, one generally observes a polypeptide which migrates more slowly on SDS-polyacrylamide gel electrophoresis than does the mature polypeptide (Maccecchini et a / . , 1979a). This larger form disappears upon a subsequent chase because it is converted to the mature protein. Such larger precursors can also be found in vitro by isolating mRNA, translating it in a reticulocyte lysate in the presence of a radioactive amino acid, and again performing immunoprecipitation. The size difference between precursor and mature forms varies widely among mitochondrial proteins and can be as much as 10 kDa (see Hay et al., 1983, for an extensive list of larger precursors). In those cases which have been directly investigated, it has been shown that the extra mass in the precursor is due to an N-terminal polypeptide extension. It is likely that N-terminal extensions will be the rule, but whether modifications also occur at the C-terminus remains to be shown. Several imported mitochondrial polypetides are synthesized without N-terminal extensions. Among these are several proteins of the mitochondrial outer membrane (Freitag et a l . , 1982b; Gasser and Schatz, 1983) and a few from other compartments; these include a matrix protein, 2-isopropylmalate synthase (Gasser et al., 1982b; Hampsey et al., 1983), an inner membrane protein, adenine nucleotide translocator (Zimmermann and Neupert, 1980), and cytochrome c , a component of the intermembrane space (Korb and Neupert, 1978; Zimmermann et al., 1979). Since some proteins can reach their final location without N-

298

GRAEME A. REID

terminal extensions, one may reasonably ask why it is that most imported proteins are made as larger precursors. The fact that the extensions are removed during or shortly after transport suggests that their role is limited to the protein’s biogenesis. It may, for example, be important in providing a “signal” that allows the precursor to be recognized by the mitochondrion. Such signals could be carried entirely within the mature protein sequence in those cases where no larger precursor is made, in a manner analogous to the noncleaved signal sequence of ovalburnin which directs ovalbumin to the endoplasmic reticulum (Meek et al., 1982). It appears that the extramitochondrial precursors of several mitochondria1 proteins are markedly different in conformation from their intramitochondrial mature counterparts. The proteclipid (subunit 9) of Neurospora ATP synthase is an extremely hydrophobic polypeptide, but it is made as a larger precursor in the cytosol. In this case the role of the N-terminal extension may be largely to confer solubility on the protein (Viebrock et al., 1982). A conformational difference between extra- and intramitochondrial forms has been demonstrated with a protein which apparently undergoes no covalent change upon transport into mitochondria. The extramitochondrial precursor of Neurospora adenine nucleotide translocator binds to hydroxyapatite, but it no longer binds after import into mitochondria (Zimmermann and Neupert, 1980). Here the initial translation product is covalently identical to the mature protein, but differs in tertiary structure and location. A very striking example of a conformational difference is seen with the precursor and mature forms of cytochrome c. Antisera raised against apocytochrome c react well with the apocytochrome but not with holocytochrome c. Non-crossreacting antibodies against holocytochrome c could also be raised (Korb and Neupert, 1978). Conformational differences between precursor and mature polypeptides are also suggested by intermolecular associations. Rat ornithine transcarbamylase is a trimeric enzyme with a sedimentation coefficient of 6 S . The in vitro synthesized precursor sediments at 14 S, although the precursor is only 3-4 kDa larger than the mature protein (Miura et al., 1981). It is not known whether the precursor forms large homooligomers or is associated with other polypeptides, nor is it known whether precursor aggregates are biologically important. Aggregates of precursors are found also for the adenine nucleotide translocator (Zirnmennann and Neupert, 1980) and for the a-and P-subunits of the F, component of yeast ATP synthase (A. S . Lewin, S. Ohta, and G . Schatz, unpublished data). The in vivo synthesized precursor of the P-subunit (56 kDa) behaves as a particle of 500 kDa on gel filtration. The mature a-and P-subunits are components of the same enzyme complex, but their precursors behave as distinct species-they can be separated by chromatography on DEAE-cellulose ( S . Ohta, unpublished observations).

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

299

B. Mitochondria1 Import Receptors Only a very specific subset of the proteins synthesized in the cytosol are imported by mitochondria, and proteins destined for mitochondria are not transported across or into other cellular membranes. There must, therefore, be an efficient sorting mechanism allowing interaction of mitochondrial protein precursors with the mitochondrion. If these precursors carry a specific “addressing signal,” there must be a receptor on the mitochondrial surface which recognizes that signal. The best-studied mitochondrial import receptor is that involved in the transport of cytochrome c (Hennig and Neupert, 1981; Hennig er ul., 1983). These detailed studies have exploited the fact that large amounts of precursor proteins can be prepared chemically. Cytochrome c does not undergo proteolytic cleavage upon import into mitochondria, but nevertheless is covalently modified. As with all c-type cytochromes, the mature protein contains a covalently attached heme group; the attachment apparently takes place in the intermembrane space. Thus, the cytochrome c precursor is equivalent to apocytochrome c, which can be prepared by chemical removal of the heme group from mature holocytochrome c. In addition, of course, radioactive apocytochrome c can be prepared in trace amounts by translating mRNA in a cell-free protein-synthesizing system. The apocytochrome c receptor has been detected by its ability to bind radioactively labeled precursor in an in vitro assay. The precursor was synthesized in vitro in a Neurosporu or rabbit reticulocyte cell-free extract in the presence of [35S]methionine,then incubated with isolated mitochondria. Under suitable conditions, the extramitochondrial apocytochrome c was converted to intramitochondrial holocytochrome c. In order to study the binding reaction in the absence of net transport, Hennig and Neupert (198 1) added deuterohemin (which blocks the attachment of heme to apocytochrome c) to the mitochondria before adding precursor proteins. After incubation of in vitro translation products with Neurosporu mitochondria in the presence of deuterohemin, about half of the apocytochrome c was found associated with reisolated mitochondria. Its submitochondrial location was probed by investigating its sensitivity to externally added protease. The labeled apocytochrome c was sensitive to added trypsin whereas endogenous cytochrome c was resistant; the mitochondrial outer membrane should prevent access of the trypsin to cytochrome c in the intermembrane space. Thus, in the presence of deuterohemin, apocytochrome c appears to accumulate on the outer surface of the mitochondrion. Are the sites to which apocytochrome c is bound (1) specific and ( 2 ) relevant to the import pathway? The answer to both questions appears to be yes. First, the apocytochrome c is very tightly bound to the mitochondria but can be released by addition of an excess of chemically prepared apocytochrome c, a finding indicating reversibility as well as specificity of the binding reaction. Second, when the

300

GRAEME A. REID

inhibitory effect of deuterohemin was reversed by addition of an excess of protohemin, apocytochrome c which had been bound at the outer mitochondrial surface was transported into the mitochondria and converted to the holocytochrome (Hennig et al., 1983). Unlabeled, chemically prepared apocytochrome c has been shown also to compete with labeled, in vitro-synthesized apocytochrome c for import into mitochondria (Hennig et al., 1983). Under conditions where the appearance of labeled cytochrome c in the mitochondria was almost completely blocked, the transport of the adenine nucleotide translocator and of the ATP synthase subunit 9 was unaffected by addition of unlabeled apocytochrome c, a finding suggesting that the latter two proteins do not require the apocytochrome c receptor for entry into mitochondria (Zimmermann et al., 1981). The binding of precursors other than cytochrome c to receptors on the mitochondrial surface has been described (Zwizinski et al., 1983; Riezman et al., 1983b). As described in Section II,C, the transport of precursor proteins into or across the mitochondrial inner membrane requires a transmembrane electrochemical potential difference. When “energization” of the inner membrane is blocked by inhibitors of oxidative phosphorylation, precursor polypeptides become associated with the external surface of the mitochondrion. This binding is tight; precursors remain bound during thorough washing of the mitochondria. When these mitochondria are reenergized, the surface-bound precursor becomes internalized. These experiments suggest that the precursor-binding sites can be used for import into the mitochondria (Zwizinski et al., 1983; Riezman et a f . , 1983b). Furthermore, it appears that transport occurs directly from these sitesthe precursor does not dissociate from the mitochondrial surface before translocation. This was shown by importing the mitochondria-bound precursor of Neurospora adenine nucleotide translocator at various dilutions of the mitochondria. No effect of concentration was observed, indicating that import occurs directly from the bound state (Zwizinski er af., 1983). To look in more detail at the precursor binding reaction in the absence of net transport, Riezman et al. (1983b) developed a precursor binding assay with yeast mitochondrial outer membrane vesicles. The isolated vesicles were shown to be sealed and to have the same orientation as outer membrane in intact mitochondria (Riezman et al., 1983a); components normally exposed on the outer mitochondrial surface are also exposed on the outer face of the outer membrane vesicles. The isolated vesicles bind labeled, in vitro-synthesized precursors of mitochondrial proteins. The binding activity is specific to the outer membrane: isolated mitochondrial inner membrane has little or no capacity to bind cytochrome b, precursor. Binding is specific for those proteins destined to be transported into mitochondria: in vitro-synthesized glyceraldehyde-3-phosphatedehydrogenase and hexokinase, two cytosolic proteins, are not bound. Binding to outer membrane vesicles is specific for the precursor forms of mitochondrial proteins:

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

301

proteolytically processed precursors do not bind. Moreover, mature cytochrome b, does not compete with precytochrome b, for binding. The binding activity must be, at least in part, governed by one or more polypeptides of the outer membrane. When the outer membrane vesicles are treated with trypsin under mild conditions, the ability of the vesicles to bind mitochondrial protein precursors is dramatically reduced. When intact mitochondria are similarly subjected to protease treatment, they lose the ability to import proteins. This correlation further suggests that these protease-sensitive binding sites are “import receptors.” One would like to know much more about the properties of this receptor. In particular, is this a general receptor for a large class of imported proteins or is its specificity more restricted? If this receptor is shared by many precursor polypeptides, one would expect that these precursors would compete for binding to the same sites. The necessary experiments have so far been largely impossible for technical reasons. To demonstrate competition one would first need to isolate large amounts of a purified precursor in its native state, as was possible in the special case of cytochrome c (see above). Those precursors which have Nterminal extensions are less readily purified, but work in this direction has begun (see Section IV,A,l). A remarkable feature of the mitochondrial protein import system is that the import receptors and other components of the transport machinery must themselves be imported from the cytoplasm. This is clearly so since rho- yeast, which are unable to synthesize proteins in the mitochondria, still import proteins into mitochondria: the necessary catalysts must therefore be present. It will be interesting to investigate the molecular details of how receptor precursors are recognized: Does a receptor recognize its own precursor?

C. Translocation and Processing Once bound to a receptor site on the mitochondrial surface, precursor polypeptides must be transported into or across the mitochondrial membranes. Since the mitochondrion is composed of four distinct compartments, it is perhaps not surprising that different import pathways exist, though we do not yet know how many, nor how they are organized. We can, at the moment, consider proteins imported to the inner membrane and matrix as one group and outer membrane proteins as another; apparently at least two routes are used to transport proteins to the intermembrane space. 1. ENERGY-DEPENDENT IMPORT Transport of proteins into or across the mitochondrial inner membrane is energy dependent. This was indicated by Nelson and Schatz (1979), who pulse-

302

GRAEME A. REID

labeled yeast spheroplasts in the presence and absence of various energy poisons. They then analyzed the radioactive polypeptides by immunoprecipitation and gel electrophoresis. In the absence of inhibitors, radioactivity appeared in the mature forms of the a-,(3-, and y-subunits of the F,-ATPase and two subunits of the cytochrome bc, complex (ubiquino1:cytochrome c reductase). When CCCP, an uncoupler of oxidative phosphorylation, was present during the labeling, radioactivity remained in the larger precursor forms of these polypeptides. Precursors accumulated in the presence of CCCP are outside the mitochondrion (Reid and Schatz, 1982b). The nature of the energy dependence of polypeptide import has been investigated in more detail using an in vitro transport assay. These experiments involved synthesis of precursor polypeptides in vitro in a reticulocyte lysate in the presence of [35S]methionine, followed by incubation of the labeled precursors with isolated mitochondria. After the incubation, mitochondria were reisolated from the suspension by centrifugation, and polypeptides in the supernatant and in the mitochondrial pellet were examined by immunoprecipitation and gel electrophoresis. Thus, it could be determined whether particular polypeptides were present as their mature or larger precursor forms. To determine whether these polypeptides were inside or outside the mitochondria, their sensitivity to externally added protease was examined; polypeptides internalized by mitochondria should be inaccessible to the protease because of the physical barrier imposed by the membranes. The import of the P-subunit of F,-ATPase was shown to be completely dependent on the presence of ATP or a substrate for respiration (Gasser et al., 1982a). In the presence of such an “energy source,” import of ornithine transcarbamylase into rat liver mitochondria (Mori et af., 1981b; Kolansky et af., 1982), the adenine nucleotide translocator, ATP synthase subunit 9 (Schleyer et af.,1982), and four subunits of the cytochrome bc, complex (Teintze et al., 1982) into Neurospora mitochondria, and several polypeptides into yeast mitochondria (Gasser et al., 1982a) could be demonstrated. In all cases import was blocked by uncouplers of oxidative phosphorylation. Uncouplers dissipate the electrochemical gradient of protons across the mitochondrial membrane, but a secondary effect of this is to stimulate ATP hydrolysis by the reverse reaction of the H -translocating ATP synthase, thereby depleting the mitochondrial matrix of ATP. One would like to know whether ATP drives transport directly or whether a transmembrane electrochemical potential is required. By investigating in vitro import in the presence of various inhibitors, Schleyer et al. (1982) and Gasser et af. (1982a) were able to answer this question. Gasser et af. (1982a) synthesized precursors in a reticulocyte lysate, then separated the precursors and other proteins from small molecules (including ATP and respiratory substrates) by gel filtration. These precursor polypeptides were then incubated with isolated yeast mitochondria in the presence of KCN (to inhibit endogenous respiration) and ATP. Under these condi+

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

303

tions a large fraction of the precursor to the P-subunit of F,-ATPase was transported into the mitochondria. This import was entirely dependent on added ATP and could be blocked by addition of either carboxyatractyloside or oligomycin. Carboxyatractyloside inhibits the adenine nucleotide translocator, thus blocking the entry of added ATP into the mitochondria. Oligomycin inhibits the proton-translocating ATPase. The inhibition of protein import by these compounds indicates that the added ATP must enter the mitochondrion and be hydrolyzed by the F,F,-ATPase. Since oligomycin inhibits hydrolysis of ATP by the ATPase, it should increase the ATP concentration in the mitochondrial matrix. The fact that oligomycin blocks protein import already suggests that ATP is not the direct source of energy for translocation. Indeed, the inhibitory effect of oligomycin could be overcome by presenting the mitochondria with a substrate for respiration, thus restoring a transmembrane electrochemical gradient. Whether supported by respiration or by ATP hydrolysis, the ability of mitochondria to import proteins always correlated with conditions where the electrochemical potential gradient would be expected to be relatively large, regardless of the ATP concentration in the matrix. Schleyer et ul. (1982) reached the same conclusion when investigating the transport of proteins into Neurosporu mitochondria. Protein import was blocked by a combination of oligomycin and antimycin, an inhibitor of electron transfer through the cytochrome bc, complex. Under these conditions, both respiration and ATP hydrolysis would be inhibited. Protein import was restored by addition of ascorbate and tetramethylphenylenediamine(TMPD), thereby allowing reduction of cytochrome c and the subsequent regeneration of a proton electrochemical potential gradient by the activity of cytochrome c oxidase. The ATP concentration should be unaffected by the presence of ascorbate and TMPD, but one would expect a significant difference in the electrochemical potential gradient in the presence and absence of this substrate combination. Thus, it appears that import of proteins to the mitochondrial matrix and inner membrane is dependent on an electrochemical gradient across the inner membrane, contrary to the initial suggestion of Nelson and Schatz (1979) that molecular ATP is the energy source. This suggestion was made partly because a mitochondrial petite (rho-) yeast mutant, which lacks a functional ATP synthase and respiratory chain, was able to import proteins in the absence but not in the presence of bongkrekate, an inhibitor of adenine nucleotide transport. In the presence of this compound ATP cannot enter the mitochondria. It was assumed that no significant electrochemical potential gradient across the mitochondrial inner membrane could be generated in this mutant since proton translocation by respiration and by ATP hydrolysis are absent. However, the movement of adenine nucleotides across the mitochondrial inner membrane is itself an electrogenic process and could thus generate a significant potential difference, albeit probably only about one-fourth of that found in normal respiring mitochondria. Despite the smaller potential

304

GRAEME A. REID

difference, rho- mitochondria can still import proteins, a finding suggesting that a smaller electrochemical potential difference across the inner membrane is required for protein transport than for ATP synthesis. Several findings show that translocation rather than proteolytic processing of precursors is the energy-dependent step in protein import. First, the import of proteins without larger precursors into the mitochondrial matrix (e.g., 2-isopropylmalate synthase; Gasser et al., 1982b; Hampsey et al., 1983) or the inner membrane (e.g., adenine nucleotide translocator; Schleyer et al., 1982) requires energy, though obviously no proteolytic processing occurs. Second, processing is still catalyzed by a partially purified matrix protease (cf. below). Third, the transport and processing of cytochrome c peroxidase are temporally separated (cf. below; Reid et al., 1982); the initial transport step is blocked by CCCP, whereas subsequent processing is not. The transport of cytochrome c into mitochondria does not require a transmembrane electrochemical potential gradient (Zimmermann et al., 1981). Cytochrome c is a component of the intermembrane space and presumably the mitochondrial inner membrane is not directly involved in the import of this protein. The insertion of proteins into the mitochondrial outer membrane similarly lacks an energy requirement (see Section II,C,4). Thus, the requirement for an energized inner membrane is found only for those proteins which are transported into or across this membrane. As described above, we now know that the transport of proteins into the mitochondrial matrix and inner membrane requires an “energized” inner membrane, but we do not know what the electrochemical potential gradient is needed for. The effect may be essentially electrophoretic, with transport being initiated by the movement of a cluster of positively charged residues of a precursor polypeptide toward the more electronegative mitochondrial matrix. It appears that the N-terminal regions of those precursors so far investigated are predominantly basic in nature (see Section IV,B). Alternatively, the energy requirement may be less direct-indeed we do not know whether energy is actually consumed during transport. The electrochemical gradient could conceivably be required to maintain a state compatible with translocation, perhaps involving the membrane lipid conformation, as suggested by Schatz and Butow (1983), or protein conformation. It will be difficult to analyze how energy facilitates protein transport using whole mitochondria. More detailed analyses may eventually be possible with reconstituted components of the import machinery. OF IMPORTED PRECURSORS 2. PROTEOLYTIC PROCESSING

Those mitochondrial polypeptides initially synthesized as larger precursors must at some stage during their maturation be processed to their mature size. Pulse-labeling and pulse-chase labeling experiments indicated that this is gener-

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

305

ally a rapid process: the half-life of the precursor of the P-subunit of F,-ATPase in yeast is about 0.5 minute (Reid and Schatz, 1982b) and rat carbamylphosphate synthase precursor disappears with a half-life of approximately 2 minutes (Raymond and Shore, 1981). This half-life reflects the overall process of transport and porteolytic cleavage. Is the precursor first transported and then cleaved, or the other way around? Since the processing protease is found in the mitochondria1 matrix (Bohni et al., 1980), at least part of the precursor must be translocated before cleavage takes place. Proteins translocated across the endoplasmic reticulum and the Escherichia coli plasma membrane can be proteolytically processed while they are being transported; since transport in these cases is at least partly cotranslational, the existence of processed nascent polypeptides showed this temporal relationship quite clearly. It is not known whether imported mitochondrial proteins can be processed during their translocation. If processing takes place after completion of translocation, one might expect to find intramitochondrial precursors. In all but one case these were not detectable, so processing must at least occur very soon after transport (Reid and Schatz, 1982b). The exception to this rule is cytochrome c peroxidase, which is processed over a period of many minutes following its import into mitochondria (Reid et al., 1982). This temporal separation of the transport and processing steps clearly demonstrates that they are not obligately coupled: Proteolytic cleavage is not required for translocation. This is also shown by the fact that several imported precursors have no N-terminal extension. 3. IMPORTOF PROTEINS TO THE INTERMEMBRANE SPACE: TWO-STEPPROCESSING

Proteins destined for the mitochondrial matrix clearly must traverse two membranes during import, but to reach the intermembrane space it would appear that only the outer membrane must be crossed. Surprisingly, the transport of at least some proteins to the intermembrane space is rather more complicated than initially imagined. Cytochrome b, and cytochrome c peroxidase are soluble proteins of the yeast mitochondrial intermembrane space (Daum et al., 1982a). Cytochrome c , is attached to the mitochondrial inner membrane as a component of the cytochrome bc, complex, but its bulk protrudes into the intermembrane space, where it interacts with cytochrome c (Li et al., 1981). It may be considered a component of the intermembrane space, particularly as its import shares many features with the import of cytochrome b, and cytochrome c peroxidase. Each of these three proteins is initially made outside the mitochondrion as a larger precursor (Gasser er al., 1982b; Maccecchini et al., 1979b; Nelson and Schatz, 1979) and subsequently imported into the mitochondrion. The import of each of these polypeptides was found to be energy dependent. Pulse-labeling of yeast spheroplasts in the presence of CCCP resulted in accumulation of label in

306

GRAEME A. REID

the precursor form of cytochrome c, (Nelson and Schatz, 1979). Similarly, the maturation of cytochrome b, and cytochrome c peroxidase was blocked by CCCP in intact yeast cells (Reid et al., 1982). The energy requirement for the import of cytochrome b, has also been investigated in vitro (Gasser et al., 1982a,b; Daum et al., 1982b) and, as with proteins transported to the matrix and inner membrane, an electrochemical potential gradient across the inner membrane is needed. Thus, the inner membrane apparently has a functional role in the transport of proteins to the intermembrane space. When yeast proteins synthesized in a reticulocyte lysate are incubated with the partially purified processing protease from the mitochondrial matrix (Section IV,A), the precursors of mitochondrial matrix and inner membrane proteins are processed to the corresponding mature polypeptides (Bohni et al., 1983; Cerletti et al., 1983). This protease does not digest any nonmitochondrial protein tested. It converts the cytochrome b, precursor (68 kDa), not to the mature size (58 kDa) but to an intermediate-size form (64 ka). This could, of course, be an in vitro artifact; indeed it was not initially expected that cytochrome b, precursor on its way from the cytosol to the intermembrane space would ever become accessible to a protease in the mitochondrial matrix. That the intermediate form is biologically significant has been shown by pulse-labeling of intact yeast cells. The cells were pulse-labeled with [35S]methioninein the presence of 20 p M CCCP, under which conditions import of cytochrome b, is blocked and all the pulse-labeled cytochrome b, is in the precursor form. When CCCP is inactivated with 2mercaptoethanol (Kaback et al., 1974) and the cells are then chased with unlabeled methionine, the accumulated cytochrome b, precursor is transported into the mitochondria. During this import, the precursor is first converted to the intermediate form and then to the mature form of the cytochrome (Fig. 2; Reid et al., 1982). The intermediate form of cytochrome b, is also found during in vitro import into mitochondria, and again it behaves as a kinetic intermediate between the precursor and mature forms (Daum et al., 1982b). The precursor of cytochrome b, is synthesized as a soluble polypeptide outside the mitochondrion (Reid and Schatz, 1982a), and the mature polypeptide is a soluble component of the intermembrane space (Daum et al., 1982a; Reid et al., 1982). The intermediate, in contrast, is firmly attached to the mitochondrial inner membrane, apparently with most of its bulk exposed to the intermembrane space. Since the precursor is cleaved on the matrix side of the inner membrane, the intermediate generated by this cleavage is presumably transmembranous. The same location was found for pulse-labeled cytochrome c peroxidase precursor which had already been imported to the mitochondrion (Reid et al., 1982). This precursor undergoes very slow processing by the matrix protease, and no intermediate form was observed in these experiments, although cytochrome c peroxidase probably does share the two-step processing pathway with cytochrome b, (see Section IV,B).

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

307

FIG. 2. Cytochrome b2 precursor is converted first to an intermediate form, then to mature cytochrome b2 in intact yeast cells. Labeled cytochrome b2 was accumulated by pulse-labeling yeast cells in the presence of CCCP. The uncoupler was inactivated by addition of 2-mercaptoethanol, and the maturation of cytochrome b2 was examined after various periods of chase (indicated above each lane, in minutes). The immunoprecipitated cytochrome b2 was analyzed by SDS-polyacrylamide gel electrophoresis (Reid er ul.. 1982). Lanes P and M contain precursor and mature cytochrome h Z , respectively,

The import and maturation of cytochrome c 1 follows a pathway similar to that of cytochrome b,. The in vitro-synthesized precursor is processed to an intermediate-size form by the matrix-located protease (Gasser et al., 1982b; Ohashi et al., 1982) and the orientation of the cytochrome c 1 intermediate is the same as that found for the cytochrome b, intermediate: attached to the inner membrane with its bulk protruding into the intermembrane space (Ohashi et al., 1982). During its maturation, cytochrome c, undergoes covalent attachment of heme. Ohashi er a / . (1982) investigated when this reaction takes place in relation to the proteolytic maturation steps. This was achieved using a heme-deficient yeast mutant, lacking 5-aminolevulinate synthase. When this mutant was pulse-labeled with [35S]methioninein the absence of heme precursors and cytochrome c 1 was subsequently immunoprecipitated and analyzed by SDS-polyacrylamide gel electrophoresis, it was found that the intermediate polypeptide was accumulated. When these pulse-labeled cells were chased with unlabeled methionine in the presence of 5-aminolevulinic acid, the accumulated intermediate form was converted to mature cytochrome c1. Thus, the second processing step in cytochrome c, maturation is dependent on the availability of heme. The two-step processing pathway involved in the maturation of cytochrome b, and cytochrome cI in yeast is summarized in Fig. 3. Cytochrome c1 has also been shown to be imported into

308

GRAEME A. REID

pre-ryt

b2

-naow&v pre-ryt

(,

CYTOPLASM

*c

MATRIX

IM

FIG.3. Suggested pathway for the maturation of cytochrome b2 and cytochrome cI in yeast. The zigzag line signifies the N-terminal peptide extension of the precursors and the arrows signify proteolytic cleavages. Noncovalently and covalently bound heme are represented by an open halfcircle and a box, respectively. The first proteolytic cleavage of each precursor generates a new Nterminus (N’). Cleavage of the membrane-bound intermediate is presumed to occur near the outer face of the inner membrane and generates the N-terminus of the mature protein (N”). This second cleavage releases cytochrome b2 in a soluble form into the intermembrane space. In contrast, cytochrome cI remains attached to the inner membrane by its hydrophobic C-terminus (Wakabayashi et al., 1980).

Neurospora mitochondria by a two-step pathway (Teintze et al., 1982). In vitro import of cytochrome c, has not been demonstrated in yeast, probably because the precursor is relatively unstable (Reid and Schatz, 1982a), but the corresponding Neurosporu polypeptide is imported into mitochondria in an energy-dependent manner (Teintze et ul., 1982). The import of cytochrome c, is not blocked

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

309

by an excess of apocytochrome c. indicating that these two hemoproteins are transported by different pathways. Little is known about the second proteolytic step in the two-step processing pathway. It is not clear whether a single enzyme catalyzes the conversion of each intermediate to the corresponding mature polypeptide, though this at present would be the most attractive possibility. No specific inhibitors of the second processing step have been found despite extensive searching, though the activity is sensitive to the detergent digitonin (Daum eta/. , 1982b). It has been suggested from in vitro experiments that the conversion of intermediate to mature cytochrome b, does not require an energized mitochondrial inner membrane (Daum et al., 1982b), but apparently the conversion of cytochrome c , intermediate to mature protein was inhibited by CCCP in Neurospora cells. The significance of these possibly contradictory results is not clear. A two-step processing pathway has also been proposed for a mitochondrial matrix protein (Mori et al., 1980), but recent findings suggest that the observed intermediate-size polypeptide may be an artifact. Rat liver ornithine transcarbamylase (OTC) is synthesized as a larger precursor (Conboy et al., 1979) and can be imported into mitochondria in vitro. Incubation of in vitro-synthesized precursor with isolated mitochondria leads to the formation, not only of a mature-size polypeptide, but also a form of OTC intermediate in size between the precursor and mature polypeptides (Mori et a/., 1980, 1981b; Morita et al., 1982a,b; Conboy and Rosenberg, 1981; Kraus et al., 1981; Kolansky et al., 1982). The conversion of the intermediate-size form of OTC to the mature protein has not, however, been demonstrated. Of some concern is the inability to detect the intermediate-size form of OTC in pulse-chase labeled, intact hepatocytes (Mori et al., 1981a; Morita et al., 1982b), suggesting the possibility that this polypeptide is an in virro artifact. Indeed, at least some of the intermediate-size form is found outside the mitochondria after incubation with in vitrosynthesized precursors (Kolansky et al., 1982), though the processing protease is found in the mitochondrial matrix (Mori et al., 1980; Miura et al., 1982a). The conversion of OTC precursor to the intermediate-size and to the maturesize polypeptide is sensitive to 1,lO-phenanthroline and other chelators of divalent metal ions, but it has recently been suggested that different enzymes are responsible for each of these proteolytic cleavages (Conboy et al., 1982). The conversion of precursor to mature OTC was found to be greatly enhanced by addition of Zn2 or Co2 , and the enzyme catalyzing this reaction was identified as a component of the mitochondrial matrix. This protease is probably analogous to that initially described by Bohni et al. (1980). The activity generating the intermediate-size polypeptide had a somewhat different submitochondrial distribution and metal-ion requirement; whether its activity is physiologically significant remains to be shown. This latter enzyme has been purified (Miura el al., 1982a) and found not to process the precursor of carbamoyl-phosphate +

+

31 0

GRAEME A. REID

synthase I which, like OTC, is imported to the mitochondrial matrix. The significance of these findings is not yet clear.

4. BIOCENESISOF THE MITOCHONDRIAL OUTERMEMBRANE In contrast to our extensive knowledge of the functions of the mitochondrial matrix and inner membrane, we know rather little of the biochemistry of the outer membrane. Some enzyme activities, such as kynurenine hydroxylase, are associated with this membrane and a pore function has recently been ascribed to a major polypeptide component (Zalman et af.,1980). However, the functions of most of the major polypeptides are currently unknown, and they are therefore described by their apparent size according to their mobility on polyacrylamide gels. It is clear, though, that the outer membrane must be important in the communication of mitochondria with the rest of the cell: it must contain receptors for proteins imported by mitochondria and possibly also components which interact with the cytoskeleton. The insertion of proteins into the mitochondrial outer membrane, unlike the transport of proteins into or across the inner membrane, does not require an electrochemical potential difference across the inner membrane, nor does it require ATP. The major polypeptides of the outer membrane [with one possible exception (Shore et a f . , 1981)] appear not to undergo proteolytic processingthey are synthesized without transient N-terminal extensions. These features indicate that the mitochondrial outer membrane proteins are transported by a pathway rather different from that (or those) involved in transport to the interior compartments of the mitochondrion. The mitochondrial outer membrane is permeable to solutes of molecular mass up to 2000-8000 (Pfaff et al., 1968; Colombini, 1979). This pore activity has been shown to reside with the most predominant band in Coomassie blue-stained SDS-polyacrylamide gels of outer membranes. This polypeptide has a molecular weight of 30,000 in rat and mung bean mitochondria and 3 1,000 in Neurospora (Zalman et al., 1980; Freitag et al., 1982a). The major polypeptide of yeast mitochondrial outer membrane has a molecular weight of 29,000 and presumably this, too, is the pore protein, which has been termed porin by analogy to a group of proteins which are found in the outer membranes of gram-negative bacteria and similarly act as nonspecific pores (Osborn and Wu, 1980). The biosynthesis of this polypeptide and its insertion into the outer membrane have recently been investigated. Mitochondria1 porin is synthesized almost exclusively on free polysomes, i.e., not on membrane-bound polysomes (Freitag et al., 1982b; Suissa and Schatz, 1982). It thus appears that it is inserted into the outer membrane posttranslationally. The initial translation product of porin mRNA has the same mobility on SDS-polyacrylamide gels as the mature polypeptide (Freitag et al., 1982b; Mihara et al., 1982; Gasser and Schatz, 1983). The N-terminal meth-

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

31 1

ionine residue remains with porin after its insertion into the membrane. Thus, no proteolytic processing appears to be necessary for the biogenesis of this protein, nor apparently for at least three other outer membrane polypeptides (Gasser and Schatz, 1983). When Neurospora or yeast proteins were synthesized in a reticulocyte lysate in the presence of [35S]methionineand incubated with mitochondria in the absence of further protein synthesis, most of the radioactive porin was found associated with the reisolated mitochondria. Whereas the soluble precursor was sensitive to added proteases, the porin associated with the mitochondria was resistant to digestion with trypsin or proteinase K (Freitag et al., 1982b; Mihara et af., 1982; Gasser and Schatz, 1983). The authentic, endogenous porin in isolated mitochondria is similarly protease-resistant. The in vitro-synthesized porin inserted not only into isolated mitochondria but also into isolated outer membrane which was essentially free of inner membrane components (Gasser and Schatz, 1983). Again the membrane-associated porin became protease resistant, a result suggesting that it had inserted into the outer membrane. The insertion of porin into isolated outer membrane vesicles indicates that its translocation is not dependent on a bulk potential across the inner membrane; no such potential can be generated across the outer membrane since small ions can readily diffuse through the pore. In agreement with this conclusion, the insertion of porin into mitochondria is not inhibited by CCCP or valinomycin under conditions where transport to the matrix and inner membrane are blocked (Freitag et al., 1982b; Gasser and Schatz, 1983). When porin is imported into Neurospora mitochondria in v i m at 25"C, essentially all of the labeled precursor quickly becomes associated with the mitochondria and immediately becomes resistant to protease. When the incubation is carried out at 4"C, protease resistance develops relatively slowly so that at early time points a large fraction of the porin associated with the mitochondria is in a protease-sensitive conformation; presumably it has not yet inserted into the outer membrane (Freitag et al., 1982b). The sites to which porin initially binds are probably different from those involved in binding and import of cytochrome b, and the P-subunit of F,-ATPase (Gasser and Schatz, 1983; Riezman et al., 1983b). When mitochondria are mildly treated with trypsin, they lose the ability to bind and import these two polypeptides, but the insertion of porin is unaffected. The binding sites for porin must, however, be specific to the mitochondria] outer membrane: whereas in vitro-synthesized porin will insert posttranslationally into isolated outer membrane, it does not insert into isolated endoplasmic reticulum membranes. This result contradicts the suggestion that the mitochondrial outer membrane is formed by differentiation of the endoplasmic reticulum (Shore, 1979). This membrane-specific, posttranslational insertion was also demonstrated with a 7O-kDa, a 45-kDa, and a 14-kDa polypeptide of the yeast mitochondria1 outer membrane (Gasser and Schatz, 1983). It was not possible to

312

GRAEME A. REID

show whether these proteins inserted correctly into the outer membrane, and this remains a major drawback in interpreting the above findings. To overcome this, detailed comparison with the orientation of the in vivo-synthesized polypeptides would be required, but little is currently known of the architecture of the outer membrane. In contrast to the findings described above, Shore et al. (198 1) reported that a 35-kDa polypeptide from rat liver mitochondria outer membrane is intially made as a slightly larger (35.5 kDa) precursor. It is not known whether this apparent difference in size is due to N-terminal processing; the results of Gasser and Schatz (1983) would suggest that this is unlikely. This 35.5-kDa polypeptide was shown to be synthesized on free polysomes and was posttranslationally imported into mitochondria in vitro (Shore et al., 1981).

D. Assembly of Imported Mitochondria1 Proteins Once imported into the mitochondrion and proteolytically or otherwise matured, a polypeptide must assume its active conformation. In many cases this involves specific interactions with other subunits, either homologous or heterologous. The assembly of a miltisubunit enzyme, such as cytochrome c oxidase, might be expected to follow a defined pathway, in which case assembly intermediates containing some but not all subunits might be formed. The techniques for detecting such intermediates have not yet been adequately developed. It is possible to investigate assembly in the absence of mitochondria1 protein synthesis either in a rho- yeast strain (Schatz, 1968) or by using specific inhibitors (e.g., de Jong et al., 1979), but such studies have not revealed specific assembly intermediates. By using the methods of in vitro mutagenesis, it should now be possible to alter a specific subunit of a complex such that assembly is no longer completed. One could then determine whether the remaining subunits become associated. Furthermore, one could isolate second-site revertants which allow assembly of the mutated polypeptide. Such reversions could be due to compensating mutations in another subunit of the complex, thus indicating an interaction between the subunits at some step in assembly; or one may perhaps find mutations in other components required for assembly: we do not know whether assembly requires catalysts. The molecular details of the assembly process will no doubt be examined in v i m , but a major difficulty so far has been to demonstrate that polypeptides imported into mitochondria in vitro become assembled into biologically active units, attaining all the characteristics of the corresponding protein in the living cell. Three approaches have been taken to determine whether this is indeed the case. The questions posed by these investigations are as follows: Does an imported protein reach the correct submitochondrial location? Does it assume a three-

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

313

dimensional structure comparable to that of the normal protein'? Does it become biologically active'? After import of in vitro-synthesized precursors into isolated mitochondria, Gasser et ul. (1982b) examined the submitochondrial distribution of various proteins. Imported cytochrome b, was found in the intermembrane space and not in the matrix, while 2-isopropylmalate synthase was found in the matrix fraction, not in the intermembrane space. These results reflect the distribution of the unlabeled mature enzymes within the mitochondrion. However, a disproportionately large amount of the labeled, imported polypeptides was isolated with the mitochondria1 membranes, for reasons which remain to be clarified. The membrane-bound forms may represent intermediate steps on the maturation pathways of these proteins (Gasser et ul., 1982b). Imported membrane proteins were found exclusively in the membrane fraction. Do in vitro-imported polypeptides assume the same conformation as the in vivo-synthesized protein? Evidence that they do indeed has come from studies of two enzymes. The adenine nucleotide translocator of Neurosporu is specifically inhibited by carboxyatractyloside, which binds tightly to the protein. When the inhibitor is present, the translocator protein does not bind to hydroxyapatite, whereas the soluble precursor does. Upon import into isolated mitochondria, however, the in virro-synthesized translocator no longer binds to hydroxyapatite in the presence of carboxyatractyloside, a finding suggesting that it has acquired the ability to bind this inhibitor (Schleyer and Neupert, 1984). The imported and processed form of ornithine transcarbamylase, but not the precursor, binds a transition state analog of carbamoyl phosphate, as does the active enzyme, thus indicating that the active conformation has been reached. Furthermore, the imported protein comigrates with the active trimeric mature protein on a gel filtration column (L. Rosenberg, personal communication). It has been suggested that in virro import of rat carbamoyl-phosphate synthase (Campbell et ul., 1982) and yeast phenylalanyl-tRNA synthase (Diatewa and Stahl, 1981) leads to new enzymatic activity. Further experimentation is required to clarify this point, the main problems being to find a system with a sufficiently low background activity and to import sufficient precursor to generate detectable enzyme activity.

111. ARE PROTEINS TRANSPORTED INTO MITOCHONDRIA COTRANSLATIONALLY OR POSTTRANSLATIONALLY?

In a series of reports, Kellems, Allison, and Butow (Kellems and Butow, 1972, 1974; Kellems et ul.. 1974, 1975) described cytoplasmic-type, 80 S ribosomes bound to the surface of yeast mitochondria. These bound ribosomes

314

GRAEME A. REID

were observed in spheroplasts and remained attached to mitochondria during isolation of the organelle. The interaction between these bound polysomes and the mitochondrial surface showed features remarkably similar to the binding of ribosomes to the rough endoplasmic reticulum. In particular, the ribosomes could be released from the mitochondria by a combination of puromycin and concentrated salt, but by neither condition alone. This finding strongly indicates that the ribosome-membrane binding is mediated by nascent polypeptides, as is the case with ribosomes bound to the endoplasmic reticulum. On the basis of these results it was proposed that transport of proteins into mitochondria is a cotranslational process: the mitochondria-bound ribosomes were considered to be directly involved in the translocation of nascent mitochondrial polypeptides across the membrane(s). It has since been clearly demonstrated that there is no obligate coupling of protein translocation across mitochondrial membranes to protein synthesis. Mitochondrial polypeptides can be synthesized in vitro by translating purified RNA in a homologous (Harmey et a l . , 1977) or heterologous (Maccecchini et a l . , 1979a) protein synthesizing system, and the finished precursor polypeptides can subsequently be imported by isolated mitochondria. The import reaction is unaffected by cycloheximide, a potent inhibitor of polypeptide chain elongation on 80 S ribosomes. Such experiments show unambiguously that polypeptides can be imported into mitochondria posttranslationally, at least in vitro. Posttranslational import has also been described in vivo in yeast. The import of most mitochondrial proteins is energy dependent and can be conveniently inhibited by CCCP, an uncoupler of oxidative phosphorylation. Yeast spheroplasts pulse-labeled with [35S]methioninein the presence of CCCP accumulate labeled precursors of mitochondrial proteins-no significant conversion to the mature proteins is observed (Nelson and Schatz, 1979). On subcellular fractionation, these labeled precursors were found outside the mitochondria (Reid and Schatz, 1982b). When the effects of CCCP were abolished by addition of 2-mercaptoethanol, the extramitochondrial precursor of the P-subunit of F, -ATPase was imported into the mitochondria and converted to the mature protein. The chase was performed in the presence of an excess of unlabeled methionine, so that labeled precursor was synthesized only during the pulse. Again the transport and maturation of this polypeptide was not inhibited by cycloheximide. These experiments demonstrated that mitochondrial protein import can occur posttranslationally even in vivo. The experiments described above relate to nonphysiological conditions, but we would like to know the sigificance of posttranslational transport into the mitochondria in a growing cell. If mitochondrial protein import is indeed posttranslational, one might expect to find pools of extramitochondrial precursors awaiting import. Such pools were first detected by double isotope pulse-chase experiments in Neurosporu (Hallermayer et a l . , 1977). These experiments dem-

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

31 5

onstrated that the appearance of pulse-labeled proteins in the mitochondria lagged behind the labeling of proteins in other cellular fractions. Incorporation of radioactive amino acids into transported proteins was rapidly stopped by addition of cycloheximide, but labeled proteins continued to appear in the mitochondrial fraction, a finding indicating transport of previously synthesized polypeptides into the organelle. In similar experiments with yeast, Ades and Butow (1980a) were unable to detect a lag in the labeling of mitochondrial proteins, a result suggesting the absence of large extramitochondrial pools of polypeptides awaiting transport. This finding does not exclude the possible existence of such pools but does define an upper limit for the pool size under the experimental conditions employed. The sensitivity of the experiments of Hallermayer et al. (1977) and of Ades and Butow (1980a) was limited by the fact that the radioactivity in immunoprecipitable mitochondrial proteins was not determined separately for precursor polypeptides and their mature forms. By fractionation of pulse-labeled yeast, it could indeed be demonstrated that extramitochondrial pools of precursors exist (Reid and Schatz, 1982b). Interestingly the pool size depends upon physiological conditions; this may at least partly explain the inability of Ades and Butow (1980a) to detect such pools. One might expect the size of an extramitochondrial precursor pool to depend on the rate of precursor synthesis and on the rate of precursor transport into mitochondria. By lowering the rate of protein synthesis with cycloheximide during pulse-labeling of intact yeast cells, the pool of the precursor to the P-subunit of F,-ATPase was lowered as well (Reid and Schatz, 1982b). Since mitochondrial protein import is probably unaffected by cycloheximide, the fewer precursor molecules being synthesized should spend less time in the cytosol. If mitochondrial protein import is indeed posttranslational in vivo, what is the function of mitochondria-bound 80 S ribosomes'? That they may have a function was suggested by analysis of the polypeptides being synthesized on these polysomes. Ades and Butow (1980b) examined the synthesis of the three largest subunits of F,-ATPase in a readout system where the nascent chains on polysomes are synthesized to completion. Each was found to be preferentially made on mitochondria-bound polysomes compared with unattached polysomes. Suissa and Schatz (1982) isolated mRNA from these two polysome populations and analyzed their in vitro translation products. The mRNAs for many mitochondrial polypeptides were enriched in the mitochondria-bound polysomes compared to the mRNAs for cytosolic proteins. These results show that the interaction between polysome and mitochondrion is specific and, as described above, that it may be mediated by nascent chains. Thus, the mitochondrial surface can recognize a specific subset of nascent polypeptides, namely, those destined to be imported into mitochondria. However, not all mitochondrial polypeptides are preferentially synthesized on these isolated mitochondria-bound polysomes, and even where enrichment is

31 6

GRAEME A. REID

greatest, at most 60% of the total mRNA for a mitochondrial polypeptide is found associated with bound polysomes (Suissa and Schatz, 1982). Indeed, some imported mitochondrial proteins are synthesized essentially exclusively on unattached polysomes. Thus, mitochondria-bound polysomes cannot account for the bulk of mitochondrial protein import. The conditions used to observe and to isolate mitochondria-bound polysomes would in fact tend to maximize their apparent significance: to prevent completion of nascent chains, with the consequent dissociation of polysomes, protein synthesis is usually frozen by addition of cyclohexirnide. This may well effect a redistribution of polysomes since nascent mitochondrial polypeptides will have time to bind to the mitochondrial surface without their synthesis being completed. Thus, the amount of mitochondria-bound polysomes found by this procedure might be much higher than that existing in growing cells. If the nascent chains on mitochondria-bound polysomes bind to a functional receptor on the mitochondrial surface, as suggested by the apparent specificity of the interaction, then these nascent chains should be en route to the mitochondria. Ades and Butow (1980b) examined the fate of these polypeptides upon completion of polypeptide chain elongation and found that mitochondrial proteins did indeed become sequestered within the mitochondria. These experiments do not, however, distinguish whether the nascent polypeptides are discharged directly into the mitochondria, or whether translocation only occurs once the polypeptide chain has been completed. In summary, cotranslational import of proteins into mitochondria is suggested by the properties of mitochondria-bound polysomes, but there is no direct evidence that nascent polypeptides can be translocated across the mitochondrial membranes. On the other hand, there is a wealth of evidence supporting a posttranslational import pathway into mitochondria, both in vitro and in vivo. Import may be exclusively posttranslational, but at the moment it is impossible to exclude that cotranslational and posttranslational import coexist, their relative importance perhaps depending on physiological factors.

IV. THE MOLECULAR APPROACH The previous sections of this article have dealt largely with the phenomenology of mitochondrial protein import: I have described what happens, but now we want to know how it happens. For this we need to look at the properties of the molecules involved in the transport process: the extramitochondrial precursor polypeptides, their receptors on the mitochondrial surface, the translocation machinery, processing enzymes, and possibly other components with as yet unidentified functions.

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

31 7

A. Isolation and Characterization of the Molecules Involved in Mitochondria1 Protein Import 1. PRECURSORS OF MITOCHONDRIAL POLYPEPTIDES

Imported mitochondrial polypeptides are synthesized in the cytosol, a sea of proteins from which they must be fished out by the mitochondria. What structural features of the precursor polypeptides are recognized by the mitochondrial import machinery'? Do different precursors share common structural features'? How different is the structure of a precursor from that of its mature counterpart'? These are some of the questions which cun only be answered by a molecular analysis of precursor polypeptides. Such studies will require the isolation of milligram amounts of precursor polypeptide, but precursors are normally only found in very small amounts. They are usually seen, either when synthesized in vitro or in pulse-labeled cells, only by virtue of incorporation of radioactive amino acids into the polypeptide chain. This problem of low abundance has been overcome in two ways: one a rather special case, cytochrome c; the other more generally applicable. Cytochrome c is made without a polypeptide extension. The only covalent difference between precursor and mature cytochrome c is the attachment of a heme group to the latter. This heme group can be removed by chemical cleavage to yield the apocytochrome, which behaves as expected of cytochrome c precursor (see Section 11,B).Since mature cytochrome c can be readily purified in large amounts, this provides a plentiful source of a pure precursor polypeptide. Cytochrome c, however, is not a typical imported mitochondrial polypeptide: apart from having no N-terminal extension, it is imported via a receptor not shared by other mitochondrial proteins tested so far and its import does not require an energized inner membrane (Zimmermann et al., 1981). The approach used to prepare cytochrome L' precursor is not generally applicable to other proteins, such as those whose maturation involves proteolytic processing. Can one find a situation where large amounts of precursor polypeptide are synthesized and accumulated'? In yeast, at least, the answer is yes. By growing yeast in the presence of CCCP (an uncoupler of oxidative phosphorylation), the energy-dependent import of proteins into mitochondria is blocked, but protein synthesis and growth continue; as a result, precursor polypeptides accumulate outside the mitochondria (Reid and Schatz, 1982a,b). The extent of accumulation is dependent on the stability of the precursor polypeptide in the yeast cytoplasm. The precursor to cytochrome c , disappears with a half-life of 10 minutes, whereas some precursors accumulate in large amounts. After growth for several hours in the presence of CCCP, rlzo- yeast contains as much precursor of the P-subunit of F,-ATPase as the corresponding mature polypeptide

31 8

GRAEME A. REID

(about 150 pg/g cell protein; Reid and Schatz, 1982a). Isolation of this precursor from such cells requires approximately 7000-fold purification; as discussed below, this has proved possible. The stability of some accumulated yeast precursors contrasts with the rapid degradation of the precursor of the mitochondrial matrix enzyme carbamoyl-phosphate synthase when import is blocked in rat liver explants (Raymond and Shore, 1981). The precursor of mitochondrial aspartate aminotransferase is similarly unstable in chick fibroblasts (Jaussi et al., 1982). When the F,-ATPase @-subunit is accumulated in the yeast cytoplasm, it remains competent to be imported into mitochondria and processed to the mature protein upon removal of the import block. S. Ohta has purified several hundred micrograms of the F,-ATPase @-subunit precursor in a denatured form from CCCP-treated rho- yeast; at least some of the denatured precursor can then be renatured such that it regains the ability to be transported into mitochondria and become proteolytically matured (Ohta and Schatz, 1984). Thus, the import of this precursor into isolated mitochondria may be studied in the absence of other precursor polypeptides, and it may be determined whether extramitochondrial factors, perhaps present in reticulocyte lysate, are also important in directing the precursor to its intramitochondrial destination. 2. IMPORTRECEPTORS It has been shown that mitochondrial protein import requires protease-sensitive components on the mitochondrial surface (Section II,B), but these receptor-like polypeptides have not yet been identified. Identification could perhaps be achieved by solubilizing and separating the components of the mitochondrial outer membrane, reconstituting them into phospholipid vesicles, and determining which proteins are able to bind precursors. The first steps in this direction have been taken, and the results suggest the feasibility of this approach. Riezman et af. (1983b) solubilized yeast mitochondrial outer membrane vesicles with the nonionic detergent octyl polyoxyethylene. The solubilized material was reconstituted into vesicles by removing the detergent by dialysis, and these vesicles were shown to retain cytochrome b, precursor binding activity comparable to that of the original outer membranes. The constituted vesicles contained many polypeptides; it remains to be determined which of these is (or are) responsible for the binding activity. 3. PROCESSING PROTEASES At least two proteases are involved in the maturation of imported mitochondrial proteins. The protease(s) catalyzing the second step of cytochrome b,, cytochrome c peroxidase, and cytochrome c , maturation has proved difficult to examine. No specific inhibitors are known, and its activity is lost in the presence of detergent (Daum et af., 1982b). Fortunately, we know somewhat more about

319

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

another protease which appears to be responsible for the single-step maturation of many precursors and which also catalyzes the first cleavage step of those proteins imported to the intermembrane space by a two-step mechanism (Section II,C,3). This protease is a soluble component of the mitochondria1 matrix in yeast, rat, and maize (Bohni et a/., 1980, 1983; McAda and Douglas, 1982; Mori et al., 1980; Miura et al., 1982a). Its activity is maximal at neutral pH, is insensitive to serine protease inhibitors, but is inhibited by chelators of divalent metal ions such as 1 ,lo-phenanthroline, EDTA, and GTP. This inhibition could be at least partially reversed by addition of an excess of Zn2+ or Co2 (Bohni et af., 1983; Conboy e t a / . , 1982) or Mn2+ (McAda and Douglas, 1982). It is not known which of these cations is normally present in the active enzyme. The protease has been partially purified from yeast mitochondria (McAda and Douglas, 1982; Bohni et a/., 1983). It behaves on gel filtration as a molecule with a molecular weight of 110,000 to 115,000. Complete purification was not achieved, but McAda and Douglas (1982) suggested that the activity of the protease, judged by its ability to convert F, -ATPase P-subunit precursor to mature form, correlated best with the presence of a band with an apparent molecular weight of 59,000 on SDS-polyacrylamide gel electrophoresis, though no such band was detected in the purest preparations of Bohni et al. (1983). Bohni et a / . (1983) demonstrated that this enzyme is responsible for the cleavage of several larger precursors of imported mitochondria1 polypeptides; it may, in fact, cleave all precursors with N-terminal extensions. In this respect the enzyme may be considered to have a broad specificity, though the structural features around the cleavage sites of different precursors are not known, but presumably share some recognizable characteristics. In other respects the protease exhibits a remarkably high degree of specificity. It does not cleave nonmitochondrial proteins, nor does it process denatured precursors. This conformational requirement indicates that the protease does not simply recognize a particular amino acid sequence, rather some three-dimensional domain of the precursor. Since the partially purified protease also cleaves in vitro-synthesized precursors in the absence of mitochondria, it appears that the conformation of at least the Nterminal precursor regions are similar in solution and during (or soon after) translocation into the matrix. The matrix-located enzyme appears to be an endoprotease since partially processed intermediates are not normally observed. A clear demonstration that this is the case might be achieved by detection of the intact N-terminal peptide after processing, but this has not yet been done. It seems then that a single cleavage is generally involved in processing, but the maturation of the proteolipid (subunit 9) of Neurospora ATP synthase may occur in two discrete steps; since each of these steps is sensitive to 1,lO-phenanthroline (unlike two step-processing of intermembrane space enzymes where the second step is insensitive to chelators), they are perhaps catalyzed by the same enzyme (W. Neupert, personal commu+

320

GRAEME A. REID

nication). The processing of in vim-synthesized precursors by the matrix protease accurately reflects their maturation in vivo in that the correct N-terminus is generated, as determined by analysis of the N-terminal amino acid sequence (Cerletti et al., 1983). Whereas the matrix-located protease processes the precursors of matrix and inner membrane proteins to the mature forms, it generates an intermediate in the maturation pathway of some proteins of the intermembrane space (Gasser et al., 1982a; Bohni et al., 1983; see Section II,C,3). The protease is present in mitochondria of rho- yeast which lack mitochondrial protein synthesis; it must, therefore, be made extramitochondrially. It is not known whether it is made as a larger precursor, in which case it would presumably cleave its own precursor. 4. CYTOSOLIC SOLUBLE FACTORS Recent studies have shown that in vitro transport of proteins into mitochondria can be stimulated by a soluble factor which is present in reticulocyte lysates and in the yeast cytosol (Ohta and Schatz, 1984; Argan et al., 1983; Miura et al., 1983). The yeast soluble factor is a protein with an apparent molecular weight of 40,000 as judged by gel filtration (Ohta and Schatz, 1984). The role of this factor in protein transport is unknown, but its purification will greatly enhance investigation of its structure and function.

B. Isolation and Characterization of the Nuclear Genes Encoding Mitochondria1 Polypeptides Recently developed methods for isolating and manipulating genes have provided a new tool with which to elucidate the molecular features of the import process. Many nuclear genes encoding mitochondrial polypeptides have been isolated, and some partially characterized. What can these genes tell us about the precursor polypeptides encoded by them? For one, the DNA sequence should give us the protein sequence. With proteins that are synthesized as larger precursors, the extra sequence in the precursor can be deduced. This will be much simpler than direct determination of the amino acid sequences of precursor polypeptides (if the N-terminal sequence of the mature protein is known). If the N-terminal regions of precursors operate as signals for import into mitochondria, it may be possible to discern recognizable features common to different precursors. However, this approach has limited use in defining the important information of the import signal; perhaps only part of the N-terminal region is responsible for providing the signal, and sequences present in the mature protein may be important, too. To define the nature of these signals, it would be useful to look at mutant forms which exhibit abnormal transport behavior. Such mutants can be generated in vitro by manipulation of the cloned precursor genes; deletions,

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

321

insertions, and point mutations can be constructed. It will also be of interest to construct gene fusions. By fusing the signal region from a mitochondrial protein to an enzyme of nonmitochondrial origin, it may be possible to deliver the enzyme to the mitochondrion: if this does occur, one can determine how much of the precursor sequence is required to provide a functional signal. Analysis of cloned genes encoding mitochondria1 proteins will also be useful in studies of their regulation; this approach has already proved fruitful with a yeast cytochrome c gene (Guarente and Mason, 1983). 1 . MOLECULAR CLONING OF MITOCHONDRIAL PROTEIN GENES

For several reasons the yeast Saccharomyces cerevisiae is generally the organism of choice for these studies. Apart from being well characterized genetically, yeast can be transformed with a variety of plasmid vectors, thus allowing the isolated genes to be returned into their homologous cell. Also, many of the biochemical studies of mitochondrial protein import have been performed with yeast. Most of the mitochondrial protein genes studied so far have indeed come from yeast. Some of the approaches used to isolate these genes are described below. A useful account of recombinant DNA technology in yeast may be found in Botstein and Davis (1982). a. Screening with u Synthetic Oligonucleotide Probe. The amino acid sequence of yeast iso- 1 -cytochrome c is known, but because of the redundancy in the genetic code the nucleotide sequence of the corresponding gene cannot be unambiguously predicted. Stewart and Sherman (1974) were able to infer the nucleotide sequence for part of this gene by analysis of a number of frameshift mutations. Montgomery et al. ( 1978) synthesized an oligonucleotide complementary to part of this region and identified a plasmid bearing the cytochrome c gene by its ability to hybridize this oligonucleotide. This approach can also be used where the nucleotide sequence in the region of interest is ambiguous; the “unknown” nucleotides can be guessed from known codon preferences in yeast (Bennetzen and Hall, 1982) or a mixed probe can be synthesized. b. Functional Complementation. Several yeast nuclear mutants have been described which are deficient in mitochondrial metabolism, particularly in oxidative phosphorylation (listed in Broach, 1981). If the chromosomal defect can be overcome by transformation of the mutant with a suitable plasmid containing a functional copy of the gene, the mutant phenotype will be suppressed. This approach has been used to isolate the yeast gene encoding the adenine nucleotide translocator (O’Malley et a!., 1982). The mutation pet9, or op, (Kovac et al., 1967; Beck et a f . , 1968), results in loss of oxidative phosphorylation and thus an inability of the mutant to grow on nonfermentable carbon sources. Several lines of evidence suggested that pet9 is the structural gene for the adenine nucleotide translocator, but the necessary confirmation of this came only with the charac-

322

GRAEME A. REID

terization of the isolated gene. It is not possible to select directly for growth of transformants on a nonfermentable carbon source since the Pet phenotype takes some time to be overcome. O’Malley et al. (1982) therefore used a two-step screening procedure. They constructed a mutant which contained not only the pet9 defect but also a mutant leu2 gene. This mutant was transformed with a plasmid capable of episomal replication in yeast which contained the selectable leu2 gene and random fragments of the yeast genome. First, transformants were selected by their ability to grow in the absence of leucine in order to select clones containing plasmid. Leu transformants were then selected for their ability to grow on a nonfermentable carbon source. In this way a plasmid containing the pet9 gene was found. The yeast DNA inserted in this plasmid was found to encode a 30-kDa protein. This protein was recognized as the adenine nucleotide translocator by its reaction with specific antiserum. Ebner et al. (1973a,b; Ebner and Schatz, 1973) and Tzagoloff et al. (1975) have described deficiencies in mitochondrial enzyme activities resulting from defined nuclear mutations. Complementation of these mutations could be used to isolate genes for subunits of the enzymes concerned. Not all of these mutations, however, are in the structural genes for components of the deficient enzymes. One group of per mutants, leading to mitochondrial cytochrome b deficiency, produces abnormal transcripts of the mitochondrial cytochrome b gene. Complementation of this mutation was used to select a nuclear gene (cbpl) responsible for correct expression of a mitochondrial gene (Dieckmann et al., 1982). Similarly, Faye and Simon (1983) isolated a gene (mss5l) involved in maturation of the mitochondrial RNA encoding subunit I of cytochrome c oxidase. If, as is very likely, cbpl and mss.51 are structural genes, their products are presumably imported into mitochondria. c. Immunological Screening. Some plasmid-borne eukaryotic genes (e.g., from yeast) can be expressed in E. coli without further genetic manipulation. It is thus possible to directly screen for E . coli colonies harbouring plasmids into which yeast gene fragments have been inserted by using antibodies to detect specific, expressed polypeptides. This method (Erlich et al., 1978; Henning et al., 1979) has been used to identify the yeast cytochrome c peroxidase gene (Goltz et al., 1982). Viebrock et al. (1982) constructed a Neurospora cDNA clone bank from which they wanted to select the gene encoding subunit 9 of the mitochondrial ATP synthase. A plasmid harboring this gene was identified after screening by in vitro translation of hybrid-selected mRNA: the plasmids were allowed to hybridize complementary RNA molecules from total Neurospora mRNA, and the hybridized RNA was translated in a wheat germ lysate. The translation products were immunoprecipitated with an antiserum against the purified subunit in order to recognize a plasmid containing the subunit 9 gene. +

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

323

d. RNA Hybridization. A more general approach to the cloning of nuclear genes encoding mitochondrial polypeptides has been developed by van Loon et al. (1982). Many mitochondrial proteins in yeast are repressed by glucose. Their regulation appears to be transcriptional, so yeast cells grown with a nonfermentable carbon source contain much higher levels of the mRNAs encoding mitochondrial polypeptides than do yeast cells grown on a glucose-containing medium. mRNA from lactate-grown yeast was radioactively labeled and hybridized to a yeast clone bank ( E . coli colonies harboring plasmids into which random pieces of the yeast genome had been inserted). The hybridization was repeated in the presence of unlabeled mRNA from glucose-repressed yeast. The hybridization of labeled RNA to most colonies was reduced by competition, but those containing genes for mitochondrial polypeptides still gave strong signals. This procedure produced an enriched clone bank which could then be screened for the presence of specific genes by translation of hybrid-selected mRNA. This procedure has been used to identify the genes encoding three subunits of the cytochrome bc, complex (van Loon et a[., 1982) and several other mitochondrial polypeptides ( H . Riezman, A. P. G. M. van Loon, G. A . Reid, M. Suissa, and G . Schatz, unpublished results).

2. WHATTHE GENESHAVETOLDUS Since the work described in Section IV,B,I is rather recent, we have so far obtained only little information from the cloned genes for mitochondrial polypeptides. This will certainly change rapidly, but it is worth considering what we have learned so far, and where we might go from here. The nucleotide sequences of a few of the cloned genes have been determined, revealing the nature of the N-terminal extensions where present. The predicted amino acid sequences of the N-terminal regions of these precursors so far examined are shown in Table I. There are no obvious homologies, but this is not so surprising. It is known that the N-terminal extensions of different imported mitochondrial proteins vary widely in apparent size (Neupert and Schatz, 1981; Hay et al., 1983) and the proteins for which the sequence has so far been determined are transported to different submitochondrial locations, so presumably have recognizably different addressing signals. There are, however, some noteworthy features of these sequences. In all but one case, the N-terminal region is predominantly basic, as had been expected from comparison of precursor and mature proteins by isoelectric focusing (Anderson, 1981; Reid et al., 1982). The N-terminal extension of cytochrome c peroxidase precursor contains three lysine, four arginine, and three histidine residues and no acidic residues (Kaput et a l . , 1982). Similarly, the cleaved Nterminal peptide of Neurospora ATP synthase subunit 9 contains 12 basic and no

TABLE I THE N-TERMINAL AMINOACIDSEQUENCES OF IMPORTED MITOCHONDRIAL POLYPEPTIDES~ Polypeptide

Location of mature protein

N-Terminal amino acid sequence of precursor

+

+ ATP s y n k subunit 9

Inner membrane

+

+

Reference

++

+

++

MASTRVLASRLASQMAASAKVARPAVRVAQVSKRTIQTGSPLQTLKR

+

(Neurosporn)

++

Viebmck ef al

( 1982)

TQMTSIVNATTRQAFQKRAYSS.

+

-

t

+

+

++ .

Cyt c reductase ICkDa subunit

Inner membrane

MPQSFTSIARIGDYILKSPVLSKLCVPVANQFINLAGYKKLCL

Cyt c reductase 17-kDasubunit

Inner membrane

MDMLELVGEYWEQLKITVVPVVAAAEDDDDNEQHEEKAA

Citrate synthase

Matrix

MSAILSTTSKSFLSRGSTRQCQNMQKALFALLNARHYSS

EF-Tu

Mamx

MSALLPRLLTRTAFKASGKLLRLSSVISRTFSQTTTSYAAA

MSS 51

Unknown (probably mamx)

MT V L Y A P S GA TQ L Y F H L LR K S P HN R L V V S HQTR R H LMG F V R N A

Cytochrome c pemxidase

Intermembrane space

MTTAVRLLPSLGRTAHKRSLYLFSAAAAAAAAAATFAYSQSHKRSSS

-

-

-

-

-_-__

+

+ +

+

+

+

+

+ +

++ Suissa er a/. ( 1984)

+

+

+

+ +

Nagata ef

a/ (1983)

+

+++

Faye and Simon (1983)

+++

+++

+

+

+--+

-

+

++

de Haan el a / . (1983)

+

Kaput ef a/. ( 1982)

SPGGGSNHGWNNWGKAAALASTTPLV

++

-

7 t

+

+

-+

++

+

++

Cytochrome c

Intermembrane space

MTEFKAGSAKKGATIFKTRCLQCHTVEKGGPHKVGPNLHGIFGRH..

7&kDa polypeptide

Outer membrane

MKSFITRNKTAILATVAATGTAIGAYYYYNQLQQQQQRGKKNT

+

+ +

+ ++

Narita and Titan,(1969); Lederer el a / . (1972)

Hase ef

(21.

(1983)

0 The single-letter code for amino acids is used. The basic amino acids arginine, lysine, and histidine are marked +; the acidic aspartate and glutamate are marked - . The proteolytic cleavage sites generating the mature forms of ATP synthase subunit 9 and cytochrome c peroxidase are marked by arrows. Cytochrome c and the 70-kDa outer membrane protein are not cleaved except that the N-terminal methionine is removed from the former. The possible cleavage sites of the other precursors are not known. All sequences other than that of ATP synthase subunit 9 (from Neurospora) are derived from yeast.

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

325

acidic amino acids (Viebrock et uf.. 1982). The signal sequences of proteins secreted across the eukaryotic endoplasmic reticulum or across prokaryotic membranes (Austen, 1979: lnouye and Halegoua, 1980) are predominantly basic and generally also contain a stretch of uncharged amino acids which may form a transmembrane segment during translocation. The precursors of the ATP synthase subunit 9, and of several other proteins contain no such hydrophobic segments near the N-terminus; indeed the ATP synthase subunit 9 precursor has a remarkably polar N-terminal region, whereas the mature protein is generally apolar (Viebrock et al., 1982). On the other hand, the 70-kDa mitochondrial outer membrane protein has an uncharged stretch of 28 amino acids near the N-terminus which may act as a membrane anchor (Hase et al., 1983). This precursor is not proteolytically processed during import into mitochondria (Gasser and Schatz, 1983). Cytochrome c peroxidase precursor also has an unusual stretch of nonpolar amino acids, which is proposed to specify membrane binding (Kaput et ul., 1982). Thus, at the moment it is difficult to propose unifying concepts describing the signals involved in directing different precursors to the mitochondria, but specific roles have been proposed for the Nterminal extension of Neurospora ATP synthase subunit 9 and of yeast cytochrome c peroxidase. Subunit 9 of ATP synthase is an extremely hydrophobic polypeptide of 81 amino acid residues (Sebald et al., 1980). Its polarity (Capaldi and Vanderkooi, 1972) is only 25.9%, making it one of the most hydrophobic proteins known. In Neurospora this polypeptide is synthesized as a precursor in the cytoplasm (Michel ef ul., 1979; Schmidt et ul., 1983a). How is it maintained in solution despite such a hydrophobic nature'?The precursor is much larger than the mature polypeptide (Michel et al., 1979); indeed, it consists of 147 amino acid residues, of which 66 are removed during maturation (Viebrock et al., 1982). This 66amino acid extension is extremely hydrophilic, with a polarity of 53%, so that the overall polarity of the precursor is similar to that of a typical water-soluble protein. It is suggested (Viebrock er al., 1982) that an important function of the N-terminal extension of this protein is to render the precursor soluble, thus allowing its posttranslational import into mitochondria. The precursor must also contain specific information directing subunit 9 to the mitochondrial inner membrane. Interestingly, the yeast ATP syntase subunit 9 is encoded on mitochondrial DNA and translated without a cleaved N-terminal extension (Macino and Tzagoloff, 1979; Hensgens et al., 1979). Cytochrome c peroxidase is a soluble protein of the mitochondrial intermembrane space which appears to be imported into mitochondria by a two-step processing mechanism involving initial translocation of the precursor such that its N-terminus protrudes across the inner membrane into the matrix space (Gasser et al., 1982b; Reid et al., 1982; see Section 11,C,3). The precursor is larger than the mature protein (Maccecchini et al., 1979b) by 68 amino acid residues (Kaput

326

GRAEME A. REID

et al., 1982). Apart from the basic nature of this N-terminal extension, there are several striking features of the sequence, e.g., a stretch of 23 uncharged amino acids including 10 consecutive alanine residues. This hydrophobic region is expected to form an a-helix and is suggested to span the bilayer of the inner membrane when the precursor is imported into the mitochondrion (Kaput et al., 1982), acting as a stop-transfer sequence (Blobel, 1980). This hydrophobic stretch is flanked on either side by basic residues, perhaps to anchor it firmly in the membrane. The predicted sequence of the precursor of the 17-kDa subunit of ubiquino1:cytochrome c reductase is striking in its content of acidic amino acid residues (van Loon et al., 1984), not only in the mature protein, but also in the N-terminal region of the precursor. It will be of interest to compare the transport of this precursor into mitochondria with the transport of other polypeptides. To define the signals directing mitochondrial protein import, considerably more data are required. We have seen so far that basic amino acid residues near the N-terminus provide a common feature of most imported polypeptides. That these may be important in protein transport has been suggested by the finding that some basic compounds can inhibit mitochondrial protein import (Miura et al., 1982b). However, the basic nature of these sequences is clearly not sufficient for either intracellular or intramitochondrial sorting. Hydrophobic stretches may determine that a protein becomes membrane bound, but what determines whether it becomes a component of the inner or the outer membrane? In addition, the information in larger precursors must specify cleavage sites for processing enzymes. The most useful approach to the study of import signals will be to investigate the import of various mutated precursor polypeptides in order to delimit those regions of the precursor sequence which are essential for correct localization and processing. It has been shown that a truncated form of the 70-kDa outer membrane protein, lacking 203 amino acids from the C-terminus, is still transported to the mitochondrion (Riezman et al., 1983c), a finding suggesting that the C terminal region is not required for correct localization of this protein. Mutants in the signal sequence of the E . coli lamB protein that are deficient in the transport of this polypeptide have been used to select second-site revertants which restore secretion (Emr et al., 1981). The compensating mutations are likely to occur in components of the secretory machinery. A similar approach could be useful in recognizing the components of the mitochondrial import machinery: receptors, translocating components, proteases, and perhaps other, as yet undefined, molecules. It may also be possible to isolate such mutants by a conventional genetic approach. Matner and Sherman (1982) have described two groups of yeast mutants deficient in cytochrome c which may be lacking in heme-attaching activity or cytochrome c transport into mitochondria. Further biochemical studies are required to confirm this suggestion. Schekman (1982) has isolated temperature-

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

327

sensitive yeast mutants deficient in protein secretion, thereby defining many genes required for function of the secretory pathway. It should be possible to isolate conditional mutants in mitochondrial protein import. This would be useful in defining how many gene products are required to direct functional import and in determining the molecular nature of these components. While the ability to analyze isolated genes will be of great importance in the study of mitochondrial protein import, such investigations cannot replace the biochemical approach. Indeed, the two lines of research should be complementary in elucidating molecular mechanisms in the import pathway.

V.

SUMMARY

Because we now have considerable information describing several important biochemical features of mitochondrial protein transport, many exciting questions can now be asked, particularly regarding the molecular mechanisms involved. We do not yet know how many different pathways are used to import proteins into mitochondria, how many molecules are required to catalyze the process, or how these molecules work to convert extramitochondrial precursor polypeptides into active intramitochondrial enzymes. At present it seems that there are at least three pathways leading to mitochondrial protein import. Cytochrome c is transported by a route different from that required for import of all other proteins tested (Zimmermann et ul., 1981; Teintze et al., 1982), including an outer membrane protein (W. Neupert, personal communication). The transport of proteins to the mitochondrial outer membrane is very different from transport to the inner membrane and matrix; the two are distinguished by protease sensitivity of mitochondrial surface components, by energy dependence, and generally by the involvement of proteolytic processing. The initial steps of the two-step pathway by which some intermembrane space polypeptides are imported are very similar to the initial steps of import to the inner membrane and matrix. Each requires a protease-sensitive receptor, an electrochemical potential difference across the inner membrane, and usually a matrix-located protease. It will be interesting to test whether the same molecules are involved in the import of these proteins to different final locations, as seems likely. This pathway then could be largely responsible for the biogenesis of the mitochondrial matrix, inner membrane, and intermembrane space. The essential feature of polypeptides imported by this route must be a signal directing the initiation of translocation across the inner membrane. Some polypeptides will be completely translocated to become components of the mitochondrial matrix, others having stop-transfer signals (Blobel, 1980) will become components of the inner membrane. When such membrane-bound proteins are cleaved by a protease at the outer face of the inner membrane (Gasser et a/., I982b), they may

328

GRAEME A. REID

become components of the intermembrane space. Other factors may also be significant in defining the ultimate topology of a polypeptide: cytochrome c , apparently binds to the inner membrane by its rather hydrophobic C-terminus; interactions with other subunits of an enzyme may also be important. The above model poses an interesting question: Why would a stop-transfer signal determine that an imported polypeptide become attached to the inner membrane and yet allow its passage through the outer membrane? The relationship between the two membranes in protein import is not well understood. It has, however, been suggested that junctions between the two membranes may be sites of protein import (Schatz and Butow, 1983). Proteins can apparently be imported to a single compartment by more than one route. Cytochrome b, is transported to the intermembrane space by an energy-dependent, two-step processing pathway, whereas cytochrome c and apparently also adenylate kinase (Watanabe and Kubo, 1982) are imported without cleavage. By isolating the components of the import pathway, it may be possible to reconstitute the process in vitro and examine the molecular details. Such a system may be useful in determining how the electrochemical gradients across the mitochondrial inner membrane are involved in protein movement. We can begin to ask how mitochondrial biogenesis is regulated and what role protein import plays in this process. Some polypeptides are synthesized in the mitochondrion and become components of multisubunit enzymes. How is the synthesis of the various subunits coordinated, particularly when some subunits of an enzyme, such as cytochrome c oxidase, are synthesized intramitochondrially and others have to be imported from the cytoplasm? The 45-kDa subunit of mitochondrial RNA polymerase is encoded on nuclear DNA and is subject to glucose repression (Lustig et al., 1982). This control at the nuclear level may then be involved in the regulation of transcription of mitochondrial genes. Genetic evidence (Dieckmann et al., 1982; Faye and Simon, 1983) indicates the importance of nuclear genes in the maturation of mitochondrial transcripts. Thus, the nucleus communicates with the mitochondrial genetic system at least in part by means of polypeptides which are imported into the mitochondrion. In this article we have discussed the import of proteins into mitochondria from very diverse organisms. There is good reason to suspect that the processes involved are well conserved among different species. The yeast F,-ATPase psubunit precursor is matured by the matrix protease from rat and from maize. The precursor of subunit 9 of Neurospora ATP synthase is imported into yeast mitochondria and correctly processed (Schmidt et al., 1983b) even though the equivalent yeast polypeptide is synthesized intramitochondrially without an N-terminal extension. Presumably the Neurospora precursor has structural features similar to those of yeast precursor polypeptides and can be imported by a pathway normally used for import of these other yeast proteins. The import of proteins into chloroplasts shares many features with mitochondrial protein import

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

329

(Chua and Schmidt, 1979): larger precursors, posttranslational and energy-dependent import, and chelator-sensitive processing by a soluble protease (Highfield and Ellis, 1978; Grossman e t d . , 1980; Ellis, 1981). There must, however, be clearly recognizable differences between the cytoplasmically synthesized precursors destined for mitochondria and those en route to chloroplasts since the two import processes must be able to coexist in the plant cell cytoplasm. ACKNOWLEDGMENTS I sincerely thank Jeff Schatz and Rick Hay for critical evaluation of the manuscript. I am particularly grateful to Jeff for his immense support and encouragement during my stay in his lab. I also thank Ilona Durring for typing this article.

REFERENCES Ades, I. Z . . and Butow, R. A. (198Oa). The products of mitochondria-bound cytoplasmic polysomes in yeast. J . Biol. Chem. 255, 9918-9924. Ades, 1. 2.. and Butow, R. A. (1980b). The transport of proteins into yeast mitochondria. Kinetics and pools. J . B i d . Chetn. 255, 9925-9935. Anderson, L. (1981). Identification of mitochondrial proteins and some of their precursors in twodimensional electrophoretic maps of human cells. Proc. Nafl. Acud. Sci. U.S.A. 78, 24072411. Anderson, S . , Bankier, A. T., Barrell, B. G., de Bruijn, M. H. L., Coulson. A. R., Drouin, J.. Eperon, I. C., Nierlich, D. P., Roe, B. A,, Sanger, F.. Schreier. P. H., Smith, A. J. H . , Staden, R.. and Young, I. G. (1981). Sequence and organization of the human mitochondrial genome. Nafrire (London) 290, 457-465. Anderson, S . , de Bruijn, M. H. L., Coulson, A. R.. Eperon, I . C., Sanger, F., and Young, I. G. (1982). Complete sequence of bovine mitochondrial DNA. Conserved features of the mammalian mitochondrial genome. J . Mol. Biol. 156, 683-7 17. Argan, C., Lusty, C. J . , and Shore, G . C. (1983). Membrane and cytosolic components affecting transport of the precursor for omithine carbamyltransferase into mitochondria. J . Biof. Chem. 258, 6667-6610. Austen. B. M. (1979). Predicted secondary structures of aminoterminal extension sequences of secreted proteins. FEES Left. 103, 308-309. Beck, J . C., Mattoon, J . R.. Hawthome. D. C., and Sherman, F. (1968). Genetic modification of energy-conserving systems in yeast mitochondria. Proc. Nut/. Acad. Sci. U.S.A. 60, 186- 193. Bennetzen. J. L., and Hall, B. D. (1982). Codon selection in yeast. J . Biol. Chem. 257, 3026-3031. Bibb, M. J . , van Etten, R. A., Wright, C. T . , Walberg, M. W., and Clayton, D. A. (1981). Sequence and gene organization of mouse mitochondrial DNA. Cell 26, 167-180. Blobel, G. (1980). lntracellular protein topogenesis. Proc. Nut/. Acad. Sci. U.S.A. 77, 1496-1500. Bohni. P. C.. Gasser. S . , Leaver, C., and Schatz, G . (1980). A matrix-localized mitochondrial protease processing cytoplasmically-made precursors to mitochondria1 proteins. In “The Organization and Expression of the Mitochondria1 Genome’’ (A. M. Kroon and C. Saccone, eds.), pp. 423-433. North-Holland Publ., Amsterdam. Bohni, P. C., Daum. G . , and Schatz, G . (1983). Import of proteins into mitochondria. Partial purification of a matrix-located protease involved in cleavage of mitochondrial precursor polypeptides. J . B i d . Chem. 258, 4937- 4943. Borst, P., and Grivell, L. A. (1978). The mitochondrial genome of yeast. Cell 15, 705-723.

330

GRAEME A. REID

Botstein, D., and Davis, R. W. (1982). Principles and practice of recombinant DNA research with yeast. In “The Molecular Biology of the Yeast Saccharomyces. Metabolism and Gene Expression” (J. N. Strathern, E. W. Jones, and J. R. Broach, eds.), pp. 607-636. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Broach, I. R. (1981). Genes of Saccharomyces cerevisiae. In “The Molecular Biology of the Yeast Saccharomyces. Life Cycle and Inheritance” (J. N. Strathern, E. W. Jones, and J. R. Broach, eds.), pp. 653-727. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Campbell, M. T., Sutton, R . , and Pollak, J. K. (1982). The import of carbamoyl-phosphate synthase into mitochondria from foetal rat liver. Eur. J . Biochem. 125, 401-406. Capaldi, R. A,, and Vanderkooi, G. (1972). The low polarity of many membrane proteins. Prur. Natl. Acad. Sci. U.S.A. 69, 930-932. Cerletti, N., Bohni, P. C., and Suda, K. (1983). Import of proteins into mitochondria. Isolated yeast mitochondria and a solubilized matrix protease correctly process cytochrome c oxidase subunit V precursor at the N-terminus. J . B i d . Chem. 258, 4944-4949. Chua, N., and Schmidt, G . W. (1979). Transport of proteins into mitochondria and chloroplasts. J . Cell Biol. 81, 461-483. Colombini, M. (1979). A candidate for the permeability pathway of the outer mitochondrial membrane. Nature (London) 279, 643-645. Conboy, J. G . , and Rosenberg, L. E. (1981). Posttranslational uptake and processing of in vitro synthesized ornithine transcarbamoylase precursor by isolated rat liver mitochondria. Proc. Natl. Acad. Sci. U.S.A. 78, 3073-3077. Conboy, J. G., Kalousek, F., and Rosenberg, L. E. (1979). I n vitro synthesis of a putative precursor of mitochondrial ornithine transcarbamoylase. Proc. Natl. Acad. Sci. U . S . A . 76, 5724-5727. Conboy, J. G., Fenton, W. A,. and Rosenberg, L. E. (1982). Processing of pre-ornithine transcarbamylase requires a zinc-dependent protease localized to the mitochondrial matrix. Biochem. Biophys. Res. Commun. 105, 1-7. Daum, G . , Bohni, P. C . , and Schatz, G. (1982a). Import of proteins into mitochondria. Cytochrome h2 and cytochrome c peroxidase are located in the intermembrane space of yeast mitochondria. J . B i d . Chem. 257, 13028-13033. Daum, G., Gasser, S. M., and Schatz, G. (1982b). Import of proteins into mitochondria. Energydependent, two-step processing of the intermembrane-space enzyme cytochrome b2 by isolated yeast mitochondria. J . Biol. Chem. 257, 13075- 13080. De Haan, M., Van Loon, A. P. G. M., Kreike, J., Vaessen, R. T. M. J., and Grivell, L. A. (1983). The biosynthesis of the ubiquinol-cytochrome c reductase complex in yeast. DNA sequence analysis of the nuclear gene coding for the I4kDa subunit. Eur. J . Biochem. 138, 169-177. DeJong, L., Holtrop, M., and Kroon, A. M. (1979). The biogenesis of rat liver mitochondrial ATPase. Subunit composition of the normal ATPase complex and of the deficient complex formed when mitochondrial protein synthesis is blocked. Biochim. Biophys. Acta 548, 48-62. Diatewa, M., and Stahl, A. J. C. (1981). Biosynthesis and transport of yeast mitochondrial phenylalanyl-tRNA synthetase. Nucleic Acids Res. 9, 6293-6304. Diechann, C. L., Pape, L. K., and Tzagoloff, A. (1982). Identification and cloning of a yeast nuclear gene (CBPI) involved in expression of mitochondrial cytochrome h. Proc. Natl. Acad. Sci. U.S.A. 79, 1805-1809. Dujon, B. (1981). Mitochondrial genetics and functions. In “The Molecular Biology of the Yeast Saccharornyces. Life Cycle and Inheritance” (J. N. Strathern, E. W. Jones, and J. R. Broach, eds.), pp. 505-635. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Ebner, E., and Schatz, G . (1973). Mitochondrial assembly in respiration-deficient mutants of Saccharomyces cerevisiae. 111. A nuclear mutant lacking mitochondrial adenosine triphosphatase. J . Biol. Chem. 248, 5379-5384. Ebner, E., Menucci, L., and Schatz, G. (1973a). Mitochondrial assembly in respiration-deficient

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

33 1

mutants of Sacchuromyces cerevisiue. 1. Effect of nuclear mutations on mitochondria1 protein synthesis. J . Eiol. ('hem. 248, 5360-5368. Ebner, E., Mason, T. L., and Schatz, G. (1973b). Mitochondria1 assembly in respiration-deficient mutants of Sacchurtrmyces cerevisiue. 11. Effect of nuclear and extrachromosomal mutations on the formation of cytochrome c oxidase. J . Eiul. Chem. 248, 5369-5378. Ellis, R. J. ( 1981). Chloroplast proteins: Synthesis, transport and assembly. Annu. Rev. P l m t Physiol. 32, 111-137. Emr, S. D.. Hanley-Way, S., and Silhavy, T. J. (1981). Suppressor mutations that restore export o f a protein with a defective signal sequence. Cell 23, 79-88. Erlich, H. A , , Cohen, S. N . , and McDevitt, H. 0. (1978). A sensitive radioimmunoassay for detecting products translated from cloned DNA fragments. Cell 13, 681-689. Faye, G . , and Simon, M. (1983). Analysis of a yeast nuclear gene involved in the maturation of mitochondrial pre-messenger RNA of the cytochrome oxidase subunit I . Cell 32, 77-87. Freitag, H., Neupert. W., and Benz, R. (1982a). Purification and characterisation of a pore protein of the outer mitochondrial membrane from Neurosporu erassu. Eur. J . Eiochem. 123,629-639. Freitag, H., Janes, M., and Neupert, W. (l982b). Biosynthesis of mitochondrial porin and insertion into the outer mitochondrial menihrane of Neurospora crassa. Eur. J . Eiochem. 126, 197-202. Gasser, S. M., and Schatz, G . (1983). Import of proteins into mitochondria. In vitro studies on the biogenesis of the outer membrane. J . Biol. Chem. 2.58, 3427-3430. Gasser, S . M., Daum, G . , and Schatz, G. (1982a). Import of proteins into mitochondria. Energydependent uptake of precursors by isolated mitochondria. J . B i d . Chem. 257, 13034- 13041. Gasser, S . M., Ohashi, A , , Daum, G., Bohni, P. C., Gibson, J., Reid, G. A,, Yonetani, T., and Schatz, G. (l982b). Imported mitochondrial proteins cytochrome bz and cytochrome c, are processed in two steps. Proc. Natl. Acad. Sci. U.S.A. 79, 267-271. Goltz, S . , Kaput, J., and Blobel, G . (1982). Isolation of the yeast nuclear gene encoding the mitochondrial protein, cytochrome c peroxidase. J . Eiol. Chem. 2.57, 1 1186-1 1190. Grossmann, A,, Bartlett, S . , and Chua, N.-H. (1980). Energy-dependent uptake of cytoplasmically synthesized polypeptides by chlomplasts. Nature (London) 28.5, 625-628. Guarente, L., and Mason, T. (1983). Heme regulates transcription of the CYC 1 gene of S. cerevisiue via an upstream activation site. Cell 32, 1279-1286. Hallermayer, G . , Zimmermann, R.. and Neupert. W. (1977). Kinetic studies on the transport of cytoplasmically synthesized proteins into the mitochondria in intact cells of Neurosporu c'rassa. Eur. J . Eiochem. 81, 523-532. Hampsey, D. M., Lewin, A. S . , and Kohlhaw, G. B. (1983). Subniitochondrial localization, cellfree synthesis, and mitochondria1 import of 2-isopropylmalate synthase of yeast. Proc. Nutl. Acad. Sci. U . S . A . 80, 1270-1274. Harmey, M. A,, Hallermayer, G., Korb, H.. and Neupert, W. (1977). Transport of cytoplasmically synthesized proteins into the mitochondria in a cell free system from Neurospora crassa. Eur. J . Eiochem. 81, 533-544. Hase, T.. Riezman, H., Suda, K., and Schatz, G. (1983). Import of proteins into mitochondria. Nucleotide sequence of the gene for a 70kd protein of the yeast mitochondrial outer membrane. EMBO J . 2, 2169-1172. Hay, R.. Bohni. P., and Gasser, S. (1983). How mitochondria import proteins. Eiochim. Eiuphys. Actu 779, 65-87. Hennig. B.. and Neupert. W. (1981). Assembly of cytochrome c. Apocytochrome c is bound to specific sites on mitochondria before its conversion to holocytochrome c. Eur. J . Eiorhem. 121, 203-2 12. Hennig, B., Koehler, H., and Neupert, W. (1983). Receptor sites involved in post-translational transport of apocytochrome c into mitochondria: Specificity, affinity and number of sites. Proc. Narl. Acad. Sci. U . S . A . 80, 4963-4967.

332

GRAEME A. REID

Henning, U., Schwarz, H., and Chen, R. (1979). Radioimmunological screening method for specific membrane proteins. Anal. Biochem. 97, 153-157. Hensgens, L. A. M . , Grivell, L. A , , Borst, P., and Bos, J. L. (1979). Nucleotide sequence of the mitochondrial gene for subunit 9 of yeast ATPase complex. Proc. Null. Acad. Sci. U.S.A. 76, 1663- 1667. Highfield, P. E., and Ellis, R. J. (1978). Synthesis and transport of the small subunit of chloroplast ribulose bisphosphate. Nature (London) 271, 420-424. Inouye, M., and Halegoua, S. (1980). Secretion and membrane localization of proteins in Escherichia coli. CRC Crir. Rev. Biochem. 7, 339-371. Jaussi, R., Sonderegger, P., Fluckiger, J., and Christen, P. (1982). Biosynthesis and topogenesis of aspartate aminotransferase isoenzymes in chicken embryo fibroblasts. The precursor of the mitochondrial isoenzyme is either imported into mitochondria or degraded in the cytosol. J . Biol. Chem. 257, 13334-13340. Kaback, H. R., Reeves, J. P., Short, S. A,, and Lombardi, F. 1. (1974). Mechanisms of active transport in isolated bacterial membrane vesicles. XVIII. The mechanism of action of carbonylcyanide m-chlorophenylhydrazone.Arch. Biochem. Biophys. 160, 2 15-222. Kaput, J., Goltz, S., and Blobel, G. (1982). Nucleotide sequence of the yeast nuclear gene for cytochrome c peroxidase precursor. Functional implications of the pre-sequence for protein transport into mitochondria. J. Biol. Chem. 257, 15054- 15058. Kellems, R. E., and Butow, R. A. (1972). Cytoplasmic type 80s ribosomes associated with yeast mitochondria. I. Evidence for ribosome binding sites on yeast mitochondria. J. Biol. Chem. 247, 8043-8050. Kellems, R. E., and Butow, R. A. (1974). Cytoplasmic type 80s ribosomes associated with yeast mitochondria. 111. Changes in the amount of bound ribosomes in response to changes in metabolic state. J. Biol. Chem. 249, 3304-3310. Kellems, R. E., Allison, V. F., and Butow, R. A. (1974). Cytoplasmic type 80s ribosomes associated with yeast mitochondria. 11. Evidence for the association of cytoplasmic ribosomes with the outer mitochondrial membrane in situ. J . Biol. Chem. 249, 3297-3303. Kellems, R. E., Allison, V. F., and Butow, R. A. (1975). Cytoplasmic type 80s ribosomes associated with yeast mitochondria. IV. Attachment of ribosomes to the outer membrane of isolated mitochondria. J. Cell Biol. 65, 1-14. Kolansky, D. M., Conboy, J. G., Fenton, W. A., and Rosenberg, L. E. (1982). Energy-dependent translocation of the precursor of omithine transcarbamylase by isolated rat liver mitochondria. J. Biol. Chem. 257, 8467-8471. Korb, H., and Neupert, W. (1978). Biogenesis of cytochrome c in Neurospora crassa. Synthesis of apocytochrome c, transfer to mitochondria and conversion to holocytochrome c. Eur. J. Biochem. 91, 609-620. KovBE, L., Lachowicz, T. M., and Slonimski, P. P. (1967). Biochemical genetics of oxidative phosphorylation. Science 158, 1564-1567. Kraus, J. P., Conboy, J. G., and Rosenberg, L. E. (1981). Pre-ornithine transcarbamylase. Properties of the cytoplasmic precursor of a mitochondrial matrix enzyme. J . B i d . Chem. 256, 1073910742. Lederer, F., Simon, A.-M., and Verdiere, I. (1972). Saccharomyces cerevisiae iso-cytochromes c: Revision of the amino acid sequence between the cysteine residues. Biochem. Biophys. Res. Commun. 47, 55-58. Li, Y., Leonard, K., and Weiss, H. (1981). Membrane-bound and water-soluble cytochrome cI from Neurospora mitochondria. Eur. J . Biochem. 116, 199-205. Lustig, A,, Levens, D., and Rabinowitz, M. (1982). The biogenesis and regulation of yeast mitochondrial RNA polymerase. J. Biol. Chem. 257, 5800-5808. McAda, P. C., and pouglas, M. G. (1982). A neutral nietallo endoprotease involved in the processing of an FI-ATPase subunit precursor in mitochondria. J. B i d . Chem. 257, 3177-3182.

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

333

Maccecchini, M.-L., Rudin. Y., Blobel, G . , and Schatz, G. (1979a). Import of proteins into mitochondria: Precursor forms of the extraniitochondrially made F,-ATPase subunits in yeast. Proc. Natl. Acad. Sci. U . S . A . 76, 343-347. Maccecchini, M.-L., Rudin, Y., and Schatz, G. (1979b). Transport of proteins across the mitochondrial outer membrane. A precursor form of the cytoplasmically made intermembrane enzyme cytochrome c peroxidase. J . Biol. Chem. 254, 7468-7471. Macino. G . . and Tzagoloff, A. (1979). Assembly of the mitochondrial membrane system. The DNA sequence of a mitochondrial ATPase gene in Succharomvces cerevisiae. J . Biol. Chem. 254, 4617-4623. Matner, R. R., and Sherman, F. (1982). Differential accumulation of two Apo-iso-cytochromes c in processing mutants of yeast. J . B i d . Chrm. 257, 981 1-9821. Meek, R . L., Walsh, K. A., and Palmiter, R. D. (1982). The signal sequence ofovalbumin is located near the NH2 terminus. J . B i d . Chem. 257, 12245-12251. Michel, R . , Wachter, E.. and Sebald, W. (1979). Synthesis of a larger precursor for the proteolipid subunit of the mitochondrial ATPase complex of Neurospora crussa in a cell-free wheat germ system. FEBS Lett. 101, 373-376. Mihara. K., Blobel, G . , and Sato, R. (1982). I n v i m synthesis and integration into mitochondria of porin. a major protein of the outer mitochondrial membrane of Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. U . S . A . 79, 7102-7106. Miura, S . , Mori, M., Amaya, Y . , Tatibana, M . . and Cohen, P. P. (1981). Aggregation states of precursors for mitochondrial carbamoyl-phosphate synthase I and ornithine carbamoyltransferase. Biochem. I n t . 2, 305-3 12. Miura, S . , Mori, M., Amaya, Y., and Tatibana, M . (1982a). A mitochondrial protease that cleaves the precursor of ornithine carbamoyltransferase. Purification and properties. Eur. J . Biochem. 122, 64-647. Miura, S . . Mori, M., Morita, T., and Tatibana, M. (1982b). A high isoelectric point of mitochondrial ornithine carbamoyltransferase precursor and inhibition of its processing in v i m by basic proteins. Biochem. Ini. 4, 201-208. Miura, S.. Mori, M., and Tatibana, M. (1983). Transport of ornithine carbamoyltransferase precursor into mitochondria. Stimulation by potassium ion, magnesium ion and a reticulocyte cytosolic protein(s). J . Biol. Chem. 258, 6671-6674. Montgomery, D. L., Hall, B. D., Gillam, S . , and Smith. M. (1978). Identification and isolation of the yeast cytochrome c gene. Cell 14, 673-680. Mori. M.. Miura, S . , Tatibana, M., and Cohen, P. P. (1980). Characterization of a protease apparently involved in processing of pre-ornithine transcarbamylase of rat liver. Proc. Natl. A c d . Sci. U . S . A . 77, 7044-7048. Mori, M., Morita, T., Ikeda, F . . Amaya, Y., Tatibana, M.. and Cohen, P. P. (1981a). Synthesis, intracellular transport, and processing of the precursors for mitochondrial ornithine transcarbamylase and carbamoyl-phosphate synthetase I in isolated hepatocytes. Proc. Natl. Acad. Sci. U . S . A . 78, 6056-6060. Mori. M.. Morita. T . , Miura, S . , and Tatibana, M. (1981b). Uptake and processing ofthe precursor for rat liver ornithine transcarbamylase by isolated mitochondria. Inhibition by uncouplers. J . B i d . Chem. 256, 8263-8266. Morita, T.. Miura, S . , Mori, M . , and Tatibana. T. (1982a). Transport of the precursor for rat-liver ornithine carbamoyltransferase into mitochondria in v i m . Eur. J . Biochem. 122, 501-508. Morita, T., Mori, M., Ikeda, F., and Tatibana, M. (1982b). Transport of carbamyl phosphate synthetase I and ornithine transcarbumylase into mitochondria. Inhibition by rhodamine 123 and accumulation of enzyme precursors in isolated hepatocytes. J . B i d . Chem. 257, 10547- 10550. Nagata, S . , Tsunetsugu-Yokota. Y . , Naito, A,. and Kaziro, Y. (1983). Molecular cloning and sequence determination of the nuclear gene coding for mitochondria] elongation factor Tu of Saccharomvces cerc4siae. Proc. Null. Acad. Sci. U . S . A . 80, 6 192-6196.

334

GRAEME A. REID

Narita, K., and Titani, K. (1969). The complete amino acid sequence in baker’s yeast cytochrome c . J . Biochem. (Tokyo) 65, 259-267. Nelson, N . , and Schatz, G. (1979). Energy-dependent processing of cytoplasmically made precursors to mitochondrial proteins. Proc. Natl. Arad. Sci. U.S.A. 76, 4365-4369. Neupert, W., and Schatz, G. (1981). How proteins are transported into mitochondria. Trends Biochem. Sci. 6, 1-4. Ohashi, A,, Gibson, J., Gregor, I., and Schatz, G. (1982). Import of proteins into mitochondria. The precursor of cytochrome cl is processed in two steps, one of them heme-dependent. J. Biol. Chem. 257, 13042-13047. Ohta, S., and Schatz, G. (1984). A Purified precursor polypeptide requires a cytosolic protein fraction for import into mitochondria. EMBO J . 3, 651-657. O’Malley, K., Pratt, P., Robertson, J., Lilly, M., and Douglas, M. G. (1982). Selection of the nuclear gene for the mitochondrial adenine nucleotide translocator by genetic complementation of the opl mutation in yeast. J. Biol. Chem. 257, 2097-2103. Osborn, M. J., and Wu, H. C. P. (1980). Proteins of the outer membrane of Gram-negative bacteria. Annu. Rev. Microbiol. 34, 369-422. Pfaff, E., Klingenberg, M., Ritt, E., and Vogell, W. (1968). Korrelation des unspezifisch permeablen mitochondrialen Raumes mit dem “Intermembran-Raum.” Eur. J . Biorhem. 5, 222232. Raymond, Y., and Shore, G. C . (1981). Processing of the precursor for the mitochondrial enzyme, carbamyl phosphate synthetase. Inhibition by p-aminobenzamidine leads to the very rapid degradation (clearing) of the precursor. J . B i d . Chem. 256, 2087-2090. Reid, G. A., and Schatz, G. (l982a). Import of proteins into mitochondria. Yeast cells grown in the presence of carbonylcyanide m-chlorophenylhydrazone accumulate massive amounts of some mitochondrial precursor polypeptides. J. B i d . Chem. 257, 13056- 13061. Reid, G. A , , and Schatz, G. (1982b). Import of proteins into mitochondria. Extramitochondrial pools and post-translational import of mitochondrial protein precursors in vivo. J . B i d . Chem. 257, 13062-13067. Reid, G. A., Yonetani, T . , and Schatz, G. (1982). Import of proteins into mitochondria. Import and maturation of the mitochondrial intermembrane space enzymes cytochrome b2 and cytochrome c peroxidase in intact yeast cells. J . B i d . Chem. 257, 13068-13074. Riezman, H., Hay, R., Gasser, S., Daum, G . , Schneider, G., Witte, C., and Schatz, G. (1983a). The outer membrane of yeast mitochondria: Isolation of outside-out sealed vesicles. EMBO J . 2, 1105-1111. Riezman, H., Hay, R., Witte, C., Nelson, N . , and Schatz, G. (1983b). Yeast mitochondrial outer membrane specifically binds cytoplasmically-synthesized precursors of mitochondrial proteins. E M B O J . 2, 1113-1118. Riezman, H., Hase, T., van Loon, A. P. G. M., Grivell, L. A., Suda, K., and Schatz, G . ( 1 9 8 3 ~ ) . Import of proteins into mitochondria: A 70 kilodalton outer membrane protein with a large carboxy-terminal deletion is still transported to the outer membrane. EMBO J . 2, 2161-2168. Schatz, G . (1968). Impaired binding of mitochondrial adenosine triphosphatase in the cytoplasmic “petite” mutant of Saccharomyces cerevisiae. J . B i d . Chem. 243, 2192-2199. Schatz, G. (1979). How mitochondria import proteins from the cytoplasm. FEBS Len. 103, 203211. Schatz, G., and Butow, R. A. (1983). How are proteins imported into mitochondria? Cell 32, 316318. Schatz, G., and Mason, T . L. (1974). The biosynthesis of mitochondrial proteins. Annu. Rev. Biochem. 43, 51-87. Schekman, R. (1982). The secretory pathway in yeast. Trends Biochem. Sci. 7, 243-246. Schleyer, M . , and Neupert, W. (1984). Transport of ADP/ATP carrier into mitochondria. Precursor

7. TRANSPORT OF PROTEINS INTO MITOCHONDRIA

335

imported in vitro acquires functional properties of the mature protein. J. Eiol. Chem. 259, 3487-3491. Schleyer, M.. Schmidt, B.. and Neupert, W. (1982). Requirement of a membrane potential for the posttranslational transfer of proteins into mitochondria. Eur. J. Eiochem. 125, 109- I 16. Schmidt, B., Hennig, B., Zimmermann, R., and Neupert, W. (1983a). Biosynthetic pathway of mitochondrial ATPase subunit 9 in Neurospora crassa. J. Cell Eiol. 96, 248-255. Schmidt, B., Hennig, B., Kohler. H . . and Neupert. W. (1983b). Transport of the precursor to Neurospora ATPase subunit 9 into yeast mitochondria. Implications on the diversity of the transport mechanism. J. Eiol. Chem. 258, 4687-4689. Sebald, W., Machleidt, W., and Wachter, E. (1980). N.N‘-dicyclohexylcarbodiimidebind specifically to a single glutamyl residue of the proteolipid subunit of the mitochondrial adenosinetriphosphatases from Neurcispora c‘rassa and Saccharoinyces cerevisiae. Proc‘. Nut/. Acad. Sci. U . S . A . 77, 785-789. Shore, G . C. (1979). Synthesis of intracellular membrane proteins in vitro. Relation between rough endoplasmic reticulum and mitochondrial outer membrane. J . Cell Sci. 38, 137- 153. Shore, G . C.. Power. F.. Bendayan, M., and Carignan, P. (1981). Biogenesis of a 35-kilodalton protein associated with outer mitochondrial membrane in rat liver. J. Eiol. Chem. 256, 87618766. Stewart, J. W.. and Sherman, F. (1974). Yeast frameshift mutations identified by sequence changes in iso- 1 -cytochrome c. In “Molecular and Environmental Aspects of Mutagenesis” (L. Prakesh, F. Sherman, M. W. Miller, C. W . Lawrence, and H. W. Tabor, eds.), pp. 102-107. Thomas, Springfield, Illinois. Suissa, M., and Schatz, G . (1982). Import of proteins into mitochondria. Translatable mRNAs for imported mitochondrial proteins arc present in free as well as mitochondria-bound cytoplasmic polysomes. J. Eiol. Chem. 257, 13048-13055. Suissa, M., Suda. K., and Schatz. G . (1984). Isolation of the nuclear yeast genes for citrate synthase and fifteen other mitochondrial proteins by a new screening approach. EMEO J . 3, 1773-1781, Teintze, M., Slaughter, M., Weiss, H., and Neupert, W. (1982). Biogenesis of mitochondrial ubiquinol: Cytochrome c reductase (cytochrome bel complex). Precursor proteins and their transfer into mitochondria. J . Eiol. Chem. 257, 10364-10371. Tzagoloff, A,, Akai, A,, and Needleman, R. B. (1975). Assembly of the mitochondrial membrane system. Characterization of nuclear mutants of Saccharomvces rerevisiae with defects in mitochondrial ATPase and respiratory enzymes. J . Eiol. Chem. 250, 8228-8235. Tzagoloff. A , , Macino, G . , and Sebald. W. (1979). Mitochondrial genes and translation products. Annu. Rev. Eiochem. 48, 419-441. Van Loon, A. P. G . M., de Groot, R . J., van Eyk. E., van der Horst. G . T. I., and Grivell, L. A. (1982). Isolation and characterization of nuclear genes coding for subunits of the yeast ubiquinol-cytochrome c reductase complex. Gene 20, 323-337. Van Loon, A. P. G . M., de Groot, R. J., de Haan. M., Dekker, A , , and Grivell. L. A. (1984). The DNA sequence of the nuclear gene coding for the 17kd subunit VI of the yeast ubiquinolcytochrome c reductase: A protein with an extremely high content of acidic amino acids. EMEO J. 3, 1039-1043. Viebrock, A., Perz, A,, and Sebald, W. (1982). The imported preprotein of the proteolipid subunit of the mitochondrial ATP synthase from Neurospora crcrssa. Molecular cloning and sequencing of the mRNA. EMEO J. I , 565-571. Wakabayashi, S . , Matsubara, H., Kim, C. H . , Kawai, K . , and King, T. E. (1980). The complete amino acid sequence of bovine heart cytochrome c I. Eiochern. Biophys. Res. Commun. 97, 1548- 1554. Watanabe, K., and Kubo, S. (1982). Mitochondrial adenylate kinase from chicken liver. Purification, characterization and its cell-free synthesis. Eur. J. Eiochem. 123, 587-592.

336

GRAEME A. REID

Zalman, L. S., Nikaido, H., and Kagawa, Y. (1980). Mitochondria1 outer membrane contains a protein producing nonspecific diffusion channels. J . B i d . Chem. 255, 1771- 1774. Zimmermann, R., and Neupert, W. (1980). Transport of proteins into mitochondria. Posttranslational transfer of ADP/ATP carrier into mitochondria in v i m . Eur. J . Biochem. 109,217-229. Zimmermann, R., Paluch, U., and Neupert, W. (1979). Cell-free synthesis of cytochrome c. FEBS Lett. 108, 141-146. Zimmermann, R., Hennig, B . , and Neupert, W. (1981). Different transport pathways of individual precursor proteins in mitochondria. Eur. J . Biochem. 116, 455-460. Zwizinski, C., Schleyer, M., and Neupert, W . (1983). Transfer of proteins into mitochondria. Precursor to the ADP/ATP carrier binds to receptor sites on isolated mitochondria. J . B i d . Chem. 258, 4071-4084.

CURRENT TOPICS IN MEMBRANES AND TRANSPORT. VOLUME 24

Chapter 8 Assembly of the Sarcoplasmic Reticulum during Muscle Development DAVID H . MACLENNAN,* ELIZABETH ZUBRZYCKA-GAARN,* A N D ANNELISE 0 . JORGENSEN' *5anting and Best Deparrment of'Medical Reseurrh Charles H . Best Institute University of Toronto Toronto. Ontario, Canada +Department of'Anatomy Medical Sciences Building University of Toronto Toronto, Ontario. Canada

I.

Introduction. . . _ _ .. .. . . . .

111.

IV. V.

VI.

...........................................

338 338 A. Structure of the Sarcoplasmic Reticulum Membrane in Adult 338 B. Composition of the Sarcoplasmic Reticulum from Adult Skeletal Muscle , . 339 Biogenesis of the Sarcoplasmic Reticulum during Muscle Cell Differentiation . . . . . . 342 A. Ultrastructure of the Sarcoplasmic Reticulum during Develop 343 B. Ultrastructure of Transverse Tubules during Development . . . . . . . . . . . . . . . . . . 344 C. Triad Formation during Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . , . . . . . . 344 D. Changes in Function and Composition of Sarcoplasrnic Reticulum during Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , . . , , . . . , , . . . . 345 E. Changes in the Activity of Phospholipid-Synthesizing Enzymes in Sarcoplasmic Reticulum during Development . . . . . . . . . . . . . . . . . . . . . . . . . . . , , . . , , . . . . , . . 348 F. Changes in Morphology of Sarcoplasmic Reticulum during Development. . . . . . 349 Models of Sarcoplasmic Reticulum Biogenesis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 I Synthesis of Specific Sarcoplasmic Reticulum Proteins . . . _ . . _ . . . , . . . . . . . . . .354 A. Synthesis of Sarcoplasmic Reticulum Proteins in Muscle Cell Cultures. . . . . . . . 354 B. Synthesis of Sarcoplasmic Reticulum Proteins in Vitro . . . . . . . . . . . , , , . , , . . . . 356 Regulation of the Biosynthesis of Sarcoplasmic Reticulum Proteins.. . . . . . . , . . 358 A. Effect of Calcium on Synthesis of Sarcoplasmic Reticulum Proteins . . . . . , . . . . 358 B. Neural Control of Transformation of Fast-Twitch to Slow-Twitch Muscle.. . . . . 360 . . . . . , , , , , . . . . , . , , . . . . . . . . . . . . , , . . . . . . 36 I C. Cloning of the ATPase Gene References . . . , , . , , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362

11. The Sarcoplasmic Reticulum

. .. . ... . .. . . ... . .. . . ,. ... . ... .

337

Copyright 0 1985 by Acadeniic Press. Inc All rights of reproduction in any form reserved. ISBN 0-12-153324-7

338

DAVID H. MACLENNAN ET AL.

I.

INTRODUCTION

The sarcoplasmic reticulum membrane from mammalian or avian skeletal muscle, like a number of other membrane systems whose synthesis has been successfully probed, has certain unique features that recommend its study. The membrane has relatively few major proteins, all of which have been well characterized with respect to their location and orientation within the membrane (MacLennan and Holland, 1975; Campbell and MacLennan, 1981). These proteins have been purified and antibodies have been raised against them. Since they exist in high density in the membrane, these proteins can be readily distinguished within muscle cells by fluorescein- or ferritin-conjugated antibody labeling (Jorgensen et al., 1977, 1979, 1982a,b, 1983). The Ca2+ transport and sequestration system of sarcoplasmic reticulum membranes is a differentiated function in muscle cells. Synthesis of the membrane is turned on during differentiation in vivo, and it is possible to follow the synthesis of the membrane in prenatal and postnatal avian or mammalian muscle (Boland et al., 1974; Sarzala et al., 1975a). Differentiation also occurs in vitro when myoblasts form myotubes in tissue culture. Because of this fact, it is possible to synchronize the initiation of synthesis of the sarcoplasmic reticulum membrane and to study the early stages of membrane assembly (Holland and MacLennan, 1976; Martonosi et al., 1977). The genetic control of the synthesis of sarcoplasmic reticulum has not been amenable to study as yet but advances in recombinant DNA technology may open the potential for a genetic approach to assembly of the sarcoplasmic reticulum (MacLennan et al., 1983).

II. THE SARCOPLASMIC RETICULUM A. Structure of the Sarcoplasmic Reticulum Membrane in Adult Skeletal Muscle The sarcoplasmic reticulum is an organelle system wholly contained within muscle cells, where it functions in the control of intracellular free Ca2+ concentrations (Hasselbach, 1964; Weber, 1966; Ebashi et al., 1969). The membrane is organized around each myofibril like a fenestrated water jacket. Its structure is related to the sarcomeric organization of the myofibril (Porter and Palade, 1957) and, in mammalian skeletal muscle, it is segmented at the level of each A-I junction so that one segment overlies the A band and another segment overlies the I band (see Fig. 1). The longitudinal tubules which form the midsections of these segments are free of luminal structure. They anastomose at both ends to form the transversely oriented, matrix-filled terminal cisternae. The terminal cisternae from adjacent segments are separated by transverse tubules, which are tubular invaginations of the plasma membranes mostly oriented per-

8. ASSEMBLY OF SARCOPLASMIC RETICULUM

339

FIG. I. A longitudinal section of extensor digitorum longus muscle (EDL) from a 30-day-old mouse (from Luff and Atwood, 1971). Thick filaments can bc seen in the center of the A band region; thin filaments are seen in the I band. The transverse tubular system and the sarcoplasmic reticulum are in the center of the picture. Transverse tubules running perpendicular to the fibers can be seen near the A-I junction. The sarcoplasmic reticulum consists, in part, of convoluted tubules overlying the A and I band regions; at the A-I junction it is thickened to terminal cisternae, which abut the transverse tubular system. The matrix in the terminal cisternae is believed to consist of calsequestrin. Reproduced from J . Cell B i d 51, 369-383 by copyright permission of the Rockefeller Univ. Press.

pendicular to the longitudinal axis of the myotubes and which carry electrical impulses into the interior of the cell. A transverse section through two terminal cistemae and a transverse tubule reveals a structure resembling three joined rings. This structure is referred to as a triad. The transverse tubules are connected to the terminal cistemae by amorphous, regularly spaced structures referred to as “feet” (Franzini-Armstrong, 1970, 1975). The feet originate in the sarcoplasmic reticulum as a series of outward dimples 25-30 nm apart, usually organized in rows; they bridge the 12-nm gap between the two membranes and usually terminate in the transverse tubules. 6. Composition of the Sarcoplasmic Reticulum from Adult Skeletal Muscle

The sarcoplasmic reticulum is composed of about one-third phospholipid and neutral lipids and two-thirds protein (Meissner and Fleischer, 1971). There are several major proteins in the sarcoplasmic reticulum and an unknown number of minor enzymatic activities that may or may not be related directly to the Ca2 regulatory mechanisms of the membrane. The Ca2 ,Mg2 -dependent ATPase enzyme (Mr, 100,000) is the major membrane protein, accounting for about two-thirds of the total protein of the +

+

+

340

DAVID H. MAC LENNAN ET AL.

membrane system. It carries out the enzymatic function of Ca2+ transport (de Meis and Vianna, 1979). Knowledge of the contributions of the ATPase to membrane structure has come from studies with the purified protein (MacLennan and Reithmeier, 1982). The purified ATPase can be incorporated into vesicular structures composed either of native phospholipid or of excess, added phospholipid. When these vesicles were freeze-fractured (MacLennan et al., 1971), they were seen to contain 8-nm intramembrane particles, identical in size to those first observed in intact sarcoplasmic reticulum and postulated to be the Ca2 ,Mg2 +-ATPase (Deamer and Baskin, 1969). These observations suggest, first, that the 8-nm particles in the sarcoplasmic reticulum are constituted of the Ca2 ,Mg2 -ATPase and, second, that the ATPase is a transmembrane protein with a considerable portion of its structure buried within the hydrophobic bilayer. Vesicles formed from the purified ATPase plus phospholipid also displayed 4nm surface particles (Migala et al., 1973; Thorley-Lawson and Green, 1973; Stewart and MacLennan, 1974) comparable to those observed earlier on intact sarcoplasmic reticulum (Inesi and Asai, 1968; Ikemoto et al., 1968). Since these structures could only represent portions of the ATPase extending into cytoplasmic regions it was apparent that the ATPase is an amphipathic molecule, part of which is buried in hydrophobic regions of the membrane and part of which is hydrophilic and exposed on the cytoplasmic surface. Allen and Green (Allen, 1977, 1980a,b; Allen et al., 1980a,b) have sequenced water-soluble peptides obtained from the purified ATPase. They have aligned the hydrophilic peptides into five sequences: a 32-amino acid amino-terminal peptide, an 8-amino acid carboxyl terminal peptide, and three internal fragments of 116, 298, and 122 amino acids, respectively. These sequences have been localized within the molecule, and it appears that the five hydrophilic peptides are separated from each other by four hydrophobic regions of about 100 amino acids each (MacLennan and Reithmeier, 1982). Since the hydrophilic peptides have been localized on the cytoplasmic surface of the membrane, the polypeptide chain seems to begin in the cytoplasm and to stitch back and forth through the membrane eight times in four transmembrane loops. The hydrophilic stretches make up the 4-nm surface particles; the hydrophobic stretches constitute the intramembrane particles. Jorgensen et al. (1979, 1982a,b) have used immunofluorescence and immunoferritin labeling techniques to show that the ATPase is localized throughout the longitudinal sarcoplasmic reticulum and nonjunctional regions of the terminal cisternae but is absent from the junctional region of the terminal cisternae, the region apposed to the terminal cisternae. This finding is in agreement with earlier freeze-fracture studies of the sarcoplasmic reticulum in skeletal muscle in situ (Franzini-Armstrong, 1975), studies which showed that 8-nm intramembrane particles were densely and uniformly distributed throughout the free sarcoplasmic reticulum membrane but were absent from the junctional membrane. +

+

+

341

8. ASSEMBLY OF SARCOPLASMIC RETICULUM

Biochemical studies showing localization of the ATPase in light and heavy fractions of the sarcoplasmic reticulum, believed to originate in longitudinal and terminal cisternal regions of the membrane, respectively (Meissner, 1975), are also consistent with the view that the ATPase is localized throughout the free sarcoplasmic reticulum membrane. Two glycoproteins are also intrinsic to the sarcoplasmic reticulum membrane (Campbell and MacLennan, 1981). These proteins are 53,000 and 160,000 Da and both contain high-mannose carbohydrate chains. The 53,000-Da protein contains 2 mol of GlcNac,:Man,, but the sugar composition of the 160,000-Da protein has not been determined. The 53,000-Da protein is clearly a transmembrane protein, with sugar residues o n the inside and with the bulk of its sequence on the cytoplasmic surface. Both glycoproteins bind ATP (Campbell and MacLennan, 1983), and there is some evidence that the 160,000-Da glycoprotein binds Ca2+ (Campbell et a / . , 1983). The contribution of these proteins to the structure of the membrane is not known. Jorgensen et al. (1981), using immunofluorescent labeling with an antibody that reacted with both the 53,000- and 160,000-Da glycoproteins, showed that one or both of these proteins were localized in the terminal cisternae. The sarcoplasmic reticulum also contains one or more low-molecular-weight proteolipids (MacLennan et a / . , 1972). One was purified by thin-layer chromatography and was shown to have a relatively hydrophilic amino acid composition but to contain one or two fatty acid moieties per mole, a finding accounting for its hydrophobic behavior. It is probable that these are also transmembrane proteins, but their contributions to structure are undefined. There are two extrinsic proteins associated with the luminal space of the sarcoplasmic reticulum: calsequestrin (MacLennan and Wong, 1971) and a highaffinity Ca2+-binding protein (Ostwald and MacLennan, 1974; Michalak et a / ., 1980). Calsequestrin is a major sarcoplasmic reticulum protein accounting for about 7% of the total membrane protein, but the high-affinity Ca2 -binding protein is present in only about one-tenth the amount of calsequestrin. Calsequestrin is a very acidic protein (Mr, 63,000) containing about 38% glutamic and aspartic acids and only 7% lysine and arginine. It binds nearly 1000 nmol of Ca2+ per milligram of protein with a dissociation constant of about 1 mM in the presence of physiological salt. In the process of Ca2+ binding, the protein undergoes vast conformational changes and, upon binding sufficient Ca2 , precipitates from solution (Ikemoto et ul., 197 1). Because calsequestrin is luminally located and because it binds so much Ca2+, it has been postulated that it acts to sequester Ca2 in the interior of the sarcoplasmic reticulum; hence the name, calsequestrin (MacLennan and Wong, 1971). Meissner ( 1975) found that sarcoplasmic reticulum vesicles could be separated on sucrose density gradients into light, intermediate, and heavy fractions. The light fractions were free of calsequestrin and of matrix structure whereas the +

+

+

342

DAVID H.MAC LENNAN ET AL.

heavy fraction contained both matrix structure and calsequestrin. These experiments suggested that calsequestrin contributes to the matrix structure and, since matrices were only observed in terminal cisternal regions of the intact sarcoplasmic reticulum, that calsequestrin was uniquely located in the terminal cisternal region of the membrane system. This postulate has been proved by both immunofluorescence and immunoferritin labeling of thin sections of skeletal muscle (Jorgensen et al., 1979, 1983). Such a localization would suggest that Ca2 pumped into the lumen of the sarcoplasmic reticulum would diffuse within this structure but be concentrated at Ca2+-binding sites provided by calsequestrin in the terminal cisternal regions. Indeed, electron microprobe analysis has shown that Ca2+ is concentrated in the terminal cisternae of resting muscle and that the concentration of Ca2 is greatly reduced in that region after periods of tetanic stimulation (Somlyo el al., 1981). The high affinity Ca2 -binding protein is a luminally located protein, (Michalak et al., 1980) of M,54,000 (Ostwald and MacLennan 1974). It binds 1 mol of Ca2+ per mole with a dissociation constant of about 4 rJ-M, similar to that of other high-affinity or “EF hand” type proteins (Kretsinger, 1976). The function of the protein is unknown. There is some evidence that the protein is concentrated in transverse tubules (Michalak et al., 1980), and this might suggest that it is found in sarcoplasmic reticulum only as a contaminant. In summary, biochemical and morphological studies have demonstrated that the two major proteins of the sarcoplasmic reticulum, the Ca2 ,Mg2 -dependent ATPase and calsequestrin, are distributed nonuniformly within the various regions of this membrane system. The ATPase is evenly and densely distributed throughout the free sarcoplasmic reticulum but is absent from the junctional sarcoplasmic reticulum. By contrast, most of the calsequestrin is confined to the lumen of the terminal cisternae. Further studies are required to determine the spatial distribution of the high-affinity calcium-binding protein as well as of the proteolipids and the 53,000- and 160,000-Da glycoproteins within the sarcoplasmic reticulum membrane system. +

+

+

+

111.

+

BlOGENESlS OF THE SARCOPLASMIC RETICULUM DURING MUSCLE CELL DIFFERENTIATION

The sarcoplasmic reticulum carries out a differentiated function in mu 1: cells. The precursors to multinucleated myotubes, the myoblasts or satellite cells, do not have an extensive organellar network for the control of Ca2+, yet such a network develops in multinucleated myotubes. Biogenesis of the sarcopolasmic reticulum has been studied either in vivo (the development of chick embryo musculature pre- and posthatching and the development of rabbit mus-

8. ASSEMBLY OF SARCOPLASMIC RETICULUM

343

culature pre- and postnatal) or in cell culture (the study of developing membrane systems in differentiating chick, rat, and rabbit muscle cells in culture).

A. Ultrastructure of the Sarcoplasmic Reticulum during Development Ezerman and Ishikawa (1967) were among the first to observe the morphological appearance of the sarcoplasmic reticulum in chick embryo muscle cells differentiating in culture. Since mononucleated myoblasts could not be distinguished from fibroblasts, they examined bi- and multinucleated myotubules which already showed myofibril formation. In these cells, the smooth endoplasmic reticulum was first observed as narrow tubular extensions of the rough endoplasmic reticulum. Similar observations have been made in a study of mouse cardiac myocytes developing in vivo (Ishikawa and Yamada, 1975). At later stages of development, the tubular projections branched and anastomosed to form a honeycomb arrangement. A honeycomb arrangement of sarcoplasmic reticulum surrounding the myofibrils in immature myotubes has been observed in studies both in situ (Walker and Schrodt, 1968; Schiaffino and Margreth, 1969; Edge, 1970; Luff and Atwood, 1971; Tomanek and Colling-Saltin, 1977) and in cell culture (Ezerman and lshikawa, 1967; Shimada et a / . , 1967). In contrast to the arrangement in older myotubes, the sarcoplasmic reticulum in early myotubes appeared to be continuous from sarcomere to sarcomere and to be undifferentiated. Electron-dense material was not apparent in the lumen of the terminal cisternae and subsarcolemmal vesicles until after the sarcoplasmic reticulum had become closely apposed to either the transverse tubules or the sarcolemma (Walker et al., 197 1; Spray et a / ., 1974). It is controversial whether myofibrils or sarcoplasmic reticulum form first. Fischman (1970, 1972) did not observe a smooth membrane system surrounding myofibrils at early stages of development of chick muscle. He believed, therefore, that the sarcotubular system could not play a formative role in the assembly of the filament lattice but rather that the myofibril would form a scaffolding around which the membrane system would be elaborated. By contrast, Walker et al. (1975) suggested that the sarcoplasmic reticulum provides a framework for myofibril formation in embryonic muscle of human and monkey on the basis of their observations that a branching and anastomosing network of endoplasmic reticulum tubules was evident prior to the appearance of myofibrils. Kilarski and Jakubowska ( 1979) observed that sarcoplasmic reticulum tubules developed at early stages from endoplasmic reticulum tubules. They also observed a close correlation between the synthesis of myofibrils and of the sarcoplasmic reticulum and, therefore, suggested that the two systems formed simultaneously from different cellular compartments.

344

DAVID H. MAC LENNAN ET AL.

8. Ultrastructure of Transverse Tubules during

Development Studies of the formation of transverse tubules in cell culture (Ezerman and Ishikawa, 1967; Ishikawa, 1968; Schiaffino et al., 1977) and in situ (Kelly, 1971) showed that the transverse tubules first appeared as short, tubular invaginations of the sarcolemma. At early stages these newly formed transverse tubules were limited to the subsarcolemmal region of the myofiber and were only later seen to extend to the center of the myofiber. While these studies suggested that transverse tubules formed by invagination of the sarcolemma, it is equally possible that they were formed by sequential addition of vesicles to the sarcolemma in a process similar to exocytosis but serving to extend the transverse tubular system toward the center of the myofiber. Although the initial stages of transverse tubule formation appeared to occur in a similar fashion in situ and in cell culture, the later stages of transverse tubule development in situ differed significantly from development in cell culture. In the cell culture system, there was an elaborate and apparently uncoordinated proliferation of the transverse tubular system during development which was similar to the elaborate proliferation of transverse tubules observed in denervated skeletal muscle (Pellegrino and Franzini, 1963). This finding suggests that coordinated development of the sarcoplasmic reticulum and the transverse tubular system, at later stages, may require appropriate innervation of the myotube.

C. Triad Formation during Development A significant aspect in the formation of a functional sarcoplasmic reticulum is the formation of triads or junctional complexes between the sarcoplasmic reticulum and the transverse tubules. Since only a few triads are formed during the differentiation of muscle cells in cell culture, morphological aspects of triad formation have so far been limited to studies in situ. In studies of development of the transverse tubular and sarcoplasmic reticulum systems in postnatal rat or mouse muscle (Schiaffino and Margreth, 1969; Luff and Atwood, 1971), early junctional contacts were seen to be longitudinally or obliquely oriented near the cell periphery while the sarcoplasmic reticulum was continuous over several sarcomeres within the cell. Transverse triadic junctions were established at the A-I band level only after 1-2 weeks. Other studies have indicated that the presence of electron-dense material in the lumen of the sarcoplasmic reticulum and the presence of feet which join the sarcoplasmic reticulum to either the transverse tubules or the sarcolemma are concurrent events (Edge, 1970; Tomanek and Colling-Salton, 1977). The fact that connections between the combined sarcoplasmic reticulum and the transverse tubular system have rarely been ob-

I I

8. ASSEMBLY OF SARCOPLASMIC RETICULUM

345

served in early stages of triad formation has led to the suggestion (Tomanek and Colling-Salton, 1977) that these two regions are formed independently and only unite to form the sarcotubular system at an advanced stage of development. In situ studies of the formation of peripheral junctions between the sarcoplasmic reticulum and the sarcolemma have shown their appearance well before the first appearance of transverse tubules (Kelly, 1971). On the other hand, studies of cells in culture did not demonstrate the presence of these couplings but showed an extensive proliferation of transverse tubules at a very early stage of development (Ezerman and Ishikawa, 1967; Ishikawa, 1968; Schiaffino et al., 1977). Coated vesicles have been observed in developing skeletal muscle cells. Some investigators have implied that they are involved in synthesis of the sarcoplasmic reticulum (Tomanek and Colling-Salton, I977), others have suggested that they are involved in the formation of transverse tubules (Schiaffino et al., 1977). To date, there has been no successful attempt to define a function for coated vesicles in organelle biogenesis in developing muscle cells by either immunocytochemical examination or biochemical isolation.

D. Changes in Function and Composition of Sarcoplasmic Reticulum during Development Since the sarcoplasmic reticulum carries out an assayable function, i.e., the ATP-dependent uptake of Ca2 , it has been possible to monitor the development of this membrane fraction in muscle samples obtained pre- and postnatally. Such studies have been carried out largely with embryonic and neonatal chick muscle and with fetal and neonatal rabbit muscle. Fanburg et al. ( 1968) studied Ca2 uptake and ATPase activity in microsomes isolated from developing chick muscle. They observed a fivefold to sixfold increase in specific activity of Ca2+ uptake during the period 3 days prior to hatching to 1 day posthatching. Total ATPase activity increased by sixfold to sevenfold about 2 days prior to the development of Ca2+ uptake capacity, but Ca2+-dependent ATPase did not show a dramatic rise at any time. Thus, they did not observe a close temporal relationship between Ca2+ uptake and Ca2+dependent ATPase. Holland and Perry ( 1969) carried out similar studies with rabbit longissimus dorsi muscle and found an increase in ATPase specific activity 8-10 days after birth, which declined to the adult value, about 25% of maximum. They did not use EGTA buffers to control Ca2 but measured Ca2 +-dependent ATPase activity as the difference between basal ATPase and ATPase stimulated during Ca2+ uptake. This “extra” ATPase activity increased fivefold between day I and adult muscle and the Ca2+ :ATP ratio also increased about fivefold, indicat+

+

+

346

DAVID H. MAC LENNAN ET AL.

ing an increased efficiency of the system. This suggested to these authors that the ATPase might form first and later be “coupled” to Ca2+ transport. Another early study by Lough et al. (1972) was concerned with the development of the calcium transport system in microsomes isolated from chick skeletal muscle maturing in tissue culture over a period of 6 days. They observed sharp increase in total ATPase activity and in Ca2+ uptake activity between days 5 and 6 , but they were unable to correlate Ca2 uptake activity with Ca2 -dependent ATPase activity during this period. Later studies of these phenomena carried out in Martonosi’s laboratory were more sophisticated, involving extensive analyses of protein composition and of phosphoprotein formation as well as measurements of Ca2 -dependent ATPase activity and Ca2+ uptake. Boland et al., (1974) studied microsomes isolated from several different chick muscles before and after hatching. They noted a rapid increase in the rate and extent of Ca2+ accumulation around the time of hatching. They also noted a large, transient increase in total ATPase activity at this time. By careful analysis they were able to dissect out the Ca2 -dependent ATPase activity and, using gel electrophoresis, to determine the concentrations of the phosphoprotein intermediate and of the ATPase protein. They noted that the temporal appearance and increase in specific activity of the Ca2 -dependent ATPase paralleled that of Ca2 transport activity. The microsomal membranes became more dense during this period, a change reflecting a decrease in the phospholipid to protein ratio with development. During the period when Ca2+ transport increased, there was a marked decrease in palmitate and an increase in the linoleate content of membrane phospholipids and a decrease in the content of cholesterol. These changes were observed in total membranes as well as in membranes isolated after Ca2 loading by centrifugation through 40% sucrose. Sarzala et al. (1975a) carried out similar studies of the development of sarcoplasmic reticulum in microsomes isolated from fetal, neonatal, and young rabbit muscle and compared them with the properties of sarcoplasmic reticulum from adult muscle. They noted that the ATPase accounted for only about 10% of the protein content of the sarcoplasmic reticulum fraction in the earliest stages of development, rising to 70% in adult muscle. While a dramatic but transient increase in total Mg2+-dependent ATPase activity occurred near the time of birth, Ca2+-dependent ATPase and Ca2 loading capacity increased in parallel from a point just prior to birth. Sarzala et al. (1975b) also noted the presence of a Ca2 -precipitable protein, presumably calsequestrin, in microsomal membranes isolated from 3-day-old rabbits. Zubrzycka et al. (1979) confirmed the presence of calsequestrin in 3-day-old rabbit muscle using immunological techniques. Sarzala et al. (1975a) also observed that the ratio of phospholipid to protein and of neutral lipid to protein decreased with the age of the rabbits and that the content of phosphatidylcholine increased over this period while the content of phosphatidylethanolaminediminished. The major fatty acids in lecithin in fetal +

+

+

+

+

+

+

+

+

8. ASSEMBLY OF SARCOPLASMIC RETICULUM

347

microsomes were found to be monoenoic species while the major fatty acids in adult tissues were found to be dienoic species (Zubrzycka-Gaarn and Sarzala, 1980). Changes in fatty acid and phospholipid composition and in phospholipid to protein ratios were postulated to affect the permeability of the membrane and the activity of transmembrane proteins in the membrane. Volpe et al. (1982) also analyzed changes in isolated microsomal membrane composition during postnatal development of rabbit fast skeletal muscle. They observed that there was an increase in Ca2+ ATPase content in the postnatal period up to 15 days after birth while calsequestrin remained constant between days 4 and 15, but in apparently lower concentration than in adult muscle. They also assayed other activities in the microsomal fraction and used Na+,K+ATPase and cholesterol as markers for transverse tubules or sarcolemma. The content of these markers dropped dramatically during the postnatal period of development, indicating that there was a progressive decrease in contamination of the microsomal fraction with the transverse tubular system and plasma membranes. Several authors have noted that proteins of 70,000-80,000 Da are prominent in rabbit and chick microsomes isolated at early developmental stages (Boland et al., 1974; Sarzala et al., 1975a; Zubrzycka e t a l . , 1979; Volpe e t a l . , 1982). The content of these proteins decreased during development. Martonosi ( 1975) and Sarzala et al. (1975a) suggested that an 80,000-Da protein might represent a precursor of the Ca2 ,Mg2+-dependent ATPase in developing muscle. However, immunological studies (Zubrzycka er al , 1979; Zubrzycka-Gaarn er al., 1985) and studies of the biosynthesis of the ATPase in vitro (Greenway and MacLennan, 1978; Chyn et al., 1979; Reithmeier et a l . , 1980) do not provide any evidence for such a precursor. Brandt et al. (1980) and Rosemblatt et al. (198 I ) found that purified transverse tubules were enriched in proteins of M, 69,000 to 87,000. It is possible, therefore, that the 80,000-Da proteins observed in microsomes isolated from muscle tissue at early developmental stages originate from the transverse tubules. Zubrzycka et al. (1979) and Zubrzycka-Gaarn et al. (1985)have examined the question of heterogeneity of microsomal fractions from differentiating rabbit muscle. Zubrzycka et al. (1979) found that microsomes from adult or differentiating muscle could be fractionated by sucrose density gradient centrifugation into two populations, the heavier of which was more active in Ca2+ transport. Zubrzycka-Gaarn et al. (1985) achieved a second fractionation by loading the fractionated vesicles with calcium oxalate and centrifuging them through layers of 50 and 55% sucrose. Vesicles penetrated the 50 and 55% sucrose layers only when they were allowed to accumulate calcium oxalate in the presence of ATP. These vesicles were clearly loaded with calcium oxalate on the basis of their appearance following negative staining. Analysis of the protein composition of this fraction showed that it was very rich in the Ca2+-ATPase, calsequestrin, and +

348

DAVID H. MAC LENNAN ET AL.

the two intrinsic glycoproteins. In fact, its composition was remarkably similar to that of sarcoplasmic reticulum isolated from adult muscle by the same procedure (Fig. 2). A second calcium-loaded fraction was isolated from the top of the 55% sucrose layer. This fraction was also enriched in the Ca2+,Mg2+ATPase but not to the extent of the fraction passing through 55% sucrose. The unloaded fraction at the top of 50% sucrose was virtually free of sarcoplasmic reticulum-specific proteins.

E. Changes in the Activity of Phospholipid-Synthesizing Enzymes in Sarcoplasmic Reticulum during Development The synthesis of membrane phospholipids occurs through the mediation of a series of enzymatic reactions, some of which occur in the cytoplasm and some of

Fro. 2. Protein composition of various fractions before and after loading with calcium oxalate. ( I ) Microsomal fraction from 3-day-old rabbit skeletal muscle before calcium oxalate loading; (2). (3) unloaded and calcium oxalate-loaded fraction from 3-day-old muscle, respectively; (4) calcium oxalate-loaded fraction from adult muscle. 160K,53K = 160,000- and 53,000-Da glycoproteins; ATPase = Ca*+ ATPase; CS = calsequestrin.

8. ASSEMBLY OF SARCOPLASMIC RETICULUM

349

which occur in membranes. Adult sarcoplasmic reticulum contains glycerophosphate acyltransferase, lysolecithin acyltransferase, choline and ethanolamine phosphotransferase, and phosphatidylethanolamine methyltransferase (Sarzala and Pilarska, 1976). These activities were found to be, in general, as high in isolated sarcoplasmic reticulum as in isolated liver microsomes. Studies of these enzymes in developing rabbit muscle microsomes has shown that all but choline and ethanolamine phosphotransferase increased dramatically postnatally, reaching a peak of activity between 5 and 10 days and declining to one- to twothirds of peak activity in sarcoplasmic reticulum of adult animals. Choline phosphotransferase activity declined in the prenatal to postnatal period but 5 days after birth began to increase, eventually reaching a level twofold higher than the postnatal level. The highest activity of glycerolphosphate acyltransferase was observed in early postnatal development so that the most active synthesis of phosphatidic acid, an intermediate in the synthesis of both neutral and phospholipid, should have occurred at the time of most rapid membrane synthesis. The highest activity of lysolecithin acyltransferase occurred at an early time during development when there were changes in the composition of fatty acids in phosphatidylcholine. The increase in phosphatidylcholine was correlated with the increase in choline phosphotransferase only at late stages of microsomal development. The highest activity of ethanolamine phosphotransferase was found in embryonic life. Comparison of the amount of phosphatidylcholine and phosphatidylethanolamine and of the activities of the corresponding phosphotransferases suggested that the methylation pathway between phosphatidylethanolamine and phosphatidylcholine might be important. This could not be proved, however. These results suggest that the sarcoplasmic reticulum in adult and developing muscle might be a major location for phospholipid synthesis. This postulate for the localization of the phospholipid-synthesizing enzymes is also not proved, since the enzymes could as well reside in a contaminating fraction of endoplasmic reticulum. Morphological studies using antibodies against some of the most active phospholipid-synthesizing enzymes would be very informative. Unfortunately, these enzymes have not yet been purified, and antibodies have not been raised against them.

F. Changes in Morphology of Sarcoplasmic Reticulum during Development Sarzala et al. (1975a) noted changes in the morphology of membranes isolated from neonatal and postnatal rabbit muscle. In early stages of development, the vesicles appeared predominantly thick-walled and free of surface particles. By 7 days these thick-walled, smooth-surfaced vesicles were seen to be interspersed

350

DAVID H. MACLENNAN ET AL.

with thin-walled vesicles whose surfaces were coated with 4-nm particles. In adult muscle, thin-walled vesicles with surface particles predominated. There was no evidence for intermediate vesicular forms. Since the 4-nm surface particles have been shown to represent cytoplasmic extensions of the ATPase (Migala et al., 1973; Thorley-Lawson and Green, 1973; Stewart and MacLennan, 1974), it appears that within 7 days after birth the ATPase had accumulated in specific regions of the reticular network that gave rise to the microsomal fraction. Zubrzycka et al. (1979) fractionated 4-day-old rabbit muscle microsomes into two fractions by sucrose density gradient centrifugation. The lighter fraction was predominantly thick in countour by negative staining, rich in phospholipid, and free of 4-nm surface particles. About 60% of the vesicles in the heavier fraction were covered with 4-nm particles. The light fraction had little Ca2+-ATPase or Ca2+ uptake capacity but did contain calsequestrin. The heavier fraction contained both the ATPase and calsequestrin and was capable of ATP-dependent Ca2 accumulation. Tillack et al. (1974) and Baskin (1974) studied freeze-fracture patterns of microsomal vesicles isolated from embryonic and postnatal chick muscle, and Tillack et al. (1974) also studied freeze-fracture of muscle in situ at different stages of development. In these studies they equated the appearance of 8-nm intramembranous particles with the incorporation of the ATPase into microsomal membranes. These studies showed that there was an increase of severalfold in the density of 8-nm particles in microsomal membranes over the differentiation period studied-1 I-day embryos to 40- to 50-day-old chicks. The increase in particle density was temporally correlated with the increase in Ca2+ uptake activity of the microsomal fraction. The design of the freeze-fracture experiments and the way that the data were presented, however, make interpretation of these experiments somewhat more complex. Mitochondria1 fragments, plasma membrane fragments, and portions of the transverse tubular system are all possible contaminants of the microsomal preparation (Baskin, 1974), but the source of these fragments was not distinguished after freeze-fracture. While both Tillack et al. ( 1974) and Baskin (1974) tried to improve the purity of the fractions by separation of calcium oxalate-loaded vesicles from the crude microsomal fraction, these experiments were only partially successful. Baskin (1974) was able to achieve a separation of membranes as judged by their protein content. He found that the percentage of sarcoplasmic reticulum vesicles that accumulated calcium oxalate increased from 3% at 14 days embryonic, through 10% at 4 days postnatal, to 30% at 6 weeks postnatal. He did not freeze-fracture the calcium oxalate-loaded pellets from differentiating muscles, however. Tillack et al. ( 1974) fractured calcium oxalate-containing vesicles which passed through 40% sucrose. These fractions had previously been shown (Boland et al., 1974) to differ in protein composition in only minor respects from +

351

8. ASSEMBLY OF SARCOPLASMIC RETICULUM

microsomes purified by differential or sucrose gradient centrifugation. Although the fraction passing through 40% sucrose contained most of the 45Ca2+ added during the loading process, it was not clear that its uptake was ATP-dependent. In this respect, it is important to note that vesicles passing through 40% sucrose but not containing any 8-nm ATPase particles on their concave faces accounted for 30-35% of the vesicles from embryonic muscles and for 6-16% of the vesicles from postnatal muscles. Correction of the data for the contribution from particle-free vesicles lowered the estimate of the increase in particle density with time of development. The finding of increased particle density with time of development was corroborated by freeze-fracture studies of embryonic and postnatal chick muscle developing in situ (Tillack et al., 1974). These studies also showed a progressive increase in the density of 8-nm particles in the concave fracture faces of sarcoplasmic reticulum developing in situ. Tillack et al. (1974) noted that the density of 8-nm particles in situ was twice that in the isolated membrane vesicles. It is not clear whether this was due to aggregation or disaggregation phenomena upon isolation or whether the membrane populations from which average density figures were derived in situ and in isolation did not correspond. Crowe and Baskin (l978), Sabbadini et al. (1978), and Scales and Sabbadini ( 1979), using either freeze-fracture or freeze-fracture stereological measurements, showed that the transverse tubular membranes have far fewer particles than does the sarcoplasmic reticulum. Analysis of microsomal fractions from adult skeletal muscle by this technique showed a heavy contamination of sarcoplasmic reticulum preparations with transverse tubular membranes (Sabbadini er al., 1978; Scales and Sabbadini, 1979). Such studies have not yet been carried out with microsomes from developing muscle.

IV.

MODELS OF SARCOPLASMIC RETICULUM BIOGENESIS

The combined biochemical and morphological studies described so far have shown that there is a gradual increase of the Ca2+ transport system in microsoma1 vesicles during development. These studies have been used to support the postulate that phospholipid-rich membranes of embryonic muscle are gradually converted during development into Ca2 transporting structures by stepwise insertion of the ATPase enzyme and other components of the transport system (Boland et al., 1974). The rationale for this postulate has been reviewed by Martonosi (1982) and Martonosi et al. (1980). Some of the experiments (e.g., those of Zubrzycka-Gaarn et al., 1985) do not support this postulate, however. Their finding that a small fraction of rather well-developed sarcoplasmic reticulum exists in microsomal fractions from developing muscle raises the pos+

352

DAVID H. MAC LENNAN ET AL.

sibility that these microsomal fractions may not be a homogeneous population undergoing a gradual conversion from multiple to limited functions by the incorporation of differentiation-specific proteins into a preexisting membrane. Instead, the microsomal fractions might be a heterogeneous population of vesicles originating from a variety of cellular organelles which include an increasing proportion of differentiated sarcoplasmic reticulum as development proceeds. Consideration of the available evidence, therefore, shows that there are at least two fundamentally different ways of interpreting the data concerning the biogenesis of the sarcoplasmic reticulum membrane. In undifferentiated muscle cells there is a well-developed system of smooth and rough endoplasmic reticulum. This system presumably carries on the synthetic and interconversion reactions associated with housekeeping functions in a cell. In this respect it is multifunctional. This membrane system is responsible, following appropriate signals to initiate differentiation, for the synthesis of the major proteins of the sarcoplasmic reticulum. Since the rough endoplasmic reticulum is the site for the synthesis of the Ca2+-ATPase and calsequestrin (Greenway and MacLennan, 1978; Chyn er al., 1979; Reithmeier et al., 1980), synthetic incorporation of the Ca2 -ATPase into this reticular network could initiate a process of transformation of the multifunctional endoplasmic reticulum into a monofunctional, calcium regulatory membrane. This process would be complete when the content of the ATPase in the membrane vastly exceeded the content of all of the proteins with housekeeping functions. This random incorporation model of sarcoplasmic reticulum biogenesis has been proposed in a number of earlier papers and reviews (Boland et al., 1974; Martonosi, 1982; Martonosi et al., 1982). An alternative view is that a differentiated sarcoplasmic reticulum membrane, uniquely capable of Ca2+ regulation, is synthesized de now within muscle cells. This idea is a consistent extension of a model of the biogenesis of the sarcoplasmic reticulum presented previously by MacLennan et al. (1978), which suggested that the sarcoplasmic reticulum may form at a growing point between a unique sarcoplasmic reticulum and the rough endoplasmic reticulum by lateral displacement of intrinsic proteins formed on membrane-bound polysomes. Extrinsic proteins, ultimately localized in the lumen of the sarcoplasmic reticulum membrane, would enter that space by circuitous cytoplasmic routes. It should be stressed that this model is not necessarily an “all-or-nothing” model. The ATPase is clearly made within the rough endoplasmic reticulum network (Greenway and MacLennan, 1978; Chyn et al., 1979). The lateral displacement process would ensure that at any given time there would exist differentiated sarcoplasmic reticulum, rough endoplasmic reticulum with ATPase molecules incorporated, and a region of accumulation between the two membranes through which the ATPase would pass. At the time of the writing of this article, it is not possible to define which of the two postulates (if either) is correct. It is only possible to evaluate the evidence at +

8. ASSEMBLY OF SARCOPLASMIC RETICULUM

353

hand in light of these proposals. Morphological studies by Ezerman and lshikawa (1967) have shown that the sarcoplasmic reticulum “buds” from a rough endoplasmic reticulum, a finding supporting the view of a growing point between differentiated sarcoplasmic reticulum and rough endoplasmic reticulum. Morphological studies using the negative staining technique (Sarzala et al., 1975a; Zubrzycka et al., 1979) are also consistent with the growing point model since they appear to show two rather distinct populations of vesicles, one of which has surface particles (ATPase containing) and one of which does not. Freeze-fracture analysis provides rather strong support for a random incorporation model for the increase in the ATPase content of a preexisting membrane. Two studies (Baskin, 1974; Tillack et al., 1974) concurred in the finding that there was an increase in 8-nm particle density (corresponding to the ATPase) both in situ and in microsomal fractions from developing chick muscle. It was pointed out earlier, however, that there were unanswered questions in the studies with isolated membranes. Tillack et al. ( I 974) attempted, but apparently did not achieve, membrane fractionation prior to freeze-fracture: Baskin ( 1 974) may have achieved membrane fractionation but did not fracture the fractionated samples. Therefore, the data that were presented may represent measurements from a heterogeneous population of membranes. Baskin (1974) was, in fact, able to identify two classes of membranes. Class A vesicles from mature microsomes had a particle diameter of 8 nm and a density of 4600/pm2 (clearly sarcoplasmic reticulum) whereas class B vesicles had a particle diameter of 1 1 nm and a density of 1880/pm2 (clearly not sarcoplasmic reticulum). Both of these classes of vesicles were observed in microsomal vesicles obtained from 4-day chick. It is of interest that vesicles of the class A type were observed in 19-day embryo microsomes. These vesicles averaged 500 particles/pm*, however, whereas the overall microsomal particle density reported for this age group averaged only 175 particles/pm2. The fact that Tillack et al. (1974) reported average particle densities of several hundred in a population where a high percentage contained no 8nm particles suggests that some vesicles were highly enriched in particle density. To fully evaluate the information available from freeze-fracture, new studies should be carried out with several objectives in mind: first, to achieve optimal purification of “microsomal” membrane classes; second, to evaluate whether there are identifiable membrane classes with respect to 8-nm particle densities; and, finally, to attempt to answer the question of whether a unique class of vesicles, highly enriched in 8-nm particles and in Ca2+ transport capacity, can be seen to be present from earliest times and to increase during development. It should be stressed that the finding of a class of rough endoplasmic reticulum vesicles with relatively few 8-nm particles would be consistent with the growing point hypothesis since it could be assumed that these particles would be ATPase molecules in transit from a point of synthesis in one membrane to a point of deposition in the sarcoplasmic reticulum. It is also possible that a unique sar-

354

DAVID H. MACLENNAN ET AL.

coplasmic reticulum could form initially with a high content of phospholipid and a low density of 8-nm intramembrane particles. As development proceeded, the vesicles could become quantitatively more protein rich without a qualitative change in protein composition, thereby leading to an increased particle density. An argument in favor of the growing point hypothesis is the finding that a fraction resembling adult sarcoplasmic reticulum with respect to protein composition and function can be isolated from crude microsomes obtained from developing muscle tissue (Zubrzycka-Gaarn et al., 1985). An unanswered question concerning this fraction is whether it might be derived from a population of muscle cells that have reached maturity in advance of other cells in the same tissue. Were this the case, then the argument could not be made in favor of the growing point hypothesis. Most other arguments that have been used in support of the gradual differentiation hypothesis are also consistent with the growing point hypothesis if one assumes that the microsomal fraction is not a homogeneous membrane fraction but rather an unfractionated collection of sarcolemma, transverse tubules, sarcoplasmic reticulum, rough and smooth endoplasmic reticulum, and Golgi apparatus. The fact that the ATPase content and the capacity for Ca2+ transport increases on this background could as well suggest the emerging predominance of a unique membrane system as the transformation of one membrane system into another by addition of novel proteins to preexisting membranes.

V. SYNTHESIS OF SPECIFIC SARCOPLASMIC RETICULUM PROTEINS A. Synthesis of Sarcoplasmic Reticulum Proteins in Muscle Cell Cultures In the normal pattern of differentiation of muscle in culture, cells divide and align during the first 48 hours. In the third day they go into a period of rampant fusion which culminates in the formation of differentiated, multinucleated myotubes that show cross striations characteristic of aligned sarcomeres and which are capable of pulsatile contraction and relaxation. Early studies showed that synthesis of muscle-specific proteins is turned on coordinately at about the time of fusion (Shainberg er al., 1971). Although it was possible by the late 1960s to identify the ATPase as a 100,000-Da phosphoprotein in gel electrophoretic profiles of microsomal fractions (Martonosi and Halpin, 1969; MacLennan, 1970), serious attempts to study the biosynthesis of the protein were not made until the mid 1970s. Holland and MacLennan (1976) and Zubrzycka and MacLennan (1976) first used antibodies against the Ca2 -ATPase and calsequestrin, respectively, to study the bio+

8. ASSEMBLY

OF SARCOPLASMIC RETICULUM

355

synthesis of these two proteins in rat skeletal muscle cells differentiating in culture. The studies of Holland and MacLennan (1976) showed that ATPase synthesis was turned on at the time of initiation of cell fusion and increased dramatically thereafter. Its synthesis was, therefore, characteristic of the synthesis of other muscle-specific proteins. By contrast, the synthesis of calsequestrin (Zubrzycka and MacLennan, 1976) was initiated much earlier, being turned on between 24 and 40 hours after cell plating and well before synthesis of the ATPase and other muscle-specific proteins was initiated. These biochemical observations were confirmed morphologically in studies using immunofluorescent antibodies against the ATPase and calsequestrin (Jorgensen et al., 1977). There was no evidence for the presence of either protein in myoblasts up to 24 hours in culture. Soon thereafter, immunofluorescent staining of calsequestrin was observed. Its staining pattern indicated that it was being synthesized in a perinuclear region that was rich in rough endoplasmic reticulum and Golgi apparatus. With time, the staining pattern became more diffuse, indicating that the protein was moving to the periphery of the cell; at late stages it was found throughout the cell. By contrast, staining for the ATPase was not apparent until just about the time when cell fusion began. The appearance of the ATPase was not dependent on cell fusion, however, since it could be seen in mono-, bi-, or multinucleated cells within the fields observed. The ATPase appeared approximately 20 hours later in small granular patches throughout the cell, a finding suggesting multiple sites of synthesis. Studies on the appearance of the ATPase, using formation of the phosphorylated intermediate as a measure of ATPase synthesis (Martonosi et al., 1977), have provided a similar picture. Very little ATPase was present in early chick myoblasts. The synthesis of the enzyme was found to increase dramatically at about the time of fusion, and patterns of synthesis were similar to those observed with immunological techniques. Later studies with the high-affinity calcium-binding protein (Michalak and MacLennan, 1980) have shown that this protein is synthesized with a similar temporal pattern to calsequestrin. The protein is probably not a glycoprotein but it does seem to be located luminally in sarcoplasmic reticulum vesicles (Michalak et al., 1980). Studies of the temporal pattern of synthesis of the two intrinsic glycoproteins of M,53,000 and 160,000, carried out in a differentiating rat muscle cell line (L6), have shown that they are formed with the same temporal pattern as the ATPase (Zubrzycka-Gaarn et a/., 1983). These studies indicate that the intrinsic and extrinsic proteins of the sarcoplasmic reticulum might be under different temporal control in the assembly of the sarcoplasmic reticulum. Studies of a genetically determined murine muscle disease, muscular dysgenesis (Platzer and Gluecksohn-Waelsh, 1972; Bowden-Essien, 1972), have suggested that the lesion in this disease might originate in development of the

356

DAVID H . MACLENNAN ET AL.

sarcoplasmic reticulum. Investigation of the biosynthesis and localization of the Ca2 -ATPase and calsequestrin in cell cultures derived from prenatal dysgenic mouse muscle showed that these proteins were synthesized and transported normally (Essien et al., 1977). Thus, the mdg mutation did not have a direct effect on the synthesis or localization of either of these two sarcoplasmic reticulum proteins. The temporal pattern of biosynthesis of the Ca2 ,Mg2 -dependent ATPase has also been studied in embryonic heart muscle cells in culture (Holland, 1979). The rate of enzyme synthesis increased linearly with proliferation of the cells, and its synthesis was dependent on the presence of serum in the culture medium. Accumulation of the ATPase in the heart muscle cell cultures, measured by the presence of a Ca2+-dependent phosphoprotein, also showed a linear increase as the cells proliferated. +

+

+

B. Synthesis of Sarcoplasmic Reticulum Proteins in Wtro Early postulates of synthetic pathways for sarcoplasmic reticulum proteins suggested that the ATPase would be formed on membrane-bound polyribosomes and that it would be incorporated cotranslationally into the membrane (MacLennan et al., 1978). It was also postulated that calsequestrin would be made on bound polyribosomes and that it would enter the lumen of the endoplasmic reticulum. From there it would travel either directly into the lumen of the developing sarcoplasmic reticulum or, in the opposite direction, into the Golgi region to be packaged and transferred via vesicle-mediated processes to the lumen of the sarcoplasmic reticulum. These postulates have been tested in part. Both the ATPase and calsequestrin are formed on membrane-bound polyribosomes (Greenway and MacLennan, 1978; Chyn et al., 1979; Reithmeier et al., 1980). The ATPase is made without a cleavable amino-terminal signal sequence (Reithmeier et al., 1980; Mostov et al., 1981). Reithmeier et al. (1980) showed that the initiator methionine at the amino terminus of the protein is acetylated cotranslationally, accounting for the acetylmethionine present in the amino-terminal position in the mature form of the ATPase protein (Allen, 1977; Tong, 1977). Moreover, Cys 12 and Ala 3, 4, and 14 are present both in the mature form and in the form synthesized in vitru (Reithmeier et al., 1980; Mostov et al., 1981). The first 32 amino acids in the protein are relatively hydrophilic (Allen, 1977). At the end of this hydrophilic stretch, there is a sequence of several basic amino acids which might interact with a negatively charged membrane surface, and this sequence is presumably followed by a hydrophobic, transmembrane stretch (Allen, 1977). The amino-terminal portion of the protein is located cytoplasmically (Reithmeier and MacLennan, 1981). These observations suggest that the signal for transmembrane insertion lies within the first transmembrane

8. ASSEMBLY OF SARCOPLASMIC RETICULUM

357

passage of the ATPase protein. The way that the remainder of the protein is inserted into the membrane bilayer is not known, but Engelman and Steitz (198 1) have proposed that helical hairpin structures can be inserted spontaneously into membranes, and this model would be appropriate for the ATPase protein. This process would probably occur cotranslationally since the polyribosomes synthesizing the ATPase are predominantly membrane bound. Calsequestrin is also formed on membrane-bound polyribosomes (Reithmeier et a / ., 1980). In contrast to the ATPase, calsequestrin is synthesized as a highermolecular-weight ( M r , 66,000) precursor with an amino-terminal signal sequence containing methionine residues at positions I , 7, and possibly 14 that would direct synthesis of the protein into the lumen of the rough endoplasmic reticulum (cf. Blobel and Dobberstein, 1975). The protein is glycosylated (Jorgensen et a l . , 1977) and the carbohydrate moiety is highly processed. This suggests that the protein must have passed through the Golgi region, which is the site of the particular processing of carbohydrate moieties that is characteristic for calsequestrin (Harpaz and Schachter, 1980). There is no evidence for the route which calsequestrin takes after passing through the Golgi, but the fact that it is found in the interior of the terminal cisternae of sarcoplasmic reticulum in adult skeletal muscle suggests that it was deposited there by vesicle-mediated transfer, perhaps involving coated vesicles (cf. Rothman and Fine, 1980). At present, it is unknown whether calsequestrin is confined to the terminal cisternae in the initial stages of development or distributed uniformly throughout the sarcoplasmic reticulum. The details of the biosynthesis of the glycoprotein and of the high-affinity calcium-binding protein are as yet unknown. The 53,000-Da glycoprotein contains a relatively unprocessed sugar moiety consisting of Man,:GlcNAc, (Campbell and MacLennan, 1981). This is a composition similar to that found in the dolichol intermediate from which carbohydrate residues are transferred to proteins (Liu et d . , 1979). Only three glucose residues are removed from the carbohydrate chain, and this processing can occur within the lumen of the rough endoplasmic reticulum (Grinna and Robbins, 1979; Hanover and Lennarz, 1980). Since the glycoprotein is a transmembrane protein (Michalak ef a / . . 1980), it is probable that it is synthesized on bound polyribosomes. Experiments to prove this point have not yet been carried out, however. The fact that the ATPase contains no carbohydrate, whereas the 53,000-Da glycoprotein contains only carbohydrate moieties which are characteristic of those retained in the lumen of the rough endoplasmic reticulum, is consistent with the view that these proteins would not move within the cell by vesicle transfer but could be retained within the confines of the rough endoplasmic reticulum and sarcoplasmic reticulum network. Studies of the biosynthesis of the sarcoplasmic reticulum create an impression of how the membrane is assembled from its component parts. Clearly more

358

DAVID H.MAC LENNAN ET AL.

information is required in this area to understand the complete assembly processes. For example, there is at present no information regarding the mechanisms whereby the ATPase becomes excluded from the region of the junctional sarcoplasmic reticulum and calsequestrin becomes confined to the lumen of terminal cisternae. While some of this information can be gleaned from biochemistry, more will have to be gleaned from morphological studies of the synthesis, transport, and assembly of individual proteins.

VI.

REGULATION OF THE BIOSYNTHESIS OF SARCOPLASMIC RETICULUM PROTEINS

A. Effect of Calcium on Synthesis of Sarcoplasmic Reticulum Proteins Extracellular Ca2 is an important determinant of the fusion of myoblasts in culture to form myotubes. If medium Ca2+ is kept below a concentration of a few hundred micromolar, fusion is inhibited (Shainberg et al., 1969). This observation has led to a variety of experiments in which manipulation of fusion has been studied as a determinant of differentiation. The effect of low Ca2+ on the synthesis of sarcoplasmic reticulum has been studied in both rat and chick muscle cells in culture. In the rat muscle system, it was observed that culture of cells from early periods in low Ca2+ was deleterious to subsequent growth. Therefore, cells were grown for 24 and 48 hours in normal Ca2 before transfer to low-Ca2+ medium (Holland and MacLennan, 1974; Zubrzycka and MacLennan, 1976; Michalak and MacLennan, 1980). This procedure resulted in healthy cultures and in prevention of cell fusion, but it did not prevent the synthesis of the ATPase, calsequestrin, or the high-affinity calcium-binding protein. Some variation was observed in the time of onset of ATPase synthesis after the switch from normal to low-Ca2+ medium and in the rates of synthesis of calsequestrin. These variations were not analyzed systematically, however. It was clear that synthesis of sarcoplasmic reticulum proteins was turned on in the absence of cell fusion, and it was concluded that cell fusion was not a determinant but only a concomitant event in differentiation. In studies with chick myoblasts (Martonosi et al., 1977), cells were cultured from the start with low Ca2+. Under these conditions, synthesis of the Ca2+ATPase, together with the synthesis of a number of other muscle-specific proteins such as myosin, troponin C, creatine kinase, actin, tropomyosin, and hemagglutinin was inhibited in low-Ca2 medium. These observations suggested that there was a class of muscle-specific proteins whose synthesis might be regulated by Ca2+ (Martonosi et al., 1977). Since external Ca2+ would not likely be a determinant of intracellular events, it seemed more reasonable that +

+

+

359

8. ASSEMBLY OF SARCOPIASMIC RETICULUM

extracellular Ca2 would influence intracellular ionized Ca2 and thereby regulate protein synthesis. It was postulated that an increase in intracellular Ca2+ either during or after fusion would promote gene transcription of a specific class of mRNA followed by increased synthesis of Ca2 -ATPase and other Ca2 induced proteins (Martonosi, 1982; Martonosi et al., 1982). The regulation was assumed to be accomplished by nuclear Ca2 -binding proteins that would serve as inducers or repressors of gene transcription. Several studies to provide evidence for this hypothesis have been carried out. The addition of the Ca2 ionophore A23 187 slightly increased the amount of ATPase, as judged by phosphoprotein formation, in 5-,7-, and 9-day-old cultures. It was not proved that intracellular Ca2+ was increased, but the deduction that it would be increased is well founded, since these ionophores increase Ca2+ flux across surface membranes (Roufa et ul., 1981) and they increased Ca2 efflux from sarcoplasmic reticulum (Martonosi et a/., 1980). The total Ca2+ content of washed, cultured muscle was measured and was found to decrease dramatically about the time of fusion, from 200 nmol/mg protein at day 3 to 59 nmol/mg protein at day 5. This result might suggest that intracellular Ca2+ was more closely regulated after day 5 than before. However, these figures do not reflect the amount of ionized Ca2+, a more important measure of Ca2+ activity. Studies on Ca2 -binding proteins in the nucleus were also carried out. The total Ca2 content of isolated liver and muscle nuclei was found to lie in the range of 818 nmol/mg protein (Schibeci and Martonosi, 1980). Since about two-thirds of this Ca2+ was free, 3-6 nmol were bound. After digestion with DNase and RNase, 90% of the Ca2 binding was lost. The remainder of the Ca2 binding sites were found in nonhistone chromosomal proteins and in the residue that was insoluble after extraction of nuclei with 0.4-2.OM NaCl. The histone fraction was free of Ca2 binding activity. Dissociation constants for the Ca2 binding were 2 mM in the insoluble fraction and 200 fl in the acidic fraction. These values would seem to be too high to permit Ca2 binding at cytoplasmic Ca2 concentrations, which range from 0.1 to 50 pA4 (Kretsinger, 1976) but might bind Ca2+ if there were a higher concentration of Ca2+ within the nucleus. This is unlikely, however, since the nuclear membrane is rather permeable (Civan, 1978). Careful studies with the Ca2+ ionophore ionomycin, as well as with the ionophore A23187, revealed a selective stimulation of the synthesis of two possible sarcoplasmic reticulum proteins of 100,000 and 80,000 Da (Wu et af., 1981). The synthesis of the 80,000-Da protein was increased fourfold and of the 100,000-Da protein 1.5-fold under conditions where synthesis of actin or of tropomyosin 1 and 2 was unchanged. The 80,000-Da protein was not unique to muscle cells but was found in chicken fibroblasts, mouse LSP cells, HeLa cells, and cells of African green monkey kidneys (Martonosi et al., 1982). Synthesis of specific proteins is also turned on in response to heat shock (Tissieres et al., 1974; Lewis et al., 1975). A recent study (Welch et al., 1983) has suggested that +

+

+

+

+

+

+

+

+

+

+

+

+

+

+

360

DAVID H. MAC LENNAN ET AL.

the 80,000- and 100,000-Da, Ca2+-dependent proteins may be identical to two minor heat-shock proteins. This view is not shared by Martonosi et al. (1982), who concluded that in muscle cells the 100,000-Da, Ca2+-dependent protein and the 100,000-Da, heat-shock proteins were different. Further studies comparing the amino acid sequences of these proteins must be carried out to determine whether they are or are not related.

B. Neural Control of Transformation of Fast-Twitch to Slow-Twitch Muscle The sarcoplasmic reticulum of slow-twitch muscle is less concentrated within muscle cells and the isolated membranes appear to contain an altered protein composition when compared with the sarcoplasmic reticulum of fast-twitch muscle. Ultrastructural and stereological analysis of slow-twitch muscle has shown the volume density of the sarcoplasmic reticulum membrane to be about one-half that observed in fast-twitch muscle (Luff and Atwood, 1971; Eisenberg et al., 1974; Eisenberg and Salmons, 1981). Morphological studies of the content of 8nm intramembrane particles (Bray and Rayns, 1976) and immunoelectron microscopic studies of the density of the Ca2 -ATPase in slow- and fast-twitch muscle (Jorgensen et al., 1982) have also suggested a reduced content of the Ca2+ATPase in the sarcoplasmic reticulum membrane of slow-twitch muscle fibers. Biochemical studies of the isolated membrane (Heilmann et al., 1977; Zubrzycka-Gaarn et al., 1982) have shown that the content of the 100,000-Da ATPase protein, the rate and capacity for Ca2+ transport, the rate of Ca2+dependent ATPase activity, and the level of Ca2 +-dependent phosphoprotein formation are all lower in slow-twitch membranes than in fast-twitch muscle membranes. Cross-innervation of fast-twitch muscle with slow-twitch nerves results in a transformation of the fast muscle phenotype to a slow muscle phenotype (Mommaerts et al., 1969; Margreth et al., 1973). Similar changes can be brought about by stimulation of fast-twitch muscle by an indirect, chronic electrical impulse simulating that of a motoneuron innervating a slow-twitch muscle. The transformation affects myofibrillar proteins, emzymes of energy metabolism, and the sarcoplasmic reticulum (Heilmann and Pette, 1979). The earliest changes were observed in the sarcoplasmic reticulum, which began to swell within 6-24 hours and to undergo a quantitative reduction in mass over a 2-week period (Eisenberg and Salmons, 1981). This led to a membrane with decreased lipid content, decreased ATPase and Ca2 transport activity, and an altered morphology, which included a diminution in the content of 8-nm intramembrane particles (Heilmann and Pette, 1979, 1980; Heilmann et al., 1981; Sarzala et al., 1982). These studies show that expression of the genes governing sarcoplasmic re+

+

361

8. ASSEMBLY OF SARCOPLASMIC RETICULUM

ticulum synthesis are regulated to some extent by the innervation and subsequent electrical activity of the muscle cell. The basis for this control is not understood, but Ca2+ is an obvious candidate for a regulator since its release and reuptake pattern is controlled by cellular electrical activity. The questions raised earlier concerning the models for membrane biosynthesis are relevant to the transformation of fast muscle sarcoplasmic reticulum to slow muscle sarcoplasmic reticulum. In this case, however, the biosynthetic process appears to be reversed. The content of the 100,000-Da ATPase protein diminishes with time, and the content of other proteins in the microsomal fraction increases. As a corollary, the content of 8-nm particles diminishes dramatically in situ and in isolated membranes. These microsomal preparations can be fractionated after calcium oxalate loading, and the fraction which is most heavily loaded is enriched in proteins of 100,000, 76,000, 54,000, 52,000, and 45,000 Da and is greatly depleted in proteins below 45,000 Da (Zubrzycka-Gaarn et af., 1982). These experiments again raise questions about whether the microsomal fractions that are referred to as sarcoplasmic reticulum are, in fact, heterogeneous populations of vesicles. These questions will have to be borne in mind as the system is studied further. Charuk and Holland (1983) have examined the question of whether blockade of spontaneous contractile activity of cultured chick muscle cells with tetrodotoxin can affect the development of the sarcoplasmic reticulum. They observed an inhibition of the accumulation of the Ca2 -ATPase protein, measured by Ca2 dependent phosphorylation, and of Ca2 uptake activity in homogenates of cells that were cultured in the presence of tetrodotoxin. +

+

+

C. Cloning of the ATPase Gene Some of the problems outlined in this article can be readily approached through recombinant DNA technology. The question of the ATPase sequence will be resolved from DNA sequencing. The question of whether there are multiple forms of the ATPase or of any of the other sarcoplasmic reticulum proteins can be readily studied by analysis of genomic DNA. Finally, analyses of the temporal patterns of synthesis of the mRNAs for the ATPase, calsequestrin, or other sarcoplasmic reticulum proteins can be carried out with cDNA probes. Cloning of these proteins has just begun and a cDNA sequence of 75 bp, of which 30 code for a known sequence in the ATPase, has been isolated (MacLennan et af., 1983). The use of recombinant DNA technology will provide ultimate proof for several postulates concerning the regulation of the synthesis of different sarcoplasmic reticulum proteins. It will not provide all the answers, however. New approaches will be necessary to answer all of the questions relating to the biogenesis of this interesting membrane system.

362

DAVID H. MAC LENNAN ET AL.

REFERENCES Allen, G. (1977). On the primary structure of the CaZ+-ATPase of sarcoplasmic reticulum. Fed. Eur. Biuchem. SUC.Meet.. Ilth, Copenhagen 45, 159-168. Allen, G. (1980a). The primary structure of the calcium-transporting adenosine triphosphatase of rabbit skeletal sarcoplasmic reticulum. Soluble tryptic peptides from the succinylated, carboxymethylated protein. Biochem. J . 187, 545-563. Allen, G. (1980b). Primary structure of the calcium ion-transporting adenosine triphosphatase of rabbit skeletal sarcoplasmic reticulum. Soluble peptides from the chymotryptic digestion of the carboxymethylated protein. Biochem. J . 187, 565-575. Allen, G., Bottomley, R. C., and Trinnaman, B. J. (1980a). Primary structure of the calcium iontransporting adenosine triphosphatase from rabbit skeletal sarcoplasmic reticulum. Some peptic, thermolytic, tryptic and staphylococcal proteinase peptides. Biochern. J. 187, 577-589. Allen, G., Trinnaman, B.J., and Green, N. M. (1980b). The primary structure of the calcium iontransporting adenosine triphosphatase protein of rabbit skeletal scarcoplasrnic reticulum. Peptides derived from digestion with cyanogen bromide, and the sequences of three long extramembranous segments. Biochern. J. 187, 591-616. Baskin, R. J. (1974). Ultrastructure and calcium transport in microsomes from developing muscle. J. Ultrastruct. Res. 49, 348-371. Blobel, G., and Dobberstein, B. (1975). Transfer of proteins across membranes. I. Presence of proteolytically processed and unprocessed nascent immunoglobulin light chains on membranebound ribosomes of murine myeloma. J . Cell Biol. 67, 835-851. Boland, R., Martonosi, A., and Tillack, T. W. (1974). Developmental changes in the composition and function of sarcoplasmic reticulum. J. B i d . Chem. 249, 612-623. Bowden-Essien, F. (1972). An in vitru study of normal and mutant myogenesis in the mouse. Dev. B i d . 27, 351-364. Brandt, N. R., Caswell, A. H., and Brunschwig, J. P. (1980). ATP-energized Ca2+ pump in isolated transverse tubules of skeletal muscle. J. Biol. Chem. 255, 6290-6298. Bray, D. F., and Rayns, D. G. (1976). Comparative freeze-etch study of the sarcoplasmic reticulum of avian fast and slow muscle fibers. J. Ulrrastrucr Res. 57, 251-259. Campbell, K. P., and MacLennan, D. H. (1981). Purification and characterization of the 53,000dalton glycoprotein from the sarcoplasrnic reticulum. J. B i d . Chem. 256, 4626-4623. Campbell, K. P., and MacLennan, D. H.(1983). Labeling of high affinity ATP binding sites on the 53,000- and 160,000-dalton glycoproteins of the sarcoplasmic reticulum with the photo-affinity probe 8-N3-[-32P]ATP. J. B i d . Chem. 258, 1391-1394. Campbell, K. P., MacLennan, D. H., and Jorgensen, A. 0. (1983). Staining of the Ca2+ binding proteins, calsequestrin, calrnodulin, troponin C and S-100, with the cationic carbocyanine dye “stains-all.” J . Biol. Chem. 258, 11267-11273. Charuk, J. H. M., and Holland, P. C. (1983). Effect of tetrodotoxin relaxation of cultured skeletal muscle on the sarcoplasic reticulum CaZ+-transport ATPase. Exp. Cell Res. 144, 143-157. Chyn, L., Martonosi, A. N., Morimoto, J., and Sabatini, D. D. (1979). In virru synthesis of the Ca2 transport ATPase by ribosomes bound to sarcoplasmic reticulum membranes. J. Biol. Chem. 76, 1241-1245. Civan, M. M. (1978). Intracellular activities of sodium and potassium. Am. J. Physiol. 234, F261F269. Crowe, L. M., and Baskin, R. J. (1978). Freeze-fracture of intact sarcotubular membranes. J . Ultrastruct. Res. 62, 147- 154. Deamer, B. W., and Baskin, R. Y.(1969). Ultrastructure of sarcoplasmic reticulum preparations. J. Cell Biol. 42, 269-301. +

363

8. ASSEMBLY OF SARCOPIASMIC RETICULUM

de Meis, L., and Vianna, A. L. (1979). Energy interconversion by the Caz+-dependent ATPase of the sarcoplasmic reticulum. Annrc. Rev. Biochem. 48, 275-292. Ebashi, S., Endo. M., and Ohtsuki, I. (1969). Control of muscle contraction. Q . Rev. Biophys. 2, 351-384. Edge, M. B. (1970). Development of apposed sarcoplasmic reticulum at the T system and sarcolemma and the change in orientation of triads in rat skeletal muscle. Dev. Biol. 23, 634-650. Eisenberg, B. R., and Salmons, S. (1981). The reorganization of subcellular structure in muscle undergoing fast to slow type transformation. Cell Tissue Res. 220, 449-471. Eisenberg, B. R., Kuda, A. M., and Peter, J. B. (1974). Stereological analysis of mammalian skeletal muscle. 1. Soleus muscle of the adult guinea pig. J . Cell Biol. 60,732-754. Engelman, D. M., and Steitz, T. A. (1981). The spontaneous insertion of proteins into and across membranes: The helical hairpin hypothesis. Cell 23, 41 1-422. Essien, F. B . , Jorgensen, A. 0.. Kalnins, V. L., Zubrzycka, E., and MacLennan, D. H. (1977). The biosynthesis and localization of sarcoplasniic reticulum proteins in dysgenic (mdglmdg) mouse cells. Lab. Invest. 31, 562-567. Ezerman. E. B., and Ishikawa, H. (1967). Differentiation of the sarcoplasmic reticulum and T system in developing chick skeletal muscle in v i m . J . Cell Biol. 35, 405-420. Fanburg, B. L.,Drachman, B. B., and Moll, D. (1968). Calcium transport in isolated sarcoplasmic reticulum during muscle maturation. Nature (London) 218, 962-964. Fischman, D. A. (1970). The synthesis and assembly of myofibrils in embryonic muscle. Curr. Top. Dev. B i d . 5, 235-280. Fischman, D. A. (1972). Development of striated muscle. In “The Structure and Function of Muscle” ( G . H., Bourne, ed.), Vol. I , pp. 75-148. Academic Press, New York. Franzini-Armstrong, C. (1970). Studies of the triad. I.Structure of the junction in frog twitch fibers. J . Cell Biol. 47, 488-499. Franzini-Armstrong, C . (1975). Membrane particles and transmission at the triad. Fed. Proc., Fed. Am. Soc. Exp. Biol. 34, 1382-1389. Greenway, D. C., and MacLennan, D. H. (1978). Assembly of the sarcoplasmic reticulum. Synthesis of calsequestrin and the CaZ+ + Mg2 -adenosine triphosphatase on membrane-bound polyribosomes. Can. J . Biochem. 56, 452-456. Grinna, L. S . , and Robbins, P. W. (1979). Glycoprotein biosynthesis. Rat liver microsomal glucosidases which process oligosaccharides. J . Biol. Chem. 254, 8814-8818. Hanover, J. A,, and Lennarz, W. J . (1980). N-Linked glycoprotein assembly. Evidence that ohgosaccharide attachment occurs within the lumen of the endoplasmic reticulum. J . Biol. Chem. 255, 3600-3604. Harpaz, N., and Schachter, H. (1980). Control of glycoprotein synthesis. Processing of asparaginelinked oligosaccharides by one or more rat liver Golgi o-mannosidases dependent on the prior action of UDP-N-acetylg1ucosamine:o-mannoside 2-N-acetylglucosaminyltransferaseI. J . Biol. Chem. 255, 4894-4902. Hasselbach, W. (1964). Relaxing factor and the relaxation of muscle. Prog. Biophys. Mol. Biol. 14, 167-222. Heilmann. C., and Pette, D. (1979). Molecular transformation in sarcoplasmic reticulum of fasttwitch muscle by electro-stimulation. Eur. J . Biochem. 93, 437-446. Heilmann. C., and Pette, D. (1980). Molecular transformation of sarcoplasmic reticulum in chronically stimulated fast-twitch muscle. In “Plasticity of Muscle’’ (D. Pette, ed.), pp. 421-440. De Gruyter, New York. Heilmann, C., Brdiczka, D., Nickel, E., and Pette, D. (1977). ATPase activities, CaZ+-transport and phosphoprotein formation in sarcoplasmic reticulum subfractions of fast and slow rabbit muscle. Eicr. J . Biochem. 81, 21 1-222. +

364

DAVID H. MAC LENNAN ET AL.

Heilmann, C., Muller, W., and Pette, D. (1981). Correlation between ultrastructural and functional changes in sarcoplasmic reticulum during chronic stimulation of fast muscle. J . Membr. Biol. 59, 143-149. Holland, D. L., and Perry, S. V. (1969). The adenosine triphosphatase and calcium ion-transporting activities of the sarcoplasmic reticulum of developing muscle. Biochem. J. 114, 161- 170. Holland, P. C. (1979). Biosynthesis of the Ca2 and Mg2 + -dependent adenosine triphosphatase of sarcoplasmic reticulum in cell cultures of embryonic chick heart. J . Biol. Chem. 254, 76047610. Holland, P. C., and MacLennan, D. H. (1976). Assembly of the sarcoplasmic reticulum: Biosynthesis of the adenosine triphosphatase in rat skeletal muscle cell culture. J. Biol. Chem. 251, 2030-2036. Ikemoto, N., Sreter, F. A., Nakamura, A., and Gergely, J. (1968). Tryptic digestion and localization of calcium uptake and ATPase activity in fragments of sarcoplasmic reticulum. J. Ultrustruct. Res. 23, 216. Ikemoto, N., Bhatnagar, G. M., and Gergely, J . (1971). Fractionation of solubilized sarcoplasmic reticulum. Biochem. Biophys. Res. Commun. 44, 1510-1517. lnesi, G., and Asai, H. (1968). Trypsin digestion of fragmented sarcoplasmic reticulum. Arch. Biochem. Biophys. 126, 469-477. Ishikawa, H. (1968). Formation of elaborate networks of T-system tubules in cultured skeletal muscle with special reference to the T-system formation. J . Cell Biol. 38, 51-66. Ishikawa, H., and Yamada, E. (1975). Differentiation of the sarcoplasmic reticulum and T-system in developing mouse cardiac muscle. In “Perspective in Cardiovascular Research” (A. Katz, ed.), Vol. 1, pp. 21-25. Raven, New York. Jorgensen, A. O., Kalnins, V. I., Zubrzycka, E., and MacLennan, D. H. (1977). Assembly of the sarcoplasmic reticulum. Localization by immunofluorescence of sarcoplasmic reticulum proteins in differentiating rat skeletal muscle cell cultures. J. Cell Biol. 74, 287-298. Jorgensen, A. O., Kalnins, V., and MacLennan, D. H. (1979). Localization of sarcoplasmic reticulum proteins in rat skeletal muscle by immunofluorescence. J. Cell Biol. 80, 372-384. Jorgensen, A. O . , Daly, P., Campbell, K. P., and MacLennan, D. H. (1981). Localization of the 53,000 dalton glycoprotein of the sarcoplasmic reticulum in adult rat skeletal muscle. J. Cell Biol. 91, 345a. Jorgensen, A. O., Shen, A. C. Y., Daly, P., and MacLennan, D. H. (1982a). Localization of Ca2+ + Mg*+-ATPase of the sarcoplasmic reticulum in adult rat papillary muscle. J . Cell Biol. 93, 883-892. Jorgensen, A. O., Shen, A. C. Y., MacLennan, D. H., and Tokuyasu, K. T. (1982b). Ultrastructural localization of the Ca2+ + Mg2+-dependent ATPase of sarcoplasmic reticulum in rat skeletal muscle by immunofemtin labeling of ultrathin frozen sections. J . Cell B i d . 92, 409416. Jorgensen, A. O., Shen, A. C.-Y., Campbell, K. P., and MacLennan, D. H. (1983). Ultrastructural localization of calsequestrin in rat skeletal muscle by immunofemtin labeling of ultrathin frozen sections. J . Cell Biol. 97, 1573-1581. Kelly, A. M. (1971). Sarcoplasmic reticulum and T tubules in differentiating rat skeletal muscle. J. Cell Biol. 49, 335-344. Kilarski, W., and Jakubowska, M. (1979). An electron microscope study of myofibril formation in embryonic rabbit skeletal muscle. Z . Mikrosk. Anur. Forsch. 93, 1159-1 181. Kretsinger, R. H. (1976). Calcium-binding proteins. Annu. Rev. Biochem. 45, 239-260. Lewis, M., Helmsing, P. J., and Ashburner, M. (1975). Parallel changes in puffing activity and patterns of protein synthesis in salivary glands of Drosophila. Proc. Nurl. Acad. Sci. U.S.A. 72, 3604-3608. Liu, T., Stetson, B., Turco, S. J., Hubbard, S. C., and Robbins, P. W. (1979). Arrangement of +

365

8. ASSEMBLY OF SARCOPLASMIC RETICULUM

glucose residues in the lipid-linked oligosaccharide precursor of asparaginyl oligosaccharides. J . Biol. Chem. 254, 4554-4559. Lough. J. W., Entman, M. L., Bossen, E. H., and Hansen, J. L. (1972). Calcium accumulation by isolated sarcoplasmic reticulum of skeletal muscle during development in tissue culture. J . Cell. Physiol. 80, 43 1-436. Luff, A. R., and Atwood, H. L. (1971). Changes in the sarcoplasmic reticulum and transverse tubular system of fast and slow skeletal muscles of the mouse during postnatal development. J . Cell Biol. 51, 369-383. MacLennan, D. H. ( 1970). Purification of adenosine triphosphatase from sarcoplasmic reticulum. J . Biol. Chem. 245, 4508-4518. MacLennan, D. H., and Holland, P. C. (1975). Calcium transport in sarcoplasmic reticulum. Annu. Rev. Biophys. Bioengineer. 4, 377-404. MacLennan, D. H.. and Reithmeier, R. A. F. (1982). The structure of the Ca2+ + MgZ+-ATPase of sarcoplasmic reticulum. Membr. Tramp. 1, 567-571. MacLennan, D. H., and Wong, P. T. S. (1971). Isolation of a calcium-sequestering protein from sarcoplasmic reticulum. Proc. Nufl. Acud. Sci. U.S.A. 68, 1231-1235. MacLennan, D. H., Seeman, P., Iles, G. H., and Yip, C. C. (1971). Membrane formation by the adenosine triphosphatase of sarcoplasmic reticulum. J . Cell Biol. 246, 2702-2710. MacLennan, D. H., Yip, C. C., Iles, G. H., and Seeman, P. (1972). Isolation of sarcoplasmic reticulum proteins. Cold Spring Harbor Symp. Quunt. Biol. 37, 469-477. MacLennan, D. H., Zubrzycka, E., Jorgensen, A. 0..and Kalnins, V. I. (1978). Assembly of the sarcoplasmic reticulum. I n “The Molecular Biology of Membranes” (S. Fleischer, Y. Hatefi, D. H. MacLennan, and A. Tzagoloff, eds.), pp. 309-320. Plenum, New York. MacLennan, D. H., Brandl, C . , Green, N. M.. Hwang, P., Powers, M., and Nilson, J. (1983). Isolation of a cDNA clone for the Ca2+ + Mg2 ATPase of sarcoplasmic reticulum. Abstr. In?. Symp. Cu2+ Bind. Proteins Health Dis. 4th. Trieste p. 30. Margreth, A., Salviati, G.. and Carraro, V. (1973). Neural control on the activity of the calcium transport system in sarcoplasrnic reticulum of rat skeletal muscle. Nature (London) 241, 285286. Martonosi, A. (1975). Membrane transport during development in mammals. Biochim. Biophys. Acta 415, 31 1-333. Martonosi, A. ( 1982). The development of sarcoplasmic reticulum membranes. Annu. Rev. Physiol. 44, 337-55. Martonosi, A., and Halpin, R. A. (1969). Sarcoplasmic reticulum. Properties of a phosphoprotein intermediate implicated in calcium transport. J . Biol. Chem. 244, 613-620. Martonosi, A , , Roufa, D., Boland, R., Reyes, E . , and Tillack, T. W. (1977). Development of sarcoplasmic reticulum in cultured chicken muscle. J . Biol. Chem. 252, 3 18-332. Martonosi, A , , Roufa, D., Ha, D.-B., and Boland, R. (1980). The biosynthesis of sarcoplasmic reticulum. Fed. Proc.. Fed. Am. Sot. Exp. Biol. 39, 2415-2421. Martonosi, A., Dux, L., Terjung, R. L., and Roufa. D. (1982). Regulation of membrane assembly during development of sarcoplasmic reticulum: The possible role of calcium. Ann. N . Y . Acad. Sci. 402, 485-514. Meissner, G. (1975). Isolation and characterization of two types of sarcoplasmic reticulum vesicles. Biochim. Biophys. Acta 389, 5 1-68. Meissner, G., and Fleischer, S. (1971). Characterization of sarcoplasmic reticulum from sekeletal muscle. Biochim. Biophys. Acru 241, 356-378. Michalak, M., and MacLennan, D. H. (1980). Assembly of the sarcoplasmic reticulum: Biosynthesis of the high affinity calcium binding protein in rat skeletal muscle cell cultures. J . Biol. Chem. 255, 1327- 1334. Michalak, M., Campbell, K. P., and MacLennan, D. H. (1980). Localization of the high-affinity +

366

DAVID H. MACLENNAN ET AL.

calcium-binding protein and an intrinsic glycoprotein in sarcoplasmic reticulum membranes. J. Biol. Chem. 255, 1317-1326. Migala, A,, Agostini, B., andHasselbach, W. (1973). Tryptic fragmentation of the calcium transport system in the sarcoplasmic reticulum. Z. Natu~orsch.28, 178-182. Mommaerts, W. F. H. M., Buller, A. H., and Seraydarian, K. (1969). The modification of some biochemical properties of muscle by cross-innervation. Proc. Natl. Acad. Sci. U.S.A. 64, 128133. Mostov, K. E., DeFoor, P., Fleischer, S . , and Blobel, G. (1981). Cotranslational membrane integration of calcium pump protein without signal sequence cleavage. Nature (London) 292, 87-88. Ostwald, T. J . , and MacLennan, D. H. (1974). Isolation of a high affinity calcium binding protein from sarcoplasmic reticulum. 1.Biol. Chem. 149, 974-979. Pellegrino, C., and Franzini, C. (1963). An electron microscope study of denervation atrophy in red and white skeletal muscle fibers. J. Cell B i d . 17, 328-349. Platzer, A. C., and Gluecksohn-Waelsch, S. (1972). Fine structure of mutant (muscular dysgenesis) embryonic mouse muscle. Dev. B i d . 28, 242-252. Porter, K. R., and Palade, G. E. (1957). Studies on the endoplasmic reticulum. 111. Its form and distribution in striated muscle cells. J . Biophys. Biochem. Cyrol. 3, 269-299. Reithmeier, R. A. F., and MacLennan, D. H. (1981). The NH terminus of the (Ca2+ + Mg2+)adenosine triphosphatase is located on the sarcoplasmic reticulum membrane. J. Biol. Chem. 256, 5957-5960. Reithmeier, R. A. F., de Leon, S., and MacLennan, D. H. (1980). Assembly of the sarcoplasmic reticulum. Cell-free synthesis of the Ca2+ + Mgz+ -adenosine triphosphatase and calsequestrin. J . Biol. Chem. 255, 11839-1 1846. Rosemblatt, M., Hidalgo, C., Vergara, C., and Ikemoto, N. (1981). Immunological and biochemical properties of transverse tubule membranes isolated from rabbit skeletal muscle. J. Biol. Chem. 256, 8140-8148. Rothman, J. E., and Fine, R. E. (1980). Coated vesicles transport newly synthesized membrane glycoproteins from endoplasmic reticulum to plasma membrane in two successive stages. P roc. Natl. Acad. Sci. U.S.A. 77, 780-784. Roufa, D., Wu, F. S., and Martonosi, A. (1981). The effect of CaZ+ ionophores upon the synthesis of proteins in cultured skeletal muscle. Biochim. Biophys. Acta 674, 225-237. Sabbadini, R. D., Scales, D., and Nixon, C. (1978). A novel technique for the isolation of transverse tubular membrane from dystrophic skeletal muscle. Proc. Int. Congr. Electron Microsc., 9th 2, 640-641.

Sarzala, M. G., and Pilarska, M. (1976). Phospholipid biosynthesis in sarcoplasmic reticulum membrane during development. Biochim. Biophys. Acta 441, 8 1-92. Sarzala, M. G., Pilarska, M., Zubrzycka, E., and Michalak, M. (1975a). Changes in the structure, composition and function of sarcoplasmic reticulum membrane during development. Eur. J. Biochem. 57, 25-34. Sarzala, M. G., Zubrzycka, E., and Michalak, M. (1975b). Comparison of some features of undeveloped and mature sarcoplasmic reticulum vesicles. In “Calcium Transport in Contraction and Secretion” (E. Carafoli, F. Clementi, W. Drabikowski, and A. Margreth, eds.), pp. 329337. North-Holland Publ., Amsterdam. Sarzala, M. G., Szymanska, G., Wiehrer, W., and Pette, D. (1982). Effect of chronic stimulation at low frequency on the lipid phase of sarcoplasmic reticulum in rabbit fast-switch muscle. Eur. J . Biochem. 123, 241-245. Scales, D., and Sabbadini, R. (1979). A stereological analysis of purified microsomes derived from normal and dystrophic skeletal muscle. J. Cell Biol. 83, 33-46. Schiaffino, S . , and Margreth, A. (1969). Coordinated development of the sarcoplasmic reticulum and T system during postnatal differentiation of rat skeletal muscle. J . Cell Biol. 41, 855-875.

8. ASSEMBLY OF SARCOPLASMIC RETICULUM

367

Schiaffino, S., Cantini, M.. and Sartore, S . (1977). T-system formation in cultured rat skeletal tissue. Tissue Cell 9, 437-446. Schibeci, A., and Martonosi, A. (1980). Cd2+-binding proteins in nuclei. Eur. J. Biochem. 113, 514. Shainberg, A., Yagil, G., and Yaffe, B. (1969). Control of myogenesis in vitro by Ca2+ concentration in nutritional medium. Exp. Cell Res. 58, 163-167. Shainberg, A., Yagil, G . , and Yaffe, D. (1971). Enzyme activities during muscle differentiation. Dev. Biol. 25, 1-29. Shimada, Y., Fishman, D. A., and Moscona, A. A. (1967). The fine structure of embryonic chick skeletal muscle cells differentiation in v i m . J. Cell B i d . 35, 446-453. Somlyo, A. V.. Gonzalez-Serratos, H., Shuman. H., McClellan, G., and Somlyo, A. P. (1981). Calcium release and ionic changes in the sarcoplasmic reticulum of tetanized muscle: An electron-probe study. J. Cell Biol. 90, 577-594. Spray, T. L., Waugh, R. A . , and Sommer, J. R . (1974). Peripheral couplings in adult vertebrate skeletal muscle. J. Cell Biol. 62, 223-227. Stewart, P. S.. and MacLennan, D. H. (1974). Surface particles of sarcoplasmic reticulum membranes. Structural features of the adenosine triphosphatase. J. Biol. Chem. 249, 985-993. Thorley-Lawson, D. A., and Green, N. M. (1973). Studies on the location and orientation of proteins in the sarcoplasmic reticulum. Eur. J. Biochem. 40, 403-413. Tillack, T. W., Boland, R., and Martonosi. A. (1974). The ultrastructureof developing sarcoplasmic reticulum. J. Biol. Chem. 219, 624-633. Tissieres, A., Mitchell, H. K., and Tracy, U. M. (1974). Protein synthesis in salivary glands of Drosophilu melanogasrer: Relation to chromosome puffs. J . Mol. Biol. 84, 389-398. Tomanek, R. J., and Colling-Saltin, A. S. (1977). Cytological differentiation of human fetal skeletal muscle. Am. J . Anar. 149, 227-246. Tong, S . W. (1977). The acetylated NH2-terminus of the Ca ATPase from rabbit skeletal muscle sarcoplasmic reticulum: A common NH2-terminal acetylated methionyl sequence. Biochem. Biophys. Res. Cornmun. 74, 1242-1248. Volpe, P., Damiani, E., Salviati, G . , and Margreth, A. (1982). Transitions in membrane composition during postnatal development of rabbit fast muscle. J. Muscle Res. Cell Motil. 3, 213--230. Walker, S. M., and Schrodt. G. R. (1968). Triads in skeletal muscle fibers of 19-day fetal rats. J. Cell Biol. 37, 564-569. Walker, S. M., Schrodt, G. R., and Edge, M. B. (1971). The density attached to the inside surface of the apposed sarcoplasmic reticular membrane in vertebrate cardiac and skeletal muscle fibres. J. Ana?. 108, 217-230. Walker, S. M., Schrodt, G . R., Currier, G . J., and Turner, E. V. (1975). Relationship of the sarcoplasmic reticulum to fibril and triadic junction development in skeletal muscle fibers of fetal monkeys and humans. J. Morphol. 146, 97- 128. Weber, A. (1966). Energized calcium transport and relaxing factors. Curr. Top. Bioenerg. 1, 203254. Welch, U . . Garrels, J. I., Thomas, P., Lin, J. J., and Feramisco. J. R. (1983). Biochemical characterization of the mammalian stress proteins and identification of two stress proteins as glucose and Ca2+-ionophore regulated proteins. J. Biol. Chem. 258, 7102-71 11. Wu, F. S., Park, Y.-C., Roufa, D., and Martonosi, A. (1981). Selective stimulation of the synthesis of an 80,000-dalton protein by calcium ionophores. J . Biol. Chem. 256, 5309-5312. Zubrzycka, E., and MacLennan, D. H. (1976). Assembly of the sarcoplasmic reticulum: Biosynthesis of calsequestrin in rat skeletal muscle cell cultures. J . Biol. Chem. 251, 7733-7738. Zubrzycka, E., Michalak. M., Kosk-Kosicka, D., and Sarzala. M. G. (1979). Properties of microsoma1 subfractions isolated from developing rabbit skeletal muscle. Eur. J. Biochem. 93, 113121.

368

DAVID H. MACLENNAN ET AL.

Zubrzycka-Gaarn, E., and Sarzala, M. G. (1980). Sarcoplasmic reticulum and sarcolemma during development. I n “Plasticity of Muscle” (D. Pette, ed), pp. 209-223. De Gruyter, New York. Zubrzycka-Gaarn, E., Korczak, B., Osinska, H., and Sarzala, M. G. (1982). Studies on sarcoplasmic reticulum from slow-twitch muscle. J. Muscle Res. Cell Moril. 3, 191-212. Zubrzycka-Gaam, E., Campbell, K. P., MacLennan, D. H., and Jorgensen, A. 0. (1983). Biosynthesis of intrinsic sarcoplasmic reticulum proteins during differentiation of the myogenic cell line L6. J. Biol. Chem. 258, 4576-4581. Zubrzycka-Gaarn, E., Cornell, R., and MacLennan, D. H. (1985). Similarities between calcium oxalate leaded sarcoplasmic reticulum vesicles from developing and mature muscle. J . B i d . Chem. (submitted).

CURRENT TOPICS IN MEMBRANES AND TRANSPORT. VOLUME 24

Chapter 9

Receptors as Models for the Mechanisms of Membrane Protein Turnover and Dynamics H.STEVEN WILEY Department of Pathology University of Utah College of Medicine Salt Lake City, Utah

...... ............. ........................................... 111. Technical Approaches for Analyzing Receptor Behavior and Function . . . . . . . . . . . . IV. Ligand Binding and Receptor Number . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Theoretical Aspects of Ligand Binding . . . . . . . . .........

369 370 372 374 374

B. Equilibrium Binding to Cells . . . . . . . . . . . . . . . . C. Steady-State Binding to Cells. . . . . . . . . . . . . . . . V. Mechanisms and Models of Receptor Internalization. A. Membrane Turnover and Endocytosis . . . . . . . . . B. Lateral Mobility of Membrane Proteins and Endocytosis. . . . . . . . . . . . . . . . . . . . C. Association of Receptors with Coated P i t s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Receptor Biosynthesis and Recycling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Evidence for Receptor Recycling.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Models of Receptor Recycling., . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Receptor Biosynthesis and Degradation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Concluding Remarks ......... .............. References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

386 392 396 397 400 403 404 405

.........

1.

INTRODUCTION

Eukaryotic cells are dynamic structures in which virtually all molecular components are in a state of continuous turnover. Even when not undergoing active growth, cells continue to synthesize and degrade proteins. As has been demonstrated both theoretically and experimentally, it is the relative rate of both synthesis and degradation of specific proteins that determines the steady-state com369 Copyright 0 1985 by Academic Pres5, Inc All nghts of reproduction In any form reserved ISBN 0-12-153324-7

370

H. STEVEN WILEY

position of cells. Since a cell’s structural composition determines its functional capability, the regulation of the turnover of individual proteins is intrinsically involved in the regulation of cellular function as a whole. However, eukaryotic cells contain a number of membranous organelles, each having a specialized protein and lipid composition (Wibo et al., 1981). The cellular mechanisms by which the individual composition of these organelles is maintained in spite of continuous turnover of their components is currently an active area of research in cell biology. One of the most difficult technical problems in investigating the dynamics of cellular membranes is identifying and quantitating a specific protein in a membrane that contains hundreds, if not thousands, of different species. The approach that I cover in this review article is the investigation of a select subset of proteins that bind specific ligands. These ligand-binding proteins are commonly referred to as receptors and include the hormone receptors, such as those for epidermal growth factor (EGF), and the metabolic product receptors, such as those for low-density lipoprotein (LDL). Since receptors can bind ligands very specifically at extremely low concentrations (down to the picomolar range), their behavior can be potentially characterized in situ. In spite of the potential usefulness of receptors as model membrane proteins, they have rarely been used as such. Most investigations on the binding of radiolabeled ligands to receptors have been directed toward investigating the subsequent cellular responses and the role of the ligand-receptor complex in generating these responses. Recent studies on the behavior of the asialoglycoprotein and low-density lipoprotein receptor have focused on the pathway that they traverse both to and from the cell surface, but these and most other studies on receptor behavior treat receptors as special cases and rarely relate their behavior to the plasma membrane as a whole (Brown et al., 1983). However, in this review I attempt to outline some of the general behavioral patterns of cell surface receptors and how this behavior compares with that observed for other membrane proteins. I hope to demonstrate that the investigation of receptor dynamics is a potentially powerful approach for understanding the mechanisms responsible for generating and maintaining the unique composition of the plasma membrane.

II. DEFINITION OF TERMINOLOGY Before proceeding further, it is necessary to define precisely what is meant by turnover in the context of this article. The turnover of cellular proteins is classically defined as a function of the synthetic and degradation rates of the proteins (Schoenheimer, 1942). This definition in essence treats the cell as a single unit in which the total composition of the cell is determined by those two processes. However, this is a very restrictive definition in the case of individual organelles

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

371

since their composition is also a function of the rate at which the proteins are transferred to them. Thus, in the context of this article, the turnover of the plasma membrane will be defined as a function of the rate at which individual proteins appear at the cell surface and the rate at which they are removed. The actual synthetic and degradation rates of individual receptor proteins will also be considered, but as a separate issue from the turnover at the plasma membrane itself. Alterations in receptor number at the cell surface subsequent to ligand addition is a very well documented phenomenon in the hormone receptor literature. The most commonly observed phenomenon is a ligand-specific reduction of receptor number referred to as “down-regulation’’ (Cuatrecasas and Hollenberg, 1976; Aharonov et al., 1978; Kosmakos and Roth, 1980). Down-regulation is an unfortunate and confusing term for a number of reasons. First, it implies that the reduction in receptor number has some sort of regulatory role in the cells’ response to a hormone. This has not been conclusively demonstrated. Second, the term does not distinguish between reduction in apparent receptor number due to enhanced internalization of the occupied receptor, a “masking” of unoccupied receptors, or a reduction in the synthetic and/or appearance rate of the receptor at the cell surface. However, because of the extremely wide usage of the term down-regulation, it would seem unadvisable to discard the term completely. For lack of a better nomenclature, I will retain the usage of the general term down-regulation for those instances in which the functional level of receptor loss is unknown. As suggested by Lloyd and Ascoli (1983), the term homologous down-regulation will be used when a ligand reduces the number of its own receptors (Heldin et al., 1982; Ward et al., 1982; Zigmond et al., 1982) and heterologous down-regulation will be used when a ligand reduces the number of receptors for an unrelated ligand (Chait et al., 1980; Ascoli, 1981; Knutson et al., 1982). The term endocytic down-regulation will be used to describe a reduction in receptor number due to an enhanced internalization of occupied receptors (Wiley and Cunningham, 1981). The term synthetic down-regulation will be used to describe those cases where the reduction in receptor number is due to a reduced synthetic rate of the receptor itself (Brown and Goldstein, 1975). These instances of down-regulation are the best documented so far, although other mechanisms may be found in the future. Kaplan (1981) has proposed that receptors be considered as members of two distinct classes. The class 1 receptors are those whose primary function is to transmit information to cells. This class would include the hormone and growth factor receptors. The class 2 receptors are those whose primary function is to facilitate the transport of large metabolically significant molecules into cells. This class includes the low-density lipoprotein, transferrin, and a,-macroglobulin receptors. While this classification scheme is somewhat arbitrary, it does serve a useful function. Since the “function” or physiological significance

372

H. STEVEN WILEY

of the two classes of receptors are different, their behavior should also reflect these differences. However, there is a danger in overemphasizing the differences between receptors and downplaying their similarities. Nevertheless, Kaplan’s receptor classification scheme is very useful in defining and discussing the behavior-function relationship of receptors and so his terminology will be used throughout this article.

111. TECHNICAL APPROACHES FOR ANALYZING RECEPTOR BEHAVIOR AND FUNCTION Virtually all of the techniques presently available for analyzing the fate of membrane receptors rely on following the bound ligand rather than the receptor itself. There are two commonly used methods for tracing the fate of the bound ligand. In the last several years ligand derivatives have been prepared that can be visualized with the light and electron microscope. These derivatives include ferritin-tagged ligands which can be directly visualized (Anderson et al., 1977a; Orci et al., 1978; McKanna et al., 1979; Gershon et al., 1981) as well as enzymatically labeled derivitives that can be subsequently visualized by histochemical procedures (Wall and Hubbard, 1981; Willingham and Pastan, 1982). Fluorescent derivatives have also been used in conjunction with image-intensification microscopes to delineate some of the general features of ligand uptake by cells. Studies using these techniques have been extensively reviewed elsewhere and will not be considered here (Pastan and Willingham, 1981; Brown et al., 1983; Haigler, 1983; Anderson and Kaplan, 1983). However, presently available visualization techniques tend to be subjective and nonquantitative in nature. There have been problems with both the sensitivity and the specificity of the techniques. Nevertheless, these techniques have indicated that the internalization of ligands proceed primarily if not exclusively through specialized regions of the cell surface termed coated pits (Brown et al., 1983). Morphometric analyses of the uptake of the EGF receptor indicated that the receptor and the ligand were taken up as a complex (McKanna et a l . , 1979). More recently, visualization techniques have indicated that internalized ligands are found in a perinuclear ring in association with the Golgi apparatus prior to their appearance in lysosomes (Willingham and Pastan, 1982; Khan et al., 1982). These types of studies have been useful in defining the overall pathway that ligands traverse subsequent to internalization. Fluorescence-labeled ligands have had a wider applicability to studies on receptor-ligand interactions. Of particular importance have been studies on the lateral diffusion of ligand-receptor complexes in the plasma membrane using the technique of fluorescence recovery after photobleaching (FRAP) (Edidin et al., 1976; Schlessinger et al., 1978). This technique uses an intense beam of laser

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

373

light to irreversibly bleach a small area of the cell surface labeled with a fluorescence-labeled ligand. The time course of recovery of fluorescence into the region is dependent upon the rate at which mobile ligand-receptor complexes diffuse into the bleached region. Using the appropriate equations, one can then estimate the lateral diffusion coefficient of the mobile ligand-receptor complexes (Soumpasis, 1983). While the number of ligand-receptor systems that have been investigated using this technique is small, the information derived from these systems has been extremely valuable in understanding the dynamics of the occupied cell surface receptor (Hillman and Schlessinger, 1982; Maxfield er al., 1981; Barak and Webb, 1982). Sklar and collaborators have developed a number of elegant approaches for quantitating the interaction of fluorescent ligands with cells (Sklar er al., 1982, 1984; Finney and Sklar, 1983). Those investigators found that when fluoresceinated chemotatic peptide was bound to its receptor on human neutrophils, the fluorescence could not be quenched by antibodies directed to the fluorescein moiety. In addition, when the pH of the medium was dropped from 7.4 to 4.0, the ligand associated with the surface receptors was immediately quenched, but the internalized ligand was quenched very slowly. These observations were used to develop a series of continuous fluorimetric assays for measuring the kinetics of binding and internalization of the chemotactic peptide. Because of the temporal resolution of these assays, Sklar et al. (1984) have been able to construct a very detailed description of ligand-cell interactions in this system. The application of these very powerful techniques to other systems would greatly expand our current knowledge of the dynamics of ligand-cell interactions. Perhaps the most popular methods for analyzing the fate of ligand bound to specific receptors are kinetic, equilibrium, and steady-state analyses of the interaction of radiolabeled ligands with cells (Carpenter et a/., 1975; Pollet et al., 1977; Connolly et al., 1981). These methods have the advantage of being very sensitive, specific, and quantitative. The difficulty with this approach is that the interpretation of the data is critically dependent upon the mathematical model used for analyzing the data. These models have been typically derived from those that describe binding reactions at equilibrium (Rodbard, 1973; Feldman, 1972; Feldman et a / . , 1972; Priore and Rosenthal, 1976). These models are rarely applicable to ligand-binding studies on cells maintained at near-physiological conditions since cells generally internalize the receptor-bound ligand (Brown et al., 1983; Anderson and Kaplan, 1983). What is needed is a mathematical description of ligand-receptor interactions that incorporates the known behavior of these systems. The construction of mechanistic models allows one to use kinetic data to discriminate between alternate models of receptor behavior. An empirical equation can fit the data, but this does not mean that the equation is a model. What is desired is a model that mechanistically details the behavior of the membrane

374

H. STEVEN WILEY

receptor. For example, if receptor occupancy is proposed to induce receptorreceptor interactions that are rate limiting in the process of receptor internalization, then this postulated step would necessarily be a second-order reaction and, thus, receptor internalization would be a function of the concentration of occupied receptors at the cell surface (Hillman and Schlessinger, 1982; Wiley and Cunningham, 1982a). The approaches that can be used to develop and validate a mechanistic kinetic model have been extensively reviewed in the enzyme kinetics literature (e.g., Lam, 1981). Nevertheless, it is important to emphasize that the ability to fit data to a kinetic model by parameter adjustments is not a test of the validity of the model. Only if the model has predictive value under a wide range of experimental conditions can it be considered a good approximation of the mechanisms involved.

IV. LIGAND BINDING AND RECEPTOR NUMBER A. Theoretical Aspects of Ligand Binding Any kinetic analysis of the rate of turnover of membrane proteins requires a quantitative determination of the number of receptors at the plasma membrane at any one moment. The most common method for quantitating receptor number and turnover rate is by binding radiolabeled ligands to the receptor and then tracing the subsequent fate of the radioactive “tag.” Because this method is indirect, it is important to know the quantitative relationship between the binding of the ligand and the actual number of cell surface receptors. Traditionally, the mathematical relationship between receptor number and binding has been modeled as equilibrium systems following the laws of mass action (Feldman er a!., 1972; Rodbard, 1973). This is the basis of familiar methods such as the Scatchard plot (Scatchard, 1949). Using equilibrium assumptions, one can express the relationship between fractional receptor occupancy and ligand concentration using the familiar Langmuir equation (Langmuir, I9 18; Colquhoun, 1979):

where pa is the fraction of the receptors that are occupied, [L] is the free ligand concentration, and Kd is the equilibrium dissociation constant. Kd is equivalent to kd/k, where kd is the first-order dissociation rate constant of the ligandreceptor complex and ka is the second-order association rate constant. Their ratio (Kd) is conventionally used as a measure of receptor affinity since it is equivalent to the concentration of ligand necessary to occupy half of the receptors in the system at equilibrium. If a ligand-receptor system indeed behaves according to the law of mass action, then the kinetics of their interaction can be analyzed

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

375

using any one of a number of mathematical formulations developed for the analysis of enzyme-substrate and antibody-antigen interactions. Unfortunately, it has become increasingly obvious in recent years that the simple assumptions regarding the application of the laws of mass action are not strictly valid when receptors are embedded in a membrane. Extensive theoretical work has been done by Berg and Purcell (1977) as well as by DeLisi and collaborators (DeLisi and Mertzger, 1976; DeLisi and Wiegel, 1981; Dower el al., 1981) cn the significance of receptor affinity in the context of the cell surface. This work can be summarized as indicating that in some, if not all, instances the density and distribution of receptors at the cell surface can have a significant effect on the rate of formation and dissociation of ligandreceptor complexes. For example, let us take the case of the significance of the second-order association rate constant of a receptor and a specific ligand. With the simple assumption of a homogeneous phase system and the laws of mass action, an equation for the diffusion-limited association rate constant can be derived (Amdur and Hammes, 1966):

k,

=

4.rra(DL + D R ) fNollOOO

where u is the sum of the radius of the ligand and receptor; D, and D, are the diffusion coefficients of the ligand and receptor, respectively; X, is Avogadro’s number, and f is an expression of the potential energy and orientation requirements of the system. Note that any alteration in the association rate constant observed with the system would usually be interpreted as reflecting a change in f (e.g., conformational change of the receptor or alteration in ionized amino acid residue in the binding site) since the other terms in the equation are more or less constant (Waelbroeck, 1982). However, Berg and Purcell(l977) have examined the implications of the receptor being fixed to a cell surface while the ligand is free in solution. Using this more realistic set of assumptions, the theoretical diffusion-limited association rate constant would be k,

=

4.rrDJuN,/(N.s

+ .rra)1000

(3)

where s is the effective radius of the receptor binding site, a is the radius of the cell, and N is the number of receptors per cell. Again, orientation and potential energy factors of the ligand-receptor interaction are included in the termf. Even though this simplified derivation ignores factors such as the shape of the cell, it is apparent that alterations in the association rate constant could in fact reflect alterations in a number of different cellular aspects extrinsic to the receptor itself. A more extensive and general treatment of the factors influencing the binding of ligands to cell surface receptors has been presented by DeLisi and Wiegel (198 I). They show that receptor distribution and orientation factors can strongly influence the kinetics of binding in some cases but not in others. In addition, any electric potential field around the cell could have a significant effect on the

376

H. STEVEN WILEY

observed rate of binding. Since a number of different cell types have been shown to generate endogenous electrical fields (e.g., Robinson, 1979), this would also be expected to alter the effective association rate. DeLisi and Wiegel (1981) conclude that even though the kinetics of association and dissociation would be altered by these secondary factors, they would have no effect on the equilibrium association rate constant. This conclusion was based on assuming a uniform distribution of receptors at the cell surface. In many cases, receptors demonstrate nonhomogeneous distributions at the cell surface (Haigler et al., 1979a; Carpentier et al., 1981; Wall and Hubbard, 1981; Brown et al., 1983). Whether a significant heterogeneity in receptor distribution would alter the slope of a “Scatchard plot” of equilibrium binding is presently not known. However, it has been demonstrated theoretically that receptor clustering significantly decreases the diffusion-limited rate of ligand binding (Goldstein and Wiegel, 1983). If the affinity of a cell surface receptor for a ligand were simply a function of a number of known and unknown properties of both the receptor and the plasma membrane, then one could lump all of these properties together under the term apparent afsinity and proceed with experimental measurements of receptor number and properties. For example, experimental determinations of turnover rates depend upon accurate measurements of receptor number as a function of time. The relative number of receptors present at any single time point is usually determined by the relative amount of radiolabeled ligand associated with the cell surface at that time. However, the time-dependent association of the ligand with the cell will reflect the relative receptor numbers only if the apparent receptor affinity remains constant, Unfortunately, the very dynamic nature of the cell surface frequently prevents this simple approach. It has been extensively documented that the binding of many ligands to cell surface receptors alters their distribution and turnover (Carpentier er al., 1981; McKanna et al., 1979). Thus the apparent affinity could also change as a function of time, complicating many types of studies.

6. Equilibrium Binding to Cells One potential approach to determining receptor number is to conduct binding studies under restricted conditions. The most common approach is to lower the temperature of the cells rapidly to O’C, which effectively blocks membrane turnover (Thilo and Vogel, 1980; Weigel and Oka, 1982) and, in some cases, receptor redistribution-(Bar& *and-Webb, 1982; Brown et al., -1983). This approach is most applicable to cultured mammalian cells. Nonmammalian cells (such as Xenopus oocytes) maintain a low but significant capacity to internalize receptor-bound ligands even at 5°C (Wallace et al., 1973). One of the most

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

377

important aspects of ligand binding at low temperatures is the speed at which the cells are cooled. Slowly lowering the temperature of cells from 37°C can result in significant alterations in the cell surface receptor number (Weigel and Oka, 1983a). In other cases, cooling the cells can result in the “shedding” of the receptors into the external medium (Kaplan and Keogh, 1982). These types of factors are often overlooked in binding studies conducted at low temperatures. Nevertheless, the procedure has proved to be generally useful in circumventing the intrinsic problems of cell surface turnover and redistribution of receptors. Once the cells are brought to a lower temperature, equilibrium binding can be conducted. The amount of ligand in association with the cells at equilibrium is then usually interpreted according to some standard mathematical transformation such as the Scatchard plot (Scatchard, 1949). In this transformation the amount of binding at equilibrium can be expressed as

Plotting the ratio of the bound and free ligand ([LR] over [L]) versus the bound ligand results in a linear plot, in the case of a single class of noninteracting receptors, with a slope of - I/& and an x axis intercept of [R]t,ta,. The maximum number of ligand molecules bound to the cell surface ([R],,,,,) is thus easily extracted from a Scatchard plot as is the apparent cellular affinity for the ligand. However, as discussed above, the cellular affinity of a ligand at equilibrium is of dubious significance since it can potentially be influenced by a host of factors whose contribution is pllesently not known. In addition, the number of ligand molecules bound can be translated to the number of receptor molecules present only if the stoichiometry involved is known. This is extremely important for investigating multivalent receptor systems such as that involved in the binding of asialoglycoproteins by cells (Connolly er al., 1981, 1982). Since the effective stoichiometry in multivalent receptor systems depends not only on the valency of the ligand, but also on extrinsic cellular parameters such as surface receptor density, distribution, and mobility (Dower et al.. 1981; Pincus et al., 1981), deriving absolute receptor numbers for such systems is difficult. As has been discussed for a number of years, the Scatchard plot and all other methods used for linearizing binding data suffer from the intrinsic problem of nonlinear weighting of experimental errors (e.g., Rodbard, 1973; Cuatrecasas and Hollenberg, 1976). This problem is generally compounded by a rise in “nonspecific” binding with increasing ligand concentrations or other factors giving rise to nonlinear, apparent “multicomponent” binding curves. Since the number of receptors on the cells is derived by extrapolating to infinite ligand concentrations, these types of errors can lead to serious over- and underestimations of receptor number. These issues have been extensively discussed in other reviews (Rodbard, 1973; DeLean and Rodbard, 1979; Hollenberg and Cuatrecasas, 1979). Perhaps the best approach to analyzing binding data is to

378

H.STEVEN WILEY

employ a nonlinear regression program in a computer (Feldman, 1972; Lam, 1981; Duggleby, 198 1). The use of computer-assisted techniques is becoming increasingly common in the analysis of binding data as a result of the rapidly decreasing costs of microcomputers and the introduction of simple, generalized, data-analysis programs. The computer-assisted approach also allows the investigator to make a reasonable determination of receptor number even in the face of complex equilibrium binding. While the significance of such apparent multicomponent binding is not always obvious, the computer-assisted approach provides a rational means by which to effectively describe the binding. Meaningful interpretation of any binding data can only be made if the specific activity of the bound ligand is known. Most ligands are generally labeled with 1251 because of the high specific activities obtained. This labeling is either directed at the tyrosine residues [such as the oxidative Chloramine-T (Hunter and Greenwood, 1962) and the IodoGen procedures (Salacinski et al., 1981)] or the free amino groups of the ligand (Bolton and Hunter, 1973). These labeling procedures almost always yield a heterogeneous pattern of modification of the protein as a result of the natural distribution of amino acid residue accessibility found in the ligand population. However, calculation of the specific activity of the labeled protein assumes that all labeled species bind identically. This is not always the case. Measuring the ability of unlabeled ligand to compete with the labeled ligand for binding will only yield the average specific activity of the ligand preparation, but will not necessarily reveal the fraction of the labeled ligand that actually binds to the cells (Calvo et al., 1983). A good example of the types of binding artifacts that can arise due to the radiolabeling procedure is that of thrombin binding to fibroblasts. In their studies, Low and Cunningham (1982) found that radioiodinated thrombin displayed a much lower capacity for binding than did native thrombin. They demonstrated that this was entirely due to the inability of labeled thrombin containing di[1251]tyrosineto bind to cells. However, the ability of thrombin containing m ~ n o [ ~ ~ ~ I ] t y r oto s i nbind e to cells was the same as native thrombin. Thus, an incorrect estimate of apparent receptor number would have resulted from the inclusion of the di[ 1251]tyrosine-containingthrombin in the specific activity calculations. Recently, several general methods for accurately determining the specific activity of radiolabeled ligands have been presented (Bowen-Pope and Ross, 1982; Calvo et al., 1983). The methods are based on the ability of increasing amounts of cell membranes or cells to deplete the “bindable” fraction of ligand from the labeled preparations. By extrapolating the amount of membranes or cells added to infinite concentrations, the fraction of labeled ligand that is actually capable of receptor binding can be readily calculated. These types of measurements are crucial for interpreting experiments in which the ability of several different ligands to bind to the same receptor are compared.

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

379

C. Steady-State Binding to Cells An alternate approach to the problem of measuring receptor number in radiolabeled ligand studies is to dispense with the notion of a static number of receptors and replace it with relative rate constants (Wiley and Cunningham, 1981). Let us first consider a cell which is at steady state with regard to its cell surface turnover. By definition, at steady state the rate at which individual proteins, such as receptor proteins, appear at the cell surface is equal to the rate at which they are removed. If the removal of the receptors is a first-order process, then at steady state

v, = kJR1

(5)

where V , is the rate at which the receptor appears at the cell surface, k, is the firstorder rate constant of removal, and [R] is the concentration of empty receptors at the cell surface. It thus follows that at steady state [R]

=

V,/k,

(6)

Therefore, the absolute level or concentration of any single protein species at the cell surface is a function of both the appearance rate and the turnover rate constant, k,. Equations (5) and ( 6 ) describe the concentration of receptors at the cell surface in the absence of any added ligand and thus describe the level of unoccupied receptors (Wiley and Cunningham, 1981). This expression is similar to those developed by Berlin and Schimke (1Y65) for describing the steady-state concentration of cellular enzymes, but here the zero-order synthetic rate is replaced with a zero-order receptor “appearance” rate at the plasma membrane and the first-order degradation rate constant is replaced by a “removal” rate constant. Equations (5) and ( 6 )simply state that the initial number of receptors at the cell surface is due to the turnover properties specific to those receptors. However, what would happen if the receptor is occupied? If we take the turnover rate constant of the occupied receptors as k, (endocytotic rate constant, Wiley and Cunningham, 1982a) then at the new steady state V , = k,[R] + k,[LR]

(7)

where [LR] is the concentration of occupied receptors at the cell surface. Thus, the number of unoccupied receptors at the cell surface would be [RI

=

(V,/k,) - (k,[LRl/k,)

(8)

Equation (8) does not imply that the value of k, is necessarily different from k,. It does make the simplest assumption that the appearance rate of new receptors is constant. If k, is equal to k,, then Eq. (8) reduces to [Rl = (V,lk,) - [LRI

(9)

380

H. STEVEN WILEY

which states that the number of empty receptors at the cell surface is equal to the total number (Vr/kt)minus the occupied ones. The assumption of a temporally constant total number of cell surface receptors (occupied plus unoccupied) is of major importance in a number of techniques for analyzing the binding of ligands to cells (such as Scatchard plots). However, it is important to note that this assumption can only be valid if k, is equal to kt. If k, is either greater than or less than k,, the total number of receptors at the cell surface will be altered when ligand is present. As shown in Eq. ( 6 ) , the number of receptors present at the cell surface at steady state is the ratio of the rate of appearance of the receptor divided by the first-order rate constant of receptor removal. By substituting rate constants for actual numbers of cell surface receptors, an equation analogous to the Scatchard equation (Scatchard, 1949) was derived for analyzing the binding of radiolabeled ligands to cells at steady state (Wiley and Cunningham, 1981): [LR]/[L]

=

-K,,[LR]

+ K,,(V,/k,)

(10)

with the term K,, for steady-state rate constant replacing the equilibrium association constant in the Scatchard equation. A comparison of Eqs. (4) and (10) shows that the total number of receptors ([R],,,a,)in the Scatchard equation is replaced by V,lk, in the steady-state equation. This is equal to the maximum number of occupied receptors present at the cell surface at steady state as shown by Eq. (7) for the case when there are no remaining unoccupied receptors, i.e., [R] = 0. Thus, a plot of bound over free ligand versus bound ligand at steady state should give a linear plot with a slope of -K,, and an x intercept of Vrlk,. This type of analysis is particularly valuable since methods for the determination of k, have been published (Wiley and Cunningham, 1982a) and thus the apparent value of V, can be readily calculated. However, note that Eq. (10) only expresses the amount of cell surface bound ligand as a function of ligand concentration. The total amount of radiolabeled ligand measured in association with cells very frequently includes an internalized component. Quantitative studies have indicated that the internalized component may represent >90% of the total amount of ligand associated with cells at steady state (Wiley and Cunningham, 1981; Lloyd and Ascoli, 1983). Thus, methods for discriminating surface-bound from internalized ligand are essential for using this type of analysis. One of the most useful aspects of the steady-state “Scatchard” equation is the apparent cellular affinity constant K,,, which is equal to (Wiley and Cunningham, 1981)

keka Kss = kt(kd + k,)

(11)

Thus, the significance of the slope of a steady-state “Scatchard plot” is quite different from the slope of an equilibrium Scatchard plot [see Eq. (4)]. It can be

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

381

formally demonstrated that 1 lKss is equal to the concentration of ligand needed to half-maximally occupy the cell surface receptors at steady state (Knauer et al., 1984). Thus, the functional significance of l l K S s as the cellular affinity for a ligand is similar to the dissociation rate constant in an equilibrium system. The concept of cellular affinity (Ks,) has particular relevance in interpreting nonideal “Scatchard plots” of steady-state binding data or when alterations are observed in cellular affinity. When the binding of radiolabeled ligands to cells yields curvilinear Scatchard plots, they have traditionally been interpreted as reflecting heterogeneity or cooperativity at the level of the receptor molecule (Rodbard, 1973; Feldman, 1972; DeMeyts, 1976). However, this is not necessarily true for those receptors where the surface clearance rates of the occupied versus unoccupied receptors are different (i.e., endocytic down-regulation). Shown in Fig. 1 is a computer-generated graph of the effect of endocytic downregulation on the apparent cellular affinity for a ligand. In those cases in which the intrinsic receptor dissociation rate constant (kd) is high (reflecting low receptor affinity), an increase in the relative clearance of the occupied receptor can have a dramatic effect on the measured cellular affinity at steady state. This “endocytic effect” on measured receptor affinity could give rise to curvilinear Scatchard plots if endocytosis is saturated at high ligand concentrations or by lowered temperatures and may in fact explain a number of reported cases of “negative-cooperativity” that cannot be directly traced to occupancy-dependent changes in receptor affinity (Pollet et al., 1977; DeLean and Rodbard, 1979). Changes in the net turnover rate of specific receptors at the cell surface can also give rise to alterations in apparent cellular affinity. In the case of class 2 receptors that do not display endocytic down-regulation, the steady-state affinity of the cell for the ligand can be expressed as

K,,

=

k,W, +

kd)

(12)

since in this case k, should be equal to k,. Since the total number of receptors at the cell surface is equal to V,lk,, a coordinate increase in V , and k, would result in an increase in turnover rate of the receptor without altering the net number of surface receptors observed at steady state. However, the increase in k, would give rise to a lower apparent cellular affinity as shown by Eq. ( 1 2). The important point is that the “affinity” of isolated receptors or cells maintained at 0°C is not intrinsically the same as the apparent affinity of a dynamic cell for a ligand. This consideration must be kept firmly in mind when attempting to interpret steady-state or kinetic binding data. If we dispense with the notion of absolute number of receptors at the cell surface, then how do we go about measuring the rate constants? Perhaps the simplest rate constants to measure are those describing the binding and internalization of the ligand molecule since following the ligand molecule requires making very few assumptions regarding the unoccupied receptors. An example

382

H. STEVEN WlLEY

10

20

30

40

50

60

70

4314

FIG. I . The effect of endocytic down-regulation on the apparent cellular affinity for ligands at steady state. This figure demonstrates the apparent increase in measured receptor affinity as a function of increasing degrees of endocytic down-regulation. The fold increase in affinity along the ordinate is the ratio of cellular affinity (Kss) to the intrinsic receptor affinity ( k , / k d ) . The degree of endocytic down-regulation is expressed along the abscissa as the ratio of the rate constant of internalization of occupied receptors (k,) to that of unoccupied receptor ( k J . The eight different curves were calculated assuming different receptor dissociation rate constants ( k d ) . The dissociation rate constants ( k d ) used in the calculations for each curve (in units of sec- I ) were (1) 2 x (2) I X 10-2;(3)5 X 10-3;(4)2.5 X l O - 3 ; ( 5 ) 1.25 X 10-3;(6)6.25 X IOC4;(7)3.l3 X 10-4;(8) 1.56 X lop4. Note that as the intrinsic strength of the receptor-ligand interaction decreases (i.e., as kd increases) endocytic down-regulation has an increasing effect on the apparent cellular affinity for ligands.

of this type of analysis is the experimental determination of the endocytic rate constant (Q. It has been demonstrated that plotting the ratio of the ligand inside the cell to that bound to the cell surface as a function of time yields a good approximation of k, (Wiley and Cunningham, 1982a). This type of analysis (InlSur plot analysis) is quite simple and fast and has been used by a number of investigators to measure internalization rates of occupied receptors (Knauer et al., 1984; Lloyd and Ascoli, 1983; Hoppe and Lee, 1984).

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

383

The development of techniques for measuring the internalization rates of occupied receptors was stimulated by the development of rapid and simple techniques for discriminating cell-surface bound ligand from that internalized by the cells. The most generally useful technique was developed by Haigler et al. (1980) for discriminating internalized EGF from that still bound to cell surface receptors. This technique was based on the observation that a mild acid treatment rapidly releases the ligand from surface receptors without significantly perturbing cellular integrity. The speed, simplicity, and wide applicability of the “acidstripping” technique have resulted in its wide use in kinetic studies of receptormediated endocytosis (Lloyd and Ascoli, 1983; Car0 et al., 1982; Olefsky and Kao, 1982). The introduction of computer-assisted techniques that correct for the incomplete removal of ligands from surface receptors has facilitated the quantitative interpretation of the data (Wiley and Cunningham, 1982a). As technical aspects of this general technique improve, it should be possible to extend the temporal resolution of the events occurring at the cell surface between the initial binding of ligands and their subsequent internalization. While the In/Sur plot and other similar techniques are very useful in determining the rate constant of occupied receptor internalization, they all suffer from certain drawbacks. As pointed out by Goldstein et al. (1981), all of the ligand bound to the cell surface may not be equally accessible to internalization. Thus the measured value of k, is really

k:

=

k,[LRl,/([LRl,

+ MI,)

(13)

where [LR], and [LR], are, respectively, the concentrations of ligand-receptor complexes either clustered at internalization sites or unclustered (Goldstein et al., 1981). If the majority of ligand-receptor complexes are “clustered” at steady state of internalization, then the measured value of k, (k:) will approximate the “true” value. There are other limitations in using the In/Sur technique such as a necessity to correct for ligand degradation, an accurate discrimination between surface-bound and internalized ligand, and the restriction of approximate steady-state conditions (Wiley and Cunningham, 1982a). Future technical improvements in analyzing the data derived from In/Sur plots should provide a more generally applicable method for quantitating receptor-mediated endocytosis and should allow a direct evaluation of the limits inherent in the technique. The steady state approach is also useful in analyzing the kinetics of ligand binding to the receptor itself at the cell surface. Consider a situation in which the binding of a ligand to a cell surface receptor is followed by either dissociation or internalization of the ligand-receptor complex. At steady state the rate of binding of the ligand will be equal to the rate of dissociation plus the rate of internalization, or

384

H. STEVEN WILEY

where [LR], is the fraction of the total receptor-ligand complexes that can dissociate and [LR], is that portion that can be internalized. If we make the simplifying assumption that all of the ligand bound to surface receptors can either be internalized or dissociated, then Eq. (14) reduces to k,[LI[Rl

= (kd

+ ke)[LR1

(15)

If we restrict ourselves to the situation in which the concentrations of neither ligand nor empty receptors change significantly with respect to the time period of the experiment, then the left side of Eq. (15) can be replaced by a zero-order term, V , (for velocity of binding). Thus, at steady state, V+,/(kd

+ k,)

=

[LR]

(16)

With this form, we can solve the differential equation for the time course of approach to steady state (Berlin and Schimke, 1965) with the conditions that to = 0 and [LR], = 0. [LRI, =

[Vb/(kd

+ ke)I (1

-

exp[-(k, + ke)tI)

(17)

This is a general solution for the time course of approach to steady-state binding to the cell surface. A very interesting aspect of Eq. (17) is the length of time it takes for the cell surface to reach half-steady-state binding. If half-steady-state binding is expressed as 2!

[LRI

=

t

[v,l(kd

4- ke)l

(18)

then substituting this into the left half of Eq. (17) and rearranging yields

t

=

t,,,

+ ke)t] = ln2/(kd + k,)

exp[-(kd

which is identical to the half-life of the ligand at the cell surface. Since the value of k, can be estimated by the In/Sur plot technique, one can thus estimate the dissociation rate constant of the ligandlreceptor complex by measuring the time required to reach half the steady-state binding. The use of Eq. (17) together with a nonlinear curve-fitting program (Duggleby, 1981) should allow one readily to estimate the value of k, and kd by simply knowing beforehand the value of k,, the ligand concentration, and the total number of receptors at the cell surface. This is only a single example of the use of steady-state assumptions in analyzing binding data. As more extensive steady-state treatments are developed for ligand-cell systems, and as experimental protocols are devised that maintain cells at a steady-state condition, their usefulness and general applicability should increase.

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

V.

385

MECHANISMS AND MODELS OF RECEPTOR INTERNALIZATION

In Section IV,C, the internalization rate of receptors at the cell surface was expressed as a function of their concentration and a rate constant [e.g., Eq. (7)J. But what does this mean in mechanistic terms? More specifically, what does k, and/or k, represent with respect to the behavior of the cell surface? Since the rate at which different receptors are internalized can vary considerably (Lloyd and Ascoli, 1983), what are the factors which can influence this process? As illustrated in Eq. (6), the specific concentration of any single membrane protein at the cell surface is dependent on the rate constant of its removal from the surface. Thus, the processes and mechanisms responsible for the differential internalization rates of receptors are potentially involved in determining the specific composition of the plasma membrane as a whole.

A. Membrane Turnover and Endocytosis Fluid-phase pinocytosis is apparently a universal occurrence in eukaryotic cells (Silverstein et al., 1977). During this process, segments of the cell surface are continuously removed with no corresponding decrease in surface area (Steinman et al., 1976). Owing to these and other observations, most current views of the mechanisms of turnover of cell surface proteins invoke endocytosislexocytosis as the means by which cells can remove or replace specific proteins in the plasma membrane. What becomes very important in considering the role of endocytosis in membrane protein turnover is the specificity of the endocytic event itself. If a random segment of the plasma membrane is incorporated in the process, then a random selection of the membrane proteins also should be incorporated. However, most current evidence regarding endocytosis is not consistent with a random process (Anderson et al., 1978; Brown et al., 1983). The best documented examples of endocytosis are those mediated through specialized “coated pit” regions of the membrane. This coat is composed primarily of the protein clathrin (Pearse, 19761, which is thought to be involved in the formation of the endocytic vesicle. Coated pits are generally distributed in groups at the cell surface (Pfeiffer et al., 1980) and in association with cytoskeletal elements (Anderson et al., 1978). Freeze-fracture studies indicate that coated pits contain a distinct population of large intramembrane particles which could represent specific membrane proteins (Orci et al., 1978). Other evidence indicates that these regions are relatively devoid of cholesterol and thus could possess a distinct lipid composition (Montesano et al., 1979). Therefore,

386

H. STEVEN WlLEY

the available evidence indicates that endocytosis through coated pits involves spatially and compositionally distinct regions of the cell surface. If there are specialized regions of the cell surface involved in endocytosis, then regulating the composition of these regions will regulate those proteins that are internalized. The mechanisms by which this compositional regulation is achieved is perhaps the central issue in understanding receptor internalization and turnover. It is appropriate to consider those mechanisms that could be involved in regulating the localization of receptors and other membrane proteins at internalization sites. These mechanisms could involve either the alteration of protein mobility in the membrane or the modulation of the ability of the proteins to migrate into specialized regions. Each of these possible mechanisms will be considered with respect to what is actually known about the kinetics of receptor internalization.

8. Lateral Mobility of Membrane Proteins and Endocytosis The fluid mosaic model of the plasma membrane implies that intrinsic membrane proteins should be free to diffuse laterally and thus should be randomly distributed at the cell surface. However, experimentally it has been found that some membrane proteins are relatively mobile whereas others are not. In a study on the mobility of surface antigens marked with specific fluorescent antibodies, Gall and Edelman (1981) found that membrane proteins could be classified as either mobile or immobile. Using the fluorescence photobleaching recovery technique in conjunction with derivatized monovalent Fab’ fragments specific for cell surface proteins, they found that specific membrane proteins that were freely mobile in one cell type were sometimes immobile in others. Importantly, surface antigens could change from mobile to immobile depending on culture conditions. Gall and Edelman (198 1) interpreted these data to indicate that the ability of membrane proteins to diffuse laterally was not intrinsically a global property of the cell surface. Instead the mobility of individual proteins could be modulated. A similar conclusion was reached by Po0 (1981) and his colleagues. They subjected cultured cells to an electrical field which resulted in the migration of membrane proteins to either end of the cell. After the current was removed, they examined the time course of back diffusion of the proteins. They found that the addition of concanavalin A immobilized the lectin-specific proteins but did not affect the diffusion of other lectin-binding proteins. This indicates that modulation of protein mobility in the membrane is specific. Even though some membrane proteins and receptors seem “diffusible” by the fluorescence photobleaching technique, the measured rates of diffusion are almost always an order of magnitude lower than predicted by the membrane

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

387

diffusion theory of Saffman and Delbruck (1975). This has been explained by either an association of the membrane proteins with cytoskeletal elements or by unknown complexities in calculating the intrinsic viscosity of the plasma membrane. However, some membrane proteins do demonstrate lateral diffusion coefficients very close to the theoretical values, an example being that of rhodopsin in the photoreceptor cells of the frog (Wey e t a / ., 198I ) . This tends to support the general features of the theory of Saffman and Delbruck (1975). Recent evidence indicates that the observed lower rate of receptor diffusion at the cell surface is mediated by cytoskeletal elements. In examining the diffusion of the LDL receptor in fibroblasts, Barak and Webb (1982) measured the diffusion coefficient of the occupied receptor both in intact cells and cells that had been induced to “bleb” by treatment with sulfhydryl agents. The blebs still possessed the LDL receptor but lacked F-actin, which is normally associated with the cytoskeleton. While the LDL receptor in the intact cells displayed a relatively low diffusion coefficient (4.5 X l o - ” cm2 sec-I), the receptors on the blebs had diffusion coefficients several orders of magnitude higher and close to that predicted by theory (1.5 X l o p 8 cm2 sec-I). It is far from clear how limitations are imposed on the lateral mobility of membrane proteins. Do the receptors or other membrane proteins actually bind transiently to the cytoskeleton or are there physical boundaries imposed by these structures? The answers are presently unknown primarily because of the complexities of the cell membrane itself. For example, there is evidence that membrane proteins can regulate the viscosity of the lipid bilayer. Jacobson and colleagues (198 1) compared the relative “viscosity” of isolated fibroblast plasma membranes and liposomes constructed from extracted plasma membrane lipids. While the diffusion coefficient of the fluorescent lipid probes used to measure membrane viscosity were nearly the same in the two types of membranes at O”C, the diffusion of the probes was significantly slower at 37°C in the isolated plasma membrane relative to the liposomes (approximately fourfold slower). Jacobson et a f . (1981) suggest that the interactions between the proteins and the membrane lipids could be a temperature-sensitive process that affects the overall viscosity of the lipid bilayer. It seems that the regulation of the lateral diffusion of membrane receptors could be extremely complex and involve a number of different types of protein-protein and protein-lipid interactions. If membrane proteins or receptors were completely immobilized in a region on the cell surface distal to an internalization site, then one would expect that their rate of internalization would be exceedingly slow. However, what would be the expected rate of internalization for a protein that diffused extremely rapidly? An upper limit can be set for the internalization rate of a randomly distributed protein by knowing the rate at which the cell surface as a whole is internalized. It has been estimated that mouse L cells internalize about 54% of their cell surface in an hour (Steinman et al., 1976). If surface proteins are randomly internalized, then

388

H. STEVEN WILEY

this would result in a rate constant of internalization of about 1.2 X lo-* min-I. Essentially the same estimate can be obtained for human fibroblasts using their measured rate of fluid uptake (Wiley and Cunningham, 1982b), their surface area (2500 pm2) and assuming that membrane is internalized through 100-nm vesicles (Orci et al., 1978). This upper rate of internalization is much lower than that observed for most occupied receptors (Goldstein et al., 1981; Wiley and Cunningham, 1981), and is actually much higher than the rate estimated for unoccupied hCG (human choriogonadotropin) and EGF receptors (Lloyd and Ascoli, 1983), a finding indicating that internalization is not a random process. Instead of assuming that receptors are randomly distributed, let us assume that they are all “clustered” at a coated pit. We can estimate the upper limit of the internalization rate constant by using the observation that coated pit regions constitute between 1 and 2% of the cell surface (Orci et al., 1978; Goldstein and Wofsy, 1981). This translates to between 800 and 1600 coated pits, each 100 nm in diameter, for an average human fibroblast. If all of the internalization of the cell surface occurs at these sites, then the calculated rate constant of their internalization would be between 0.6 and 1.2 min-’ and the mean lifetime of a coated pit would be 0.8-1.7 minutes. Of course, if fluid uptake by cells involved other types of endocytotic vesicles besides those derived from coated pits, as seems very likely (Orci et al., 1978; Haigler et al., 1979b; Silverstein et al., 1977), then the upper limit of the coated pit internalization rate would be lower and their mean lifetime would be longer. Nevertheless, the receptors or membrane proteins that are localized exclusively in coated pits should display internalization rates equal to that of the coated pits themselves. Anything less than exclusive localization at these structures should result in lower rates of internalization. How does the theoretical upper limit of receptor internalization compare with that actually observed? Table I provides a tabulation of the apparent internalization rate constants for occupied receptors; the values have either been published or calculated from published data. The highest internalization rate constant observed is for mannose glycoconjugates in alveolar macrophages (1.23- 1.30 min- l ) . While this value compares favorably with the theoretical upper limit, most receptors display rate constants significantly below this limit. This is especially apparent for the different class I receptors which are internalized at widely divergent rates. It is somewhat difficult to compare all of the values of receptor internalization rates since they were obtained for a number of different cell types under different experimental conditions. Nevertheless, the observed rates of internalization are certainly compatible with the hypothesis that there is an actual upper limit of receptor internalization which is comparable to the rate at which a coated pit itself is internalized. Even though some receptors seem to be internalized at rates approaching that

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

TABLE I RATECONSTANTS OF INTERNALIZATION FOR LICIANDS IN

389 CULTURED CELLSN

ke

(min - 1 )

0.04 0.16

Ligandh

0.16-0.25

hCG EGF EGF

0.29 0.05-0. I7 0.07 0.07 0.11 0.15 0.24

EGF EGF Insulin Insulin Insulin Chemotactic peptide Chemotactic peptide

0.17 0.11 1.23-1.30 0.20 0.25 0.47 0.91 0.17 0.08-0.41

LDL a-Macroglobulin-trypsin Mannose glycoconjugates Asialoorosomucoid Asialoorosomucoid Asialoorosomucoid Transfernin Transfernin Vitellogenin

Cell type

Class 1 receptors Leydig tumor cells Leydig tumor cells Human fibroblasts KB cells A431 cells

Hepatoma cells Adipocytes Hepatocytes Leukocytes Neutrophils

Class 2 receptors Human fibroblasts Alveolar macrophage Alveolar macrophage Hepatocytes Hepatocytes Hepatoma cells A431 cells Hepatoma cells Xenopus oocytes

References

Lloyd and Ascoli (1983) Lloyd and Ascoli (1983) Wiley and Cunningham ( 1982a) Wiley (unpublished) Wiley (unpublished) Ciechanover et a/. (1983) Olefsky and Kao (1982) Car0 et al. (1982) Zigmond et al. ( 1982) Sklar ef a/. (1984) Goldstein et al. (1981) Kaplan and Keogh (198 I ) Hoppe and Lee (1983) Bridges et a/. ( 1982) Weigel and Oka (1982) Ciechanover et a/. ( I 983) Wiley (unpublished) Ciechanover et a / . (1983) L. Opresko (personal com. munication)

The values of k, were either directly taken from the indicated sources or calculated from the published data using the method of Wiley and Cunningham (1982). All values of k, were for cells at 37°C except for Xenopus oocytes which were maintained at 22°C. hCG, Human choriogonadotropin; EGF, epidermal growth factor; LDL, low-density lipoprotein.

of the coated pit, this is clearly not the case for all receptors. This can be due either to an exclusion of the receptor from the coated pit itself (Bretscher et al., 1980)or to other factors such as low lateral diffusion rates of the receptors in the plane of the membrane (Goldstein et al., 1981). For example, if the length of time necessary for a receptor to “find” a coated pit by a random walk in the plane of the membrane is long compared to the lifetime of a coated pit, then the receptor would be captured and internalized at a lower rate than the coated pit itself. Thus, by regulating the lateral mobility of a membrane protein or receptor, a cell could potentially regulate the rate of their internalization. Since modulation in lateral mobility of surface proteins has been reported (Gall and Edelman, 1981), this type of regulatory mechanism could be operant. The theoretical relationship between receptor internalization rates and their

390

H. STEVEN WlLEY

lateral mobility in the plasma membrane has been explored in the elegant study of Goldstein and his collaborators (198 1). They derived the relationship between the diffusion-limited forward rate constant of receptor “clustering” in a coated pit and the mean lifetime of the coated pit itself. If the lifetime of a coated pit structure is infinite, then one can calculate the mean capture time (t,) of a receptor if one knows the diffusion coefficient of the receptor and the size and density of coated pits at the cell surface (Goldstein et al., 1981; Berg and Purcell, 1977): t, = (b2/2D)[ln(b/s) -

$1

(21)

where b is the effective radius of the area of the cell surface containing a single coated pit, s is the radius of the coated pit, and D is the diffusion coefficient of the receptor in the membrane. The diffusion-limited first-order forward rate constant of receptor capture by each pit (in units of pm2 pit-’ sec-’) would then be

K d + = lIPt,

(22)

where P is the density of coated pits at the cell surface (Goldstein et af., 1981). Since the rate of internalization and the lateral diffusion coefficient of the occupied EGF receptor have been measured at 37°C (Wiley and Cunningham, 1982a; Hillman and Schlessinger, 1982), one can calculate that the mean capture time of this receptor is about 17 seconds. This is significantly shorter than the limiting lifetime of the coated pit of 53 seconds as calculated above or the mean lifetime of 350 seconds of the cell surface EGF receptor (Wiley and Cunningham, 1982a). Thus, the lateral diffusion of the EGF receptor in the plasma membrane is apparently not rate limiting in its internatization. However, other receptors and membrane proteins have diffusion coefficients lower than that observed for the EGF receptor (Gall and Edelman, 1981; Barak and Webb, 1982). In these cases diffusion could potentially become rate limiting. The point at which diffusion theoretically becomes limiting for internalization of surface proteins depends upon the set of assumptions used for modeling the dynamics of the coated pit itself. If the coated pit disappears and subsequently reappears at a random spot on the cell surface, then the capture probability of a slowly diffusing protein will become greater since there is a chance that a new coated pit will “appear” next to the protein. To illustrate this principle, Fig. 2 shows the theoretical relationship between the mean time to capture of a randomly distributed cell surface protein at a coated pit and the lateral diffusion coefficient of the protein (Goldstein, Griego, and Wofsy , personal communication). Three different models of coated pit dynamics are presented: (1) pits with infinite lifetimes, (2) pits that disappear and reappear at the same spot, and (3) pits that disappear and reappear at random spots on the cell surface. In all cases, it is assumed that the number of coated pits at the cell surface remains constant.

1’

391

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

10.10

10-11

I0.12

10-13

10.14

10.15

D (cm*/sec)

FIG. 2. The theoretical relationship between the mean capture time of a receptor at a coated pit (ordinate) and the lateral diffusion coefficient of the receptor in the membrane (abscissa). The mean capture time is taken as the time necessary for a receptor diffusing by Brownian motion to encounter a coated pit. Three different models of coated pit dynamics are presented. Curve 1 was derived assuming that the pits are stationary at the cell surface and infinitely long lived. Curve 2 assumes that the coated pits disappear and reappear in the same place. Curve 3 assumes that the pits disappear and then reappear at a random location. The parameters used in the calculations are those that characterize human fibroblasts: the mean lifetime of a coated pit is 4 minutes; the mean recycling time of a coated pit is 4 minutes; and the surface density of coated pits at 37°C is 0.3/pm2. This figure is courtesy of B. Goldstein, R. Griego, and C. Wofsy (unpublished results).

Consequently, in the second and third models the total number of coated pits per cell is greater than in the static first model. The mean time to capture of a membrane protein in this context is the average time that the protein will encounter the coated pit by random diffusion. Let us take the mean lifetime of a coated pit to be approximately 250 second which is the estimated value for human fibroblasts (Goldstein eta!. , 1981).Then the relationship in Fig. 2 indicates that, for those receptors or membrane proteins that have lateral diffusion coefficients of l o - ” cmz sec- I or less, diffusion becomes rate limiting in internalization even though they may be effectively captured by coated pits. However, it is not known if this type of mechanism actually operates in those cases where the internalization of receptors is slow [such as the hCG receptor (Lloyd and Ascoli, 198311 since data regarding the lateral diffusion of receptors are very limited. Nevertheless, it is important to consider this possibility rather than simply assuming that selective proteins are “excluded” from the coated pit itself.

392

H. STEVEN WlLEY

C. Association of Receptors with Coated Pits The available information on receptors displaying very high internalization rates indicates that this is mediated through a preferential association of the receptors with coated pit structures (Brown et al., 1983; Anderson and Kaplan, 1983). What are the mechanisms by which receptors are preferentially localized to coated pit structures? Perhaps the most information can be derived from a comparison of the mechanisms of internalization of the class 1 EGF receptor and the class 2 LDL receptor in human fibroblasts. The EGF receptor undergoes endocytic down-regulation upon occupancy (Wiley and Cunningham, 198 1) while the LDL receptor is regulated by synthetic down-regulation (Brown and Goldstein, 1975). Morphological studies of the surface distribution of LDL receptors labeled with a ferritin-LDL conjugate revealed that between 70 and 95% of the receptors are localized in coated pits (Anderson et al., 1977a). Since this distribution was observed at both low temperatures, at 37°C and in prefixed cells, it was concluded that the observed distribution reflected the steady-state surface distribution of the receptor in both the presence and absence of the ligand. Thus, the observed rate of ligand internalization presumably reflects the normal internalization rate of the receptor itself which is high because of a constitutive preferential interaction with coated pit regions of the cell surface. On the other hand, the normal surface distribution of the EGF receptor is more or less random (Gorden et al., 1978; McKanna et al., 1979) and the basal internalization rate of the unoccupied receptors is relatively slow (Wiley and Cunningham, 1981; Knauer et al., 1984). However, when the EGF receptor is occupied with its ligand, it is internalized at a rate in human fibroblasts that is equal to the LDL receptor. Morphological studies indicate that this endocytic down-regulation is accompanied by a change from a random surface distribution to one preferentially associated with coated pits (McKanna et al., 1979). Thus, it would appear that the class 2 LDL receptor constitutively associated with coated pits while the class 1 EGF receptor inducibly associates with coated pits. Other class 1 receptors have also been demonstrated to undergo an occupancy-induced redistribution at the cell surface which is probably related to endocytic downregulation (Carpentier et al., 198 I). Using the terminology introduced in Section IV,C, we see that for class 2 receptors such as that for LDL, k, = k,, whereas for class 1 receptors, such as those for EGF, k, > k,. The molecular mechanism of endocytic down-regulation or an occupancyinduced association with a coated pit probably involves a conformational change in receptor structure. When EGF binds to its receptor at O’C, there is no observed preferential association with coated pit structures (McKanna et al., 1979) even though the occupied EGF receptor is freely mobile in the membrane at that temperature (Hillman and Schlessinger, 1982). However, a brief warming to 37°C is sufficient to “induce” clustering (McKanna et al., 1979; Willingham

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

393

and Pastan, 1982). A similar requirement for warming has been observed in order for EGF occupancy to stimulate the endogenous receptor protein kinase activity in solubilized receptors (Cohen et al., 1980). Importantly, the EGF receptor seems to undergo a change in tertiary conformation after occupancy as revealed by alterations in proteolytic susceptibility (Linsley and Fox, 1980). Thus, an increase in temperature subsequent to binding probably facilitates an allosteric conformation change in the EGF receptor. This allosteric change in the receptor would then result in a preferential association with coated pit regions of the cell surface. The hypothesis that receptor structure dictates the ability to associate with coated pits has received support from investigations on “internalization defective” fibroblasts (Anderson et al., 1977b). These cells were derived from a patient suffering from familial hypercholesterolemia. The LDL receptors on these fibroblasts were able to bind the ligand in a normal fashion but internalized it at a very low rate. Morphological studies on the surface distribution of these receptors revealed that they were essentially randomly distributed with no preferential association with coated pits. This lack of association is not due to an altered lateral mobility of the receptor since it is at least as rapid as the normal LDL receptor (Barak and Webb, 1982). In addition, the internalization of other receptors (such as the EGF receptor) by these cells is normal (Carpentier et a l ., 1982). It has been postulated that the genetic defect in these cells is in the LDL receptor itself that results in an altered receptor structure (Anderson et al., 1977b). The altered receptor would thus be unable to associate with coated pits with a consequent lowered internalization rate. These studies also elegantly illustrate the connection between receptor distribution in the plasma membrane and internalization rates. Postulating that tertiary structure of receptors is involved in their preferential association with coated pits does not provide a mechanism for the association. Indeed virtually nothing is known about the physical principles or molecular mechanisms involved. However, in spite of the paucity of information regarding the events occurring during most receptor-coated pit interactions, it seems clear that some type of preferential association occurs with both the EGF and LDL receptor. A preferential association with coated pits could occur either through binding of the receptor to some structural element of the coated pit or could be due to a phase partitioning of the receptor into a coated pit-specific lipid domain. Either of these hypotheses is consistent with current experimental evidence. In freeze-fracture studies of the coated pits on human fibroblasts, it was observed that they contained membranous particles that were significantly larger than those of the membrane as a whole (Orci et al., 1978). These particles could represent either unique ‘‘binding proteins” in coated pits or simply prelocalized receptors. Anderson et al. (1982) found that treatment with a nonionic detergent does not remove all prelocalized LDL receptors from coated pits as visualized by

394

H. STEVEN WILEY

anti-receptor antibodies. The clathrin coat could conceivably provide a structure to which receptors could attach. It has also been observed that coated pits have less cholesterol than the plasma membrane as a whole, which argues for a unique lipid composition of these structures (Montesano et al., 1979). A unique lipid composition or structure could also make these regions more resistant to detergent extraction. Since the presence of specialized lipid domains in the cell surface has been strongly indicated by a number of studies (Karnovsky et al., 1982), it seems plausible that coated pits could represent one of these domains. If there are structural elements in coated pits that function to bind receptors, then one would expect that most receptors possess a common site for binding these elements. The evidence for this supposition is that most if not all receptors seem to be internalized in the same coated pit (Brown et al., 1983). A specific binding site for each particular receptor would pose a formidable steric and complexity problem for such an apparent common internalization point. Further evidence against receptor-specific binding sites in the coated pits comes from the studies of Baumann et al. (1980) in which the asialoglycoprotein receptor from hepatocytes was transferred to mouse fibroblasts. Even though the recipient cells do not normally have the asialoglycoprotein receptor, when transplanted it was capable of multiple rounds of rapid internalization. It is difficult to envision how this could occur if receptor-specific binding sites are present in coated pits. Instead, it seems more likely that some general property of the coated pit facilitates the association of receptors of the proper structure and/or conformation. There is evidence in at least one receptor system that ligand binding results in an association of the receptor with cytoskeletal elements. Jesaitis et al. (1984) found that the binding of N-formyl chemotactic peptide to human granulocytes was rapidly followed by the association of the ligand-receptor complex to cytoskeletal elements. This cytoskeletal linkage was transient and preceded the internalization of the ligand. They postulate that this cytoskeletal association of the receptor may be involved either in signal transduction or ligand-receptor internalization. The evidence that the cytoskeletal association is involved in internalization is very suggestive. Both the internalization and the cytoskeletal association are stimulated by receptor occupancy. The linkage is also blocked by low temperatures as is ligand-induced receptor clustering in other systems (Haigler et a l . , 1979a; McKanna er al., 1979). Importantly, the cytoskeletal linkage was transient with a half-life ranging from 30 seconds to 4 minutes (Jesaitis et a l . , 1984). Since the half-life of coated pits seems to be significantly longer than that value (Goldstein et al., 1981), the transient nature of the linkage would neatly explain why occupied receptors are not exclusively localized to coated pits. Although occupancy-induced association of receptors with structural elements of coated pits is an appealing hypothesis, there are several pieces of evidence that are inconsistent with its involvement in internalization. When Jesaitis et al. ( 1984) isolated the receptor-cytoskeletal complex, they found a relative enrichment of F-actin, but no associated clathrin. Schecter and Bothwell (198 1) have

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

395

described a similar cytoskeletal association of a subclass of nerve growth factor receptors in PC12 cells. However, they found no ligand-induced conversion from one form to another, although nerve growth factor can induce the internalization of its receptor. The association of both the chemotactic peptide and the nerve growth factor receptors with cytoskeletal elements was associated with an increase in receptor affinity. However, the kinetics of binding, clustering, and internalization of the EGF receptor show no evidence of alterations in receptor affinity (Wiley and Cunningham, 1981; Knauer et af., 1984). Thus, it is possible that the association of receptors with cytoskeletal elements may be involved in signal transduction and not internalization. Postulating that the specific lipid composition of coated pits favors the preferential association of receptors would seem to offer a simple explanation of their selectivity. It has been hypothesized that a specialized lipid composition of the coated pit could facilitate the invagination process required for vesicle formation during endocytosis (Anderson and Kaplan, 1983). However, a unique lipid composition or domain in the coated pit could also effectively partition specific membrane proteins and receptors into the region. Membrane proteins are necessarily surrounded by an annulus of lipid that directly interacts with the protein itself (Brotherus et af., 1981). It has been demonstrated that some proteins exhibit preferential binding of classes of phospholipid, apparently due to ionic interactions between the charged head groups of the phospholipids and charged amino acid residues of the membrane proteins (Knowles et af., 1981). If the coated pit possessed a lipid composition that favored increased hydrogen bonding or ionic interactions with receptor protein, then the average residency time of the receptor in the coated pits would increase. This would result in an increased steady-state association of the receptor with the internalization site. A “phase partition” mechanism for coated pit association would be dependent on the conformational state of the protein since this would dictate which amino acid residues of the protein are in direct contact with the lipid annulus. There are several other features of this type of mechanism that can be predicted. Lateral diffusion of the receptor could potentially be limiting for internalization since a number of “hits” or partitioning events would be required at the coated pit prior to its internalization in order to reach a steady-state receptor partitioning (Weaver, 1983). The lipid in the coated pit would have to be effectively immobilized or localized so that the rate of lipid exchange of the region is lower than the rate of receptor diffusion into the region (Brotherus et al., 1981). The major difficulty in any phase partition model of receptor clustering is that the partition coefficients needed to concentrate receptors in a small area such as a coated pit would have to be extremely high (Sklar et al., 1979; Sklar, 1984). Such a selectivity of membrane proteins for lipids has never been demonstrated. However, coated pits have a geometry that may facilitate any phase partitioning. The outer edge of a coated pit is necessarily convex whereas the inner surface is concave. Since the local membrane free energy is dependent upon curvature,

396

H. STEVEN WILEY

lipid composition, and conformation of the intramembranous particles, the precise geometry of a coated pit could facilitate the clustering of receptors if their presence resulted in a lowering of the local free energy (Markin, 1981). It has already been postulated that high membrane curvatures (such as those found at the tips of microvilli) are responsible for localization of specific membrane proteins at those structures (Gordon and Marquardt, 1975; de Petris, 1978). This type of partitioning mechanism of receptor localization over coated pits is speculative but is consistent with current experimental evidence. One of the major difficulties in determining which mechanism of receptorcoated pit association operates is that the process has never been adequately described. For example, it is not known whether some receptors or membrane proteins demonstrating low internalization rates are actually excluded from the coated pits, are immobilized at distal areas, have low lateral diffusion coefficients, or have no preferential association with internalization sites. On the other hand, it is not known whether receptors that display high internalization rates are irreversibly captured by coated pits or simply have a longer average residence time in that area of the cell surface. Most of the data describing the association of receptors with coated pits come from the LDL receptor system. It has been found that most, but not all, of the LDL receptors are associated with coated pits at steady state. If receptors are essentially captured by coated pits and diffusion is not rate limiting in internalization, then one would expect that virtually all of the receptors would be in the coated pits at steady state (Weaver, 1983). When the EGF-induced receptor redistribution is examined in A43 1 cells, the number of receptors in the coated pits increases approximately 50-fold after occupancy (Haigler, 1983). However, the great majority of the occupied EGF receptors were still observed outside of the coated pits. A similar situation has been observed in human fibroblasts (Gorden et al., 1978). The A431 cells have been used extensively to examine the interaction of EGF with its receptor since they possess enormous numbers of EGF receptors (approximately 3 X lo6 per cell) (Haigler et af., 1978). However, the rate at which these cells actually internalize either the occupied or unoccupied EGF receptor compared to ‘‘normal’’ cells has never been documented. Since it has been observed by Anderson et al. (1981) that A431 cells are internalization defective with respect to the LDL receptor, this is far from a trivial concern. What is needed is a more detailed knowledge of what altered receptor association with coated pits actually involves. A detailed comparison between normal cells and “internalization defective” cells is potentially a valuable approach.

VI.

RECEPTOR BIOSYNTHESIS AND RECYCLING

As discussed in Section IV,C, the total number of cell surface receptors at steady state is a function of both the rate of receptor internalization and the rate of

9 . RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

397

receptor appearance. Although the absolute rates of receptor internalization are inevitably coupled to the number of receptors at the cell surface, much less is known regarding the source of receptors inserted in the plasma membrane. The newly inserted receptors could represent newly synthesized ones or those that had previously been internalized or a combination of the two types. The concept of receptor “recycling” or reutilization is an area of intense current study since the mechanisms involved in the reinsertion of previously internalized membrane proteins should provide insight into the general mechanisms of membrane dynamics in living cells. Unfortunately, investigations into the source of newly appearing cell surface receptors have been hampered by a lack of suitable techniques for quantitating unoccupied receptors. As discussed in Section 111, the great advantage of utilizing receptors as model membrane proteins is that tagged ligands can be readily employed as markers of occupied receptors. Unfortunately, these advantages do not extend to the quantitation of receptors in transit from an intracellular compartment. Ligand binding has been utilized in assays for measuring receptor content in membranes derived from fractionated cells (Krupp and Lane, 1982; Marshall et al. , 1981; Krupp et al., 19821, but these assays are rarely quantitative in a strict sense. This is primarily due to the necessity of solubilizing the membranes prior to the assays so that receptors initially facing into the lumen of vesicles are accessible to ligand binding. The solubilization step can result in receptor inactivation (Cohen et al., 1982) or large decreases in receptor affinity (Carpenter, 1979) that can complicate the interpretation of the data. There have been recent methodological advances in analyzing unoccupied and/or intracellular receptors that involve monospecific or monoclonal antibodies directed against receptors (Harrison et al., 1979; Haigler and Carpenter, 1980; Anderson et al., 1982; Stoscheck and Carpenter, 1984). The further development of these techniques promises to alleviate many of the problems encountered in receptor biosynthesis-recycling studies. However, most studies to date have relied on indirect kinetic analyses of receptor dynamics at the cell surface to imply the source and pathway of intracellular receptor transfer. While these data can only suggest which events are actually occurring, they do provide a foundation for constructing models of receptor dynamics that should be amenable to direct experimental testing in the near future.

A. Evidence for Receptor Recycling The strongest evidence for receptor recycling comes from studies in which ligand internalization was quantitated in the presence of inhibitors of protein synthesis. For example, Xenopus oocytes were observed to internalize the yolk precursor protein, vitellogenin, at high rates for up to 24 hours in culture in the presence of concentrations of cycloheximide sufficient to completely inhibit

398

H. STEVEN WILEY

protein synthesis (Wallace and Ho, 1972). In studies on the LDL receptor (Brown et al., 1983) and the a,-macroglobulin receptor (Kaplan, 1980), it was found that inhibiting protein synthesis does not lead to a corresponding decrease in ligand internalization rates. Since the number of ligand molecules internalized greatly exceeded the number of receptors present initially at the cell surface or the number which were estimated to reside intracellularly, it was concluded that each receptor molecule was capable of mediating multiple rounds of ligand internalization. Since it now seems extremely likely that ligand internalization occurs as a result of the internalization of ligand-receptor complexes (Brown et al., 1983), the conclusion that some receptors recycle seems inescapable. Besides the aforementioned studies on the vitellogenin, LDL, and a2-macroglobulin receptor systems, evidence has accumulated on the recycling of receptors for asialoglycoproteins (Steer and Ashwell, 1980), mannose-glycoconjugates (Stahl et al., 1980), and lysosomal enzymes (Gonzalez-Noriega et al., 1980). It is noteworthy that all of these receptors can be considered class 2 receptors whose function is to internalize ligands. The evidence for recycling of the class 1 hormone receptors is a little less certain. Krupp and Lane (1982) utilized the heavy isotope density-shift technique to quantitate the rate of synthesis of the insulin receptor in chick hepatocytes. They found that the rate of receptor-mediated insulin internalization was 50 times the rate of receptor synthesis. Thus, they concluded that the insulin receptor was recycled. However, their conclusion was dependent upon the accuracy of the assay for intracellular insulin receptors and an unaltered protein biosynthetic rate in the presence of heavy amino acids which have been found to be toxic for some cultured cells (Pollack et al., 1981). Since it is not clear that these requirements were fully met, it is difficult to state that those experiments conclusively demonstrate recycling of the insulin receptor. More recently, Knutson et al. (1983) have presented persuasive evidence that insulin receptors recycle in 3T3-C2 fibroblasts. The addition of insulin to those cells did not alter the synthetic rate of the receptor but did increase its degradation rate after a lag of about 2 hours. Importantly, they demonstrated that the rate of receptor internalization was far greater than the increased rate of degradation. In addition, when insulin was removed prior to the onset of increased degradation, the cells fully recovered their surface complement of receptors at a rate greater than that of receptor synthesis. These data are fully consistent with a recycling model for the insulin receptor. Other investigators have also produced evidence that the insulin receptor recycles in adipocytes since cycloheximide does not block the reinsertion of new receptors into the cell surface (Marshall et al., 1981; Deutsch et al., 1982). Even though these investigations are not conclusive because of the above mentioned difficulties in quantitating surface and intracellular receptor number, the evidence is certainly supportive of insulin receptor reutilization.

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

399

Zigmond et al. (1982) have found that the kinetics of the chemotactic peptide receptor internalization and reappearance is consistent with a recycling model also. However, there is evidence that other class 1 receptors, such as those for acetylcholine (Gardner and Fambrough, 1979), do not recycle. In the cases of the EGF receptor and the platelet-derived growth factor receptor, inhibiting protein synthesis prevents the appearance of new receptors (King er al., 1980; Heldin et al., 1982). Treatment of Leydig tumor cells with EGF dramatically decreases the number of receptors for hCG but only after a lag of 6 hours (Lloyd and Ascoli, 1983). Removal of the EGF results in a restoration of the number of hCG receptors but, again, only after a lag of 6 hours. This and other evidence on the internalization rate of the hCG receptor (Lloyd and Ascoli, 1983) is consistent with the hypothesis that EGF regulates the synthetic rate of the hCG receptor in that cell type, but it would be difficult to explain if the hCG receptor recycled to any appreciable extent. When '251-labeled EGF was cross-linked to its surface receptor prior to internalization, proteolytic processing of the internalized complex was observed (Das and Fox, 1978). again inconsistent with receptor recycling. More recently, Stoscheck and Carpenter (1984) have directly measured the degradation rate of the EGF receptor. This was accomplished by labeling the receptor in vivo with radioactive amino acids, transferring the cells to nonradioactive medium, and then following the degradation of the labeled receptor by immunoprecipitation. They found that the half-life of the receptor was about 10 hours and was decreased to 1.2 hours in the presence of EGF. The basal degradation rate is somewhat less than the rate at which the unoccupied receptor has been estimated to be internalized in the absence of EGF (-4 hours; Knauer et al., 1984). The rate at which the receptor is degraded in the presence of EGF was again about one-half the rate at which internalized EGF has been reported to be degraded (Carpenter et al., 1975; Wiley and Cunningham, 1981; Knauer et al., 1984). These differences may be due to the culture conditions or exact strain of human fibroblasts used in the different studies. Nevertheless, it seems clear that if the EGF receptor does recycle in this cell type, it does so to a very small extent. Evidence both for and against receptor recycling is scarce for other class 1 receptors, apparently because of the relatively slow rate at which these receptors appear at the cell surface in comparison to class 2 receptors. Most current studies on receptor recycling utilize inhibitors such as amines and cycloheximide or incubations at lowered temperatures. The slow rate of class 1 receptor appearance necessitates lengthy incubation times under conditions that are less than optimum for cell viability. This contributes to the difficulties in interpreting the experimental data. The issue of the extent of recycling of class I receptors will probably remain unresolved until other experimental approaches are developed.

400

H. STEVEN WILEY

B. Models of Receptor Recycling Over the last several years there has been a number of different models proposed for receptor recycling. Virtually all of the kinetic models are empirical in nature since they attempt to describe the dynamics of a single system under restricted conditions. There have been a number of qualitative models described, but these generally suffer from system specificity and are difficult to test quantitatively. As mentioned in Section VI,A, the basic problem in testing any of these models is the inherent difficulty in simultaneously following the fate of occupied and unoccupied receptors both at the cell surface and in intracellular compartments. Nevertheless, it is useful to summarize some of the models that have been proposed to date. Since space will not permit a description of all of the models suggested for all of the receptor systems investigated, I will describe three of the models proposed for the recycling of the asialoglycoprotein receptor, a system that is currently under intense investigation by a number of different laboratories. 1. The model of Schwartz et al. (1982) proposes that unoccupied asialoglycoprotein receptors do not undergo significant turnover and are present almost exclusively at the cell surface. The occupancy of the receptor induces its rapid internalization followed by ligand discharge prior to fusion with the lysosomes and a subsequent rapid receptor reinsertion at the cell surface. 2. The model of Weigel and Oka (1983b) proposes that there are significant internal pools of the asialoglycoprotein receptor and that at least part of this pool is continuously exchanging with the cell surface pool. The occupancy of the receptor accelerates its internalization. The internalized, occupied receptor transfers its ligand to an intracellular receptor and then reinserts into the cell surface. The recycling of the occupied receptor is perceived to follow a pathway different from that of the unoccupied receptor. 3. The model of Bridges et al. (1982) proposes that the occupied asialoglycoprotein receptor undergoes rapid internalization but that an unoccupied receptor appears for every receptor that is internalized. This implies that turnover of the receptor is constitutive and is not altered by occupancy. The ligandreceptor complex dissociates prior to lysosomal delivery and the receptor then enters the intracellular receptor pool from which it is eventually returned to the cell surface. Thus, there is no explicit difference between the recycling of occupied and unoccupied receptors.

The differences between all of these models can be classified as the perceived difference between the behavior of the occupied versus the unoccupied receptor. If receptors are to be useful models for investigating the dynamics of membrane proteins, then it is crucial to define those aspects of receptor function that are

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

401

specific to that class of protein and those that are general properties of membrane proteins. For example, there is now a wealth of evidence that many membrane proteins exist both at the cell surface and in an intracellular pool and that there is a continuous exchange between the two pools (e.g., Thilo and Vogel, 1980; Fishman and Cook, 1982). Investigating the recycling of receptors could potentially provide information on the mechanisms by which this continuous exchange is accomplished by cells. However, models 1 and 2 described above postulate that the recycling of the occupied asialoglycoprotein receptor occurs through specialized pathways; and thus, information derived from this system may only apply to receptors. On the other hand, model 3 above does not propose any specialized pathway for the occupied receptor; and, if correct, the information derived from the asialoglycoprotein receptor system could be generally applicable to other membrane proteins. Is it necessary to propose receptor-specific intracellular pathways in recycling? This is a very difficult question to answer at present. Extremely simple kinetic models of asialoglycoprotein receptor turnover do not seem to indicate a recycling scheme where the internalized occupied receptor enters a common cell surface precursor pool (Schwartz et al., 1982; Weigel and Oka, 1983b). However, these simple models are based on ligand binding studies which do not always provide accurate ligand-receptor stoichiometries. For example, asialoglycoproteins may bind to multiple receptors as a multivalent ligand (Connolly er al., 1982). Since the stoichiometry of the binding may depend upon the specific ligand used, it is perhaps not surprising that different laboratories find different numbers of receptors in the same cell type (Weigel and Oka, 1983a). Importantly, the affinity and stoichiometry of multivalent binding depend upon the distribution and lateral mobility of the cell surface receptors and, thus, can be strongly temperature dependent (Dower et al., 1981). Since temperature shifts also seem to alter the absolute number of surface receptors, comparing the “number” of receptors at 37 and 0°C is difficult (Weigel and Oka, 1983a). Thus, it is not at all clear whether the addition of asialoglycoproteins to cells reduces or has no effect on the number of surface receptors. As demonstrated in Section IV ,C, establishing this point is crucial when proposing that receptor occupancy facilitates receptor internalization. If methods and techniques are developed for accurately quantitating the number of receptors (both occupied and unoccupied) in different cellular compartments, then there is still the necessity of developing suitable kinetic models of receptor recycling so that postulated pathways can be quantitatively tested. An important step in this direction has been taken by the work of Goldstein and Wofsy (198 I ) . These investigators developed a general model for analyzing simple recycling systems that include receptor synthesis and degradation. Using a slight modification of their nomenclature, the change in receptor number at the cell surface with respect to time can be expressed as

402

H. STEVEN WILEY

where the term t indicates at that moment of time, S, is the zero-order biosynthetic rate of the receptor, and kt is the rate constant of receptor recycling. S(t) is the rate of receptor recruitment from any intracellular pool that has kinetic properties different from those of the pool of recycling receptors. This could include recruitment from the lysosomal or other secondary intracellular compartments. In their original model, Goldstein and Wofsy (1981) set S(t) as being proportional to the intracellular pool of recycling receptors prior to an experimental perturbation of the system since it was mathematically convenient to model the transition from one state to another as a shift from one type of kinetic pool to another. The function k,(t - x ) is the probability per unit time that a receptor that is internalized at time x will be recycled to the surface at time t. Thus, k,(t) dt is the probability that a receptor that was internalized at time zero will recycle to the surface at time t . The probability that a receptor that is internalized will eventually recycle to the surface is k,(t) dt

=

1 -

E

(24)

where E is the probability that a receptor will be lost in the process, i.e., irreversibly degraded in some way. As recycling approaches 100% efficiency, then E approaches 0. From these equations one can show that the average time it takes an undegraded receptor to be recycled to the cell surface is equal to rm

T,

=

J

tk,(t)dt/(l - E) 0

If this general model is applicable, then at steady state when recycling greatly exceeds receptor biosynthesis and degradation, the ratio of receptors at the cell surface to the total cellular content of receptors is equal to

The very favorable feature of this general model is that it is based on what is known about the recycling of general membrane proteins and thus has a firm experimental foundation. Since it is a general model, it can be used to analyze any number of different receptor systems. Importantly, it makes very specific predictions regarding the kinetic behavior of receptors that behave in this fashion, and all of the kinetic constants are potentially measurable. But perhaps the greatest advantage of this model is that it explicitly incorporates terms for recep-

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

403

tor biosynthesis, degradation, and recycling rates. If appropriate techniques are developed for quantitating the absolute rates of these processes, then it should be possible to determine their relative contribution to the steady-state receptor distribution within cells. This would greatly facilitate the use of receptors as model membrane proteins. Of course, the equations presented are the simplest case of recycling. Goldstein and Wofsy ( I 98 I ) presented other more complex multicompartment models that are also potentially testable. If one derived parallel equations for the recycling of occupied versus unoccupied receptors, such as substituting k, for k, in Eq. (23), then the model would be applicable to receptors demonstrating endocytic down-regulation. Finally, if one utilized equations relating ligand concentration to receptor occupancy, then the model would become a very powerful tool for analyzing the kinetics of ligand binding and the subsequent receptor dynamics.

C. Receptor Biosynthesis and Degradation The relationship between the rate of receptor biosynthesis-degradation and their steady-state number at the cell surface is perhaps the least understood aspect of receptor dynamics. It is clear that in a number of systems, altering the rate of receptor synthesis or degradation alters the absolute number of cell surface receptors present at steady state (Gardner and Fambrough, 1979; Reed et al., 1981; Knutson et al., 1982). What is less clear is the quantitative relationship between these parameters. An increase in receptor synthetic rate could expand a pool of intracellular receptors from which those appearing at the plasma membrane are recruited. Newly synthesized receptors could also be directly transported to the cell surface and then join an intracellular receptor pool only after an initial round of internalization. An increase in receptor degradation rate could be due either to an increase in the relative rate of receptor removal from an intracellular pool or simply to expansion of the internal receptor pool by enhanced internalization. Understanding the regulation of cellular composition will require methods for measuring the rate of flow of receptors between different compartments. Studies which have focused on receptor synthetic-degradation rates have generally treated the cell as a single unit in which synthesis and degradation are the only routes of entry and exit (Reed et al., 1981; Krupp et al., 1982). While these studies have indicated that there are control mechanisms that serve to regulate the total cellular content of receptors by modulating both receptor synthetic and degradation rates, they reveal little concerning the regulation of the individual cellular compartments such as the plasma membrane. Since the cell surface is the interface at which receptors presumably mediate their functions, the regulation of the composition of this organelle could be more crucial than the regulation of total cellular receptor levels.

404

H. STEVEN WlLEY

Recent studies on the relationship between receptor internalization and degradation rates have provided evidence that receptor-mediated endocytosis may directly regulate total cellular receptor levels. Knutson et al. (1983) have investigated the relationship between the ligand-induced internalization of the insulin receptor and its rate of degradation. They conclude that insulin accelerates the rate of receptor degradation by inducing a steady-state redistribution of the receptors between the cell surface and an intracellular pool. They postulate that the rate of receptor degradation is proportional to the size of the intracellular pool. By inducing endocytic down-regulation of the insulin receptor [increasing k, in Eq. (26)], insulin treatment would lead to an increase in the intracellular pool of receptors that can be degraded. Krupp et al. (1982) have reported that the occupancy of the class 1 EGF receptor in A-431 cells (which also undergo endocytic down-regulation) results in an increased rate of receptor degradation. Stoscheck and Carpenter (1984) have presented similar findings for the EGF receptor in human fibroblasts. However, neither of these studies quantitatively related the increased degradation rate to the size of the intracellular pool of receptors or to the occupancy-enhanced receptor internalization rate. In the case of the class 2 asialoglycoprotein receptor, it has been reported that receptor occupancy and ligand internalization has no effect on the degradation rate of the receptor (Tanabe et al., 1979). This could be due to the lack of endocytic down-regulation of class 2 receptors. However, as discussed in Section VI,B, there is presently no agreement as to whether occupancy of the asialoglycoprotein receptor results in an alteration in the cellular distribution of the receptor. It is, therefore, not presently known whether different receptors have different degradative pathways and regulatory mechanisms or instead whether there is some general intracellular compartment from which receptors destined for degradation are randomly recruited. The resolution of this question will depend upon experiments that relate the biosynthetic and degradation rates of receptors to the absolute rates of their appearance and removal from individual cellular compartments.

Vll. CONCLUDING REMARKS I have attempted to demonstrate in this article the approaches that can be used in analyzing receptors as model systems for examining membrane protein dynamics. It is perhaps inappropriate at this point to state what investigations on receptor dynamics have actually revealed about the dynamics of other membrane proteins since relatively few comparative studies have been done. The great promise of receptor studies is that the binding of specific ligands to specific membrane proteins is a very powerful method for analyzing their behavior in situ. Unfortunately, the complexities of the observed binding reactions necessi-

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

405

tate the development and experimental testing of adequate kinetic models of the process. This is not an impossible task. However, it is necessary to consider each of the steps in the binding, internalization, and possibly recycling of receptors separately and then to reconstruct the entire sequence of events. The increasing use of computers in this process should facilitate the development of integrated models of receptor dynamics. However, there are clear dangers inherent in facile interpretations of ligand binding studies. The cell surface is very complex and this is reflected in the complexities inherent in ligand binding. While the myopia inherent in analyzing ligand binding is a necessary evil at the present time, it is probable that the development of other technologies for quantitating receptor number and distribution will provide new dimensions to current investigations. It is a truism in science that the greater the number of techniques available for analyzing a phenomenon, the greater the probability of converging to an accurate analysis. Thus, the further development and utilization of ligand binding analyses should increase our understanding of the dynamics of the cell surface. REFERENCES Aharonov, A,. Pruss, R. M . , and Henchman, H. R. (1978). Epidermal growth factor: Relationship between receptor regulation and mitogenesis in 3T3 cells. J. Biol. Chern. 253, 3970-3977. Amdur, I . . and Hammes, G . G . (1966). Theories of chemical kinetics. I n “Chemical Kinetics. Principles and Selected Topics,” pp. 59-64. McGraw-Hill, New York. Anderson, R. G . W., and Kaplan, J . (1983). Receptor-mediated endocytosis. Mod. Cell Biol. 1, I 52. Anderson, R. G . W., Brown, M. S . , and Goldstein, J . L. (1977a). Role of the coated endocytic vesicle in the uptake of receptor-bound low density lipoprotein in human fibroblasts. Cell 10, 351-364. Anderson, R. G . W., Goldstein, J . L., and Brown. M. S . (1977b). A mutation that impairs the ability of lipoprotein receptors to localize in coated pits on the cell surface of human fibroblasts. Nature (London) 14, 695-699. Anderson, R. G . W., Vasile, E., Mello, R. J . , Brown, M. S . , and Goldstein, J . L. (1978). Immunocytochemical visualization of coated pits and vesicles in human fibroblasts: Relation to low density lipoprotein receptor distribution. Cell 15, 919-933. Anderson, R. G . W., Brown, M. S . , and Goldstein, J . L. (1981). Inefficient internalization of receptor-bound low density liproprotein in human carcinoma A-43 I cells. J . Cell Biol. 88,441452. Anderson, R. G . W., Brown, M. S . , Biesiegel, U . , and Goldstein, J . L. (1982). Surface distribution and recycling of the low density lipoprotein receptor as visualized with antireceptor antibodies. J . Cell Biol. 93, 523-531. Ascoli, M. (1981). Regulation of gonadotropin receptors and gonadotropin responses in a clonal strain of Leydig tumor cells by epidermal growth factor. J . Biol. Chern. 256, 179-183. Barak, L. S . , and Webb, W. W. (1982). Diffusion of low density lipoprotein-receptor complex on human fibroblasts. J . Cell Biol. 95, 846-852. Baumdnn, H., Hou, E.. and Doyle, D. (1980). Insertion of biologically active membrane proteins from rat liver into the plasma membrane of mouse fibroblasts. J . Biol. Chern. 255, 1000110012.

Berg, H. C., and Purcell, E. M. (1977). Physics of chemoreception. Biophys. J. 20, 193-219.

406

H. STEVEN WILEY

Berlin, C. M., and Schimke, R. T. (1965). Influence of turnover rates on the responses of enzymes to cortisone. Mol. Pharmacol. I, 149-156. Bolton, A. E., and Hunter, W. M. (1973). The labeling of proteins to high specific activities by conjugation to a 12JI-containingacylating agent. Biochem. J . 133, 529-538. Bowen-Pope, D. F., and Ross, R. (1982). Platelet-derived growth factor. 11. Specific binding to cultured cells. J . Biol. Chem. 257, 5161-5171. Bretscher, M. S.,Thomson, J. N., and Pearse, B. M. F. (1980). Coated pits act as molecular filters. Proc. Natl. Acad. Sci. U.S.A. 77, 4156-4159. Bridges, K., Harford, J . , Ashwell, G., and Klausner, R. D. (1982). Fate of receptor and ligand during endocytosis of asialoglycoproteins by isolated hepatocytes. Proc. Natl. Acad. Sci. U.S.A. 79, 350-354. Brotherus, J . R., Griffith, 0. H., Brotherus, M. O., Jost, P. C., Silvius, J. R., and Hokin, L. E. (1981). Lipid-protein multiple binding equilibria in membranes. Biochemistry 20, 5261 -5267. Brown, M. S.,and Goldstein, J. L. (1975). Regulation of the activity of the low density lipoprotein receptor in human fibroblasts. Cell 6, 307-316. Brown, M. S.,Anderson, R. G. W., and Goldstein, J . L. (1983). Recycling receptors: The roundtrip itinerary of migrant membrane proteins. Cell 32, 663-667. Calvo, J . C., Radicella, J. P., and Charreau, E. H. (1983). Measurements of specific radioactivities in labeled hormones by self-displacement analysis. Biochem. J . 212, 259-264. Caro, J. F., Muller, G . , and Glennon, J. A. (1982). Insulin processing by the liver. J . Biol. Chem. 257, 8459-8466. Carpenter G. (1979). Solubilization of the membrane receptor for epidermal growth factor. Life Sci. 24, 1691-1698. Carpenter, G., Lemhach, K. J . , Morrison, M. M., and Cohen, S. (1975). Characterization of the binding of IZ5I-Labeled epidermal growth factor to human fibroblasts. J . Biol. Chem. 250, 4297-4304. Carpentier, I. L., Obberghen, E. V., Gorden, P., and Orci, L. (1981). Surface redistribution of 1251insulin in cultured human lymphocytes. J . Cell Biol. 91, 17-25. Carpentier, J . L., Gorden, P., Anderson, R. G. W., Goldstein, J. L., Brown, M. S., Cohen, S., and Orci, L. (1982). Co-localization of 1251-epidermal growth factor and ferritin-low density lipoprotein in coated pits: A quantitative electron microscopic study in normal and mutant human fibroblasts. J . Cell Biol. 95, 73-77. Chait, A,, Ross, R., Albers, J. J . , and Bierman, E. L. (1980). Platelet-derived growth factor stimulates activity of low density lipoprotein receptors. Proc. Narl. Acad. Sci. U.S.A. 77, 4084-4088. Ciechanover, A., Schwartz, A. L . , and Lodish, H. F. (1983). The asialoglycoprotein receptor internalizes and recycles independently of the transfenin and insulin receptors. Cell 32, 267275. Cohen, S.,Carpenter, G., and King, L., Jr. (1980). Epidermal growth factor (EGF)-receptor protein kinase interactions: Co-purification of receptor and EGF-enhanced phosphorylation activity. J . Biol. Chem. 255, 4834-4842. Cohen, S., Ushiro, H., Stoscheck, C., and Chinkers, M. (1982). A native 170,000epidermal growth factor receptor-kinase complex from shed plasma membrane vesicles. J . Biol. Chem. 257, 1523-1531, Colquhoun, D. (1979). The link between drug binding and response: Theories and observations. In “The Receptors: A Comprehensive Treatise” (R.D. O’Brien, ed.), pp. 93-142. Plenum, New York. Connolly, D. T., Hoppe, C. A., Hobish, M. K., and Lee, Y. C. (1981). Steady state and kinetic analysis of the binding of asialoorosomucoid to the isolated rabbit hepatic lectin. J . Biol. Chem. 256, 12940-12948.

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

407

Connolly, D. T . , Townsend, R. R., Kawaguchi, K., Bell, W. R., and Lee, Y. C. (1982). Binding and endocytosis of cluster glycosides by rabbit hepatocytes. J . B i d . Chem. 257, 939-945. Cuatrecasas, P., and Hollenberg, M. D. (1976). Membrane receptors and hormone action. Adv. Prof. Chem. 30, 251-451. Das, M., and Fox, C. F. (1978). Molecular mechanism of mitogen action: Processing of receptor induced by epidermal factor. Proc. Narl. Acud. Sci. U.S.A. 75, 2644-2648. DeLean, A , , and Rodbard, D. (1979). Kinetics of cooperative binding. In “The Receptors: A Comprehensive Treatise” (R. D. O’Brien, ed.). pp. 143-192. Plenum, New York. DeLisi. C., and Metzger, H. (1976). Some physical chemical aspects of receptor-ligand interactions. Immunol. Commun. 5, 417-436. DeLisi, C., and Wiegel, F. W. (1981). Effect of nonspecific forces and finite receptor number on rate constants of ligand-cell bound-receptor interactions. Proc. Nut!. Acad. Sci. U.S.A. 78, 55695572. DeMeyts, P. (1976). Cooperative properties of hormone receptors in cell membranes. J . Supramol. Struct. 4, 241-258. De Petris, S. (1978). Preferential distribution of surface immunoglobulins on microvilli. Nature (London) 272, 66-68. Deutsch, P. J . , Rosen, 0. M., and Rubin, C. S. (1982). Identification and characterization o f a latent pool of insulin receptors in 3T3-LI adipocytes. J . B i d . Chem. 257, 5350-5358. Dower, S . K . , DeLisi, C., Titus, J. A,, and Segal, D. M. (1981). Mechanism of binding of multivalent immune complexes to Fc receptors. I . Equilibrium binding. Biochemistry 20, 6326-6334. Duggleby. R. G. (1981). A nonlinear regression program for small computers. Anal. Biochem. 110, 9-18. Edidin, M., Zagyansky, Y., and Lardner, T. J. (1976). Measurements of membrane protein lateral diffusion in single cells. Science 191, 466-468. Feldman, H. A. (1972). Mathematical theory of complex ligand-binding systems at equilibrium: Some methods for parameter fitting. Anal. Biochem. 48, 317-338. Feldman, H., Rodbard, D., and Levine, D. (1972). Mathematical theory of cross-reactive radioimmunoassay and ligand-binding systems at equilibrium. Anul. Biochem. 45, 530-556. Finney, D. A , , and Sklar, L. A. (1983). Ligand/receptor internalization: A kinetic, flow cytometric analysis of the internalization of N-formyl peptides by human neutrophils. Cytometry 4, 54-60. Fishman, J. B., and Cook, J. S. (1982). Recycling of surface sialoglycoconjugates in HTC and HeLa cells. J . Biol. Chem. 257, 8122-8129. Gall, W. E., and Edelman, G . M. (1981). Lateral diffusion of surface molecules in animal cells and tissues. Science 213, 903-905. Gardner, J. M., and Fambrough, D. M. (1979). Acetylcholine receptor degradation measured by density labeling: Effects of cholinergic ligands and evidence against recycling. Cell 16, 661 674. Gershon, N. D., Smith, R. M.. and Jarett, L. (1981). Computer assisted analysis of ferritin-insulin receptor sites on adipocytes and the effects of cytochalasin B on groups of insulin receptor sites. J . Membr. Biol. 58, 1.5-160. Goldstein, B., and Wiegel. F. W. (1983). The effect of receptor clustering on diffusion-limited forward rate constants. Biophys. J . 43, 121-125. Goldstein, B., and Wofsy, C. (1981). Analysis of coated pit recycling of human fibroblasts. Cell Biophys. 3, 25 1-277. Goldstein, B., Wofsy, C . , and Bell, G. (1981). Interactions of low density lipoprotein receptors with coated pits on human fibroblasts: Estimate of the forward rate constant and comparison with the diffusion limit. Pror. Nut/. Acad. Sci. U.S.A. 78, 5695-5698. Gonzalez-Noriega, A,, Grubb, I . H., Talkad, V . , and Sly, W. S. (1980). Chloroquine inhibits

408

H. STEVEN WILEY

lysosomal enzyme pinocytosis and enhances lysosomal enzyme secretion by impairing receptor recycling. J. Cell Biol. 85, 839-852. Gorden, P., Carpentier, J.-L., Cohen, S., and Orci, L. (1978). Epidermal growth factor: Morphological demonstration of binding, internalization, and lysosomal association in human fibroblasts. Proc. Natl. Acad. Sci. U.S.A. 75, 5025-5029. Gordon, J. A,, and Marquardt, M. D. (1975). Erythrocyte morphology and clustering of fluorescent anti-A immunoglobulin. Nature (London) 258, 346-347. Haigler, H. T. (1983). Epidermal growth factor: Cellular binding and consequences. In “Growth and Maturation Factors” (G. Guroff, ed.), Vol. I , pp. 117-154. Wiley, New York. Haigler, H. T., and Carpenter, G. (1980). Production and characterization of antibody blocking epidermal growth factor: receptor interactions. Biochim. Biophys. Acta 598, 3 14-325. Haigler, H., Ash, J. F., Singer, S. 1.. and Cohen, S. (1978). Visualization by fluorescence of the binding and internalization of epidermal growth factor in human carcinoma cells A-431. Proc. Natl. Acad. Sci. U.S.A. 75, 3317-3321. Haigler, H. T., McKanna, J. A,, and Cohen, S. (1979a). Direct visualization of the binding and internalization of a ferritin conjugate of epidermal growth factor in human carcinoma cells A-431. J. Cell B i d . 81, 382-395. Haigler, H. T., McKanna, J . A , , and Cohen, S. (1979b). Rapid stimulation of pinocytosis in human carcinoma cells A-431 by epidermal growth factor. J. Cell Biol. 83, 82-90. Haigler, H. T., Maxfield, F. R., and Pastan, I. (1980). Dansylcadaverine inhibits internalization of ‘2sI-epidermal growth factor in BALB 3T3 cells. J . Biol. Chem. 255, 1239-1241. Harrison, L. C., Flier, J . , Itin, A., Kahn, C. R., and Roth, J . (1979). Radioimmunoassay of the insulin receptor: A new probe of receptor structure and function. Science 203, 544-547. Heldin, C. H., Wasteson, A., and Westermark, B. (1982). Interaction of platelet-derived growth factor with its fibroblast receptor. Demonstration of ligand degradation and receptor modulation. J. Biol. Chem. 257, 4216-4221. Hillman, G. M., and Schlessinger, J. (1982). Lateral diffusion of epidermal growth factor complexed to its surface receptors does not account for the thermal sensitivity of patch formation and endocytosis. Biochemistry 21, 1667- 1672. Hollenberg, M. D., and Cuatrecasas, P. (1979). Distinction of receptor from nonreceptor interactions in binding studies. In “The Receptors: A Comprehensive Treatise” (R. D. O’Brian, ed.), Vol. I , pp. 193-214. Plenum, New York. Hoppe, C. A., and Lee, Y. C. (1983). The binding and processing of mannose-bovine serum albumin derivitives by rabbit alveolar macrophages. Effect of the sugar density. J . Biol. Chem. 258, 14193- 14199. Hunter, W. M., and Greenwood, F. C. (1962). Preparation of iodine-131 labeled human growth hormone of high specific activity. Nature (London) 194, 495-496. Jacobson, K., Hous, Y., Derzko, Z., Wojcieszyn, J., and Organisciak, D. (1981). Lipid lateral diffusion in the surface membrane of cells and in multibilayers formed from plasma membrane lipids. Biochemistry 20, 5268-5275. Jesaitis, A. J . , Naemura, J . R., Sklar, L. A , , Cochrane, C. G., and Painter, R. G. (1984). Rapid modulation of N-formyl chemotactic peptide receptors on the surface of human granulocytes: Formation of slowly dissociating ligand-receptor complexes in transient association with cell cytoskeleton. J . Cell B i d . 98, 1378-1387. Kaplan, J. (1980). Evidence for reutilization of surface receptors for u2-macroglobulin-protease complexes in rabbit alveolar macrophages. Cell 19, 197-205. Kaplan, J. (198 1). Polypeptide-binding membrane receptors: Analysis and classification. Science 212, 14-20. Kaplan, J., and Keogh, E. A. (1981). Analysis of the effect of amines on inhibition of receptormediated and fluid-phase pinocytosis in rabbit alveolar macrophages. Cell 15, 925-932.

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

409

Kaplan, J., and Keogh, E. A. (1982). Temperature shifts induce the selective loss of alveolarmacrophage plasma membrane components. J . Cell Biol. 94, 12- 19. Karnovsky, M. J., Kleinfeld, A. M., Hoover, R. L., and Klausner, R. D. (1982). The concept of lipid domains in membranes. J . Cell B i d . 94, 1-6. Khan, M. N., Posner, B. I . , Khan, R. .I., and Bergeron, J . J. M. (1982). Internalization of insulin into rat liver golgi elements: Evidence for vesicle heterogeneity and the path of intracellular processing. J . B i d . Chem. 257, 5969-5976. King, A. C . , Willis, R. A , , and Cuatrecasas, P. (1980). Accumulation of epidermal growth factor within cells does not depend on receptor recycling. Biochem. Biophvs. Res. Commun. 97, 840845. Knduer, D. J., Wiley, H. S . , and Cunningham, D. D. (1984). Relationship between epidermal growth factor receptor occupancy and mitogenic response: Quantitative analysis using a steady state model system. J . Biol. Chem. 259, 5623-5631. Knowles, P. F., Watts, A , , and Marsh, D. (1981). Spin-label studies of head-group specificity in the interaction of phospholipids with yeast cytochrome oxidase. Biochemistry 20, 5888-5894. Knutson, V. P., Ronnett, G . V., and Lane, M. D. (1982). Control of insulin receptor level in 3T3 cells: Effect of insulin-induced down-regulation and dexamethasone-induced up-regulation on rate of receptor inactivation. Proc. Nurl. Acatf. Sci. U.S.A. 79, 2822-2826. Knutson, V. P., Ronnett, G. V., and Lane. M. D. (1983). Rapid, reversible internalization of cell surface insulin receptors: Correlation with insulin-induced down-regulation. J . Biol. Chem. 258, 12139- 12142. Kosmakos, F. C., and Roth, J . (1980). Insulin-induced loss of the insulin receptor in IM-9 lymphocytes. J . Biol. Chem. 255, 9860-9869. Krupp. M. N., and Lane, M. D. (1982). Evidence for different pathways for the degradation of insulin and insulin receptor in the chick liver cell. J . Biol. Chem. 257, 1372-1377. Krupp, M. N., Connolly, D. T., and Lane, M. D. (1982). Synthesis, turnover and down-regulation of epidermal growth factor receptors in human A431 epidermoid carcinoma cells and skin. J . Biol. Chem. 257, 1148- 1149. Lam, C. F. (1981). Systematic and practical procedures for modeling enzyme kinetic mechanisms. In “Techniques for the Analysis and Modeling of Enzyme Kinetic Mechanisms” (D. W. Hill, ed.), pp. 213-262. Wiley, New York. Langmuir, 1. (1918). The adsorption of gasses on plane surfaces of glass. mica and platinum. J . Am. Chenr. Soc. 40, 1361-1374. Linsley, P. S., and Fox, C. F. (1980). Controlled proteolysis of EGF receptors: Evidence for transmembrane distribution of the EGF binding and phosphate acceptor sites. J . Suprumol. Struct. 14, 461-471. Lloyd, C. E., and Ascoli, M. (1983). On the mechanisms involved in the regulation of the cellsurface receptors for human choriogonadotropin and mouse epidermal growth factor. J . Cell Biol. 96, 521-526. Low, D. A., and Cunningham, D. D. (1982). A novel method for measuring cell surface-bound thrombin. J . Biol. Chem. 257, 850-858. McKanna, J . A., Haigler, H. T . . and Cohen, S . (1979). Hormone receptor topology and dynamics: Morphological analysis using femtin-labeled cpidermal growth factor. Proc. Nor/. Acud. Sci. U . S . A . 76, 5689-5693. Markin, V. S. (1981). Lateral organization of membranes and cell shapes. Eiophys. J . 36, 1-19. Marshall, S . , Green, A,, and Olefsky, I. M. (1981). Evidence for recycling of insulin receptors in isolated rat adipocytes. J . Biol. Chem. 256, 11464-1470. Maxfield, F. R . , Willingham, M. C., Haigler, H. T., Dragsten, P.. and Pastan. 1. H. (1981). Binding, surface mobility, internalization, and degradation of rhodamine-labeled a2-macroglobulin. Biochernisrrj 20, 5353-5358.

41 0

H. STEVEN WILEY

Montesano, R., Perrelet, A., Vassalli, P., and Orci, L. (1979). Absence of filipin-sterol complexes from large coated pits on the surface of culture cells. Proc. Nut/. Acad. Sci. U.S.A. 76, 63916395. Olefsky, J . M., and Kao, M. (1982). Surface binding and rates of internalization of 1251-insulinin adipocytes and IM-9 lymphocytes. J. B i d . Chem. 257, 8667-8673. Orci, L., Carpentier, J. L., Perrelet, A., Anderson, R. G. W., Goldstein, J . L., and Brown, M. S. (1978). Occurrence of low density lipoprotein receptors within large pits on the surface of human fibroblasts as demonstrated by freeze-etching. Exp. Cell Res. 113, 1-13. Pastan, I. H., and Willingham, M. C. (1981). Receptor-mediated endocytosis of hormones in cultured cells. Annu. Rev. Physiol. 43, 239-250. Pearse, B. M. F. (1976). Clathrin: A unique protein associated with intracellular transfer of membranes by coated vesicles. Proc. Natl. Acad. Sci. U.S.A. 73, 1255-1259. Pfeiffer, J. R., Oliver, J. M., and Berlin, R. D. (1980). Topographical distribution of coated pits. Nature (London) 30, 727-729. Pincus, M. R., Lisi, C. D., and Rendell, M. (1981). Ligand binding to multiple equivalent sites with steric hindrance. Biochim. Biophys. Acta 163, 392-396. Pollack, L. R., Tate, E. H., and Cook, J . S . (1981). Turnover and regulation of Na-K-ATPase in HeLa cells. Am. J . Physiol. 241, C173-CI83. Pollet, R. J., Standaert, M. L., and Haase, B. A. (1977). Insulin binding to the human lymphocyte receptor. Evaluation of the negative cooperativity model. J . Biol. Chem. 252, 5828-5834. Poo, M. (1981). In situ electrophoresis of membrane components. Annu. Rev. Biophys. Bioeng. 10, 245-276. Priore, R. L., and Rosenthal, H. E. (1976). A statistical method for the estimation of binding parameters in a complex system. Anal. Biochem. 70, 23 1-240. Reed, B. C., Ronnett, G. V., Clements, P. R., and Lane, M. D. (1981). Regulation of insulin receptor metabolism. Differentiation-induced alteration of receptor synthesis and degradation. J. Biol. Chem. 256, 2917-3925. Robinson, K. R. (1979). Electrical currents through full-grown and maturing Xenopus oocytes. Proc. Nail. Acud. Sci. U.S.A. 76, 837-841. Rodbard, D. (1973). Mathematics of hormone-receptor interaction. I. Basic principles. Adv. Exp. Med. Biol. 36, 289-341. Saffman, P. G., and Delbruck, M. (1975). Brownian motion in biological membranes. Proc. Natl. Acad. Sci. U.S.A. 72, 3111-3113. Salacinski, P. R. P., McLean, C., Sykes, I. E. C., Clement-Jones, V. V., and Lowry, P. J . (1981). Iodination of proteins, glycoproteins, and peptides using a solid-phase oxidizing agent, I ,3,4,6Tetrachloro-etc. (Iodogen). Anal. Biochem. 117, 136-146. Scatchard, G. (1949). The attraction of proteins for small molecules and ions. Ann. N . Y . Acad. Sci. 51, 660-672. Schecter, A. L., and Bothwell, M. A. (1981). Nerve growth factor receptors on PC12 cells: Evidence for two receptor classes with differing cytoskeletal association. Cell 24, 867-874. Schlessinger, J . , Schechter, Y.,Cuatrecasas, P., Willingham, M. C., and Pastan, I. (1978). Quantitative determination of the lateral diffusion coefficient of the hormone-receptor complexes of insulin and epidermal growth factor. Proc. Natl. Acad. Sci. U.S.A. 75, 5353-5357. Schoenheimer, R. (1942). “The Dynamic State of Body Constituents.” Harvard Univ. Press, Cambridge, Massachusetts. Schwartz, A. L., Fridovich, S. E., and Lodish, H. F. (1982). Kinetics of internalization and recycling of the asialoglycoprotein receptor in a hepatoma cell line. J . B i d . Chem. 257, 42304237. Silverstein, S. C., Steinman, R. M., and Cohn, Z. A. (1977). Endocytosis. Annu. Rev. Biochem. 46, 669-722.

9. RECEPTORS: MODEL FOR MEMBRANE PROTEIN TURNOVER

41 1

Sklar, L. A . (1984). Fluorescence polarization studies of membrane fluidity. Where do we go from here‘?In “Biomembranes” (M. Kates and L. A . Manson. eds.), pp. 99-131. Plenum, New York. Sklar, L. A., Miljanich, G. P., and Dratz, E. A. (1979). Phospholipid lateral phase separation and the partition of cis-parinaric acid and trans-parinaric acid among aqueous, solid lipid, and fluid lipid phases. Biochemistry 18, 1707- 1716. Sklar, L. A., Jesaitis, A. J . , Painter, R. G., and Cochrane, C. G. (1982). Ligandlreceptor internalization: A spectroscopic analysis and a comparison of ligand binding, cellular response, and internalization by human neutrophils. J. Cell. Eiochem. 20, 193-202. Sklar, L. A,, Finney, D. A,, Oades, 2. G . , Jesaitis, A. J., Painter, R. G . , and Cochrane, C. G. ( 1984). The dynamics of ligand-receptor interactions. Real-time analyses of association, dissociation and internalization of an N-formyl peptide and its receptor on the human neutrophil. J. Eiol. Chem. 259, 566 1-5669. Soumpasis. D. M. ( 1983). Theoretical analysis of fluorescence photobleaching recovery experiments. Biophys. J. 41, 95-97. Stahl, P., Schlesinger. P. H., Sigardson, E., Rodman, J. S., and Lee, Y. C. (1980). Receptormediated pinocytosis of mannose glycoconjugates by macrophages: Characterization and evidence for receptor recycling. Cell 19, 207-215. Steer, C. J . , and Ashwell, G. (1980). Studies on a mammalian hepatic binding protein specific for asialoglycoproteins: Evidence for receptor recycling in isolated rat hepatocytes. J. Eiol. Chem. 255, 3008-30 13. Steinman, R. M., Brodie, S. E., and Cohn, Z. A. (1976). Membrane flow during pinocytosis. A stereologic analysis. J. Cell Eiol. 68, 665-687. Stoscheck, C. M., and Carpenter, G . (1984). Down-regulation of epidermal growth factor receptors: Direct demonstration of receptor degradation in human fibroblasts. J . Cell Biol. 98, 1048- 1053. Tanabe, T., Pricer, W. E., Jr., and Ashwell, G. (1979). Subcellular membrane topology and turnover of a rat hepatic binding protein specific for askaloglycoproteins. J. B i d . Chem. 254, 1038- 1043, Thilo. L., and Vogel, G. (1980). Kinetics of membrane internalization and recycling during pinocytosis in Dictyostelium discoideum. Proc. Natl. Acad. Sci. U.S.A. 77, 1015-1019. Waelbroeck, M. (1982). The pH dependence of insulin binding: A quantitative study. J. Eiol. Chem. 257, 8284-8291. Wall, D. A,, and Hubbard, A . L. (1981). Galactose-specific recognition system of mammalian liver: Receptor distribution on the hepatocyte cell surface. J. Cell Eiol. 90, 687-696. Wallace, R. A , , and Ho, T. (1972). Protein incorporation by isolated amphibian oocytes. 11. A survey of inhibitors. J. Exp. Zool. 181, 303-317. Wallace, R. A,, Jared, D. W., Dumont, J . N., and Sega, M. W. (1973). Protein incorporation by isolated amphibian oocytes. 111. Optimum incubation conditions. J. Exp. Zool. 184, 321-333. Ward, J . H., Kushner, J. P., and Kaplan, J. (1982). Regulation of HeLa cell transfenin receptors. 1. Eiol. Chem. 257, 10317-10323. Weaver, D. L. (1983). Diffusion-mediated localization on membrane surfaces. Biophys. J. 41, 8 I 86. Weigel, P. H., and Oka, J. A. (1982). Endocytosis and degradation mediated by the asialoglycoprotein receptor in isolated rat hepatocytes. J. Biol. Chem. 257, 1201-1207. Weigel, P. H., and Oka, J. A. (1983a). The surface content of asialoglycoprotein receptors on isolated hepatocytes is reversibly modulated by changes in temperature. J. B i d . Chem. 258, 5089-5094. Weigel, P. H.. and Oka, J. A. (1983b). The large intracellular pool of asialoglycoprotein receptors functions during the endocytosis of asialoglycoproteins by isolated rat hepatocytes. J. Eiol. Chem. 258, 5095-5 102.

41 2

H. STEVEN WILEY

Wey, C. L., Cone, R. A,, and Edidin, M. A. (1981). Lateral diffusion of rhodopsin in photoreceptor cells measured by fluorescence photobleaching and recovery. Biophys. J . 33, 225-23282. Wibo, M . , Thines-Sempoux, D., Amar-Costesec, A,, Beaufay, H., and Godelaine, D. (1981). Analytical study of microsomes and isolated subcellular membranes from rat liver. VIII. Subfractionation of preparations enriched with plasma membranes, outer mitochondria1 membranes, or Golgi complex membranes. J . Cell Biol. 89, 456-474. Wiley, H. S., and Cunningham, D. D. (1981). A steady state model for analyzing the cellular binding, internalization and degradation of polypeptide ligands. Cell 25, 433-440. Wiley, H. S.. and Cunningham, D. D. (1982a). The endocytotic rate constant: A cellular parameter for quantitating receptor-mediated endocytosis. J . Biol. Chern. 257, 422-4229. Wiley, H. S., and Cunningham, D. D. (1982b). Epidermal growth factor stimulates fluid phase endocytosis in human fibroblasts through a signal generated at the cell surface. J . Cell. Biochem. 19, 383-394. Willingham, M. C., and Pastan, 1. H. (1982). Transit of epidermal growth factor through coated pits of the Colgi system. J. Cell B i d . 94, 207-212. Zigmond, S. H . , Sullivan, S. J., and Lauffenburger, D. A. (1982). Kinetic analysis of chemotactic peptide receptor modulation. J . Cell Biol. 92, 34-43.

CURRENT TOPICS IN MEMBRANES AND TRANSPORT. VOLUME 24

Chapter 10 The Role of Endocytosis and Lysosomes in Cell Physiology YVES-JACQUES SCHNEIDER, JEAN-NOEL OCTAVE, AND ANDRE TROUET Laboratoire de Chitnie Physiologique International Institiire of Cellular and Molecular Pathologv and UniversitP Catholiyue de Louvain Brussels. Belgium

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Role in Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Types of Endocytosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. The Fate of the Plasma Membrane in the Course of Endocytosis. . . . . . . . . . . . . . . . . A . Binding, Uptake, and Processing of Anti-PM IgG.. . . . . . . . . . . . . . . . . . . . . . . . B. Binding. Uptake, and Processing of Fluorescein-Labeled Control IgG.. . . . . . . . C. Evidence for Recycling of Plasma Membrane Constituents. . . . . . . . . . . . . . . . . . 111. The Role of Endocytosis and Lysosomes in Transfernin Iron Uptake.. . . . . . . . . . . . . A. Transfernin Iron Uptake by Mammalian Cells.. . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Model for Transfemn Iron Uptake .......... IV. The lntracellular Sorting of Ligands Tak A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Intracellular Fate of Polymeric IgA. Galactosylated Serum Albumin, and Hemoglobin-Haptoglobin Taken up by the Liver . . . . . . . . . . . . . . . . . . . . . . . . . V. Summary and Perspectives.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I.

A.

1.

413 413 415 418 418 420 424 430 432 438 440 440 446 45 1 452

INTRODUCTION

A. Role in Cells Endocytosis (Jacques, 1969; de Duve et d., 1974; Silverstein et al., 1977) is a mechanism by which macromolecules present in the extracellular medium can be taken up by a living cell. This process is present in most, if not all, eukaryotic cells and represents, with the possible exception of some toxins or viruses (Sil413

Cupyright 10 1985 by Academic Precs. Inc All rights 01 rrproducuon in any form reserved ISBN 0-12-153124-7

41 4

YVES-JACQUES SCHNEIDER ET AL.

verstein et al., 1978), the only way in which these macromolecules can enter the cells. Endocytosis starts with the invagination of the plasma membrane around the substances to be captured and with their enclosure into an endocytic vesicle. The vesicle formed at the expense of the cell surface then migrates into the cytoplasm and fuses usually with a lysosome into which its content is delivered and digested by acid hydrolases. In some circumstances the content of the endocytic vesicle may appear in the Golgi region (Herzog and Farquhar, 1977; Ottosen et al., 1980), and in some cell types, such as epithelial cells, the vesicle may cross the cytoplasm and release its content at another pole of the cell (Kraehenbuhl and Kuhn, 1978). In endothelial cells, there is a continuous bidirectional exchange of macromolecules between the plasma and the interstitial fluid. This is obtained by a very efficient coupling of endocytosis and exocytosis; in this particular case, the terms diucytosis (Jacques, 1969) or trunscytosis (Simionescu , 1981) are often preferred. Until a few years ago, endocytosis was considered mainly as a process by which macromolecules present in the extracellular medium were captured and delivered to lysosomes for digestion. It was also generally accepted that concomitantly the membrane surrounding the endocytic vesicle was either incorporated into the lysosomal membrane or digested by lysosomal enzymes (Daems and Van Rijsel, 1961; Miller and Palade, 1964; de Duve and Wattiaux, 1966). During the last few years, however, it has become increasingly evident that endocytosis plays a key function in cell physiology. In particular, it appears that the presence of receptors at the cell surface allows the specific uptake of extracellular ligands and that lysosomes are not solely a metabolic dead end but have a crucial role in the exchanges of substances between the cell and its extracellular environment. Endocytosis first plays a role in cell nutrition through the nonspecific uptake of various macromolecules which are conveyed to lysosomes to be digested, thus providing the cell with amino acids, carbohydrates, and lipids as energy sources or for biosynthetic purposes. This process could, however, be of little importance on a quantitative basis. In parallel with this largely nonspecific mechanism, many substances are rapidly interiorized after specific interaction with receptors present at the cell surface. Receptor-mediated endocytosis allows the selective uptake of transport proteins, peptide hormones, glycoproteins, plasma proteins, or enzymes and has a fundamental role in growth, nutrition, and differentiation of animal cells (for review, see Goldstein et al., 1979; Pastan and Willingham, 1981). Another crucial role of endocytosis is the selective clearance of various endogenous or exogenous substances present in the bloodstream; this is chiefly carried out by the reticuloendothelial system and to a lesser extent by the liver parenchymal cells. For instance, endocytosis allows the uptake by macrophages of microorganisms, such as parasites [Leishmania donovani (Chang and Dwyer,

10. ROLE OF ENDOCMOSIS AND LYSOSOMES

41 5

1978) or Trypanosoma cruzi (Nogueira and Cohn, I976)], bacteria; toxins, such as diphtheria and Pseudomonas toxin or ricin [Sandvig and Olsnes (1980); Draper and Simon (198O)l; viruses [Semliki Forest virus (Helenius et al., 1980; Marsh and Helenius, 1980), fowl plague virus (Matlin et al., 1981)l; and antigens in the form of complexes with antibodies and complement (Silverstein et al., 1978). Finally, endocytosis allows the intracellular recovery of the large amounts of membrane brought to the cell surface during the exocytic event at the presynaptic endings of neurons (Heuser, 1978) or in secretory cells of the endocrine and exocrine glands (Herzog and Farquhar, 1977; Ottosen et a/., 1980).

B. Types of Endocytosis In this section, endocytosis has been tentatively divided into several categories on the basis of different criteria such as the size of the substances ingested by the cells. Phagocytosis (large particles) is compared to pinocytosis (fluids or small particles). Also, a comparison of how substances enter the cell in the fluid content of endocytic vesicles (fluid-phase endocytosis) or bound to its membrane, (adsorptive endocytosis) is made, the latter being further subdivided into receptor and non-receptor-mediated processes. 1. PHAGOCYTOSIS Phagocytosis refers to the uptake, by specialized “phagocytic cells” such as macrophages or polymorphonuclear leukocytes, of large particles such as bacteria, erythrocytes, and latex beads, with relatively little fluid. It is generally accepted that only particles greater than 1 p-m in diameter are captured by phagocytosis. Although substances can be captured without prior attachment to plasma membrane binding sites (fluid-phase phagocytosis), this process seems to be triggered by the interaction of the ingested particle with the plasma membrane of the phagocytic cells (receptor-mediated phagocytosis), a finding implying specific receptors for immunoglobulins (Fc receptors) and complement (C, receptors) (Silverstein et a l . , 1978). Phagocytosis also appears to have energy requirements different from those for pinocytosis and to be much more dependent on the cytoskeleton (Silverstein et al., 1977, 1981). 2. PINOCYTOSIS Pinocytosis is a ubiquitous process, found in almost all cells, that leads to the interiorization of fluids, solutes, and small particles dissolved in the extracellular medium (Silverstein et a l . , 1977). This process allows the transport of macromolecules from the exterior to intracellular compartments.

41 6

WES-JACQUES SCHNEIDER

ET AL.

3. FLUID-PHASE ENDOCYTOSIS Fluid-phase endocytosis is a completely nonselective process by which substances are taken up by the cells and delivered to the lysosomes without prior binding to the cell surface. Accordingly, the principal characteristic of fluidphase endocytosis is that uptake of substances by the cells must be strictly proportional to their concentration in the extracellular medium and must be minimal at low temperature at which endocytosis is almost completely impaired (Silverstein et al., 1977; Steinman et al., 1978). Although fluid-phase endocytosis is a completely nonspecific mechanism, there are, nevertheless, differences in the endocytic rates between different cell types, thus permitting some degree of selectivity. For instance, using the same experimental conditions, Steinman el al. (1976) have reported that cultured macrophages take up by fluidphase pinocytosis nine times more horseradish peroxidase (HRP) than do cultured L cells.

4. ADSORPTIVE ENDOCYTOSIS Adsorptive endocytosis refers to the uptake of macromolecules or particles for which there are binding sites on the plasma membrane. However, depending on affinity, the selectivity of these sites for the ligands, and the biological consequences, investigators have further subdivided this process. a. Receptor-Mediated Endocytosis. As reviewed (Goldstein et al., 1979; Pastan and Willingham, 1981), several systems have been described in which cells use endocytosis to interiorize substances that have become bound to surface receptors. Receptor-mediated endocytosis has become recognized as an important and general mechanism by which cells take up nutritional, regulatory substances or particles (viruses, opsonized bacteria) from the extracellular medium. During recent years receptor-mediated endocytosis has been demonstrated for transport proteins such as low-density lipoprotein (LDL; which carries cholesterol used for membrane and hormone biosynthesis), transcobalamin I1 (which and transferrin (which contains iron), and for polypeptransports vitamin B tide hormones (insulin, chorionic gonadotropin, a-melanotropin), growth factors (epidermal, nerve, and platelet-derived growth factors), glycoproteins exposing terminal galactose, mannose, mannose 6-phosphate, or N-acetylglucosamine, enzymes (lysosomal enzymes or pancreatic ribonuclease), plasma proteins (a2macroglobulin, IgG, polymeric IgA), toxins, and viruses. All these systems share properties that collectively define receptor-mediated endocytosis: i. The binding site on the cell surface is a receptor in the strict sense. It is present at the cell surface in relatively small numbers, and therefore the specific

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

41 7

binding is a saturable process dependent on the ligand concentration. For some ligands (LDL, insulin, transferrin) receptors will be present at the plasma membrane of many different cell types but in different numbers, whereas for other ligands the presence of specific receptors will be restricted to one organ or within the organ to one cell type. For many receptors, binding to the receptor will be restricted to one or several ligands sharing a common chemical group or moiety, and the interaction of one ligand with the receptor will be inhibited competitively by analogs. The affinity constant of the ligand for the receptor must be related to the physiological concentration of the ligand in the extracellular medium. This criterion is important in that it will determine the degree of saturation in physiological conditions. For a regulatory role of receptor-mediated endocytosis, it is important that the saturation concentration is not greatly in excess of the physiological concentration of the ligand. For other receptors, the binding of the ligand to the receptor allows the achievement of a biological effect, e.g., nutrition, differentiation, or growth of the target cell. ii. Once the ligand is bound to the receptor, it is rapidly endocytosed, the half-time of interiorization ranging usually from seconds to a few minutes. Endocytosis of the ligand allows its trapping even if it dissociates from the receptor. After interiorization, the ligands are usually delivered to lysosomes, where they are digested. The intralysosomal digestion or the exposure of the ligand to an acidic pH appear in some cases to be required for the achievement of the biological effect. Occasionally the ligands are delivered to other cellular organelles or are even released at another pole of the cell. ... 111. In all cases for which ultrastructural data are available, prior to endocytosis the receptor-bound ligands are collected into specialized portions of the plasma membrane which have been named coated pits because of their particular morphological appearance.

b. Non-Receptor-Mediated Adsorptive Endocytosis. In addition to the binding of ligands to plasma membrane receptors, several substances can attach nonspecifically to a large number of low-affinity sites on the cell surface. This is the case, for example, for cationized ferritin, which is used to trace the fate of plasma membrane anionic sites during endocytosis in different cell types (Farquhar, 1978). However, physiological substances may also be taken up by nonreceptor-mediated adsorptive endocytosis. In addition to its uptake via the highaffinity LDL receptor pathway, plasma LDL seems also to be captured by such a process (Goldstein and Brown, 1977). Although this mechanism requires high LDL plasma concentrations to achieve important uptake of cholesterol, in hypercholesterolemic patients and even in most people of developed countries in which the mean plasma level of LDL is about five times that required to saturate the specific receptors, it can induce the overloading of cells with cholesteryl esters which could become components of atherosclerotic plaques.

41 8

WES-JACQUES SCHNEIDER ET AL.

II. THE FATE OF THE PLASMA MEMBRANE IN THE COURSE OF ENDOCYTOSIS With the aim of exploring the possible use of antibodies as target-specific carriers for chemotherapeutic drugs, according to the general lysosomotropic model proposed by Trouet and co-workers (Trouet et al., 1972; de Duve et al., 1974; Trouet, 1978), we initiated over 10 years ago, in collaboration with P. Tulkens, experiments on the interaction with cultured fibroblasts of antibodies directed against plasma membrane antigens (anti-PM IgG). As a control for these experiments, we used IgG from nonimmune rabbits (C IgG). This experimental model provided us with the opportunity to test the recycling hypothesis proposed by Steinman er al. (1976). We were able to demonstrate that during endocytosis in cultured fibroblasts, not only plasma membrane fragments but also substances firmly attached to them, like anti-PM IgG, gain access to the lysosomes but escape digestion and are recycled back to the cell surface. A more detailed account of these data as well as the experimental procedures have been published in Schneider et al. (1977, 1978, 1979a,b, 1981a,b) and Tulkens et al. (1977a-c, 1978, 1980).

A. Binding, Uptake, and Processing of Anti-PM IgG Figure 1 illustrates the main kinetic parameters of the interaction of anti-PM IgG with cultured fibroblasts. Both at 4°C (Fig. 1A) and at 37°C (Fig. lB), the uptake is saturable, with the antibody concentration reaching a similar plateau level of about 8 pg IgG/mg cell protein. This value is very high since it represents about lo7 molecules per cell. The binding of anti-PM IgG is specific since it is about 10 times higher than that of C IgG (not shown). Binding is also stable (Fig. 1A) since, if after binding of the antibody at 4"C, the cells are transferred into a fresh medium at 4"C, the label remains almost entirely associated with the cells. As a function of time (Fig. IC), the binding of anti-PM IgG to the cells is progressive and surprisingly slow, taking about 10 hours before reaching a plateau. The kinetics of uptake are similar at 4°C (not shown) and 37"C, although at this temperature, the plateau level is higher by about 15%. The cell-associated, labeled material consists almost entirely of intact IgG, as indicated by gel filtration, SDS-polyacrylamide gel electrophoresis, or reaction with anti-rabbit IgG antibody. In parallel to the accumulation of IgG by the cells, labeled material soluble in trichloroacetic acid appears in the culture medium. Its concentration increases proportionally to the duration of the incubation; but after 48 hours it only amounts to less than 5% of the cell-associated material. The subcellular localization of the cell-associated antibody has been studied by cell fractionation. As indicated in Fig. 2, after 36 hours incubation, anti-PM IgG

41 9

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

8

t

0 -0

C

B

A

50

Anti-PM IgG (pg/rnl)

r t

8

I00

50 Anil-

I00

P M IgG (pg/ml)

u 24

48

Duration of incubolion (hours)

FIG. I . Interaction of anti-PM IgG with cultured fibroblasts. (A), (B)Cells ( - I mg of protein) for 1 hour at 4°C in fresh medium incubated for 16 hours at 4°C and then reincubated ( 0 )or not (0) or 24 hours at 37°C (B)in the presence of different concentrations of 3H-labeled anti-PM IgG. (C) Cells incubated for different durations with 20 pgiml of .lH-anti-PM IgG (A)or preincubated for 24 hours with 20 kglml of anti-PM IgG and reincubated in fresh culture medium (V).The appearance in the medium of low-molecular-weight "-labeled degradation products (A)was assayed after precipitation of the protein in 15% trichloroacetic acid and assay of the supernatant for the presence of radioactive material. At the end of the incubation, the cells were washed, dissolved in I % (w/v) sodium deoxycholate, and analyzed for radioactivity and protein content. All results are expressed in micrograms of IgG equivalents as determined by the radioactivity measurements. Mean results k SD of three independent experiments.

accompanies closely the plasma membrane marker 5'-nucleotidase; similar results were obtained (1) after 6 hours incubation, (2) when cells incubated for 24 hours with anti-PM IgG were reincubated for 48 hours in fresh culture medium, or (3) when anti-PM F(ab') fragments were presented to the cells instead of IgG. The equilibration profiles strongly suggest the association of anti-PM IgG with structures bearing 5'-nucelotidase, i.e., the plasma membrane or endocytic vesicles. This localization is confirmed by examination with a fluorescence microscope of fibroblasts that had been incubated with fluorescein-labelled anti-PM IgG: fluorescence was seen associated mainly with the pericellular membrane, delineating the cell periphery with formation of occasional patches (not shown). Taken together, all these data indicate that anti-PM IgG combines with antigens present at the cell membrane and remains attached for at least several days with no appreciable evidence of interiorization or transfer to lysosomes and only little digestion and release in the form of low-molecular-weight material. Compared with observations made on other cell types or with other ligands, our results show two surprising features. The first one is the extreme slowness of the binding process, which takes some 12 hours to be completed, suggesting that the antigens are not readily accessible to the antibody. Whatever the exact nature of the barrier, its uncovering can hardly depend on metabolism, cell movement, or membrane remodeling, since the binding kinetics are practically the same at 4

420

YVES-JACQUES SCHNEIDER ET AL.

PROTEIN

5*-NUCsOTlDASE

N-ACETVL - 0

-

GLUCOSAMINIDASE 5

n CVTOC H R O M E OXIDASE

-

t U

z

2o

W

3 O W

II. LL

10

DENSITY

FIG. 2. lsopycnic centrifugation of a postnuclear supernatant from fibroblasts incubated for 36 hours at 37°C with 3H-anti-PM IgG freed from IgG cross-reacting with lysosomal antigens.

and 37°C. Also surprising is the apparent stability of the binding, which does not seem to result in capping and subsequent interiorization or shedding.

B. Binding, Uptake, and Processing of Fluorescein-Labeled Control IgG Figure 3 illustrates the main kinetic features of the interaction of fluoresceinlabeled control IgG (FC IgG) with cultured rat fibroblasts. At 4"C, FC IgG binds to fibroblasts in an almost immediate process (not shown) which is saturable with FC IgG concentration (Fig. 3A) and which is completely reversed when cells are transferred into a fresh medium. These results indicate the presence at the cell surface of a large number of binding sites (about 6.7 X lo6 per cell), halfM. After 30 minutes incubation at 37°C (Fig. 3A), saturated at about 2 X the accumulation of labeled material by the cells is equal to that observed after 1 hour at 4°C. When the incubation is prolonged (Fig. 3C) the accumulation of labeled material is first proportional to the length of the incubation but levels off after a few hours without, however, reaching a true plateau level. Concomitantly, but after a lag phase of a few hours, low-molecular-weight degradation products are recovered in the culture medium. If cells cultured for 24 hours in the presence of FC IgG are then transferred into a fresh medium, only about 50% of the label reappears in the medium in the form of degradation products. As

42 1

10. ROLE OF ENDOCMOSIS AND LYSOSOMES

indicated by Fig. 3B, after 4 hours and after 24 hours, there is no complete saturation in the accumulation of the label as a function of the extracellular FC IgG concentration. Considering, however, the impressive uptake (cell accumulated label + degradation products in the medium) of FC IgG at high concentrations (not shown), this results probably from a partial impairment of the digestive process and therefore from an overaccumulation of unreleasable label. The distribution patterns recorded for cell-associated FC IgG after various times of exposure to the IgG are shown in Fig. 4. After 30 minutes the equilibration profile resembles that of 5’-nucleotidase, with a shoulder toward high densities. Later, however, it shifts progressively to an essential unimodal distribution mimicking closely that of lysosomal hydrolases and clearly separated from 5 ‘ nucleotidase and galactosyltransferase, marker enzymes of the plasma membrane and the Golgi structures (not shown). The association of FC IgG with lysosomes is further confirmed by the effect of chloroquine on their buoyancy. In the presence of this drug, which considerably accumulates within lysosomes (Wibo and Poole, 1974; Ohkuma and Poole, 1981), the equilibrium density of these organelles in sucrose gradients is significantly decreased, and the cell-associated fluorescent material closely accompanies the lysosomal marker in its density shift. Such a pattern is also observed if cells incubated for 24 hours with FC IgG are then transferred for 24 hours into a fresh culture medium (not shown). On the A

C

B

50

50-

c

-

-

lo00 FC IgG (pg/ml)

2000

1000

FC IgG (pg/rnl)

2000

24

48

Duration of incubation (hours)

FIG. 3. Interaction of FC IgG with cultured fibroblasts. (A), ( B ) Cells (-1 mg of protein) or at 37°C for 30 minutes ( A , a),4 hours (B, 0). and 24 hours incubated at 4°C for 1 hour ( A , 0) (B, ) . in the presence of different concentrations of FC IgG. (C) Cells incubated for different durations with 100 pg/ml of FC IgG (A)or preincubated for 24 hours with 100 pg/ml of FC IgG and reincubated in fresh culture medium (V).The appearance in the medium of low-molecular-weight, fluorescein-labeled degradation products was assayed by gel filtration of the medium through a Sephadex C-100 column (A). At the end of the incubation, the cells were washed, dissolved in 1 % (w/v) sodium deoxycholate, and analyzed for fluorescence and protein content. Results are expressed in niicrograms of FC IgG equivalents as determined by the fluorescence measurements. Mean results ? SD of three independent experiments.

422

YVES-JACQUES SCHNEIDER ET AL.

S-NVCLEOTIDASE

107

115

123

FC IpG Uluoro.cmcol

W7

116

123

N.ACETVLbOWCOgAUIWID*3E

107

115

(13

DENSITY

FIG. 4. Isopycnic centrifugation of postnuclear supernatant fractions from fibroblasts incubated at 37°C for 30 minutes, I hour, and 3 hours with 250 pg/ml of FC IgG or for 24 hours with 100 pglml of FC IgG/ml.

other hand, when cells incubated with FC IgG are examined in the fluorescence microscope, they showed numerous brightly lit cytoplasmic granules (not shown). As shown in Fig. 5, when cells exposed to a mixture of anti-PM and FC IgG at high concentrations are fractionated, each type of IgG maintains its characteristic location. However, the presence of an anomalous bulge of 5’nucleotidase and anti-PM IgG in the high-density region of the gradient should be noted. The nature of the labeled material that remains trapped within lysosomes has been studied by gel filtration and immunoprecipitation after different times of exposure of the cells to FC IgG. The IgG molecules are progressively trans-

423

10. ROLE OF ENDOCMOSIS AND LYSOSOMES

formed into F(ab'), and F(ab') fragments and part of them remains accumulated intralysosomally in an undigested form. The results summarized above indicate that FC IgG binds to a large number of low-affinity binding sites of the plasma membrane and that it is taken up continuously by the cells, where it accumulates in amounts almost proportional to the extracellular IgG concentration over a large range of concentration. About 75% of the cell-associated IgG molecules are completely digested to low-molecularweight labeled products that diffuse out of the cell and are recovered in the culture medium. About 25% of the IgG molecules remain incompletely digested in lysosomes, in which they progressively accumulate in the form of F(ab'), and F(ab') fragments. The mechanism whereby FC IgG is taken up by fibroblasts appears to be rather complex. Although the relation between the accumulation of the fluorescent label by the fibroblasts and the extracellular concentration of FC IgG is linear over a wide range of concentrations, FC IgG is very unlikely to be taken up by fluid-phase pinocytosis since the experimental clearance ranges from 10 to 20 kl per hour per milligram of cell protein, which is 60 times higher than the values observed for horseradish peroxidase (HRP) and inulin (not shown). Since FC IgG binds at 4°C to plasma membrane, the simplest explanation would be that after adsorption on the cell surface, FC IgG is rapidly interiorized and transferred to lysosomes and digested. In view of such a mechanism and comparing the high

10

0 W

LT LL

J

2ol i n PA € O X I D A S E

O CFV T -O C I

't J107

115

0

1

t

10

0 123

107

115

123

107

115

123

DENSITY

FIG. 5 . Isopycnic centrifugation of a postnuclear supernatant fraction from fibroblasts incubated for 36 hours at 37°C simultaneously with ?H-anti-PM IgC (40 pg/ml) and FC IgC (250 Fglml).

424

YVES-JACQUES SCHNEIDER ET AL.

clearance of FC IgG with its binding at 4°C to the cell surface, we have to conclude that the binding sites have to be restored at least once every hour. In addition, protein synthesis seems not to be essential in this process since cycloheximide at a concentration which almost completely inhibits protein synthesis (25 p,g/ml) depresses only slightly after 24 hours incubation both the uptake of FC IgG at 37°C (by 20%) and its binding at 4°C (by 16%). Taken together, these results strongly suggest the hypothesis that the plasma membrane constituents involved in the uptake of FC IgG could be, after endocytosis, recycled back to the cell surface. This finding is similar to the results of Steinman et al. (1976), who reported that fibroblastic cells taking up HRP interiorize 48% of their plasma membrane per hour without modification of the relative surface area of cell surface, pinocytic vesicles, and lysosomes.

C. Evidence for Recycling of Plasma Membrane Constituents The results presented above show the almost entire lack of interaction between the uptake of anti-PM IgG and that of FC IgG. It means, for one thing, that coating of a major part of the membrane with tightly bound antibody does not interfere with whatever binding or clustering of FC IgG is involved in its uptake, nor does it interfere with the various processes of invagination and membrane fusion that determine interiorization and delivery of the FC IgG to the lysosomes. The only disturbance is shown by the shoulder of 5‘-nucleotidase and anti-PM IgG toward the densities at which lysosomes equilibrate (Fig. 5). Equally remarkable is the fact that the anti-PM IgG remains attached to the membrane for days on end in cells which in the accomplishment of their pinocytic activity would have to interiorize their surface area at least once hourly. Both types of observations would seem to suggest that membrane patches involved in endocytosis somehow exclude the surface antigens that bind the anti-PM IgG. Such a possibility would be comparable with the hypothesis presented by Tsan and Berlin (197 l), who reported that while macrophages interiorize large proportions of their plasma membrane during phagocytosis, specific membrane transport sites for adenine, adenosine, or amino acids remain present at the cell surface. This model would also be compatible with more recent observations which indicate that the receptors, which bind ligands such as LDL, epidermal growth factor (EGF), and hormones, prior to endocytosis (Goldstein et al., 1979) are clustered into restricted areas of the plasma membrane, referred to as “coated pits.” These observations could therefore suggest that endocytosis involves exclusively small areas of the cell surface (amounting to a small percentage of the plasma membrane) which could be devoid of the antigens to which the anti-PM IgG attaches. An alternative possibility is that antibody-coated membrane areas

425

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

do, in fact, participate in endocytic uptake, but are continuously recycled or shuttled back to the cell surface. As pointed out by Steinman et al. (1976), the very magnitude of membrane interiorization makes the hypothesis of recycling almost mandatory. Our findings on the uptake of IgG by fibroblasts offered an opportunity to put the recycling model to an experimental test. Implicit in this model is the assumption that the membrane-bound IgG molecules which seem to remain on the cell surface actually participate in a succession of endocytic cycles, and thus periodically “see” the inside of lysosomes for a brief interval of time during the transient fusion events that take place between pinocytic vesicles and lysosomes. The occurrence of such exposures could be detected if the lysosomes should happen to contain a material with which the membrane-bound IgG are capable of interacting specifically, for instance, anti-IgG antibodies. As a result of such interactions, part of the intralysosomal antibodies could become bound to the membrane and accompany it to the cell surface. At the same time, some of the membrane-bound antibodies could be stripped off and transferred to the lysosomes. Possibly even membrane could be immobilized with the lysosomes and its orderly return to the cell surface prevented. As previously described (Schneider et al., 1979a,b), these predictions were tested in the following way, and the results are illustrated in Figs. 6 and 7 and Table I. During a first phase (experiments B and C), the cells were allowed to take up and store within their lysosomes for 24 hours at 37°C fluorescein-labeled goat antibodies directed against rabbit IgG (F anti-R IgG). They were then TABLE 1 UPrAKE A N D PROCESSING OF IGG Experiment First incubation (h) Goat IgG Fluorescent goat IgG (type) Amount introduced (p,g/ml) Uptake (pglmg cell protein)“ Second incubation (h; Rabbit anti-PM IgG) Amount introduced (p,g/ml) Uptake (pg/mg cell protein) Fluorescence in medium High MW Low MW

BY

FIBROBLASTS

A

B

C

24 C 100 8.30

24 anti-R I00 5.90

24 anti-R

24 20 4.09

24 10

2.03

100

6.95 4

+ 20c 7.5 I .48

(%)b

26 26

0 26

I1 24

~~~

a

Estimated from total fluorescence of cells + medium at the end of second incubation. Percentage of total fluorescence of cells + medium. Four hours in the presence of rabbit anti-PM IgG followed by 20 hours in fresh medium.

426

YVES-JACQUES SCHNEIDER ET AL.

washed and incubated during a second phase in the presence of rabbit anti-PM IgG, either for 24 hours at 37°C (experiment C) or for 4 hours at 37°C followed by a 20-hour reincubation at 37°C in fresh culture medium. As a test of the immunological specificity of any observed interaction, a similar experiment (experiment A) was carried out on cells preloaded with fluorescein-labeled goat control IgG (FC IgG). The uptake of rabbit anti-PM IgG by the cells is unaffected by the previous storage of goat F IgG. As shown in Fig. 6B and C, the digestion of the latter is uninfluenced by the rabbit IgG, but the cells that had stored goat F anti-R IgG suffered an additional substantial loss of fluorescence in the form of high-molecular-weight material when exposed to rabbit IgG. This phenomenon did not occur in cells preloaded with goat FC IgG (Fig. 6A). The material released in this way preceded authentic IgG upon gel filtration and consisted most probably of soluble immune complexes between the goat F anti-R IgG unloaded from the cells and rabbit IgG present in the culture medium. The subcellular distribution patterns recorded in the different experiments are shown in Fig. 7. In the cells preloaded with goat FC IgG (Fig. 7A), the distribution of rabbit anti-PM IgG is similar to that observed previously. This pattern is, however, altered in the cells preloaded with goat F anti-R IgG and reincubated in the presence of anti-PM IgG (Fig. 7B and C). The alteration consists in the appearance of a second peak or shoulder of fluorescent label in the lower-density region occupied by 5'-nucleotidase and occurs without a comparable change in the distribution of the lysosomal marker. Preloading of the cells with goat F antiR IgG (Fig. 7B and c ) brings about a significant shift of radioactivity from the plasma membrane region to the lysosomal region of the gradient, thus causing

100

I-

40

r

,.

EXTRACE LLULAR 0 Low M W

=

High M W

- - - - - - - - - - - -...- ..- - - ..... INTRACELLULAR

LA

d0

nn

20

Plasma Membrane Lysosomes

a Unassigned

0

FIG.6. Fate of goat F IgG accumulated by fibroblasts described in Table I and Fig. 7. Properties of intracellular IgG accompanying lysosomes and membrane computed by fitting IgG distribution to those of marker enzymes N-acetyl-P-glucosaminidase and 5'-nucleotidase as described in Schneider et al. (1979a) (see text).

427

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

B

A 5-nucleotidase

l(

R a b b i 1 A IgG l(

JL

C

10

L

10

> V

z

W

3

Goat F IgG

0 W

I

1c

N acel)

a Y

10

I

n

0-qIu

c o sa m i n idase

A

10

10

0 1.11

1.19

1.27

1.11

1.19

1.27

1.11

1.19

1.27

DENSITY

FIG. 7. lsopycnic centrifugation of postnuclear supernatant fractions from fibroblasts incubated as described in Table I and in the text.

the distribution to become clearly bimodal. A similar but slighter shift affects also 5’-nucleotidase. In agreement with predictions based on the recycling hypothesis, our results demonstrate a considerable translocation of stored goat IgG, accompanied by a lesser but significant change in the cellular distributions of the rabbit IgG in cells that are exposed to rabbit anti-PM IgG after a preliminary incubation in the presence of goat F anti-R IgG. Nothing happens if goat FC IgG is substituted for anti-R IgG in the first incubation, making it clear that the observed effects are the result of a specific immunological interaction between the antibodies. An alternative explanation of our results which does not require membrane recycling is

428

YVES-JACQUES SCHNEIDER ET AL.

that F anti-R IgG unloading occurs by exocytic regurgitation, a process which could be triggered off by an extracellular or juxtacellular interaction between the two antibodies concerned. To test such a possibility, cells were offered HRP in addition to goat F IgG during the first incubation period and were then reincubated in the presence or the absence of rabbit anti-PM IgG. The results (not shown) indicate that the unloading of F anti-R IgG brought about by anti-PM IgG is selective and is not accompanied by a simultaneous unloading of stored HRP. According to the recycling hypothesis, the interaction between the rabbit IgG added to the culture medium and the cell-associated goat anti-R IgG is assumed to take place intracellularly, within the lysosomes since the data presented above have pointed out clearly that, except for the initial passage through phagosomes, the lysosomes are the sole site for the storage and processing of fluoresceinlabeled IgG or undigestible remnants. An important question to answer is whether the recycling process is restricted to pinocytosis or whether it also concerns phagocytosis. Muller et al. (1980) have developed a method which permits one to label in vivo the phagolysosomal membrane. They allowed macrophages to ingest lactoperoxidase covalently linked to latex beads into phagolysosomes and then, at 4°C added 1251 and an extracellular peroxide-generating system which leads to the incorporation of label into material precipitable by trichloroacetic acid. Electron microscopic investigations indicated that immediately after iodination of the cells the label is associated with the phagolysosomal membrane, but that 15 minutes after return to culture at 37"C, large proportions of the label are found associated with the plasma membrane and to a much lesser extent with intracellular vesicles or with phagolysosomes. Concurrently with a membrane flow from the phagolysosomes to the cell surface, they observed that there is a reciprocal movement which counterbalances it and that endocytosis of HRP or Thorotrast is not impaired. This movement is required since the appearance of fragments of the phagolysosomal membrane at the cell surface occurs without shrinkage of the phagolysosomes. Particularly intriguing are the phenomena that may be associated with the return of the membrane patches from the lysosomes to the cell surface. Studying pinocytosis of cationized ferritin and HRP in mouse fibroblasts, Van Beurs and Nilausen (1982) have recently observed the presence of double-labeled endocytic vesicles. They further reported that numerous tiny vesicles containing HRP pinch off from large vacuoles and move to the cell surface to fuse with the plasma membrane, providing evidence in favor of a shuttle between cell surface and lysosomes in the form of closed vesicles. The data we have obtained on fibroblasts using anti-PM IgG as a probe as well as those of Muller et al. (1980) and others demonstrate that during pinocytosis and phagocytosis there is a continuous bidirectional membrane flow between the cell surface and the phagolysosomes. Our results further indicate that the pieces

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

429

of membrane interiorized during pinocytosis pass through the lysosomal compartment and return to the cell surface. They do so in such a way as to be able to pick up ligands from the lysosome contents and to deliver them to the extracellular medium. On the basis of these considerations, it becomes obvious that contrary to what has long been believed, lysosomes are not a metabolic dead end from which trapped molecules can escape only in the form of small diffusible digestible product or, exceptionally in certain cell types or under certain welldefined physiological or pathological circumstances, through bulk exocytic discharge. They are connected with the extracellular medium by means of mobile membrane patches acting as some sort of endless moving belt and are capable of transporting materials in both directions. When moving inward through endocytosis, the belt brings in from the outside various substances that have become bound to cell surface. Such substances, however, will be delivered to the lysosomal compartment; there they undergo digestion but only to the extent that they become detached from their receptor, for instance by exposure to the lysosomal acidity or through enzyme action or by some other agency prevailing in the intralysosomal medium. If this detachment fails to occur, the membranebound substance will return to the cell surface, either intact or possibly altered in some ways by its exposure to the lysosomal content. The anti-plasma membrane antibody used in these experiments provides a particularly striking example of strongly bound ligands that cycle almost endlessly in this manner with only very little transfer to lysosomes and digestion. Membrane constituents may be expected to behave similarly, which could explain their relatively very long halflife (from 20 to 100 hours). Other examples of receptor-bound ligands which could pass through lysosomes but escape digestion will also be presented later in this review. In addition to membrane-bound ligands, endocytosis also brings to the lysosomes droplets of extracellular fluid and certain molecules such as the fluorescein-labeled IgG that are taken up with a remarkable selectivity without the participation of specific receptors. Cationized fenitin used by other investigators (Farquhar, 1978) seems also to be such an example. If kinetic and morphological data from different laboratories have confirmed that plasma membrane fragments gain access to lysosomes before being recycled back to the cell surface, recent results indicate that membrane recycling may take place before the plasma membrane patches reach lysosomes. Besterman et al. (1981) reported that the accumulation by macrophages or fibroblasts of I4Clabeled sucrose, a nonmetabolizable substrate of fluid-phase endocytosis, is not linear with the incubation time. Their kinetic data indicate that this results from the exocytosis of sucrose out of the cells. Sucrose could gain access to at least two intracellular compartments, a first one supposed to be endocytic vesicles from which it is recycled with a t,,* of 5-8 minutes and a second one, possibly lysosomes, turning over very slowly: 3 hours (macrophages) and 10 hours (fibroblasts). Calculation on these kinetic data allowed these authors to postulate that

430

WES-JACQUES SCHNEIDER ET AL.

more than 50% of the sucrose taken up by fluid-phase endocytosis is very rapidly recycled back from endocytic vesicles to cell surface and exocytosed. Very recently, Burgert and Thilo (1983) reported results obtained while studying internalization and recycling of plasma membrane labeled glycoconjugates in the macrophage cell line P388D. Kinetic studies indicate that plasma membrane is internalized and gains access to two intracellular compartments, probably endocytic vesicles and secondary lysosomes from which it is recycled. The equivalent of the plasma membrane is endocytosed every 21 minutes in the form of endocytic vesicle, where it resides for about 3 minutes before being recycled back to cell surface. In addition, in these cells, about 3% of the amount of the internalized plasma membrane enters the lyosomal compartment, where it stays for about 50 minutes before being recycled. Therefore, it becomes increasingly evident that membrane recycling is a complex phenomenon involving at least endocytic vesicles and lysosomes. In addition, it is particularly important to know whether other transit stations exist, such as the Golgi system. Certain experimental data suggest that exogenous ligands gain access to the Golgi apparatus (Herzog and Farquhar, 1977; Farquhar, 1978), where further processing or repair of transported ligands or membrane constituents may take place. It has recently been reported, for example, by Regoeczi et af. (1982) that after binding to the receptor specific for desialylated glycoproteins, asialotransferrin is neither degraded nor transported across the hepatocytes and released into the bile but is returned to the blood with a proportion of its carbohydrate side chains being resialylated.

111. THE ROLE OF ENDOCYTOSIS AND LYSOSOMES IN TRANSFERRIN IRON UPTAKE Iron participates in diverse metabolic processes as different as, for example, the electron flow in bioenergetic pathways, the activation of oxygen, nitrogen, and hydrogen, the decomposition of peroxide and superoxide ions, or DNA synthesis. As reviewed by Bothwell et al. (1979), in early times when molecules essential to life were synthesized, because of reducing atmospheric conditions, large amounts of ferrous iron were available. Later, as a result of an increase in atmospheric oxygen, the metal became almost exclusively present as hydrolyzed ferric oxides and hydroxides which have considerably decreased its bioavailability. Consequently, microorganisms have designed sophisticated organic molecules that they secrete in their extracellular environment and that are able to complex iron and facilitate its uptake. In mammals, a small proportion of iron present in food is absorbed by the mucosal cells of the upper intestine and this process seems mediated by a specific protein. Within the body, iron may be associated with two functional pools. The first

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

431

one consists of iron that is biologically active and is bound, for example, to hemoglobin or myoglobin for oxygen transport and storage or to enzymes as a catalyst of biochemical reactions. The second pool consists of storage iron, which is incorporated within ferritin (a protein present mainly in cytosol and lysosomes) and hemosiderin. When one considers that for a human being of 75 kg, only 1 mg of iron is absorbed and excreted every day but that in the same time the plasma turnover of iron reaches about 36 mg/day (Bothwell ef ul., 1979), it becomes obvious that within the body, iron must be reutilized and recycled. It is also evident that there are considerable exchanges and transport of the metal between the body compartments, where it is absorbed (intestine), recycled (liver, spleen), utilized (bone marrow, erythrocyte precursors, but also all the cells in the organism), excreted (gastrointestinal tract, etc.), and that all these processes must be highly regulated. Since in addition, at the pH, PO,, and ionic strength of the physiological fluids, iron is almost totally insoluble, specialized proteins have been designed for the transport of the metal in plasma and extravascular spaces and for its recovery after intravascular hemoiysis. Among these proteins, transferrin plays a central role. As recently reviewed (Williams, 1982), transferrin is a glycoprotein of 80,000 Da, formed by a single polypeptide chain. It has two specific iron-binding sites which under physiological conditions hold the metal with higher affinity than most other iron-binding molecules (&, M).Transferrin seems to be the only iron source for the biosynthesis of hemoglobin by red cell precursors as well as for almost all mammalian cells, except for hepatocytes and macrophages which take up additional amounts of the metal from the hemoglobin-haptoglobin and heme-hemopexin complexes, ferritin or lactoferrin. The interaction of a transferrin molecule with a cell starts with its binding to a specific plasma membrane receptor. The presence of receptors for transferrin has been detected at the cell surface of many different normal or tumor cell types. The transferrin receptor density is increased in actively growing cells such as many cancer cells or red cell precursors, a finding suggesting a direct correlation between the number of receptors and the iron requirement of the cells. The transferrin receptor which has been isolated from different human cell types with the help of monoclonal antibodies consists of a disulfide-linked dimer formed by two 90,000-MW subunits, each of them being able to bind one transferrin molecule (Newman ef al., 1982). The receptor appears to be an integral protein which spans the lipid bilayer and has a cytoplasmic portion. By analogy to what has been proposed for the receptor specific for low-density lipoproteins (Goldstein et al., 1979), the transmembranous structure of the transferrin receptor could suggest that it is constituted of two active sites. First, a binding site is located at the external face of the plasma membrane, allowing the attachment of transferrin. Another site would be present at the cytoplasmic face

432

YVES-JACQUES SCHNEIDER ET AL.

and be responsible for the clustering of the ligand-receptor complexes into coated pits, a morphological entity formed by the plasma membrane surrounded by a “coat” mainly composed of a specific protein named clathrin (Pearse, 1980). Recent observations (Harding et al., 1982) suggesting that transferringold particles bind to reticulocyte plasma membrane and are associated with clathrin-coated pits could support this hypothesis. However, it is not yet established whether transferrin receptors are randomly distributed at the cell surface like many other receptors or whether they are preclustered into coated pits.

A. Transferrin Iron Uptake by Mammalian Cells Until recently the mechanism by which cells take up iron from transferrin remained largely unelucidated. That transferrin is reutilized after delivery of its iron to the cells was already apparent from the pioneering observation of Katz (1961), who reported that after intravenous injection of 59Fe-loaded, 1311-labeled transferrin to human volunteers, radiolabeled iron was cleared six times more rapidly than radioiodine. To explain these results it was proposed that diferric transferrin binds to the cell surface of immature red cells and that this attachment permits the active removal of iron while the iron-depleted transferrin would be preferentially displaced by other molecules of plasma transferrin loaded with iron. An alternative model was proposed on the basis of morphological observations of the interaction of ferritin or peroxidase-conjugated transferrin with reticulocytes or bone marrow cells. After binding to its receptor, transferrin could enter the cell by endocytosis, release its iron intracellularly, and return to the extracellular medium by some sort of reverse endocytosis (Sullivan et al., 1976; Hemmaplardh and Morgan, 1977). This mechanism could be compared to the pathway of receptor-mediated endocytosis. To gain better insight into transferrin iron uptake, in collaboration with J. C . Sibille, P. Hoffmann, and R. R. Crichton, we initiated some years ago experiments on the interaction of 59Feloaded 3H-labeled rat transferrin with cultured rat fibroblasts, erythroblasts, and hepatocytes. A more detailed account of these data as well as the experimental procedures have been published in Octave et al. (1979, 1981a,b, 1982a,b 1983), Sibille et al. (1983), and Schneider et al. (1982). At 4”C, a temperature at which the endocytic pathway is almost completely inhibited, 3H- and 59Fe-loadedtransferrin bind to rat fibroblasts in low and equal amounts and a plateau is reached after a few hours (Fig. 8A). However, as a function of the transferrin concentration, increasing amounts of 3H label are bound, and a Scatchard analysis of these parameters permits one to estimate the presence of about 90 X lo3 receptors per cell (Ka, 1.1 X lo7 M - l ) and also the presence of a rather large number of low-affinity binding sites from which

433

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

2

I

4

6

8

P

I 6 6

1 12 12

I 18 18

0 2I4 24

30 -

20

-

18

24

TIME (h)

FIG.8. Cells ( I mg protein) were incubated in 20-cm2 petri dishes for different durations at 4°C (A), 15°C (B), or 37°C (C) in 1 ml of culture medium containing 100 p g h l (A) or 10 pg/ml (B,C) of 'YFe-loaded, 3H-labeled transfenin (specific radioactivity of about 10,000dpm X pg protein- I in both isotopes). At the end of the incubation, the cells were washed, dissolved in sodium deoxychoCellular uptake of syFe; (O), uptake of late, and analyzed for radioactivity and proteins. (0). [3HH]transfenin. Mean results C SD of three independent experiments.

434

YVES-JACQUES SCHNEIDER ET AL.

transferrin molecules can be easily detached. At 15"C, a temperature at which lysosomal proteolysis is impaired but at which FC IgG is still taken up by fibroblasts and accumulated within lysosomes (not shown), the uptake of 59Fe is six times higher than that of 3H-labeled transferrin, but reaches a plateau after about 6 hours (Fig. 8B). After 4 hours incubation at 15"C, 35% of the cell-bound 3H label is accumulated intracellularly since it can no longer be released from the cells by a short treatment with trypsin at 4°C. At 37"C, fibroblasts take up 59Fe from doubly labeled transferrin in much greater amounts than at 4" and 15°C in a time-dependent continuous process for up to 24 hours (Fig. 8C) that is almost saturable with transferrin concentration (not shown). In contrast, the uptake of 3H label is much less than that of 59Fe (Fig. 8C). When fibroblasts are incubated in the presence of transferrin concentrations lower than those required for saturation of the receptor, no degradation products can be detected. However, at higher concentrations closer to the physiological situations, a time- and concentrationdependent release of transferrin degradation products from the cells was observed (not shown). Therefore, summing up the amounts of cell-bound 3H label and degradation products released in the culture medium, the uptake of [3H]transferrin is found to be continuous over the incubation period but not proportional to the transferrin concentration in the culture medium. More than 60% of the radiolabeled iron accumulated by the cells during an 18-hour incubation at 37°C is incorporated in the cytosol ferritin, as determined by reaction of cell lysate with anti-ferritin antibody and cell fractionation experiments. On the other hand, the distribution of the cell-bound 3H label after isopycnic centrifugation indicates that it can be associated to a large extent with lysosomes. These results demonstrate that in conditions close to the physiological situation, transferrin is endocytosed by cultured fibroblasts and gains access to lysosomes, where it is digested. After a delay, 3H-labeled degradation products are released from the cells into the culture medium. This endocytic uptake and lysosomal processing of [3H]transferrin could be attributed primarily to fluid-phase endocytosis but also to adsorptive endocytosis involving the low-affinity binding sites for transferrin. However, our kinetic studies which demonstrate that at 37°C the uptake of radiolabeled iron is much greater than that of 3H label clearly indicate that there is also another mechanism in parallel, whereby the iron-depleted transferrin is returned in an intact form to the culture medium. Short-term kinetic experiments (Fig. 9A) further show that when cells incubated for 1 minute at 37°C with doubly labeled transferrin and washed for 1 minute at 37°C are reincubated in a fresh medium for up to 30 minutes, almost all the radiolabeled iron remains associated with the cells, whereas 3H label is released with a t,,, of about 1.5 minutes in a form that reacts with anti-transferrin antibody. Other data indicate also that this mechanism allowing the return of intact iron-depleted transferrin to the extracellular medium is saturable with transferrin concentration, half-satura-

435

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

20 10

';

.c

40

c

-

30

0,

. g

20

E

10

t

50 40

30 20

lo

t 0

10

20

30

TIME ( m i d

FIG.9. Fibroblasts ( I mg protein) preincubated in 20-cm2 petri dishes for 16 hours at 37°C in the absence (A) or in the presence of 10 mM methylamine (B) or 100 ph4 chloroquine (C) were incubated for I minute at 37°C in 0.6 ml of culture medium containing 10 pg/ml of 59Fe-loaded 3H-labeled transfemn (specific activity as in Fig. 8) in the absence (A) or in the presence of 10 mM methylamine (B) or 100 ph4 chloroquine (C). After four washings with 0.6 mi PBS during I minute at 37"C, the cells were reincubated for different durations at 37°C in a fresh culture medium without serum, in the absence (A) or the presence of 10 mM methylamine (B) or 100 p I 4 chloroquine ( C ) .At the end of the reincubation, the cells were washed four times with PBS, dissolved in sodium deoxycholate. and cell-bound "Fe. analyzed for radioactivity and protein. (a),Cell-bound Wlabeled transferrin; (O), Mean results ? SD of three independent experiments.

436

WES-JACQUES SCHNEIDER ET AL.

tion being reached at about 0.12 pJ4, a value close to the half-saturation of the transferrin receptor. To further investigate the site where iron is released from transferrin before the latter recycles to the extracellular medium, we studied the effect of different substances known to interfere with the endocytic process and to increase the pH of lysosomes (Ohkuma and Poole, 1978; Poole and Ohkuma, 1981) and other intracellular acidic compartments (Maxfield, 1982). In the presence of 10 mM methylamine (Fig. IOB), 100 pkl chloroquine (Fig. lOC), or 2.5 pkl monensin (not shown), the uptake of 59Fe from transferrin is inhibited by 65% (chloroquine, methylamine) to 90% (monensin), whereas that of [3H]transferrin is only slightly reduced. In addition, when fibroblasts were pretreated for 16 hours at 37°C prior to short-term kinetic experiments, 49% of the 59Fe initially bound to the cells is released after a 30-minute washout and is found associated with transferrin; when the cells have been pretreated with chloroquine or methylamine, respectively, 35 and 43% of the 3H label is detached after a 30-minute reincubation with a t,,* much longer than in control experiments (Fig. 9B and C). To investigate whether these processes could be generalized, we extended our studies to two other cell types: erythroblasts, which are the precursors of red cells and therefore have a much greater requirement for iron, and hepatocytes, which play a central role in iron metabolism. Largely comparable results were obtained with rat erythroblasts (Octave et al., 1982a), except that the number of receptors appears greater (about 500,000 per cell). In the case of rat hepatocytes, the mechanism of transferrin iron uptake appears to be more complex. Using cultured rat hepatocytes, we observed that if receptors for transferrin are present at the sinusoidal membrane, most of them are in fact occupied by transfenin molecules synthesized and secreted by the hepatocytes (Sibille et al., 1983) and that extensive washings of the cells and a 4hour incubation at 4°C are required to detect their presence. In addition, it appears that iron uptake from transferrin by hepatocytes is much lower than in erythroblasts and fibroblasts. As a consequence, at rather high transferrin concentrations, close to the physiological ones, the uptake of transferrin iron by nonreceptor-mediated processes becomes predominant. At transferrin concentrations higher than those required for half-saturation of membrane receptors, the uptake of 59Fe from double-labeled transferrin is proportional to both incubation time and transferrin concentration and is comparable to the accumulation of inulin by hepatocytes. These results suggest that transferrin is taken up by fluid-phase endocytosis. In contrast, however, the uptake of 3H-labeled transferrin is much lower than that of 5yFeand inulin, and we were unable to detect the presence of labeled degradation products in the culture medium. These results are difficult to reconcile with the classic mechanism of endocytosis since they imply that although transfenin is taken up by a nonselective process, probably fluid-phase endocytosis, iron-depleted transferrin is recycled back in an intact form to the

437

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

80 60 40

__--- - -

20

I

-

'5

g

B 80

60

0

f

2 40

7

0

s

c

20

6

12 18 TIME (h)

24

FIG. 10. Cells (about 300 pg protein) were incubated in 9-cm2 petri dishes for different durations at 37°C in 0.5 ml culture medium containing 100 kg/ml of SyFe-loaded 'H-labeled transferrin (specific radioactivity as in Fig. 8) in the absence (A) or the presence of 10 mM methylamine (B) or 100 phf chloroquine (C). At the end of the incubation, the cells were washed, dissolved in sodium deoxycholate, and analyzed for radioactivity and protein. The culture media were analyzed for the Cellular uptake of syFe; (-0). cell-bound 'H-labeled presence of degradation products. (O), material; (0-- -O),total uptake of transferrin (cell-bound + digestion). Mean results f SD of three independent experiments.

438

YVES-JACQUES SCHNEIDER ET AL.

cell surface. Using rat liver perfused in the presence of double-labeled transferrin, we observed that after 20 minutes part of 59Fe and 3Hlabels are associated with endocytic vesicles equilibrating around 1.12 g/ml after isopycnic centrifugation on sucrose gradient (Sibille et al., 1984). In addition, part of 59Fe present in these vesicles can no longer be retained by anti-transferrin antibody, indicating that iron has been released from transferrin in those vesicles.

8. Model for Transferrin Iron Uptake On the basis of our experimental results and of morphological data from the literature, we proposea some years ago a model for transferrin iron uptake by mammalian cells. It was based on the current view of receptor-mediated endocytosis and on our previous evidence (see above) indicating that ligands tightly bound to the plasma membrane are endocytosed, gain access to the lysosomes, but may then be recycled in an intact form to the cell surface. This model also took into account that whereas at physiological pH iron is very firmly bound to transferrin, at acidic pH (around 5.0), it promptly dissociates from the protein. According to this model, transferrin would be taken up by receptor-mediated endocytosis and iron would be detached from the protein within lysosomes before the iron-depleted protein is returned in an intact form to the cell surface and detached from its receptor. The intralysosomal detachment of iron from transferrin was mainly based on the inhibition of iron uptake by methylamine and chloroquine, which could result, not only from an increase of the intralysosomal pH from 4.8 to 6.2 by these substances (Ohkuma and Poole, 1978; Poole and Ohkuma, 1981), but also from an inhibition of the delivery of macromolecules taken up by endocytosis to lysosomes and from a decrease in the rate at which plasma membrane is recycled back to the cell surface (Schneider and Trouet, 1981). Comparable results were obtained in mouse teratocarcinoma cells (Karin and Mintz, 1981) and rabbit reticulocytes (Morgan, 1981) and a similar model was proposed. However, although there was considerable evidence in favor of lysosomes, the precise intracellular site where iron is released from transferrin was not unequivocally established. Furthermore, recent morphological and biochemical data (Willingham and Pastan, 1980; Wall et al., 1980) indicate that in hepatocytes or fibroblasts, shortly (1-2 minutes) after binding to their receptor, ligands appear in large uncoated structures close to the cell surface and which can be separated by isopycnic centrifugation from the other subcellular organelles, and can be detected within lysosomes only after 15-30 minutes. More recently it has been reported that the pH of these vesicles, named “endosomes” or “receptosomes,” is around 5.0 (Tycko and Maxfield, 1982) and that it can be increased to above 6.0 in the presence of 140 pA4 chloroquine or 10 mM methylamine (Maxfield, 1982).

10. ROLE OF ENDOCYTOSISAND LYSOSOMES

439

It has also been reported (Van Renswoude et a / . , 1982) that in K562 cells, a human erythroid leukemia cell line, iron could dissociate from transferrin in acidic endocytic vesicles, which can be separated from lysosomes by centrifugation on a Percoll gradient. Accordingly, it has been proposed that after release of iron in a nonlysosomal intracellular acidic compartment, the iron-depleted transferrin would be recycled back to cell surface to be detached as a result of a lower affinity for the membrane receptor at neutral pH (Klausner et a / . , 1983). Combining all these recect observations with new results from our laboratory on fibroblasts and hepatocytes, we propose a more complex model which takes into account most of the experimental results. Transferrin would be endocytosed after binding to plasma membrane receptor and nonspecific binding sites or as a result of fluid-phase endocytosis. On a quantitative basis, the relative importance of these processes would vary from one cell type to another, the receptormediated one being maximal in erythroblasts and minimal in hepatocytes. Transferrin molecules would be enclosed in endocytic vesicles and then collected in what are called “endosomes,” “receptosomes,” or “CURL” (Geuze et a / . , 1983). These vesicles or others deriving from them would thereafter fuse with lysosomes. However, all along this journey of the endocytic vesicle within the cytoplasm, two concurrent events would occur. First, as a result of a proton pump present in the membrane of the endosome, the pH would progressively decrease; second, membrane patches would segregate from endosomes or lysosomes and be recycled back to the cell surface in the form of small vesicles. Consequently and as a result of the low affinity of iron for transferrin, as soon as the pH falls to around 5.0, iron would detach; in addition and concurrently, transferrin molecules loaded or not with iron would be recycled back to the extracellular medium. Finally, transferrin molecules gaining access to lysosomes would be digested by hydrolytic enzymes insofar as they are not attached to their receptors, whereas the receptor-bound molecules would escape proteolysis as a result of membrane recycling originating from lysosomes. In contrast to the previous model which postulated that iron release and recycling of iron-depleted transferrin take place either in endosomes or in lysosomes, we suggest that these two processes could occur continuously along the journey of the endocytic vesicle within the cytoplasm. In summary, it has become increasingly evident that transfenin is taken up by endocytosis, that iron is detached within intracellular acidic compartments, and that iron-depleted transferrin is recycled back to the extracellular medium in an intact form. However, it appears that the intracellular compartment in which iron detaches from transferrin (lysosomes or endosomes) is not yet unequivocally established. On the other hand, it also appears that in addition to a receptormediated process, nonspecific uptake resulting from nonspecific binding or fluid-phase endocytosis could also be important and even predominant for some cell types in physiological conditions.

440

WES-JACQUES SCHNEIDER ET AL.

IV. THE INTRACELLULAR SORTING OF LIGANDS TAKEN UP BY RECEPTOR-MEDIATED ENDOCYTOSIS

A. Introduction Our understanding of receptor-mediated endocytosis substantially originates from a few systems which have been carefully investigated during recent years. One of them is the LDL pathway, which has been extensively studied by Brown, Goldstein, and their co-workers in human fibroblasts (for reviews, see Goldstein et al., 1979; Brown and Goldstein, 1979). LDL transports about two-thirds of the total plasma cholesterol of normal subjects and delivers it to most cells through receptor-mediated endocytosis. After binding to its specific plasma membrane receptor, LDL is rapidly conveyed to lysosomes. The digestion of LDL allows the release of cholesterol, which is the central agent mediating complex systems of feedback control that stabilize the cellular content of cholesterol. Since the pioneering work of Ashwell and Morel1 (1974), who observed that asialoglycoproteins are rapidly cleared from the circulation and digested in the liver, the concept has become widely accepted that exposed sugar residues on glycoproteins serve as determinants for in vivo clearance and in vitro uptake. Different distinct carbohydrate-specific pathways for receptor-mediated endocytosis of glycoconjugates have been described. Among them three have been particularly well characterized: the galactose-specific glycoprotein recognition by hepatocytes (for reviews, see Neufeld and Ashwell, 1980; Harford and Ashwell, 1982), the mannose-N-acetylglucosaminespecific uptake of glycoproteins and enzymes by macrophages (Stahl and Schlessinger, 1980), the mannose 6phosphate recognition of lysosomal hydrolases by fibroblasts (Sly, 1982). The different steps involved in receptor-mediated endocytosis will be briefly summarized below. 1 . BINDING TO THE RECEPTOR

On the basis of genetic, morphological, and biochemical studies, it has been proposed that the receptor for LDL (Goldstein et al., 1979) and that for glycoproteins exposing galactose moieties (Harford and Ashwell, 1982) span the lipid bilayer of the plasma membrane and are composed of two active sites: a binding site located at the external face of the plasma membrane and an internalization site at the cytoplasmic face. After synthesis in the endoplasmic reticulum, processing in the Golgi and insertion in the plasma membrane, the receptor migrates in the plane of the membrane until it reaches a coated pit. The anchorage of the receptor within the coated pit results probably from the interaction of the internalization site with a cytoplasmic protein such as clathrin, the major protein constituent of the coat (Pearse, 1980).

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

44 1

As reported in the review by Pastan and Willingham (1981), most of the receptors present at the cell surface are randomly distributed. However, using either a LDL-ferritin conjugate or a monoclonal antibody raised to the LDL receptor and coupled to ferritin, Anderson et al. (1982) reported that about 70% of the LDL receptors present at the cell surface of formaldehyde-prefixed fibroblasts are associated with coated pits, an observation which suggests that the unoccupied receptors cluster spontaneously in the coated pits before binding of the LDL. On the other hand, Pastan and co-workers (for a review, see Pastan and Willingham, 1981) have reported that primary amines such as monodansylcadaverine and methylamine block interiorization of different receptor-bound ligands, which in the presence of the drug become unable to cluster in coated pits and therefore are endocytosed at a decreased rate. On the basis of biochemical studies, they have proposed that primary amines are analogs of lysine and act as competitive inhibitors of transglutaminase, an enzyme which could be involved in the formation of the coated pits. 2. ENDOCYTOSIS OF THE LICAND

Almost immediately after binding to its receptor, the ligand or its complex with the receptor is interiorized. Endocytosis allows first the trapping of the ligand, even if the dissociation rate of the ligand-receptor is rapid. Then, it promotes its delivery to the intracellular site where it can exert part or all of its biological activity. Finally, it induces the removal of the receptor from the cell surface, thereby facilitating its down-regulation. As reported in the review by Harford and Ashwell (1982), it seems also that in the case of the receptor specific for glycoproteins exposing galactose residues, the binding of the ligand to the plasma membrane receptor and its endocytosis could be separate phenomena. In particular, Baenziger and Fiete (1980) have reported that the in vitro binding properties of glycopeptides obtained from different glycoproteins for isolated receptors and their uptake through endocytosis by living cells are not necessarily interconnected since glycopeptides with affinities for the isolated receptor differing as much as 750-fold display similar kinetics of endocytosis by hepatocytes. To explain these results, they have proposed a model in which multivalent glycans with correct spacing of determinants could induce a conformational change in the receptor which would permit interaction with cytoplasmic constituents and promote endocytosis. Using LDL covalently coupled to ferritin, Goldstein et al. (1979) have observed that rapidly after its association with a coated pit, the LDL particle is endocytosed. After 5-10 minutes, the LDL-ferritin conjugate is seen in lysosomes. Biochemical studies with 12sI-labeled LDL have indicated that the LDL particle is rapidly digested and its cholesterol content released in the cytosol to be used for biosynthetic purpose and to exert its regulatory effects. In the case

442

YVES-JACQUES SCHNEIDER ET AL.

of glycoproteins exposing galactose, Wall et al. (1980) observed that almost immediately after their binding to the receptor, the ligands are localized in coated pits. About 30 seconds later, the glycoprotein is transferred to a larger irregular nonlysosomal vesicle near the sinusoidal membrane of the hepatocytes. The coats are lost from the vesicle before the fusion takes place. By 5 minutes, much of the ligand is transported to the Golgi-lysosome region. These morphological data are in agreement with biochemical studies (for a review, see Harford and Ashwell, 1982) which indicate that 5 minutes after injection asialoglycoproteins are associated with plasma membrane-enriched fractions and after 13 minutes with lysosomes. When all these results are combined, it appears that the transfer of the ligands to lysosomes has a half-time of about 7 minutes. As reported in the review by Pastan and Willingham (1981), comparable observations have been made with other ligands as different as LDL, EGF, a2macroglobulin, P-galactosidase, viruses, the only difference being the time course which varies from one cell type to another. A few minutes after binding, all these ligands appear localized in large uncoated and irregular electron lucent vesicles which have been named “endosomes,” “CURL,” or “receptosomes” according to the authors. 3. RECYCLING OF THE RECEPTORS It has been proposed in many different systems that receptors may be reused after a round of interiorization. For example, in hepatocytes each receptor for glycoproteins exposing galactose delivers to lysosomes from 1000 molecules of ligand in the case of cultured rat hepatocytes (Warren and Doyle, 1981) to 4000 molecules for the rat in vivo (Tanabe et al., 1979). The receptor for glycoproteins with terminal mannose is able to endocytose 25 molecules of ligand per hour (Stahl and Schlessinger, 1980) and the LDL receptor 8 molecules per hour (Goldstein et al., 1979). However, most of the evidence that receptors are recycled after endocytosis is based upon the observation that ligand uptake proceeds continuously over extended periods of time even in the absence of protein synthesis and far exceeds the number of receptors that can be detected at the cell surface. The length of time between internalization and recycling of the receptor is not known precisely, but all the experimental data suggest that it must be rather short, in the range of a few minutes. One should also remember that membrane recycling can be extremely rapid since at neuromuscular junctions the membrane of the presynaptic vesicles recycles within seconds (Heuser, 1978). In the case of the interaction of EGF with fibroblasts, the initial uptake of the hormone by receptor-mediated endocytosis leads to an 80% depletion in the number of surface receptors (Carpenter and Cohen, 1976) which is not compensated by reappearance of new receptors. This suggests that endocytosis of the

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

443

receptor could allow its down-regulation. One could therefore suggest that receptor recycling could be related to the physiological role of the ligand. In the case of growth factors or hormones for which the achievement of the physiological function requires only a few molecules, the absence of receptor recycling would permit down regulation of the activity of the ligand. In contrast to other ligands such as transport proteins (LDL, transfenin) or clearance systems (asialoglycoproteins, enzymes) receptor recycling would be rapid and almost complete to allow a continuous intake of ligands. In the case of the LDL pathway (Goldstein and Brown, 1977), cholesterol released by lysosomal digestion of the LDL particle regulates receptor synthesis, decreasing the number exposed at the cell surface; this process, however, does not seem to affect directly receptor recycling. Recent results of Basu et a / . (1981) suggest, however, that in normal human cultured fibroblasts there are two types of functionally distinct receptors for LDL. First, there are receptors which are continuously endocytosed and recycled back to the cell surface, even in the absence of the ligand. It seems, however, that the remaining 50% of the LDL receptors can be induced to be endocytosed and subsequently recycled only in the presence of LDL. It would be interesting to know whether this observation is restricted to the LDL system or whether it could be generalized. As is the case for membrane recycling (see above), the return of receptors to the cell surface is inhibited by lysosomotropic agents as well as by substances such as monensin, a carboxylic ionophore that catalyzes the exchange of Na+ and H+ across biological membranes (Basu et al., 1981).

4. INTRACELLULAR FATEOF THE RECEPTOR-LIGAND COMPLEX Except for a few cases which will be discussed later, it appears that the receptor-ligand complex exists only transiently within the cell and that very often an acidic pH promotes its rapid dissociation. Among the different structures involved in the vacuolar apparatus, two have such an acidic pH and could therefore provide the appropriate environment for a pH-mediated dissociation of ligand-receptor complexes. Lysosomes have a pH around 4.8 (Ohkuma and Poole, 1978, 1981), which is increased to around 6.2 in the presence of lysosomotropic drugs such as chloroquine, methylamine, and ammonium chloride (Poole and Ohkuma, 1981). It has been reported that the endocytic vesicles themselves could have an acidic pH. Tycko and Maxfield (1982), using fluorescein-labeled a,-macroglobulin as a marker of receptor-mediated endocytosis, have estimated that 15-20 minutes after exposure to the ligand, the pH of the fluorescein-labeled a,-macroglobulin environment is around 5 .O. In comparable conditions, they localized colloidal gold a,-macroglobulin in uncoated endocytic vesicles. A first example of a pH-mediated dissociation has already been discussed in

444

YVES-JACQUES SCHNEIDEA ET AL.

this review in the case of the mechanism of transferrin iron uptake. Although at acid pH transferrin does not detach from its membrane receptor, iron is easily released from the protein. It has also been demonstrated that lowering the pH in vitro enhances the dissociation of ligand from the receptor recognizing the mannose 6-phosphate moiety of lysosomal enzymes (Gonzalez-Noriegaet al., 1980). However, much of the experimental evidence in favor of the role of an acidic environment for receptor-ligand dissociation is indirect and has been obtained with the help of weak bases which increase the pH of lysosomes (Ohkuma and Poole, 1978; Poole and Ohkuma, 1981) and of endocytic vesicles (Maxfield, 1982). Gonzalez-Noriega et al. (1980) observed that increasing pH with lysosomtropic agents prevents the dissociation of lysosomal enzymes from their receptor. In addition, this inhibits receptor recycling and consequently impairs endocytosis. Stahl and co-workers (Stahl and Schlessinger, 1980; Tietze et al., 1980, 1982) reported that lysosomotropic agents produce a time-dependent inhibition of the uptake by macrophages of glycoproteins exposing terminal mannose by causing a reversible loss of cell surface receptor. They propose that weak bases, by raising pH, inhibit receptor-ligand dissociation and the return of free receptors to the cell surface. Others have also emphasized the potential role of acid pH in ligand delivery through the galactose-specific receptor of the hepatocytes (Tolleshaug and Berg, 1979; Baumann and Doyle, 1980). In this particular case, however, after receptor-ligand dissociation, a membrane potential would further promote translocation of the receptor to the outside of the lysosomal membrane (Tanabe et al., 1979). It is also clear from the work of Helenius and colleagues (Helenius et al., 1980; M a t h et al., 1981; Marsh et al., 1983) and Wehland et al. (1982) that the penetration of many viruses into the cytosol of the host cell results from the fusion of the virus with the membrane of an intracellular acidic compartment. For the LDL pathway in which a role of an acidic pH in the dissociation of LDL from the receptor has not been demonstrated, results of Anderson et al. (1982), who used antibody to the LDL receptor, have indicated that the dissociation of the ligand from its receptor occurs within the cell rather than on the cell surface. Although many authors have proposed that ligands could dissociate from their receptors within the lysosomes which provide an acidic environment and are the intracellular site for their digestion, more recent data suggest that dissociation could take place before fusion of the endocytic vesicle with lysosomes. Debanne and Regoeczi (198 1) have reported that from 3 minutes to 1 hour after injection of low doses of asialotransferrin, this protein, which is taken up by the galactosespecific receptor, is found partly associated with vesicles which equilibrate in sucrose gradients around 1.1 I g/ml, and wherein it stays undigested. However, at higher doses at which degradation occurs, asialotransferrin behaves as asialoorosomucoid and can be localized within lysosomes. Baenziger and Fiete (1982) have reported that hepatocytes, incubated in a medium devoid of Na+ but

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

445

containing 0.15 M K + are still able to bind asialo-orosomucoid. However, in these conditions, the ligand is not delivered to lysosomes but remains associated with vesicles which could derive from the plasma membrane. Since in these conditions the uptake of this protein appears unimpaired, they conclude that the delivery of the ligand to lysosomes is not necessary for receptor reutilization. Recently, Harford et al. (1983) reported that after a treatment of cultured hepatocytes with ammonium chloride, asialo-orosomucoid accumulates in a prelysosomal fraction, as separated by fractionation on a Percoll gradient. They interpreted these results in terms of a pH-mediated dissociation of the ligand from its receptor within acidic endosomes. On the other hand, Marsh et al. (1983) reported recently that during infection of BHK cells by Semliki Forest virus, the viral genome is released into the cytosol of the host cell within 5-7 minutes, whereas the delivery of viruses to the lysosomes requires 15-20 minutes. From these data and a few others, it has been proposed that ligands could dissociate from their receptors before the endocytic vesicle fuses with lysosomes. The large uncoated and electron-lucent vesicles with which ligands are found associated (“receptosomes,” “endosomes,” “CURL”) could facilitate receptor-ligand dissociation, permit a sorting between the endocytosed ligands, and promote receptor recycling. Hepatocytes provide an appropriate model to analyze the different intracellular mechanisms which carry endocytosed substances to the lysosomes, to the Golgi apparatus or even to cross the cell. First of all, the hepatocyte is an epithelial cell which is responsible for the transport of polymeric IgA (pIgA) from the blood to the bile (Birbeck ef al., 1979). IgA is synthesized by plasmocytes as dimers linked by a “joining chain.” It then binds to the secretory component (SC), a specific receptor which is present at the plasma membrane of various epithelial cells (Orlans et al., 1978; Socken et a / . , 1979; Kiihn and Kraehenbuhl, 1979), and crosses the epithelial cells to be recovered in the secretions in the form of secretory IgA (sIgA) resulting from the covalent linkage of an SC molecule to a pIgA molecule during the transcellular transport. On the other hand, concentrations of sIgA were found to be much higher in rat bile than in the serum (Lemaitre-Coelho et al., 1977) and the active and specific transfer to pIgA as sIgA in the bile has been demonstrated after intravenous injection of labeled pIgA (Orlans et af., 1978; Jackson et al., 1978). In collaboration with J. N . Limet, J. Quintart, P. J. Courtoy, C. OtteSlachmuylder, P. Baudhuin, and J . P. Vaerman, we further studied receptormediated endocytosis either using cultured rat hepatocytes or rat liver in vivo. As discussed above, liver cells are involved in the clearance of various substances from the plasma through receptor-mediated endocytosis. One well-documented example is the uptake by the hepatocytes of glycoproteins exposing galactose moieties (Ashwell and Morell, 1974; Hubbard et al., 1979). It has been observed (Quintart et al., 1981) that galactosylated bovine serum albumin (gal-SA) pre-

446

WES-JACQUES SCHNEIDER ET AL.

pared by reductive amination of serum albumin using lactose and sodium cyanoborohydride, binds specifically to cultured hepatocytes, is taken up, and digested by these cells. We have also examined the uptake by the liver of the hemoglobin-haptoglobin (hem-hap) complex which allows the selective recovery of hemoglobin released in the circulation during intravascular hemolysis. Data have shown the presence of a receptor for this complex at the cell surface of the hepatocytes (Kino et a/., 1980). Finally we have used HRP as a marker for fluid-phase pinocytosis. We have compared the fate of pIgA, gal-SA, and hemhap and that of HRP in cultured hepatocytes or in the liver after intravenous injection into rats or mice. In a second step, we have studied the subcellular localization of the ligands and of HRP by means of cell fractionation techniques. A more detailed account of these results as well as the experimental procedures have been published by Limet et al. (1980, 1981, 1982a-c) and Schneider ef al. (1980, 1982).

B. lntracellular Fate of Polymeric IgA, Galactosylated Serum Albumin, and Hemoglobin-Haptoglobin Taken up by the Liver When injected intravenously into rats, 3H-labeled pIgA, gal-SA, or hem-hap disappear from the plasma with half-lives of, respectively, 3, 6, and 7 minutes. In the three cases, most of the labeled material is found associated with the liver, wherein it is progressively transformed into degradation products. Competition experiments with asialofetuin and in vitro experiments carried out on cultured hepatocytes (Quintart et al., 1981) indicate that gal-SA is selectively taken up through the presence of receptors specific for galactose residues and that this mechanism is saturable with concentration. The hemoglobin-haptoglobin complex is also selectively taken up by the liver in a partially saturable process. Data of Kin0 et al. ( 1 980) have indicated that in vivo hem-hap is cleared through specific receptors located at the plasma membrane of hepatocytes. Finally, data of Birbeck et al. (1979) and our results have confirmed that the uptake of pIgA is mediated by the hepatocytes through the presence of the secretory component exposed at their cell surface (not shown). Concurrently with the determination of the uptake of labeled material by the liver and of the presence of degradation products, we have investigated the appearance of label in the bile. As illustrated in Fig. 11, after intravenous injection in rats, respectively 3.94 ? 0.34%of the injected dose of hem-hap, 2.79%of gal-SA, and 60.1 -t 1.6%of pIgA, 13.8% of anti-SC IgG (not shown) but less than 1% of HRP (not shown) appear in the bile during the 180 minutes following the bolus injection. All these proteins appear in the bile with the same kinetics. Analysis by isokinetic centrifugation on sucrose gradients reveal that

447

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

I

A

1 .o

W

8 0

w c

g

0.5

7

z

6 $ 20

10

80

160

MIN

FIG. 1 1 . Appearance of labeled material in the bile. Rats were injected intravenously with 20 bg of he~n-[~H]hap (A), 60 bg ['HIgal-SA (B), or 150 p.g ['HIpIgA (C), and the appearance of 3H label in the bile was determined.

after injection of pIgA, almost all the labeled material transferred into the bile sediments as slgA whereas after injection of hem-hap and gal-SA respectively 58.9 and 61.5% of the labeled material behave as molecules with the same apparent molecular weight as the native glycoproteins. After injection of anti-SC IgG, respectively 20-40 and 60-80% of the label sediments as material of lower and of higher molecular weight than native IgG, a finding which is in agreement with the data of Lemaitre-Coelho et al. (1981). At different times after intravenous injection of labeled hem-hap, gal-SA, pIgA, or HRP, rat livers were fractionated. Homogenates were first separated into nuclei, MLP [combination of heavy mitochondrial (M), light mitochondrial (L), and microsomal (P) fractions], and soluble fractions as described in Limet et al. (1982a-c). Table I1 indicates the repartition of the marker enzymes and of the accumulated labeled proteins between these fractions. Galactosyltransferase and cathepsin B are found almost entirely in MLP fractions, whereas 5'-nucleotidase is associated to

448

WES-JACQUES SCHNEIDER ET AL.

TABLE 11 REPARTITION OF MARKER ENZYMESAND 3 H - L MATERIAL ~ ~ ~ AWER ~ ~ ~ DIFFERENTIAL CENTRIFUGATION OF HOMOGENATES PREPARED FROM THE LIVEROF TIMESAbTER INTRAVENOUS INJECTION OF HEM-[3H]HAP, RATS, AT DIFFERENT [3H]G~~-SA OR, [ 3 H ] ~ I ~ A Fractions (% of total)a

N

MLP

S

Galactosyltransferaseb 5 ’-Nucleotidaseb Cathepsin Bb

8.1 f 2 . 9 27.0 1.7 1.5 0.1

*

*

88.2 ? 1.7 49.3 0.8 96.2 1 . 1

* *

3.7 2 2.0 23.7 f 1.3 2.3 2 1.0

hem-hap 5 minutes 20 minutes 45 minutes

0.9 0.0 1 .O

90.9 84.1 43.8

8.2 15.9 55.2

Gal-SA

5 minutes 20 minutes 45 minutes

1.3 0.7 I .4

89.4 78.8 80.2

9.3 20.5 18.4

HRP

5 minutes 20 minutes 45 minutes

3.2 7.4 5.3

49.0 78.3 80.9

47.8 14.3 13.8

pIgA

5 minutes 20 minutes 45 minutes I hour 2 hours 3 hours

1.9 0.0 1.7 5.1 4.0 6.3

69.3 73.7 75.3 57.9 52.1 50. I

28.8 26.3 23.0 37.0 43.9 43.6

N, Nuclei; MLP, combination of heavy mitochondria1 (M), light mitochondrial (L), and microsomal (P) fractions; S , soluble. Mean of three independent experiments t SD.

about one-half with this fraction and to one-fourth with both nuclei and soluble fractions. The 3H label accumulated in the liver after injection of hem-hap is found, for the largest part, in the MLP fraction after 5 and 20 minutes whereas, after 45 minutes, 55% of the label is present in the soluble fraction, which is probably related to the appearance in the liver of large amounts of degradation products diffusing in the cytosol. Whatever the time after injection of gal-SA, the label is largely found in the MLP fraction. After injection of pIgA, the label is to a large extent found in the MLP fraction, but the proportion progressively decreases after 2 and 3 hours. [’“CIHRP is mostly found associated with the MLP fraction, except after 5 minutes, at which time 48% of the label is recovered in the soluble fraction. This could represent material either loosely bound to the

449

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

plasma membrane of Kupffer cells or released during the homogenization procedure or centrifugation of the MLP fraction. The MLP fractions were further fractionated by isopycnic centrifugation. Five minutes after injection (Fig. 12A) hem-hap, gal-SA, and pIgA are found at a median density of about I . 13 g/ml and their distributions dissociate clearly from that of cathepsin B, but also partially from those of galactosyltransferase and 5 ‘ nucleotidase. The presence of labeled material at these light densities could result either from a subcellular component devoid of these marker enzymes or from material released from vesicles damaged during the fractionation process. Flotation experiments (not illustrated) clearly rule out this possibility; in such condition, the bulk of the radioactive label is also found at the light densities, a finding suggesting its association with intact vesicles rather than with solubilized material. Twenty minutes after injection (Fig. 12B), the distributions of the labeled material associated with the MLP fractions become clearly bimodal after isopycnic centrifugation. From one-half (hem-hap and gal-SA) to two-thirds (pIgA) of the label still equilibrates at light densities, but material is also recovered at higher densities, where 5’-nucleotidase and cathepsin B activities are detected. Forty-five minutes after injection (Fig. 12C), the equilibration profile of gal-SA

HEMOGLOBINt i APTOGLOBIN

GALACTOSVLATED SERUM ALBUMIN

POLYMERIC IOA

PEROXIOASE

DENSITY

FIG. 12. Isopycnic centrifugation of MLP fractions prepared from liver homogenates from rats 5 minutes (A), 20 minutes (B), or 45 minutes (C) after injection of 25 p g he~n-[~H]hap, 60 &H]galSA. or 150 pg[3H]pIgA or 500 p g I4C. For these experiments, the samples were introduced at the top of the gradients.

450

WES-JACQUES SCHNEIDER ET AL.

is almost superposable to that of cathepsin B; the bulk of hem-hap is also associated with this enzyme, but in addition, some label is found at lighter densities. The distribution of pIgA is, in contrast, still bimodal with large amounts of labeled material found at the light densities and with label equilibrating at higher densities. However, if cytoplasmic extracts obtained from rats 1015 minutes after injection of plgA and hem-hap are further fractionated into M, L, and P fractions, much of the label recovered in the M fractions [from 5% (plgA) to 10% (hem-hap)] accompanies cathepsin B , a finding which strongly suggests its association with lysosomes (not illustrated). In addition, from 1 to 3 hours after injection, the distribution of pIgA remaining associated with the liver becomes closely similar to that of cathepsin B (not shown). In contrast to these three ligands, 5 minutes after injection of [14C]HRP,the distribution of the label closely accompanies that of 5’-nucleotidase (Fig. I2A). Twenty minutes after injection, the profile of the I4C label is largely similar to that of cathepsin B (Fig. 12B) and after 45 minutes, it becomes almost superimposable (Fig. 12C). Five minutes after injection, from 70 to 90% of the labeled material accumulated by the liver after receptor-mediated endocytosis is present in the MLP fraction. After isopycnic centrifugation, the labels can be associated with structures which equilibrate around a density of 1.13 g/ml and which dissociate from the marker enzyme of lysosomes but also from Golgi elements, plasma membrane, and the distribution of HRP taken up by fluid-phase endocytosis. The precise nature of these subcellular components, however, remains questionable. Since labeled material can be associated with them shortly after injection in the animals, they could consist of plasma membrane fragments bearing very little 5’nucleotidase activity but rather high concentrations of receptors, such as the “coated pits” which have been shown to be involved in receptor-mediated endocytosis (Goldstein et al., 1979) and with which pIgA is associated after its binding to cultured hepatocytes or its uptake by hepatocytes in vivo (Courtoy et al., 1981, 1982a,b). However, comparing these results with recent biochemical and morphological data from several laboratories, these structures consist most probably of “endosomes,” “receptosomes,” and “CURL,” in which many different ligands are collected after their binding to membrane receptors. If 5 minutes after injection the distribution of the three proteins but also that of anti-SC IgG (not shown) is largely similar, important differences can, however, be observed afterward. Between 20 and 45 minutes after injection, almost all the hem-hap complex and gal-SA are delivered to lysosomes, whereas small amounts appear in the bile, partly in the form of intact protein. In contrast, from 10 to 45 minutes after injection, only small proportions of pIgA or anti-SC IgG accumulated by the liver appear in lysosomes, whereas the bulk of both antibodies is transferred into the bile in the form of secretory IgA or, for anti-SC IgG, in the form of immune complexes between SC and anti-SC antibody or of partially degraded IgG. Longer times after injection, plgA, which remains associated with the liver, is found within the lysosomes.

I

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

451

In brief, polymeric IgA, galactosylated serum albumin, and hemoglobinhaptoglobin complexes are taken up by hepatocytes through receptor-mediated endocytosis. Whereas more than 90% of the captured gal-SA and hem-hap are rapidly delivered to lysosomes, only one-third of pIgA accumulates in these granules, two-thirds being transferred into the bile. The key question is to determine the sorting site of these different ligands recognized by receptors and the sinusoidal membrane. By combining morphological and biochemical methods, Courtoy et al. (198 1 , 1982a,b) have been able to rule out surface discrimination as a major sorting mechanism. Their data and those presented in this article indicate that all these ligands are rapidly interiorized and collected into vesicles which equilibrate at densities around 1.13 g/ml after isopycnic centrifugation on sucrose gradients, and which resemble “endosomes,” “receptosomes,” or “CURL” on the basis of morphological examination. That sorting of the ligands could take place within endosomes has been suggested by Courtoy et al. (1982a,b). They injected rats with pIgA and gal-SA conjugated to HRP, isolated the vesicles equilibrating around 1.13 g/ml on sucrose gradients, treated part of these fractions with diaminobenzidine and H202, and centrifuged them again on sucrose gradients. They observed that whereas shortly after injection both ligands coequilibrated and shifled together to higher densities after diaminobenzidine-H,02 treatment, at later times, they progressively dissociate. However, although these results strongly suggest that sorting could occur early after endocytosis and before endosome-lysosome fusion takes place, participation of lysosomes in the sorting process cannot be ruled out. Results presented in this review have indeed indicated that about one-third of pIgA taken up by the liver ends up in lysosomes and that anti-SC IgG which is bound by the same receptor is transferred into bile in part in the form of F(ab)’ fragments, as a result of a partial hydrolysis which could be achieved in lysosomes but not in endosomes. Accordingly, we propose that, although sorting of the different ligands is most probably initiated within endosomes, this process could take place all along the journey of the endocytic vesicle in the cytoplasm and even after the fusion of endosome with lysosome. This hypothesis allows us to explain our experimental data and becomes more plausible since recycling can take place from the lysosomal compartment.

V. SUMMARY AND PERSPECTIVES All the data discussed throughout this review article have pointed out that endocytosis and lysosomes play a central role in cell physiology. They indicate that macromolecules such as hormones, immunoglobulins, glycoproteins, enzymes, or transport proteins are taken up by receptor-mediated endocytosis. Morphological observations from several laboratories have indicated that after

452

YVES-JACQUES SCHNEIDER ET AL.

binding to specific receptors the ligands become collected into restricted areas of the plasma membrane, the coated pits. Morphological and biochemical studies have pointed out that this event allows a rapid interiorization of the ligands and their delivery to the intracellular compartments where they can exert their biological effects. Several results indicate, however, that concurrently to this highly specialized process, endocytosis also permits the interiorization of large amounts of plasma membrane, thereby allowing membrane recovery and the nonselective uptake of various substances present in the extracellular medium or bound to nonspecific binding sites at the plasma membrane. The data we have collected during recent years have first of all confirmed these observations. In addition, however, they have also outlined original concepts which seem important for our understanding of cell physiology. 1. Substances which enter the cell bound to the membrane of the pinocytic vesicle are not necessarily delivered to lysosomes, even if they gain access to this intracellular compartment. Our results have indeed demonstrated that antibodies bound to plasma membrane antigens are endocytosed and arrive in lysosomes but escape digestion and are thereafter recycled back to the cell surface still associated with the membrane of the endocytic vesicle. 2. The acidic pH encountered within lysosomes and endocytic vesicles plays a key role in endocytosis. For example, a receptor-bound ligand such as ironloaded transferrin exposed to an acidic pH releases iron while remaining bound to its receptor and being thereafter recycled back to the extracellular medium. 3. Lysosomotropic weak bases, which accumulate within lysosomes and probably endocytic vesicles, increase the pH of these compartments and considerably affect endocytosis. As a consequence, for example, iron uptake from transferrin is largely inhibited. 4. In the hepatocytes, ligands taken up at the sinusoidal membrane are, after interiorization, collected into specialized vesicles which allow them either to be transported to lysosomes (galactosylated serum albumin, hemoglobin-haptoglobin complex) or to cross the cytoplasm and be released into the bile (secretory IgA). Recent data suggest that the sorting between these ligands could take place in endocytic vesicles and/or in lysosomes.

Although our understanding of the endocytic process has been substantially increased during the past few years, many problems remain to be solved. In particular, the relative importance of endocytic vesicles and lysosomes in the detachment of ligands from their receptors and in the sorting of different ligands should be further investigated. REFERENCES Anderson, R . G . W., Brown, M. S . , Beisiegel, U . , and Goldstein, J . L. (1982). Surfacedistribution and recycling of the low density lipoprotein receptors as visualized with antireceptor antibodies. J . Cell Biol. 93, 523-531.

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

453

Ashwell, G., and Morell, A. G. (1974). The role of surface carbohydrates in the hepatic recognition and transport of circulating glycoproteins. Ad\’. Enzvmol. 41, 99- 128. Baenziger, J . U . , and Fiete, D. (1980). Galactose and N-acetyl-galactosamine specific endocytosis of glycopeptides by isolated rat hepatocytes. Cell 22, 61 1-620. Baenziger, I . U., and Fiete, D. (1982). Recycling of the hepatocyte asialoglycoprotein receptor does not require delivery of ligand to lysosomes. J . Biol. Chem. 257, 6007-6009. Basu. S. K., Goldstein, J. L., Anderson, R. G. W., and Brown, M. S. (1981). Monensin interrupts the recycling of low density lipoprotein receptors in human fibroblasts. Cell 24, 493-502. Bauniann, H.. and Doyle, D. (1980). Metabolic fate of cell surface glycoproteins during immunoglobulin-induced internalization. Cell 21, 897-907. Besterman, J . M., Airhart, J . A , , Woodworth, R. C., and Low, R. B. (1981). Exocytosis of pinocytosed fluid in cultured cells: Kinetic evidence for rapid turnover and compartmentation. J . Cell B i d . 91, 716-727. Birbeck, M. S. C., Cartwright, P., Hall, J . G., Orlans, E., and Peppard, J . (1979). The transport by hepatocytes of immunoglobulin A from blood to bile visualized by autoradiography and electron microscopy. Immimology 37, 477-484. Bothwell, T. H., Charlton, R. W., Cook, J . D., and Finch. C. A. (1979). “Iron Metabolism in Man,” pp. 576. Blackwell, Oxford. Brown, M. S . , and Goldstein, J. L. (1979). Receptor mediated endocytosis: Insights from the lipoprotein receptor system. Proc. Narl. Acud. Sci. U . S . A . 76, 3330-3337. Burgert. H. G., and Thilo. L. (1983). Internalization and recycling of plasma membrane glycoconjugates during pinocytosis in the macrophage cell line P388D. Exp. Cell Res. 144, 127- 142. Carpenter, G., and Cohen, S. (1976). 1251 labelled human epidermal growth factor. Binding, internalization and degradation in human fibroblasts. J . Cell Eiol. 71, 159-171. Chang, K. P., and Dwyer, D. M. (1978). Leishmania donovani. Hamster macrophage interactions in vitro: Cell entry, intracellular survival, and multiplication of amastigotes. J . Exp. M e d . 147, 515-530. Courtoy, P. J . , Limet. J . N . . Baudhuin, P . , Schneider, Y . - J . , and Vaerman, J . P. (1981). Receptormediated endocytosis of polymeric IgA by cultured rat hepatocytes. Cell Eiol. Int. Rep. 5, 57. Courtoy, P. J., Limet, J. N., Quintart, J . , Schneider, Y.-J., Vaerman, P. J . , and Baudhuin, P. (1982a). Transfer of IgA into rat bile. Ultrastructural demonstration. Ann. N . Y . Acad. Sci. 409,. 799-802. Courtoy, P. J., Quintart, J . , Limet, J . N . , De Roe. C., and Baudhuin. P. (3982b). Intracellular sorting of galactosylated proteins and polymeric IgA in rat hepatocytes. J . Cell Eiol. 95, 425a. Daems, W . T., and Van Rijsel, T. G . (1961). The fine structure of the peribiliary dense bodies in mouse liver tissue. J . Ultrastruct. Res. 5, 263-290. Debanne, M. T., and Regoeczi, E. (1981). Subcellular distribution of human asialotransferrin type 3 in the rat liver. J . Eiol. Chem. 256, 11266-1 1272. De Duve, C., and Wattiaux. R. (1966). Function of lysosomes. Annu. Rev. Physiol. 28, 435-492. De Duve. C., de Barsy, Th., Poole, B., Trouet, A., Tulkens, P., and Van Hoof, F. (1974). Lysosomotropic agents. Eiochem. Pharmacol. 23, 2495-253 1. Draper, R. F., and Simon, K. ( 1980). The entry ofdiphteria toxin into the mammalian cell cytoplasm: Evidence for lysosomal involvement. J . Cell Biol. 87, 329-339. Farquhar, M. G. (1978). Recovery of surface membrane in anterior pituitary cells. J . Cell Eiol. 78, R35-R42. Geuze. H. J . . Slot, J. N., Strous, G. J . A. M., Lodish, H. F., and Schwartz, A. L. (1983). Intracellular site of asialoglycoprotein receptor ligand uncoup1ing:Double label immuno-electron microscopy during receptor mediated endocytosis. Cell 32, 277-287. Goldstein, J . L., and Brown, M. S. (1977). The low density lipoprotein pathway and its relation to atherosclerosis. Annu. Rev. Biochem. 46, 897-930. Goldstein. J. L., Anderson, R. G . W., and Brown, M. S. (1979). Coated pits, coated vesicles and receptor mediated endocytosis. Nature (London) 279, 679-685.

454

YVES-JACQUES SCHNEIDER ET AL.

Gonzalez-Noriega, A., Grubb, J.. M., Taklab, U., and Sly, W. S. (1980). Chloroquine inhibits lysosomal enzyme pinocytosis and enhances lysosomal enzyme secretion by impairing receptor recycling. 1. Cell Biol. 85, 839-852. Gundelfinger, E. D., Krause, E., Melli, H., and Dobberstein, B. (1983). The organization of the 7SL-RNA in the signal recognition particle. Nucleic Acids Res. 11, 7363-7374. Harding, C., Heuser, J., and Stahl, P. (1982). Receptor mediated endocytosis of transfenin and recycling of the transferrin receptor in rat reticulocytes. J . Cell Biol. 97, 329-339. Harford, J., and Ashwell, G. (1982). The hepatic receptors for asialoglycoproteins. In “The Glycoconjugates” (M. I. Honvitz, ed.), Vol. 4. Academic Press, New York. Harford, J., Bridges, K., Ashwell, G., and Klausner, R. D. (1983). Intracellular dissociation of receptor bound asialoglycoproteins in cultured hepatocytes. J . Biol. Chem. 258, 3 191-3197. Helenius, A,, Kartenbeck, J., Simons, K., and Fries, E. (1980). On the entry of Semliki Forest Virus into BHK 21 cells. J . Cell Biol. 85, 404-420. Hemmaplardh, D., and Morgan, E. W. (1977). The role of endocytosis in transferrin uptake by reticulocytes and bone marrow cells. Br. J . Haematol. 36, 85-96. Herzog, V., and Farquar, M. G. (1977). Luminal membranes retrieved after exocytosis reaches most golgi cistemae in secretory cells. Proc. Nafl. Acad. Sci. U.S.A. 74, 5073-5077. Heuser, J. E. (1978). Synaptic vesicle exocytosis and recycling during transmitter discharges from the neuromuscular fraction. Dahlem Konf., pp. 445-464. Hubbard, A. L., Wilson, G., Ashwell, G . , and Stuhenbrok, H. (1979). An electron microscope autoradiographic study of the carbohydrate recognition systems in rat liver. I . Distribution of 1251ligands among the liver cell types. J. Cell Biol. 83, 47-64. Jackson, G. D. F., Lemaitre-Coelho, I., Vaerman, J . P., Bazin, H., and Beckers, A. (1978). Rapid disappearance from serum of intravenously injected rat myeloma IgA and its secretion into bile. Eur. J . Immunol. 8, 123. Jacques, P. J. (1969). Endocytosis. In “Lysosomes in Biology and Pathology” (J. T. Dingle and H. B. Fell, eds.), Vol. 2, pp. 395-420. North Holland Publ. Co., Amsterdam. Karin, M., and Mintz, B. (1981). Receptor-mediated endocytosis of transferrin in developmentally totipotent mouse teratocarcinoma stem cells. J. Biol. Chem. 256, 3245-3252. Katz, J. H. (1961). Iron and protein kinetics studied by means of doubly labelled human crystalline transferrin. J . Clin. Invest. 40, 2143-2152. Kino, K., Tusnoo, H., Higa, Y.,Takamo, M., Hamaguchi, H., and Nahajima, H. (1980). Hemoglobin-haptoglobin receptor in rat liver plasma membrane. J . Biol. Chem. 255, 9616-9620. Klausner, R. D., Ashwell, G., van Renswoude, J., Harford, J . B., and Bridges, K. R. (1983). Binding of apotransferrin to K562 cells: Explanation of the transferrin cycle. Proc. Narl. Acad. Sci. U.S.A. 80, 2263-2266. Kraehenbuhl, 1. P., and Kiihn, L. (1978). Transport of immunoglobulins across epithelia. Dahlem Konf. pp. 213-228. Kiihn, L. C., and Kraehenbuhl, J . P. (1979). Role of secretory component, a secreted glycoprotein, in the specific uptake of IgA dimer by epithelial cells. J . Biol. Chem. 254, 11072-1 1081. Lemaitre-Coelho, I., Jackson, G. D. F., and Vaerman, J. P. (1977). Rat bile, a convenient source of secretory IgA and free secretory component. Eur. J. Immunol. 7, 588-590. Lemaitre-Coehlho, I . , Meykens, R., and Vaerman, J . P. (1981). Anti-receptor antibodies: A comparison in the rat of plasma to bile transfer of purified IgG, F(ab’)2 and Fab‘ antibodies against rat secretory component (SC) Prof. B i d . Fluids 29, 419-422. Limet, J . N., Schneider, Y.-J., Vaerman, J. P., and Trouet, A. (1980). Interaction of rat IgA with cultured rat hepatocytes: Binding sites, drug effects. Toxicology 18, 187- 194. Limet, J. N., Schneider, Y.-J., Vaerman, J. P.,and Trouet, A. (1981). Receptor mediated endocytosis of polymeric IgA by cultured rat hepatocytes. Prof. Biol. Fluids 29, 423-426. (1982a). Receptor mediated Limet, J. N., Quintart, J . Otte-Slachmuylder, C., and Schneider, Y.-J.

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

455

endocytosis of hemoglobin-haptoglobin, galactosylated serum albumin and polymeric IgA by the liver. Ac/u Biol. Med. Germ. 41, 113-124. Limet, J . N., Schneider, Y.-J., Trouet, A., and Vaerman, J . P (1982b). Binding, uptake and processing of polymeric IgA by cultured rat hepatocytes. Curr. Top. Vet. Med. Anim. Sci. 12, 49-68. Limet, J. N., Schneider, Y.-I., Vaerman, J . P., and Trouet, A. ( 1 9 8 2 ~ ) Binding, . uptake and intracellular digestion and processing of polymeric rat IgA in cultured rat hepatocytes. Eur. J . Biochem. 125, 437-443. Marsh, M., and Helenius, A. (1980). Adsorptive endocytosis of Semliki forest virus. J . Mol. B i d . 142, 439-454. Marsh, M., Balzam, E., and Helenius, A. (1983). Penetration of Semliki Forest Virus from acidic prelysosomal vacuoles. Cell 32, 931 -940. Matlin, K. S., Reggis, H.. Helenius, A , . and Simons, K . (1981). Infectious entry pathway of influenza virus in a canine kidney cell line. J . Cell Biol. 91, 601-613. Maxfield, F. R. (1982). Weak bases and ionophores rapidly and reversibly raise the pH of endocytic vesicles in culture of mouse fibroblasts. J . Cell Biol. 95, 676-682. Miller, F.. and Palade, G . E. (1964). Lytic activities in renal protein. Absorption droplets. An electron microscopical cytocheniical study. J . Cell Biol. 23, 5 19-552. Mizijenski, J . G . , and Allen, R. P. (1974). Imniunotherapeutic suppression in transplantable solid turnours. Nufure (London) 250, 50-52. Morgan, E. H. (1981). Inhibition of reticulocyte iron uptake by NH4CL and CH3NH2. Biochim. Biophys. Acfa 642, 119-134. Muller. W. A , , Steinman, R. M., and Cohn, Z. A. (1980). The membrane proteins of the vacuolar system 11. Bidirectional flow between secondary lysosomes and plasma membrane. J . Cell B i d . 86, 304-314. Neufeld. E. F.. and Ashwell, G . (1980). Carbohydrate recognition systems for receptor mediated pinocytosis. In “The Biochemistry of Glycoproteins and Proteoglycans” (W. J . Lennarz, ed.), pp. 241-266. Plenum, New York. Newman, R . , Schneider, C., Sunderland, R . , Vadinelick, L . , and Greaves, M. (1982).The transferrin receptor. Trends Biochem. Sci. 7 , 397-400. Nogueira, N., and Cohn, Z. A. (1976). Trypanosoma cmzi: Mechanism of entry and intracellular fate in mammalian cells. J . Exp. Med. 143, 1402-1420. Octave, J . N., Schneider, Y.-J., Hoffmann, P., Trouet, A , , and Crichton, R. R. (1979). Transferrin and iron uptake by cultured rat fibroblasts. FEBS Left. 108, 127- 130. Octave. J . N., Schneider, Y.-J., Crichton, R. R., and Trouet, A. (1981a). Transferrin uptake by cultured rat embryo fibroblasts. The influence of temperature and incubation time; subcellular distributions and rapid kinetic studies. Eur. J . Biochem. 115, 61 1-618. Octave, I. N., Schneider, Y.-J., Crichton, R. R., and Trouet. A. (1981b). Evidence forendocytosis of iron loaded transfemn by cultured rat embryo fibroblasts. Prof. B i d . Fluids 29, 447-450. Octave, J . N., Schneider, Y.-J., Crichton, R. R., and Trouet, A. (1982a). Transfenin protein and iron uptake by isolated rat erythroblasts. FEES Leu. 137, 119-123. Octave, J . N . , Schneider, Y . - J . , Hoffmann, P., Trouet, A , , and Crichton, R . R. (1982b). Transferrin uptake by cultured rat embryo fibroblasts. The influence of lysosomotropic agents, iron chelators and colchicine on the uptake of iron and transferrin. Eur. J . Biochem. 123, 235-240. Octave, J . N., Schneider, Y.-J., Trouet, A , , and Crichton, R. R. (1983). Iron uptake and reutilization by mammalian cells. 1. Cellular uptake of transferrin and iron. Trends Biochem. Sci. 8, 2 17-220. Ohkuma, S., and Poole, B. (1978). Fluorescence probe measurement of the intralysosomal pH in living cells and the perturbation of pH by various agents. Proc. Nufl. Acud. Sci. U.S.A. 75, 3327-333 1 .

456

YVES-JACQUES SCHNEIDER ET AL.

Ohkuma, S., and Poole, B. (1981). Cytoplasmic vacuolation of mouse peritoneal macrophages and the uptake into lysosomes of weakly basic substances. J. Cell Biol. 90, 656-664. Ohkuma, S., Moriyama, Y . , and Takano, T. (1982). Identification and characterization of a proton pump on lysosomes by fluorescein isothiocyanate-dextran fluorescence. Proc. Natl. Acud. Sci. U.S.A. 79, 2758-2762. Orlans, E., Peppard, J . , Reynolds, J . , and Hall, J . (1978). Rapid active transport of immunoglobulin A from blood to bile. J . Exp. Med. 147, 588-592. Ottosen, P. D., Courtoy, P. J . , and Farquhar, M. G. (1980). Pathways followed by membrane recovered from the surface of plasma cells and myeloma cells. J. Exp. Med. 152, 1-19. Pastan, I. H., and Willingham, M. C. (1981). Journey to the center of the cell: Role of the receptosome. Science 214, 504-509. Pearse, B. (1980). Coated vesicles. Trends Eiochem. Sci. 5, 131-134. Poole, B., and Ohkuma, S. (1981). Effect of weak bases on the intralysosomal pH in mouse peritoneal macrophages. J. Cell Biol. 90, 665-669. Quintart, J . , Limet, J . N., and Baudhuin, P. (1981). Receptor mediated endocytosis of glycosylated derivatives of bovine serum albumin: Targeting based on sugar recognition. P ror. Eiol. Fluids 29, 389-392. Regoeczi, E., Chindemi, P. A,, Debanne, M. T., and Charlwood, P. A. (1982). Partial resialylation of human asialotransferrin type 3 in the rat. Proc. Natl. Acud. Sci. U.S.A. 79, 2226-2230. Sandvig, K., and Olsnes, S . (1980). Rapid entry of nicked diphteria toxin into cells at low pH. J . Cell Eiol. 87, 828-832. Schneider, Y.-J., and Trouet, A. (1981). Effect of chloroquine and methylamine on endocytosis of fluorescein-labelled control IgG and of anti-(plasma membrane) IgG by cultured fibroblasts. Eur. J. Eiochem. 118, 32-38. Schneider, Y.-J., Tulkens, P., and Trouet, A. (1977). Recycling of fibroblast plasma membrane antigens internalized during endocytosis. Eiochem. SOC. Trans. 5, 1164-1 167. Schneider, Y.-J., Tulkens, P., and Trouet, A. (1978). Recycling of plasma membrane during endocytosis. Dahlem Konf. pp. 181-195. Schneider, Y.-J., Tulkens, P., de Duve, C . , and Trouet, A. (1979a). The fate of plasma membrane during endocytosis. I. Uptake and processing of non specific and anti-membrane and control immunoglobulins by cultured fibroblasts. J . Cell Eiol. 82, 449-465. Schneider, Y.-J., Tulkens, P., de Duve, C., and Trouet, A. (1979b). The fate of plasma membrane during endocytosis. 11. Evidence for recycling (shuttle) of plasma membrane constituents J. Cell Biol. 82, 466-474. Schneider, Y.-J., Octave, J . N., Limet, J. N., and Trouet, A. (1980). Functional relationship between cell surface and lysosomes during pinocytosis. Int. Cell Eiol., pp. 590-600. Schneider, Y.-J., de Duve, C., and Trouet, A. (1981a). Fate of plasma membrane during endocytosis. Ill. Evidence for incomplete breakdown of immunoglobulins in lysosomes of cultured fibroblasts. J. Cell Eiol. 88, 380-387. Schneider, Y.-J., Octave, J . N . , Limet, J . N., and Trouet, A. (1981b). Functional relationship between cell surface and lysosomes during pinocytosis. Int. Cell Eiol. pp. 590-601. Schneider, Y.-J., Limet, J . N . , Octave, J . N., Slachmuylder-Otte, C., Crichton, R. R., and Trouet, A. (1982). The role of receptor-mediated endocytosis in iron metabolism. I n “Membranes in Growth and Development” (J. F. Hoffman, G. M. Giebisch, and D. Bolis, eds.), Vol. 91, pp. 495-521. Sibile, J. C., Octave, J . N., Schneider, Y.-J., Crichton, R. R., and Trouet, A. (1983). Transferrin protein and iron uptake by cultured rat hepatocytes. FEES Lett. 150, 365-369. Sibille, J . C., Octave, J . N., Schneider, Y.-J., Crichton, R. R., and Trouet, A. (1984). Transferrin protein and iron uptake by liver parenchymal cells. Subcellular distribution in rat livers. Arch. Int. Physiol. Eiochim. 92, 359-360.

10. ROLE OF ENDOCYTOSIS AND LYSOSOMES

457

Silverstein, S . C . , Steinman, R. M.. and Cohn, Z. A. (1977). Endocytosis. Annu. Rev. Biochem. 46, 669-722. Silverstein, S . C., Michl, J., and Sung, S . S . J. (1978). Phagocytosis. Dahlem Konf. pp. 245-264. Silverstein, S . C., Michl, J . , and Loihre, J . D. (1981). Studies on the mechanism of phagocytosis. Int. Cell B i d . pp. 604-612. Simionescu, N. ( 1981). Transcytosis and traffic of membrane in the endothelial cell. In!. Cell Biol. pp. 657-672. Sly, W. S . (1982). The uptake and transport of lysosonial enzymes. Glycoconjugates IV, 3-25. Socken, S. J., Jeejeebhoy. K. W., Bazin, H., and Underdown, B. J. (1979). Identification of secretory component as an IgA receptor on rat hepatocytes. J . Exp. Med. 150, 1538-1548. Stahl. P. P., and Schlessinger, P. H. (1980). Receptor mediated pinocytosis of mannoselN acetyl glucosamine terminated glycoproteins and lysosomal enzymes by macrophages. Trends Biochem. Sci. 5 , 194-196. Steinman, R. M., Brodie. S . E., and Cohn, Z. A. (1976). Membrane flow during pinocytosis: A stereologic analysis. J . Cell B i d . 68, 665-687. Steinman, R. M., Silver, I. M., and Cohn, Z. A. (1978). Fluid phase pinocytosis. Dahlem Konf. pp. 167- 180. Sullivan, A. L., Crasso, J. A,, and Weintraub, L. R. (1976). Micropinocytosis of transferrin by developing red cells. An electronmicroscopy study utilizing femtin conjugated transfenin and femtin conjugated antibodies to transferrin. Blood 47, 133-143. Tanabe, J . , Pricer, W. F., and Ashwell, G . (1979). Subcellular membrane topology and turnoverof a rat hepatic binding protein specific for asialoglycoproteins. J . B i d . Chem. 254, 1038- 1043. Tietze. C . , Schlessinger, P., and Stahl, P. (1980). Chloroquine and ammonium ion inhibit receptor mediated endocytosis of mannose glycoconjugates by macrophages: Apparent inhibition of receptor recycling. Biochem. Biophys. Res. Commun. 93, 1-8. Tietze, C . , Schlessinger, P., and Stahl, P. ( I 982). Mannose specific endocytosis receptor of alveolar macrophages: Demonstration of two functionally distinct intracellular pools of receptor and their roles in receptor recycling. J . Cell B i d . 92, 417-424. Tolieshaug, H., and Berg, T. (1979). Chloroquine reduces the number of asialo-glycoprotein receptors in the hepatocyte plasma membrane. Biochem. Pharmacol. 28, 2919-2922. Trouet, A. (1978). Increased selectivity of drugs by linking to carriers. Eur. J . Cancer 14, 105- I I 1. Trouet, A , , Deprez-De Campeneere, D., and de Duve, C. (1972). Chemotherapy through lysosomes with a DNA-daunorubicin complex. Nature (London) New B i d . 239, 110- 112. Tsan, M. F.. and Berlin, R. B. (1971). Effect of phagocytosis on membrane transport of non electrolytes. J . Exp. Med. 134, 1016-1035. Tulkens, P., Schneider, Y.-J., and Trouet, A. (1977a). Studies on the mechanism of endocytosis in fibroblasts with the use of specific antibodies. In “Intracellular Protein Catabolism 11” (V. Turk and N. Marks, eds.), pp. 73-84. Stefan Institute, Ljubljana and Plenum, New York. Tulkens, P., Schneider, Y.-J., and Trouet, A. (1977b). The fate of plasma membrane during endocytosis. Biochem. Soc. Trans. 5 , 1809-1815. Tulkens, P.. Schneider, Y.-J., and Trouet, A. (1977~).Flow and shuttle of plasma membrane during endocytosis. Acta Biol. Med. Germ. 36, 1681-1690. Tulkens, P., Schneider, Y.-J., and Trouet, A. (1978). A shuttle between cell surface and lysosomes during endocytosis. In “Protein Turnover and Lysosome Function” (H. Segal and D. Doyle, eds.), pp. 719-738. Academic Press, New York. Tulkens, P., Schneider. Y.-J., and Trouet, A. (1980). Membrane recycling (shuttle?) in endocytosis. In “Mononuclear Phagocytes: Functional Aspects” (R. M. Van Furth, ed.), pp. 613-647. Nijhoff, The Hague. Tycko, B., and Maxfield, F. R. (1982). Rapid acidification of endocytic vesicles containing alpha2macroglobulin. Cell 28, 643-651.

458

WES-JACQUES SCHNEIDER ET AL.

Van Beurs, B., and Nilausen, K. (1982). Pinocytosis in mouse L fibroblasts: Ultrastructural evidence for a direct membrane shuttle between the plasma membrane and the lysosomal compartment. J. Cell Biol. 94, 279-286. Van Renswoude, J., Bridges, K. R., Harford, J. B., and Klausner, R. D. (1982). Receptor-mediated endocytosis of transferrin and the uptake of Fe in K562 cells: Identification of a non-lysosomal acidic compartment. Proc. Nutl. Acad. Sci. U.S.A. 79, 6186-6190. Wall, D. A., Wilson, G., and Hubbard, A. L. (1980). The galactose specific recognition system of mammalian liver: The route of ligand internalization in rat haptocytes. Cell 21, 79-93. Warren, R., and Doyle, D. (1981). Turnover of the surface proteins and the receptor for serum asialo-glycoproteins in primary cultures of rat hepatocytes. J. Biol. Chem. 256, 1346-1355. Wibo, M., and Poole, B. (1974). Protein degradation in cultured cells. 11. The uptake of chloroquine by rat fibroblast and the inhibition of cellular protein degradation of cathepsin B. J. Cell B i d . 63, 430-440. Wehland, J., Willingham, M. C., Gallo, M. C., and Pastan, 1. (1982). The morphologic pathway of exocytosis of the vesicular stomatitis virus of protein in culture of fibroblasts. Cell 28, 831-841. Williams, J. (1982). Evolution of transferrin. Trends Biochem. Sci. 7, 394-397. Willingham, M. C., Pastan, I. (1980). The receptosome: An intermediate organelle of receptor mediated endocytosis in cultured fibroblasts. Cell 21, 67-77.

CURRENT TOPICS IN MEMBRANES AND TRANSPORT, VOLUME 24

Chapter 7 1 Regulation of Glucose Transporter and Hormone Receptor Cycling by Insulin in the Rat Adipose Cell I A N A . SIMPSON AND SAMUEL W . CUSHMAN Experimental Diabetes, Metabolism and Nutrition Section Molecular. Cellular and Nutritional Endocrinology Branch National Institute of Arthritis. Diabetes, and Digestive und Kidney Diseases National Institutes of Health Bethesda, Maryland

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ......................................... 11. Glucose Transport. . . . . . . . . . A. Stimulation by Insulin in t Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

459 460 460 B. Stimulation by Insulin in Subcellular Membrane Fractions . . . . . . . . . . . . . . . . . . 462 C. Modulation by Hormones Other Than Insulin.. . . . . . . . . . . . . . . . . . . . . . . . . . . . 474 D. Chronic Regulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Structure of the Glucose Transporter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482 F. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ill. Insulin-Like Growth Factor 11 Binding . . . . . . . . . . . . . . . . . . IV. Insulin Binding., . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488 A. Structure of the Insulin Receptor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488 B . Internalization of Insulin.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488 C. Internalization of the Insulin Receptor., . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 489 D. Relationship to Insulin Action.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494 495 V Summary.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496

1.

INTRODUCTION

The ability of insulin to stimulate glucose metabolism in rat adipose tissue was first observed by Winegrad and Renold (1958). Using the same tissue preparation, Vaughan (1961) and subsequently Crofford and Renold (1965a,b) then indirectly demonstrated that this stimulatory effect of insulin occurred at the 459 ISBN 0-12-153324-7

460

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

glucose transport level and that the stimulation of glucose transport by insulin was characterized by a change in the maximum rate of transport (V,,) and not by a change in the apparent affinity (K,) of the transporter for glucose. Not until well after the successful preparation of isolated rat adipose cells by Rodbell (1964), however, were these kinetic effects of insulin directly confirmed at the cellular level by Vinten et al. (1976). Nevertheless, kinetic studies alone were not able to distinguish whether the change in V,, was the result of an increase in the intrinsic activity of the individual transporters or the number of transporters capable of transporting glucose. To resolve this problem, our group at the National Institutes of Health recently developed a method for quantitating the number of glucose transporters in plasma membranes prepared from isolated rat adipose cells (Wardzala et al., 1978). This methodology is based on the specific binding of cytochalasin B, a compound previously shown to be a specific competitive inhibitor of glucose transport in both the human erythrocyte and the rat adipose cell. Using this compound, we demonstrated an increase in the number of glucose transporters in the plasma membranes prepared from cells treated with insulin compared to that in equivalent membranes prepared from control cells (Wardzala et al., 1978). We subsequently showed that these additional transporters were derived from a large intracellular pool of transporters (Cushman and Wardzala, 1980). Concurrently and independently, Suzuki and Kono (1980) at Vanderbilt University used completely different methodologies for both cell fractionation and the assessment of glucose transporter distribution and arrived at essentially the same conclusion. These initial observations led to the proposal of the “translocation hypothesis” for insulin’s stimulatory action on glucose transport, a proposal which will constitute the principal focus of this article. Two other proteins in the rat adipose cell have now been shown to change their subcellular distribution in response to insulin: the type I1 insulin-like growth factor (IGF-11) receptor and the insulin receptor itself. While less information is available on their translocations, some striking similarities exist between the movement of these two receptor proteins and the glucose transporter, which will be discussed at the end of this article.

II. GLUCOSE TRANSPORT

A. Stimulation by Insulin in Intact Cells A vast literature has been amassed on the transport of glucose and its analogs by the rat adipose cell such that certain aspects fall outside the scope of this article. For example, we refer readers to an excellent review by Gliemann and

46 1

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

Rees ( 1983) on methods for assessing the kinetics of glucose transport in this cell tY Pe. The direct measurement of glucose uptake into rat adipose cells is clearly complicated by the cell's ability to rapidly metabolize this transport substrate. This problem has been overcome by using nonmetabolizable glucose analogs such as L-arabinose, L-xylose, 2-deoxyglucose, and 3-0-methylglucose, of which the latter is now considered the analog of choice (Gliemann and Rees, 1983). Using a 3-0-methylglucose (3-OMG) uptake method to assess glucose transport activity (Karnieli et ul., 1981 b), Fig. 1 shows a typical time course for the ability of insulin to stimulate the rate of 3-OMG transport activity in the rat adipose cell. The onset of insulin's action is very rapid although lag times of up to 45 seconds at 37°C have been reported (Haring et ul., 1978; Ciaraldi and Olefsky, 1979). The maximum response, a 10- to 40-fold increase above the basal rate of transport, is achieved within 15 minutes, with a half-time of 2-3 minutes. Figure 2 demonstrates the insulin concentration dependence for the stimulation of 3-OMG transport. In contrast to insulin binding (Fig. 2, inset), which characteristically displays negative cooperativity , the insulin action curve is clearly positively cooperative, with 5 9 5 % of the stimulated activity expressed within concentrations of insulin ranging from 0.1 to 1.0 nM. Furthermore, the concentration of insulin which elicits a half-maximal response (0.3 nM) corresponds to a receptor occupancy of 10%. This discrepancy between the cell's binding capacity for insulin and the amount of insulin required for action was first observed by Kono and Barham (1971) and has given rise to the concept of spare receptors, one of the major enigmas in the study of insulin action. This is further discussed in Section IV,C, where the translocation of insulin receptors is described.

-

4 Time ( m i d

1.

Time course of the stimulation of glucose transport activity by insulin. From Karnieli et al. (1981b). FIG.

462

IAN A. SIMPSON AND SAMUEL W. GUSHMAN

4.0

3

.

5.5

3.0

Total Insulin (nM)

K

0.1

I

I

1

10

Insulin (nM1

FIG. 2. Insulin-concentration dependence of the stimulation of glucose transport activity. Inset: Insulin concentration dependence of insulin binding to rat adipose cells.

B. Stimulation by Insulin in Subcellular Membrane Fractions 1. PREPARATION OF SUBCELLULAR MEMBRANE FRACTIONS

A reproducible fractionation procedure is a prerequisite for demonstrating changes in the subcellular distribution of a particular membrane protein. In our approach (Simpson et al., 1983b), the subcellular membrane fractions are obtained by differential ultracentrifugation using a modification of the method of McKeel and Jarett (1970), as is shown schematically in Fig. 3. Kono and coworkers (Kono et al., 1981, 1982), on the other hand, have used a linear sucrose gradient to separate the various subcellular membrane fractions. Neither methodology allows for complete separation of distinct organelles (Kono et al., 1982; Simpson et al., 1983b). Nevertheless, as shown in Fig. 4, the distribution of characteristic marker enzyme activities clearly indicates an enrichment of the plasma membrane fraction with plasma membranes, the high-density microsoma1 membrane fraction with endoplasmic reticulum, and the low-density microsomal membrane fraction with membranes of the Golgi apparatus. In addition, preincubation of intact cells with insulin has no effect on the distribution or yield of any of the marker enzymes studied using either fractionation methodology (Suzuki and Kono, 1980; Simpson et al., 1983b). Both methodologies appear to give comparable resolution of mitochondria, plasma membranes, and membranes of endoplasmic reticulum and Golgi, although the differential ultracentrifugation is perhaps better suited for large-scale preparations whereas the sucrose gradient gives better yields when fewer cells are used.

463

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

2. MEASUREMENT OF GLUCOSE TRANSPORTERS a. Cyrochalasin B Binding. Cytochalasin B, a mold metabolite, was initially shown to be a competitive inhibitor of the glucose transporter in human erythrocytes; it is by far the most potent in the cytochalasin family (Wardzala, 1979). However, like many cytochalasins, it also interacts specifically with other cellular proteins, particularly actin. This becomes very evident when a Scatchard plot of the binding of [3H]cytochalasin B to plasma membranes is performed (Kasahara and Hinkle, 1976; Wardzala et al., 1978). Computer analysis of the curve reveals at least three distinguishable components, each with differing

Isolated Adipose Cells

Incubate in KREH Wash 2 x in TES Homogenize Homogenate 15 min-16,000 gmar

Supernatant Resuspend in TES 20 min-16,000 gmax

1 15 min-48.000 gma,

PI ?t

PC

Resuspend in TES 65 rnin-101,000 gmox (1.12 M sucrose cushion)

t

Supernatant

Resuspend in TES 30 min-48.000 gmar

75 min-300.000 gmax

P let

Resuspend in TES 60 min-365,000 gmax

I F

Interface

Resuspend in TES 30 min-100.000 gmax

It

Resuspend in TES 15 min-16,000 gmax

Pellet

Resuspend in TES 30 rnin-48.000 gmax

Plasma Membranes (PM)

!t MitochondrialNuclei IM/Nl

Pl t High- isity Microsomes (HDM)

P let Low ensity Microsomes (LDM)

FIG. 3. Subcellular fractionation procedure. KRBH, Krebs-Ringer bicarbonate/HEPES buffer; TES. Tris/EDTA/sucrose buffer.

464

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

OL PLASMA HIGH-DENSITY LOW-DENSITY MEMBRANES MICROSOMES MICROSOMES

FIG.4. Distribution of marker enzyme activities among membrane fractions. 5’-Nucleotidase is a marker enzyme for plasma membranes; rotenone-insensitive NADH-cytochrome c reductase is a marker enzyme for membranes of the endoplasmic reticulum; UDP-ga1actose:N-acetylglucosamine galactosyltransferase is a marker enzyme for membranes of the Golgi apparatus.

affinities and capacities. To resolve the binding component representing the specific binding to the glucose transporter, cytochalasin E must be added. Cytochalasin E (2 p M ) does not inhibit glucose transport activity in either intact cells or isolated plasma membranes (Kasahara and Hinkle, 1976), but does dramatically reduce the binding of [3H]cytochalasin B to proteins other than the glucose transporter. Figure 5A demonstrates the Scatchard plots obtained when equilibrium binding of [3H]cytochalasin B to plasma membranes prepared from basal or insulintreated cells is performed in the presence of cytochalasin E and in the presence or absence of D-glucose (500 mM). The Scatchard plots are still complex, although more binding in the absence of D-glucose is observed in the plasma membranes from insulin-treated cells. However, in the presence of D-glucose, binding in the plasma membranes from basal and insulin-treated cells is reduced to the same level. By subtraction of the respective curves obtained in the presence of Dglucose from those obtained in the absence of D-glucose along radial axes of constant free cytochalasin B concentration, the so-called derived Scatchard plots shown in Fig. 5B are observed. Here, the binding of [3H]cytochalasin B (Dglucose-inhibitable) comprises a single component with the same dissociation constant (-100 nM) in the plasma membranes from both basal and insulinstimulated cells. The binding capacity, however, is increased from -4 pmol/mg in the plasma membranes from basal cells to -25 pmol/mg in those from insulinstimulated cells. This same assay procedure has been applied to measuring the

465

1 1 . REGULATION BY INSULIN IN RAT ADIPOSE CELL

concentration of glucose transporters in the other subcellular fractions described above, with only the centrifugation conditions adjusted to ensure recovery of the different membrane species (Simpson et al., 1983b). At the time when this assay was initially performed, the evidence that this increased cytochalasin B binding actually represented an increase in the concentration of glucose transporters was based on the following criteria: (1) the increased concentration of transporters detected by cytochalasin B binding in the plasma membranes from insulin-stimulated cells compared with that in basal cells was clearly correlated with increased glucose transport activity measured directly in the same membrane preparations; (2) the Kifor inhibition of glucose transport by cytochalasin B in isolated plasma membranes closely corresponded to the Kd for cytochalasin B binding (- 100 IN) and ;(3) the sugar specificity for the inhibition of cytochalasin B binding was comparable to that seen for the transport of the same sugars into the intact cell. Similarly, the inhibitory effects of different cytochalasins on the specific binding of cytochalasin B corresponded to their inhibitory effects on transport. b. Reconstitution of Glucose Transport Activiry. The most compelling evidence that the changes in cytochalasin B binding corresponded to changes in the distribution of glucose transporters was provided, however, by Kono and coworkers (Suzuki and Kono, 1980; Kono et al., 1981, 1982), who directly measured changes in the glucose transport activity of reconstituted liposomes. Kasahara and Hinkle (1976) were first to use a reconstitution technique to

=B

INSULIN -D.GLUCOSE

0.20

0.16

\ -

'.

Ei

\ '0

\

INSULIN

0.12 -

fA

BASAL W w

=

:

0.10

0.08

2

3

8

\

-

I

\\ BASAL

\Q

0.05

0.00

0

2 0 4 0

60

NS

BOUND CYTOCHAIASIN B (pmol/mg protein)

FIG. 5 . (A) Scatchard and (B) derived Scatchard plots of cytochalasin B binding to plasma and insulin-stimulated (---) cells. Derived Scatchard plots are membranes prepared from Basal (---) obtained from Scatchard plots by subtracting the curves obtained in the presence of 500 mM D-glucose from the respective curves obtained in the absence of D-glucose (see text).

466

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

investigate the human erythrocyte glucose transporter. Shanahan and Czech (1977) subsequently applied this methodology to the glucose transporter in the adipose cell. The technique requires the solubilization of membrane with a detergent such as deoxycholate or (3-octylglucosideand the subsequent replacement of the detergent with phospholipidsderived from an egg lecithin preparation. These mixtures are then repeatedly sonicated and freeze-thawed until the liposomes are of sufficiently large size (volume) to accurately measure the accumulation of labeled glucose. To ensure reproducibility, a precise procedure has been carefully documented by Robinson et al. (1982). A variation of this methodology has been developed by Baldwin et al. (198 1); it depends on ensuring the reconstitution of less than one transporter unit per liposome. By achieving this, the number of transporters can be more readily determined by the proportion of liposomes which rapidly equilibrate with labeled glucose due to the presence of a transporter. This end-point assay is preferable to measuring small changes in the rate of transport. The recent results obtained by Gorga and Lienhard (1984) using the latter assay fully support those obtained using cytochalasin B binding. 3. SUBCELLULAR DISTRIBUTION OF GLUCOSE TRANSPORTERS The subcellular distribution of glucose transporters in membrane fractions prepared from basal and insulin-treated cells, as measured by cytochalasin B []ti] Cytochalasin 6 Dissociation Constant (nM) Basal Insulin

I 93k14.8

LDM MIN PM HDM FIG. 6. Distribution of glucose transporters among subcellular membrane fractions prepared from basal (open bars) and insulin-stimulated (closed bars) cells. PM, Plasma membranes; HDM, high-density microsomes; LDM, low-density microsomes; M/N, mitochondria/nuclei; B.D., below detection.

467

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

0

1

I

I

I

I

20

40

60

80

100

CYTOCHALASIN B BINDING SITES (pmol/mg protein)

FIG.7. Comparison between reconstitutable o-glucose transport activity and the concentration of A,0) and insulin-stimulated glucose transporters in membrane fractions prepared from basal (0, (0,A,H)cells. (0, 0 )Plasma membranes; (A,A) high-density microsomes; (0, H)low-density microsomes. From Cushnian et a / . (1984).

binding, is shown in Fig. 6. In the membranes prepared from basal cells, the plasma membranes and high-density microsomes contain relatively few glucose transporters (-7 pmol/mg), with substantially more (-80 pmol/mg) being detected in the low-density microsomes, the so-called intracellular pool of glucose transporters. This distribution is dramatically changed in the membranes prepared from cells preincubated with a maximally stimulating concentration of insulin. As previously seen in Fig. 5 , the concentration of transporters detected in the plasma membranes is increased approximately fivefold. An apparent doubling of the concentration of transporters present in the high-density microsomal membrane fraction is also observed, whereas that in the low-density microsomes is decreased by -60%.

468

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

To confirm these results and further establish the parallelism between the cytochalasin B binding and reconstitution methods for assessing the distribution of glucose transporters, we have measured both parameters in the same set of membranes. The results are shown in Fig. 7. The correlation between the two methodologies is high even when the reconstitution is measured at different protein concentrations.

4. TIMECOURSEAND REVERSIBILITY OF INSULIN ACTION The data presented so far describe only the steady-state distribution of transporters in basal and maximally insulin-stimulated cells. It is equally important, however, to establish that the translocation mechanism is compatible with the rapidity with which insulin is known to stimulate glucose transport as seen in Fig. 1. Figure 8 illustrates the time course for the translocation process in response to insulin and for its reversal induced by anti-insulin antibody. The data indicate that the increase in the number of transporters seen in the plasma membranes following exposure of the cells to insulin is closely mirrored by a decrease in the number of transporters in the intracellular pool, with half-times for both events being -2 minutes. Similarly, the loss of transporters from the plasma membranes during the reversal of insulin action by anti-insulin antibody is again mirrored by the reappearance of transporters in the low-density microsomal

01 ' 0

"

' ''' ''I ' 5

10

"

' I ' J1''

15 " 30

" "

I '

"

35

"

40

'''

' 1

45

'

"

"' ' 50

' ' 1

80

"

' ' 1 70

TIME Iminj

FIG. 8. Time course of stimulation of glucose transporter translocation by insulin and its reversal by anti-insulin antiserum. (0)Plasma membranes; (0) low-density microsomes. From Karnieli et af. (1981b).

469

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

0

5

10

15 45

0

10

20

30

TIME (mid

FIG.9.

Comparison between (A) the time courses of the stimulation of glucose transport activity

(0)and the translocation of glucose transporters to the plasma membrane (0)by insulin and (B) the time courses of their reversal by anti-insulin antiserum. From Karnieli et al. (1981b).

-

fraction, with the half-time for both events being 10 minutes. A comparison of the changes in glucose transport activity, measured immediately prior to homogenization of the cells, and the distribution of glucose transporters subsequently measured in the isolated plasma membranes (Fig. 8) is illustrated in Fig. 9. For the stimulation by insulin, the appearance of glucose transporters seems to precede the appearance of glucose transport activity, a finding suggesting that transporters are associated with the plasma membranes for a finite time before their activity is expressed. In contrast, on reversing the stimulation with antiinsulin antibody, the loss of transporters from the plasma membranes directly corresponds to the decrease in transport activity, suggesting that the retranslocation of transporters may proceed by a different mechanism. 5 . STOICHIOMETRY

To determine if these changes in distribution represent a stoichiometric translocation of transporters from the pool of transporters in the low-density microsomes to the plasma membranes, the yields and recoveries of the transporters present in the different fractions must be taken into account. Unfortunately, several problems are encountered in such a calculation (Simpson er al., 1983b). First, neither the cytochalasin B binding nor reconstitution methods of assaying the number of glucose transporters is applicable for determining the number of transporters present in the initial homogenate. Thus, an overall recovery of

470

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

transporters is unattainable. The use of marker enzyme activities appears to provide a viable alternative for assessing plasma membrane recovery and, consequently, the recovery of glucose transporters associated with the plasma membranes. By using this approach, at least 60% of the transporters detected in the high-density microsomes can also be accounted for. However, the same approach cannot be applied to assessing the recovery of those transporters associated with the intracellular pool; although the low-density microsomes are enriched in both glucose transporters and galactosyltransferase activity, a comparison of their distributions (Figs. 4 and 6) reveals that these activities do not parallel each other in the other fractions. This suggests that the intracellular vesicles containing the transporters represent either a specialized subfraction of the Golgi or a unique membrane species. Thus, in the absence of a marker enzyme specific for the intracellular vesicles containing the glucose transporter and the inability to measure an overall recovery, a final balance sheet is at present not possible. A ramification of these problems was highlighted by Carter-Su and Czech (1980), who attempted to perform reconstitution experiments using “total” membranes prepared from basal and insulin-treated cells. They observed that greater glucose transporter activity was reconstituted per milligram of membrane protein in membranes from insulin-treated cells than from basal cells, and they therefore concluded that insulin induces an activation of the glucose transporter rather than a reciprocal translocation. These observations now, however, appear to be the result of the relatively poor recovery of the transporters present in the low-density microsomes, which require relatively high centrifugal forces (>200,000 g for 1 hour), compared with the essentially quantitative recovery of plasma membranes. Consequently, in the basal state, where the majority of transporters resides in the intracellular pool and relatively few are present in the other membrane fractions, the overall “total” recovery was less than that seen from membranes prepared from insulin-treated cells. HYFJOTHESIS 6. A MODELFOR THE TRANSLOCATION On the basis of the data so far presented, we proposed the model illustrated in Fig. 10 for how insulin stimulates glucose transport in the rat adipose cell (Karnieli et al., 1981b). The first two steps, the binding of insulin to its receptor (step 1) and the generation of a signal (step 2), still represent a black box. Several hypotheses have been put forward as to the nature of this signal, including generation of a small peptide (Lamer et al., 1979; Seals and Czech, 1980; Kiechle et al., 1981), a change in redox state (Czech, 1976a), changes in membrane fluidity (Armatruda and Finch, 1979; Pilch et al., 1980), phosphorylation of the receptor and/or other proteins (Kasuga er al., 1982), internalization of the receptor itself (Simpson and Hedo, 1984), and/or combinations of all of the

471

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

Intracellular Pool

FIG. 10. Schematic representation of a hypothetical mechanism of insulin’s stimulatory action on glucose transport. From Karnieli et a!. (198 1b).

above. We have, therefore, depicted the signal by a question mark and await the outcome of the intense investigations currently under way in several laboratories. In response to the signal, the intracellular vesicles become associated with the plasma membranes (steps 3-5). The depiction of this vesicle association and subsequent fusion with the plasma membrane as separate steps was initially based on the time-course data presented in Fig. 9 and by analogy with more firmly established secretory processes. However, as will be discussed later, further evidence for at least a two-step process in the association of the intracellular vesicles with the plasma membrane is now available. Following the fusion step (step 5 ) , the transporters become exposed to the extracellular medium and at this point become capable of transporting glucose (step 6) and bringing about the observed increase in the maximum transport velocity. Upon removal of the insulin (step 7), the transporters rapidly return to the intracellular pool (step 8) with no apparent processing. Since the publication of this model, several embellishments have been added. One of the most important is the energy requirement for the translocation process determined by Kono etal. (1981). These investigators demonstrated that both the

472

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

forward and reverse translocations of transporters may be blocked by agents such as KCN and NaN,, a reaction which can effectively freeze the transporters at any stage of the translocation process by reducing cellular ATP levels. Equally important were the observations that insulinomimetic agents as diverse as trypsin, H,O,, sodium vanadate, and p-chloromercuriphenylsulfonate, known to stimulate glucose transport in the adipose cell by mechanisms not requiring insulin binding, achieve this stimulation by inducing translocations of transporters comparable to that seen with insulin (Kono et al., 1982). This finding provides further support for the conviction that the translocation of glucose transporters is the primary mechanism for eliciting the stimulation of glucose transport.

7. MECHANISTIC CONSIDERATIONS A major problem in the translocation hypothesis is the disparity between the magnitude of insulin’s stimulatory action on glucose transport in intact cells, usually 10- to 40-fold above the basal rate of transport, and the degree of insulin stimulation measured in isolated plasma membranes either by reconstitution, in which only 3- to 5-fold increases are seen, or by the binding of cytochalasin B , where only 4-to 7-fold increases are observed. Part of this discrepancy is clearly due to the cross-contamination of the plasma membranes with membranes derived from the intracellular pool. As pointed out earlier, no marker enzyme activity is available to correct for this contamination. As little as 5% contamination would increase the number of transporters in the plasma membranes from basal cells by as much as twofold, thus reducing the apparent stimulation by insulin. Nevertheless, contamination would still not necessarily account for all of the discrepancy, and the possibility of an additional, direct activation of the transporter by insulin cannot be discounted. Yet another explanation for part of this discrepancy has recently been observed with interesting ramifications as to the actual site of insulin action. The existence of a “nonfunctional” form of the glucose transporter associated with the plasma membrane was initially proposed from the time-course studies of Karnieli et al. (1981b). In this form, glucose transporters are “bound” to, but not “fused” with, the plasma membrane (Fig. 10, step 4).More recently, in studies of the effects of Tris (a common buffering agent with known lysosomotrophic characteristics) and of low temperatures on insulin action, we have observed that a subcellular redistribution of glucose transporters to the plasma membrane can be induced in the absence of insulin by exposing the cells to either 40 mM Tris or incubation temperatures below 16°C (Simpson et al., 1982, 1983~).This apparent translocation can be detected, however, only by measurements of cytochalasin B binding and not by measurements of glucose transport activity in either the intact cell, isolated plasma membranes, or more importantly but unexpectedly,

473

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

reconstituted plasma membranes (unpublished observations). While the effect of Tris might be explained simply by a change in the fractionation characteristics of the membranes containing the intracellular pool, such an effect of low incubation temperatures seems unlikely. Thus, a state of the transporters exists which is clearly distinct from that of the transporters found in the plasma membranes from insulin-treated cells or in the intracellular pool from basal cells. However, in both cases, insulin is capable of eliciting an increase in transport activity, a finding suggesting that it acts at the level of the plasma membrane to convert inactive to active transporters (Fig. 10, step 5). Further insights into the individual steps in the translocation process are provided by investigating the time course of insulin action and its reversal at 16"C, as illustrated in Fig. 11A and B, respectively (Simpson et al., 1983~).In this study, we stimulated the cells with a saturating concentration of insulin (700 nM) at 16"C, arrested the stimulatory process with I mA4 KCN, and subsequently measured the rate of 3-OMG transport at 37°C. Even at 16"C, insulin is still capable of rapidly and fully stimulating glucose transport with a half-time only approximately twofold greater than that seen at 37°C. Indeed, insulin remains capable of fully stimulating 3-OMG transport at temperatures below 16°C; at these lower temperatures, however, the half-times to achieve full stimulation become significantly greater (Ezaki and Kono, 1982). These data should now be contrasted with those in Fig. 1 I B , which show the reversal of insulin-stimulated glucose transport at 16°C. In these experiments, collagenase was added to and left with the cells to remove bound insulin (Kono et al., 1982) and the same KCN technique was used to arrest translocation. The amount of insulin bound, expressed in Fig. 11B as loo%, represents the miniI

1 I

I

I

1

- 100

-80

g) C

C

2

,

M-

M U 1 8 0 8 0 1 0 0

TIME Iminl

TIME (minl

FIG. 1 I . Time courses at 16°C of (A) the stimulation of glucose transport activity by insulin and ( 8 ) its reversal (m) and the removal of bound insulin (A) by collagenase. From Simpson er a/. ( 1984b).

474

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

mum required to achieve a fully stimulated response. By this technique, the decrease in insulin binding (t,,2, -30 minutes) can clearly be distinguished from the reversal of glucose transport activity (t,,*, -4-5 hours). These data, therefore, provide more conclusive evidence for the distinction between the translocation process involved in the stimulation of transport activity (Fig. 10, steps 3-5) and the retranslocation of transporters from the plasma membrane to the intracellular pool involved in its reversal (Fig. 10, step 8). The latter appears to be more sensitive to lower temperatures than is the former (Simpson et al., 1984b).

C. Modulation by Hormones Other Than Insulin 1 . LIPOLYTICHORMONES

In the adipose cell, the counterregulatory action of insulin on catecholaminestimulated lipolysis has long been recognized. Only comparatively recently, however, have the counterregulatory effects of catecholamines on insulin action been reported (Taylor et al., 1976; Kashiwagi and Foley, 1982; Kashiwagi et al., 1983; Simpson et af., 1983a, 1984b; Kirsch et al., 1983). Now, ACTH (Taylor et al., 1976), glucagon (Green, 1983), and catecholamines have all been shown to inhibit basal and insulin-stimulated glucose transport in the rat adipose cell. However, these inhibitory actions are only manifested when extracellular adenosine is removed from the incubation medium with the enzyme adenosine deaminase. Indeed, in the absence of this enzyme, isoproterenol has been shown in several studies to actually stimulate basal glucose transport (Ludvigsen et af., 1980; Kashiwagi and Foley, 1982). More recently, further evidence from our own laboratory has suggested that the inhibitory actions of these classic adenylate cyclase stimulators are mediated through a CAMP-independent mechanism (Kuroda et al., 1984). Isoproterenol (200 nM) in combination with adenosine deaminase (1 U/ml) elicits a 70% inhibition of insulin-stimulated glucose transport activity. This inhibition is characterized by a decrease in the maximum transport velocity (VmaX)without altering the transporter’s apparent affinity (K,) for 3-OMG. In view of this observation, experiments were undertaken to ascertain whether this effect was due to an alteration in the extent of translocation of glucose transporters induced by insulin. Figure 12 compares the inhibitory effects of isoproterenol and adenosine deaminase in combination, and of adenosine deaminase alone on basal and insulin-stimulatedglucose transport, with their effects on the distribution of glucose transporters between the plasma membranes and low-density microsomes under the same conditions. Qualitatively, the reduced numbers of transporters in the plasma membranes prepared from basal and insulin-treated cells incubated with isoproterenol and adenosine deaminase appear to parallel the corresponding decreases in glucose transport. Quantitatively, however, the magnitude of the changes in the number of translocated transporters in both cases is

475

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL A

C

6(

5c

m m

81E

BASAL

INS

INS

20

1 INS

BASAL

INS

INS

INS

BASAL

INS

INS

INS

+

t

t

t

t

t

ADA

ADA

ADA

ADA

ADA

ADA

+

IS0

t

+

IS0

IS0

FIG. 12. Effects of isoproterenol (ISO) and adenosine deaminase (ADA) on (A) glucose transport activity and (B,C) the distribution of glucose transporters between the plasma membranes and lowdensity microsomes in insulin (INS)-stimulated cells.

significantly smaller than would be predicted from the degree of inhibition of glucose transport. These data would, therefore, suggest a regulatory mechanism which not only modulates the translocation process but also appears to either alter the intrinsic activity of the transporter or prevent the expression of transporters associated with the plasma membrane (Fig. 10, step 5) (Simpson et al., 1983a). The effects so far described are only partially mimicked by dibutyryl-CAMP in the absence of adenosine deaminase. Furthermore, other modulators of CAMP levels such as forskolin and phosphodiesterase inhibitors have been shown, not only to alter the maximum transport velocity, but also to directly interact with the glucose transporter, thereby causing a decrease in the apparent affinity (K,) of the transporter for glucose and its analogs (Kashiwagi et al., 1983). The interaction of adenosine with glucose transport activity is clearly a complicated one. In our studies, adenosine deaminase induces a 30% decrease in the V,,, of insulin-stimulated glucose transport and is a prerequisite for expression of the inhibition of transport by either glucagon or catecholamines. These observations suggest that adenosine itself appears to exert an insulin-like action. This concept is supported by studies with N6-phenylisopropyladenosine,an analog of adenosine that specifically interacts with the adenosine receptor but is not degraded by adenosine deaminase. This adenosine analog is capable of almost completely reversing the inhibition induced by either adenosine deaminase alone

476

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

or adenosine deaminase in combination with isoproterenol. Indeed, in the presence of adenosine deaminase alone, N6-phenylisopropyladenosine actually stimulates basal glucose transport. A further complexity in the role of adenosine was demonstrated by Green (1983) and confirmed in our laboratory. Green found that adenosine deaminase also induces a shift to the right in the dose-response curve for the stimulation of glucose transport by insulin. This effect may well be related to the isoproterenolinduced loss of insulin receptors recently observed by Pessin et al. (1983) and Lonnroth and Smith ( 1 983). Thus, both the signal generation and the response to insulin appear to be under counterregulatory control, a conclusion adding a new dimension to the already complex role of insulin in regulating adipose cell function. 2. GROWTHHORMONE The precise role of growth hormone in regulating carbohydrate metabolism remains to be defined. In vivo experiments in both rat and man have indicated that excess circulating growth hormone is associated with insulin resistance, while growth hormone deficiencies are accompanied by an increased sensitivity to insulin. In 1966, Goodman observed that incubation of adipose tissue from hypophysectomized rats with growth hormone elicited an increase in glucose oxidation which he subsequently ascribed to an increase in glucose transport capability. More recently, Schoenle et al. (1979a,b) looked specifically at glucose transport and observed that the basal rate of transport in adipose cells from hypophysectomized rats was increased almost to the insulin-stimulated level in cells from normal rats. Simultaneously, the ability of insulin to further augment this activity was lost. These investigators further observed that this enhanced basal activity and diminished responsiveness to insulin could be reversed by the administration of growth hormone. These data were interpreted to suggest that a growth hormone-dependent factor which is present in control cells suppresses transport in the basal state and that the inhibition of this suppressive factor by insulin results in transport stimulation. One problem with assessing the effect of growth hormone is the relative inability to mimic in vilro the conditions induced by hypophysectomy in vivo. For example, Maloff et al. (1980), using adipose tissue in primary culture, were able to demonstrate an inhibitory action of growth hormone on basal transport but not changes in insulin sensitivity or responsiveness. 3. GLUCOCORTICOIDS The interaction of glucocorticoids with adipose tissue and their effects on insulin sensitivity have long been recognized (Munck, 1962). A 1- to 2-hour

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

477

exposure to the glucocorticoid derivative dexamethasone has been shown by several laboratories to inhibit glucose utilization, an effect subsequently attributed to inhibition of basal glucose transport by Foley et ul. (1978). Czech and Fain (1972) and Olefsky (1975) further demonstrated that dexamethasone caused a diminished stimulatory effect of insulin on both glucose oxidation and transport. At present, no data are available on how the effects of glucocorticoids might alter either the levels or activity of the glucose transporters. The time lag before the expression of dexamethasone’s inhibitory action suggests a potential role of protein synthesis in the mediation of its action. It remains to be tested whether these effects represent a more acute manifestation of the loss of glucose transporters seen in certain pathophysiological situations, as described in the next section.

D. Chronic Regulation Various pathophysiological conditions in both rat and man are associated with alterations in the ability of insulin to stimulate glucose transport in the adipose cell. Most of the situations so far studied have been ones in which the response to insulin has been impaired. These include the high-fat-fed rat, the aged, obese rat, the streptozotocin-induced diabetic rat, and the starved rat. We shall describe in some detail the data obtained in the first two conditions. More recently, however, we have identified two conditions, the starved, refed rat, and the chronic hyperinsulinemic rat, in which the adipose cells show an exaggerated stimulation of glucose transport activity by insulin. 1. INSULIN RESISTANCE I N THE

HIGHFAT/LOWCARBOHYDRATE-FED RAT In this model of insulin resistance, rats were fed ad libitum from weaning on equicaloric diets containing either a high fat/low carbohydrate composition, 50:30% by calories, or a low fat/high carbohydrate composition, 9:71% by calories. The protein content of both diets was maintained at a constant 20% of total calories. The latter is essentially identical in calorie composition to the standard chow used throughout our studies. The rats maintained for 3 weeks on the respective diets ate the same quantities of calories, grew at identical rates, and at the time of sacrifice contained epididymal adipose cells of the same size. The ability of insulin to stimulate glucose transport activity in adipose cells from these animals is shown in Fig. 13A. A decrease of 25-30% is seen in basal 3-OMG transport in the cells prepared from the high-fat-fed rats as compared to those from the high-carbohydrate-fed rats, but this difference is not statistically significant (p > 0.05) over three experiments. A 5 1 % reduction in the maximally

478

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

C

-In 60

INSULIN:

-

+

HIGH CARBOHYDRATE

-

t

HIGH FAT

-

t

HIGH CARBOHYDRATE

-

t

HIGH FAT

-

+

HIGH CARBOHYDRATE

-

+

HIGH FAT

FIG. 13. Effects of dietary composition on glucose transport activity (A) and the distribution of glucose transporters between the plasma membrane (B) and low-density microsomes ( C ) in basal (-) and insulin-stimulated (+) cells. From Hissin ef a / . (1982b).

insulin-stimulated rate of 3-OMG transport in these cells, however, is very reproducible. Figure 13B and C illustrate the concentrations of glucose transporters, expressed in pmol/mg of membrane protein, in the plasma membranes and low-density microsomes, respectively, prepared from cells which have been incubated in the absence or presence of a maximally stimulating concentration of insulin. No changes are observed in either the yields of plasma membrane protein or the distributions of marker enzymes in either membrane fraction. However, the microsomal membrane protein yield and intracellular water space in the cells from the high-fat-fed rat are decreased by 21 and 28%, respectively. The concentrations of glucose transporters detected in the plasma membranes prepared from basal cells are essentially identical (Fig. 13B). In contrast, the increase in concentration of transporters in the plasma membranes from insulintreated cells from high-fat-fed rats is reduced by 53% as compared to the equivalent control membranes. In the low-density microsomes, on the other hand, reductions in the concentrations of glucose transporters of 48 and 37% are observed in the membranes prepared from basal and insulin-treated cells, respectively, from the high-fat-fed rats. Thus, the loss of insulin-stimulated transport activity in the intact cell with high fat feeding appears to be directly accounted for by the reduced concentration of transporters present in the plasma membrane. However, this decreased concentration does not appear to be due to an impairment of the translocation mechanism, but rather to the decrease in the concentration of glucose transporters in the intracellular pool in the basal state (Hissin et al., 1982b).

479

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

2. INSULIN RESISTANCEI N THE AGED, OBESE,MALERAT The relationship between basal and insulin-stimulated glucose transport activity and adipose cell size in the aging male rat model of obesity is illustrated in Fig. 14. In these experiments, we have used an L-arabinose uptake technique to assess glucose transport activity (Foley et al., 1978) and expressed the data either per cell (Fig. 14A) or per unit cellular surface area (Fig. 14B). On a per cell basis, insulin-stimulated glucose transport activity remains fairly constant with increasing cell size whereas basal transport activity progressively increases. However, on a per unit cellular surface area basis, the exact converse applies, namely, that basal transport activity appears to remain constant while the insulinstimulated activity progressively declines with increasing cell size. Irrespective of the mode in which the data are expressed, the incremental increase in glucose transport activity in response to insulin declines with increasing cell size. Similar results have been reported by Livingston and Lockwood (1974) and Foley et al. (1980). An entirely comparable relationship has also been observed for insulin's ability to stimulate glucose oxidation (Czech, 1976b; Olefsky, 1976). We have therefore investigated the effects of insulin on the subcellular distribution of glucose transporters at both extremes of cell size, from young, lean rats (-180 g) with a mean cell size of 0.08 pg of lipid/cell and from old, obese rats (-800 g) with a mean cell size of0.94 pg of lipid/cell (Hissin et al., 1982a). The basal and insulin-stimulated rates of 3-OMG transport per cell and per unit cellular surface area are shown for both sets of cells in Fig. 15A and B and the respective distributions of glucose transporters between the plasma membranes and low-density microsomes, expressed per milligram of membrane protein, in

B

40

-G

3.0

X

3"\

E

B I

:r

E ?

0

2.0

0) -

-2

10

I

-1

0.0

0.2

0.4

0.6

0.0 0.0

Adipose cell size

0.2

0.4

0.6

1pg lipidkell)

glucose transport FIG. 14. Effects of adipose cell size on basal ( 0 )and insulin-stimulated (0) activity expressed ( A ) per cell and (B) per unit of cellular surface area in the aging male rat model of obesity. From Hissin et a / . (1982a).

480

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

ZINSU~IN - + LEAN

- t

- t LEAN

OBESE

- + OBESE

60

0 INSUUN:-

+

LEAN

-

+

OBESE

- +

- +

LEAN

OBESE

FIG. 15. Glucose transport activity per cell (A) and per unit cell surface area (B) and the distribution of glucose transporters between the plasma membranes (C) and low-density microsomes (D)in basal and insulin-stimulated small and large cells from young, lean and aged, obese rats, respectively. From Hissin et al. (1982a).

Fig. 15C and D. Before discussing these data, it should be pointed out that despite clear differences in marker enzyme specific activities between the small and large cells, their distributions among the various subcellular fractions were not significantly different. In addition, the recoveries of marker enzyme activities in the various fractions from the starting homogenates were also not affected by cell size. Membrane protein, however, was recovered in each fraction in proportion to cellular surface area; thus, approximately 20 pg/cell and 13 pg/cell of plasma membrane and low-density microsomal protein, respectively, were recovered from the small cells and approximately 146 pg/cell and 67 pg/cell, respectively, from the large cells (Hissin et af., 1982a). The 3-OMG transport activity measured immediately prior to homogenization essentially confirms the results obtained using L-arabinose and shown previously (Fig. 14): basal glucose transport activity in the large cells is elevated per cell but constant per unit cellular surface area compared to the small cells, and the incremental increase in transport activity induced by insulin is markedly diminished. In addition, the concentrations of transporters in the plasma membranes from basal and insulin-treated large cells are the same, reflecting the lack of an effect of insulin on 3-OMG transport observed in the intact cell, and are com-

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

48 1

parable to that seen in the plasma membranes prepared from basal small cells. Similarly, no change is induced by insulin in the concentration of glucose transporters in the low-density microsomes from the large cells; however, the concentrations are dramatically decreased when compared to the equivalent membranes obtained from small cells. Several conclusions may be drawn from these observations: (1) The absence of changes in the concentrations of glucose transporters in the plasma membranes and low-density microsomes in response to insulin directly correlates with the lack of response to insulin observed in the intact cell. ( 2 ) When the concentrations of glucose transporters in the membrane fractions are corrected for the recoveries of membrane protein from the original homogenates, then the estimated total number of transporters/cell in the large cell is actually approximately 2.5-fold more than that in the small cell; the estimated number in the plasma membranes of the large cell in either the basal or insulin-stimulated state is approximately 3-fold larger than that seen in the fully insulin-stimulated small cell. The latter observations suggest that either the intrinsic activity of the plasma membrane transporters or their fusion with the plasma membrane in the large cell is also diminished. (3) The cause of the failure to observe an insulin-stimulated translocation of glucose transporters in the large adipose cell is still far from clear. The obese rat is generally hyperinsulinemic, and it has therefore been suggested that this condition might induce a permanent translocation which is not reversible as seen in the small cell. Alternatively, while the level of transporters present in the plasma membrane is maintained with enlargement of the adipose cell and expansion of the plasma membrane, the total number of intracellular transporters is not. Thus, the absence of a response to insulin might be due to a relative depletion of this intracellular pool similar to that seen in the high fat-fed rat.

3. ADDITIONAL ANIMAL MODELSOF INSULINRESISTANCE Two other conditions of insulin-resistant glucose transport examined in our laboratory, the streptozotocin-diabetic rat (Karnieli et al., 1981a) and the fasted rat (Kahn and Cushman, 19841, are also accompanied by a specific decrease in the size of the intracellular pool and a diminished translocation of glucose transporters to the plasma membrane in response to insulin. It is of interest that the levels of circulating insulin are diminished in three of the four situations we have studied, as a result of either altered diet or perturbed insulin secretion. A possible role for insulin itself in controlling the level of intracellular glucose transporters is thus suggested. Further support for this concept is provided by at least one of the two conditions we have examined in which a hyperresponsive glucose transport activity is observed, namely, the hyperinsulinemic rat (Kahn et al., 1984). In this condition, the total number of transporters per cell is clearly increased

482

IAN A. SIMPSON AND

SAMUEL W. CUSHMAN

although the proportion residing intracellularly, compared to the plasma membrane, remains comparable to the control situation. The adipose cell is a target tissue for a vast array of circulating hormones of which we have discussed only a few which specifically affect the ability of the cell to transport glucose. It is, therefore, not surprising that both acute regulation of the ability of the adipose cell to transport glucose and convert it into an alternative energy store and more chronic forms of regulation of glucose transport activity that may reflect the animal’s overall well-being are observed. Until recently, however, the latter were not so readily amenable to study in vitro. With the advent of primary culture systems (Smith, 1974; Maloff et al., 1980; Marshall, 1983; Simpson et al., 1984a) and model cell lines such as the mouse 3T3-Ll fatty fibroblast (Resh, 1982), new insights into the chronic regulation of glucose homeostasis in the whole animal using the isolated adipose cell as an in v i m model should be forthcoming.

E. Structure of the Glucose Transporter In contrast to the human erythrocyte glucose transporter, which has been successfully purified and characterized as a 45,000-Da glycoprotein possessing one cytochalasin B binding site per monomer (Kasahara and Hinkle, 1977; Gorga et al., 1979; Sogin and Hinkle, 1980; Baldwin and Lienhard, 1981), little information has been available until recently regarding the structure of the rat adipose cell glucose transporter. A major reason for this paucity of data is the low concentration of glucose transporters (-0.5%) in the plasma membranes from insulin-stimulated adipose cells, approximately one order of magnitude less than that found for erythrocytes where glucose transporters constitute -5% of the total membrane protein. A second problem confronting identification of the rat adipose cell glucose transporter has been the lack of availability of a suitable probe, either covalent or with sufficient affinity for the glucose transporter, to use in monitoring a purification procedure. However, two such techniques have very recently been developed. For the first technique, antisera to the purified human erythrocyte glucose transporter have been produced in rabbits which, under well-defined conditions, cross-react with the rat adipose glucose cell transporter (Wheeler et al., 1982; Lienhard et al., 1982). The affinity of these antisera for the rat adipose cell transporter appears to be 1/ 1000 of that for the erythrocyte transporter but has enabled detection by a Western blot technique of one, and possibly two, protein(s) of molecular weight 45,000-55,000 which fulfill(s) the criteria for the glucose transporter. An example of the use of this technique is shown in Fig. 16. These antisera have also been used to confirm the translocation of a specific protein from the low-density microsomal membranes to the plasma membrane in

-

483

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

INSULIN: -t SUBCELLULAR PLASMA FRACTION : MEMBRANES

-t

LOW-DENSITY MICROSOMES

FIG. 16. Cross-reactivity of an affinity-purified rabbit IgG prepared against the purified human erythrocyte glucose transporter with the plasma membranes and low-density microsomes from basal (-) and insulin-stimulated (+) cells, as assessed by the Western blot technique. From Cushman et al. (1984).

response to the exposure of the intact adipose cell to insulin (Fig. 16). Unfortunately, the antisera currently available cannot be used to directly immunoprecipitate functional rat adipose cell glucose transporters since they appear to recognize only a denatured form of the protein. However, they do provide a relatively simple method for monitoring more conventional purification methodology. The second approach, developed independently by Shanahan et al. (1982) and Czech and co-workers (Carter-Su et al., 1982), has been to photolyse the cytochalasin B-glucose transporter complex with ultraviolet radiation. This photolysis induces covalent binding of the [3H]cytochalasin B to the transporter, presumably by activating an amino acid in the protein. Using the same criteria as initially used for the equilibrium binding of cytochalasin B (Wardzala et a[., 1978), i.e., the ability to inhibit covalent binding with D-glucose in the absence of an effect of cytochalasin E, these investigators have identified two proteins of

484

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

molecular weight 45,000 and 54,000 whose disposition in the plasma membranes and low-density microsomes is altered in response to insulin stimulation of the intact cell in a manner consistent with the translocation hypothesis (Pessin et al., 1984). This approach appears to fulfill the long-standing need for an affinity label for the glucose transporter and should facilitate the further purification and characterization of the adipose cell glucose transporter. A variation of this procedure has been developed by Horuk et al. (1983a) in which [3H]cytochalasin B is attached to the glucose transporter by a cross-linking technique somewhat analogous to that used for cross-linking insulin to its receptor. In this case, however, a photoaffinity cross-linker has been used. The principal advantage of this technique, aside from a slightly greater incorporation of [3H]cytochalasin B , is the avoidance of the nonspecific cross-linking between proteins that can occur when high-intensity ultraviolet photolysis is used. Using this technique in combination with an anti-human erythrocyte glucose transporter antiserum, these investigators have resolved at least four cytochalasin B-binding proteins in the low-density microsomes that cross-react with the antiserum. These proteins have molecular weights in the 45,000-50,000 range and isoelectric points of pH 6.4, 5.4, 4.5, and 4.2. The reasons for this heterogeneity are presently being investigated, but the possibility that it may be the result of different levels of glycosylation would be compatible with the known chemistry of the erythrocyte transporter (Gorga et al., 1979). Alternatively, this heterogeneity could be the result of differential phosphorylation. The latter would be consistent with the effects of catecholamines on glucose transport activity described earlier. The availability of these labeling techniques now offers new opportunities, not only to purify the rat adipose cell glucose transporter and thus raise specific antibodies to this protein, but also to provide a molecular basis for the regulation of glucose transport.

F. Summary In the preceding pages, we have described the basis for the mechanism by which glucose transport is regulated in the isolated rat adipose cell. Since the initial observations by Kono and co-workers and ourselves, the “translocation hypothesis” has been substantiated in several different laboratories using a variety of techniques. It is therefore with some confidence that we propose that the translocation of glucose transporters from an intracellular location to the plasma membrane is the major mechanism by which insulin regulates glucose transport in this cell type. The ubiquity of this mechanism has been enhanced by comparable observations in both human (Cushman et al., 1982) and guinea pig (Horuk et al., 1983b) adipose cells and, perhaps more importantly, in rat muscle (Wardzala and Jeanrenaud, 1981, 1983). The latter clearly represents the most important

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

485

tissue (by mass) in which insulin exerts control of glucose utilization in the intact cell. The insulin-induced translocation of glucose transporters in the rat adipose cell has now provided an impetus to study the disposition of two other membrane proteins, the type I1 insulin-like growth factor (IGF-11) receptor and the insulin receptor itself. These two proteins will be discussed in the remainder of this article. Many aspects of our working model of this translocation process (Fig. 10) require both confirmation and clarification. For example, little is known regarding the intracellular disposition and composition of the vesicles containing the glucose transporter. Similarly, the mechanism of insulin's stirnulatory action still remains to be determined, as do the mechanisms through which insulin's action is modulated by counterregulatory hormones and a variety of pathophysiological conditions. Nevertheless, a basic mechanism has been described for the process through which insulin regulates the passage of glucose into a cell whose complexity, reflecting its central importance in mammalian physiology, continues to expand with further investigation.

111.

INSULIN-LIKE GROWTH FACTOR It BINDING

The presence of type I1 insulin-like growth factor (IGF-11) receptors in rat adipose cells was first demonstrated by Schoenle et al. (1976). To date, however, no specific function has been ascribed to these receptors in the rat adipose cell. Nevertheless, the IGF-I1 receptor has been characterized as a single polypeptide chain of molecular weight 260,000, distinct from the IGF-I receptor, which is not found in the rat adipose cell (Zapf et al., 1978; Kasuga et a(., 198 I ; Massague and Czech, 1982; Rechler et al., 1982). Insulin has been shown by several laboratories to stimulate the binding of tracer 1251-labeledIGF-I1 to rat adipose cells (Schoenle et al., 1977; King et al., 1980, 1982; Oppenheimer et al., 1983). More recently, this stimulatory effect of insulin has been attributed to a change in the affinity (K,)of the receptor for IGFI1 (King er al., 1982; Oppenheirner et al., 1983). In contrast to these observations, however, subcellular fractionation, as originally performed by Oppenheimer et al. (1983) and confirmed in Fig. 17 using our own fractionation procedure (Fig. 3), suggests that insulin induces a redistribution of IGF-I1 receptors from an intracelIular membrane pool to the plasma membrane in a manner entirely analogous to the movement of the glucose transporter. In basal cells, the distribution of IGF-I1 receptors, expressed per milligram of membrane protein, between the plasma membranes and low-density microsomes is 1:4.6; in insulinstimulated cells, this distribution is shifted to 1:2.4. The apparent Ka'sof the receptors in the two membrane fractions are similar (0.2 nM- I ) and unaffected by insulin.

486

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

These data on the subcellular distribution of the IGF-I1 receptor are clearly at variance with the results obtained when IGF-I1 binding to the intact cell is measured at steady state, as shown in Table I. In the basal state, IGF-I1 binding to the intact cell (Table I) appears to greatly exceed the number of receptors detected in the plasma membrane fraction (Fig. 17) when the latter is corrected for the recovery of membrane protein from the original homogenate (-90 pg/cell; see Table V in Simpson et al., 1983b). In addition, total binding to the cells, calculated from Scatchard analysis, appears unaltered in response to insulin whereas the K , of the receptor appears to be increased approximately twofold (Table I). These discrepancies led Oppenheimer et al. (1983) to conclude that the data obtained from the subfractionation studies were artifacts of the homogenization procedure. By implication, these investigators further suggested that similar phenomena might, in fact, be occurring in the case of the glucose transporter. We have now reexamined IGF-I1 binding to the intact rat adipose cell under conditions in which the potential recycling of IGF-I1 receptors is blocked by 1 mM KCN (Wardzala et a l . , 1984). This approach is similar to that developed by Kono et al. (1981) to study the movement of the glucose transporter. Following an initial preincubation with either 0 or 7.0 nM insulin, IGF-I1 binding is then measured in the presence of KCN to stop receptor recycling. Under these conditions, the amount of IGF-I1 bound is dramatically reduced, the apparent K , for IGF-I1 is increased to that observed in the isolated membrane fractions, and an increase of approximately fivefold in the number of IGF-I1 receptors is observed in response to insulin (Table I). Furthermore, when KCN is added prior to A 12,5

10.0

INSULIN: FIG.

-

+

B

r

i n -

+

17. Distribution of insulin-like growth factor I1 receptors between the plasma membranes

(A) and low-density microsomes (B) from basal (-) and insulin-stimulated (+) cells.

487

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

Treatment with KCN None State of cells

Ro

Basal Insulin-treated

0.73 0.69

(1

After insulin K'l

0.030 0.06X

Before insulin

Ro

K,

RO

K.1

0.02s 0.160

0.18 0.24

0.030 0.030

0.20 0.20

Ro, Receptor number (in attomoles per cell); K,, receptor affinity (in nM-

I).

stimulation by insulin, the response to insulin is completely inhibited. Thus, as seen for glucose transport, the stimulatory action of insulin on IGF-I1 binding is energy dependent. We have interpreted these results to indicate that IGF-I1 receptors of constant affinity recycle between the plasma membrane and an intracellular, low-density microsomal pool in both basal and insulin-treated cells. In both cases, IGF-I1 is internalized by the recycling receptors and accumulates within the cell, thus effectively masking the redistribution of receptors induced by insulin. Further support for this concept is provided by assessing the sensitivity of bound IGF-I1 to trypsin. When binding to either basal or insulin-stimulated cells is conducted in the absence of KCN, not only is more IGF-I1 found associated with the cell, but also greater than 75% of the bound tracer is insensitive to exposure to trypsin. Of the IGF-I1 bound to cells in the presence of KCN, on the other hand, greater than 60% is trypsin sensitive. The binding data obtained with the intact cell in the presence of KCN would, therefore, appear to be compatible with that obtained by subcellular fractionation, with the IGF-11 receptor continually recycling and insulin altering the steady-state distribution of receptors between the two subcellular locations. At present, because of the absence of a high-affinity or covalent probe for the transporter, it has not been possible to determine whether the glucose transporter undergoes a similar recycling. Nevertheless, the analogies between the actions of insulin to modulate the subcellular distributions of both of these proteins are striking, including similar time courses ( t l l Z 1-3 minutes) and insulin sensitivities (Karnieli ef d.,1981b; King et d.,1982). Thus, it is tempting to speculate that the mechanisms by which glucose transporter and IGF-I1 receptor translocation is initiated and achieved are clearly very similar even if these two proteins are not, in fact, localized to the same intracellular vesicles. The study of IGF-I1 receptor translocation may very well provide a valuable system for understanding the mechanism of insulin action. Such studies have already highlighted the need for caution when measuring equilibrium binding to mobile receptors. ^I

4aa

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

IV. INSULIN BINDING A. Structure of the Insulin Receptor The final protein to be discussed whose subcellular distribution is altered by incubation of the intact rat adipose cell with insulin is the insulin receptor itself. Considerably more information regarding the structure of this protein is available than for the previous two. The insulin receptor is believed to exist as a tetramer composed of two a subunits of molecular weight 135,000 and two @ subunits of molecular weight 95,000, held together by interchain disulfide bridges (Massague and Czech, 1980). Both subunits have been shown to be glycoproteins with distinct complex carbohydrate side chains (Cuatrecasas and Tell, 1973; Hedo et al., 1981a,b; Simpson et af., 1984a). Based on the cross-linking of Iz5Ilabeled insulin and photoaffinity insulin probes (Yip et al., 1978; Pilch and Czech, 1979, 1980; Brandenburg etal., 1980; Wang et ul., 1982, 1983; Berhanu et al., 1982), the a subunit appears to possess the binding site for insulin, but does not appear to completely traverse the plasma membrane (Hedo et al., 1982). More recently, the p subunit has been shown to possess a tyrosine kinase activity which is stimulated by insulin to phosphorylate both itself (autophosphorylation) and artificial substrates (Kasuga et al., 1982; Haring et al., 1982; Petruzzeli et a l . , 1982; Avruch et al., 1982; Roth and Cassels, 1983; Simpson and Hedo, 1984). This kinase activity and the transmembrane nature of this protein are compatible with the f3 subunit’s proposed function in signal transduction. These structural characteristics of the insulin receptor, if not initially observed in the adipose cell, have since been confirmed in this and several other cell types, and we refer readers to reviews now available in which a comprehensive description is given (Czech, 1980; Jacobs and Cuatrecasas, 1981; Kahn, 1983; Kahn et al., 1983). In contrast to the apparent strict conservation of insulin receptor structure among cell types, the subcellular distribution of receptors appears to vary widely in the relatively few cell types so far investigated. In the rat hepatocyte, for example, an intracellular pool of receptors has been detected comprising 3060% of the total (Posner et al., 1978, 1980; Desbuquois et al., 1982; Fehlmann et al., 1982). In the rat adipose cell in the absence of insulin, very few (1015%), if any, intracellular receptors have been detected (Sonne and Simpson, 1984; Marshall, 1983).

B. Internalization of Insulin In almost all of the cell types so far investigated, insulin has been shown to be internalized and degraded through receptor-mediated endocytosis. This was first

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

489

demonstrated in the adipose cell by Kono et al. (1975, 1977) and has since been confirmed in several other laboratories. Morphological studies using 1251-labeled insulin have been used in several cell types to chart the course of insulin internalization. However, the adipose cell poses special problems due to the cell’s unique architecture, comprising a large lipid droplet surrounded by a thin rim of cytoplasm. Consequently, relatively few electron microscopic studies have been performed using this particular cell type. Jarett and co-workers (Hammons and Jarett, 1980; Smith and Jarett, 1982a,b) have used monomeric ferritin-insulin as a probe and demonstrated its receptor-specific internalization via pinocytic, noncoated imaginations. Internalized ferritin-insulin first appears in noncoated cytoplasmic vesicles, then in multivesicular bodies, and ultimately in lysosomes. Some internalized insulin may return to the plasma membrane. The relatively nonspecific lysosomotrophic agent chloroquine caused an arrest of the passage of ferritin-insulin through the multivesicular bodies, dense bodies, and lysosomes, thereby causing accumulation of ferritin-insulin in these organelles (Smith and Jarett, 19R2b). This effect is believed to result from a change in the intravesicular pH of these organelles, a change which prevents proteolytic degradation of the ferritin-insulin complex. The action of chloroquine leading to the increase in cell-associated insulin has also been attributed to an alteration of the affinity of the receptor for insulin and to the accumulation of insulin in prelysosomal vesicles that cofractionate with elements of the Golgi (Marshall and Olefsky, 1980; Iwamoto et al., 1981; Posner et a l ., 1982). All of these studies have been confined to the cellular localization of insulin and provide little or no precise information as to the fate of the insulin receptor itself.

C. Internalization of the Insulin Receptor Unlike many cell types, chronic exposure of the rat adipose cell to insulin results in very little apparent loss of cell surface insulin receptors-so-called down-regulation (Gavin et al., 1974). However, these cells can be induced to respond in this manner if the incubation with insulin is performed in the presence of Tris, a common buffering agent with known lysosomotrophic effects. Using Tris to promote down-regulation, either a photoaffinity insulin analog or direct insulin binding to assay for receptors, and trypsin digestion to distinguish cell surface from internalized receptors, Olefsky and co-workers (Marshall and Olefsky, 1981; Berhanu et al., 1982; Green and Olefsky, 1982; Olefsky et al., 1983) have been able to demonstrate that a significant proportion of total adipose cell receptors is internalized in response to insulin. Furthermore, they have also demonstrated a relatively slow and progressive degradation of the photoaffinitylabeled a subunit complex and the inhibition of this degradation by chloroquine (Berhanu et al., 1982). However, all of these actions would appear to be at-

490

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

tributable to the presence of Tris since both Marshall and Olefsky (1981) and Rennie and Gliemann (198 1) have been unable to reproduce these observations when cell incubations are carried out in either phosphate or HEPES buffer in the absence of Tris. BY LIGANDBINDING 1. ASSESSMENT

Initially, our own approach to studying the subcellular localization of the insulin receptor was simply to measure directly the ability of subcellular membrane fractions (plasma membranes and high- and low-density microsomes) to bind tracer insulin, and to monitor alterations in the binding capacity of these membrane fractions with time of exposure of the intact cell to insulin (6 nM) (Sonne and Simpson, 1984). The results of such a study are illustrated in Fig. 18. In the absence of insulin (represented by the zero time point), the subcellular distribution of insulin binding corresponds very closely to those of plasma membrane marker enzyme activities such that the levels of receptor binding in the two intracellular membrane fractions are completely accounted for by plasma membrane contamination. These results suggest that few, if any, of the basal cell’s insulin receptors are intracellular. Following exposure to insulin, receptor binding in the plasma membranes rapidly decreases ( i l l * = 2-3 minutes), reaching an apparent steady-state decrease of -30% within 10 minutes. A parallel increase in receptor binding is observed in the low-density microsomal membrane fraction, thus accounting for approximately one-third of the number of receptors lost from the plasma membranes. However, to detect these internalized receptors in the low-density micro-

10 0

5

1

0

1°

5

2

0

2

5

3 V0

TIME (mid

FIG. 18. Time course of the effect of insulin treatment of intact cells on the distribution of insulin binding activity among subcellular membrane fractions. (O), Plasma membrane; (W) high-density microsomal; (A) low-density microsomal fractions. From Sonne and Simpson ( I 984).

49 1

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

1.

0.0 1.0 0

10

30

Time h i n J

M, = 138K

Time (min) Temperature

O 2 3 0 3 0 C o n

37

37

37

16

("C) FIG.19. Time course and temperature dependence of the effect of insulin treatment of intact cells on the covalent binding of I2'I-labeled Bz9-Napa-insulin to the 135K a subunit of the insulin receptor (M,= 138K band) in the low-density microsomes ( 0 . 3 7 " C ; 0, 16°C). Con indicates labeling in the presence of a high concentration of nonradioactive insulin. From Wang ei a / . (1983).

somes, it was necessary to expose the membrane vesicles to a low concentration of digitonin (0.01%)to render them permeable to insulin. The latter observation suggested that insulin receptors are inverted during the internalization process. These results have subsequently been confirmed by using the photoaffinity insulin analog B,,-Napa-insulin ( '2sI-labeled) in experiments in which both the time course of internalization (Fig. 19) and the action of digitonin on both the plasma membranes and low-density microsomes (Fig. 20) can be monitored directly by labeling the 135K (Y receptor subunit (Wang et af., 1983). We have used the same approach to demonstrate the insulin concentration dependence for the internalization process; internalization can thus be shown to parallel receptor occupancy (max,,, = 3 nM), but not the profile for insulin action (max, , = 0.1

492

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

Plasma Membranag

Low Density Microsomes

FIG. 20. Effects of digitonin on the covalent binding of 1251-labeledBzg-Napa-insulin to the 135K subunit of the insulin receptor (138Kband) in the plasma membranes from basal cells and lowdensity microsomes from insulin-treated cells. Con indicates labeling in the presence of a high concentration of nonradioactive insulin. From Wang et al. (1983).

nM). Figure 19 also illustrates the inhibitory effect of low incubation temperatures on this process.

2 . ASSESSMENT BY RECEPTOR LABELING Neither of the two approaches described above, however, could explain why only 30-40% of the total number of receptors lost from the plasma membranes were detected in the low-density microsomes and why no change in binding capacity was observed in the high-density microsomes, even when binding was carried out in the presence of digitonin. To resolve this problem, we adopted an approach originally developed for the study of insulin receptors in IM-9 lymphocytes in which both the a and p subunits of the insulin receptor are labeled with either tritiated sugars (Hedo et al., 1981a) or labeled amino acids (Van Obberghen ef al., 1981). In the example of this approach illustrated in Fig. 21, we have maintained adipose cells for 24 hours in primary culture in a medium containing [3H]glucosamine, exposed the cells to 0 or 700 nM insulin, fractionated the cells, solubilized the individual subfractions with 1 % Triton X-100, and quantitatively immunoprecipitated the insulin receptors with an anti-insulin receptor antiserum obtained from patients with severe insulin resistance and

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

493

acanthosis nigricans (Simpson e t a / ., 1984a). (For a review on the properties and applications of these antisera, see Van Obberghen and Kahn, 1981.) The immunoprecipitated proteins were then separated by polyacrylamide gel electrophoresis and visualized by fluorography. This technique revealed both the a and p subunits of the insulin receptor and a 30% decrease in both subunits in the plasma membranes from cells exposed to insulin. In addition, and also in agreement with the previous studies, an increase in the presence of both subunits is seen in the low-density microsomes which, by densitometric scanning, corresponds to 30-4070 of the receptors lost from the plasma membranes. However, in contrast to the other studies, an insulin-induced increase in both the a and p subunits is also seen in the high-density microsomes, the magnitude of which almost entirely accounts for the missing receptors from the plasma membranes. An explanation for the inability of the receptors appearing in the high-density rnicrosomes to bind insulin, even in the presence of digitonin, has not yet been found. These receptors do not appear to have been structurally modified since both receptor subunits seem to be identical in size to those seen in the other two fractions. One potential explanation has been suggested by the work of Harrison and co-workers (Saviolakis et al., 1981; Clark and Harrison, 1982, 1983), who have observed the existence of insulin covalently bound to the receptor via

FIG. 2 I . Effects of insulin treatment of intact cells on the distribution of specifically immunoprecipitable (Anti-R) insulin receptor subunits, labeled biosynthetically in primary culture with [3H]glucosamine, among subcellular membrane fractions.

494

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

disulfide linkages. In the rat adipose cell, covalent-bound insulin appears to account for 10-15% of the total cell receptors, essentially the same proportion as is found in the high-density microsomes following exposure of the cells to saturating insulin concentrations. We have since been able to confirm these biosynthetic labeling studies using a Na1251/lactoperoxidaselabeling technique to iodinate receptors either in the intact cell or in the isolated membrane fractions (Hedo and Simpson, 1984). This approach has also confirmed the inverted disposition of the insulin receptor in both of the internalized fractions (previously inferred from the use of digitonin) and demonstrated the transmembrane nature of the p subunit and the cell surface disposition of the (Y subunit. These data suggest, therefore, that the insulin receptor rapidly internalizes in response to insulin. In addition, the accumulation of more insulin in the cell in the presence of chloroquine than can be accounted for by a unidirectional internalization of insulin-receptor complexes further suggests that the receptor actually recycles (Sonne and Simpson, 1984). Thus, as previously demonstrated for IGF-11, the binding of insulin to intact cells is clearly complicated by its very rapid internalization and subsequent degradation. In most cases where insulin binding to the intact cell has been studied, this binding has been characterized by curvilinear Scatchard plots which have been interpreted to represent either negative cooperativity or the existence of two or more different insulin binding sites. Clearly, a third alternative must now be considered, namely, that bound insulin comprises not only ligand associated with cell surface receptors, but also internalized ligand and ligand-receptor complexes whose steady-state redistribution in response to insulin occurs well within the time required for total bound insulin to achieve its own apparent steady state.

D. Relationship to Insulin Action The rapidity with which the insulin receptor is internalized in the rat adipose cell parallels the stimulation of both glucose transport and the redistribution of IGF-I1 receptors by insulin. These observations suggest that the recycling of the insulin receptor may have more significance than simply the removal of surfacebound insulin. One possible ruison d’ktre is that the internalization process serves as a mechanism for terminating the signal generated by insulin. Alternatively, internalization could represent part of the actual signaling mechanism itself. The latter becomes particularly attractive when the disposition of the protein kinase activity is considered. Since internalization occurs through an endocytic mechanism, the inner portion of the (3 receptor subunit containing the kinase activity remains exposed to the cytoplasm during the internalization process. Thus, insulin would stimulate, not only the kinase activity itself, but also the translocation of this activated kinase from the plasma membrane into intra-

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

495

cellular regions of the cytoplasm containing organelles not usually exposed to plasma membrane enzyme activities. In a preliminary study, we have obtained evidence suggesting that the acute phosphorylation of the receptor induced by insulin is not essential per se for internalization of the receptor (Simpson and Hedo, 1984). Anti-insulin receptor antiserum B- 10 induces internalization of the receptor and stimulates glucose transport in the rat adipose cell but does not appear to affect the phosphorylation state of the p receptor subunit. Under the same experimental conditions, however, this antiserum does induce alterations in the phosphorylation state of several intracellular membrane proteins in a fashion directly paralleling the effects of insulin. A third possibility is that this receptor recycling mechanism serves as a system for delivering insulin to intracellular organelles such as the nucleus (Goldfine and Smith, 1976) or even to other cell types. The latter might be envisaged for the transcellular transport of insulin across capillary endothelia.

V.

SUMMARY

We have described here the effects of insulin on the subcellular disposition of three integral membrane proteins in the isolated rat adipose cell: the glucose transporter, the IGF-I1 receptor, and the insulin receptor itself. Of these, only the insulin receptor appears to conform to the established pathways of membrane protein movement, that of receptor-mediated endocytosis. Given the paucity of coated pits and vesicles in this cell type (Smith and Jarett, 1982a), however, some doubt exists as to whether even the latter mechanism is entirely conventional. The glucose transporter and IGF-I1 receptor appear to represent a new class of integral membrane proteins [for review see Lienhard ( 1983)] whose intracellular movement is hormonally regulated. Their reversible translocation would appear to be most analogous to the exocytic/endocytic movement observed in secretory processes in response to a given stimulus. Indeed, the secretion of insulin induced by glucose and the accompanying incorporation of counterregulatory somatostatin receptors into the p cell plasma membrane show a striking parallelism to the processes described here (Mehler et al., 1980). Thus, given the scope of the potential interactions among the glucose transporter, and IGF-I1 and insulin receptors, studies of these three proteins in the rat adipose cell offer a unique opportunity for elucidating both the mechanism of insulin action and many of the underlying processes involved in membrane protein recycling. ACKNOWLEDGMENTS The authors wish to thank their many colleagues, both former and current, for their indispensable contributions to the concepts and experimental results described here. These investigators include

496

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

Kenneth C. Appell, Jose A. Hedo, Paul J. Hissin, Richard Horuk, Barbara B. Kahn, Eddy Karnieli, Masao Kuroda, Matthew M. Rechler, Lester B. Salans, Ole Sonne, Ulf Smith, Chih-chen Wang, Lawrence J. Wardzala, Thomas J. Wheeler, Dena R. Yver, and Mary Jane Zarnowski. The authors also wish to thank Drs. Appell, Kahn, and Kuroda for their critical comments regarding this article and Louie Zalc for her patience and expertise in typing the manuscript.

REFERENCES Armatruda, J. M., and Finch, E. D. (1979). Modulation of hexose uptake and insulin action by cell membrane fluidity. J. Eiol. Chem. 254, 2619-2625. Avruch, J., Nemonoff, R. A,, Blackshear, P. I . , Pierce, M. W., and Osathanondh, R. (1982). Insulin-stimulated tyrosine phosphorylation of the insulin receptor in detergent extracts of human placental membranes. J . Eiol. Chem. 157, 15162-15166. Baldwin, J. M., Gorga, J. C., and Lienhard, G. E. (1981). The monosaccharide transporter of the human erythrocyte. J . Eiol. Chem. 256, 3685-3689. Baldwin, S. A,, and Lienhard, G. E. (1981). Glucose transport across plasma membranes: Facilitated diffusion systems. Trends Eiochem. Sci. 6, 208-21 1. Berhanu, P., Olefsky, J. M., Tsai, P., Saunders, D. T., Thamm, P., and Brandenburg, D. (1982). Internalization and molecular processing of insulin receptors in isolated rat adipocytes. Proc. Natl. Acad. Sci. U.S.A. 79, 4069-4073. Brandenburg, D., Diaconescu, C., Saunders, D. T., and Thamm, P. (1980). Covalent linking of photoreactive insulin to adipocytes produces a prolonged signal. Nature (London) 186, 82 1822. Carter-Su, C., and Czech, M. P. (1980). Reconstitution of D-glucose transport activity from cytoplasmic membranes. J. B i d . Chem. 255, 10382- 10386. Carter-Su, C., Pessin, J. E., Mora, E., Gitomer, W., and Czech, M. P. (1982). Photoaffinity labeling of the human erythrocyte D-glucose transporter. J. Eiol. Chem. 257, 5419-5425. Ciaraldi, T. P., and Olefsky, J. M. (1979). Coupling of insulin receptors to glucose transport: A temperature-dependent time lag in activation of glucose transport. Arch. Eiochem. Eiophys. 193, 221-231. Clark, S., and Harrison, L. C. (1982). Insulin binding leads to the formation of covalent (-S-S-) hormone receptor complexes. J . Eiol. Chem. 257, 12239-12244. Clark, S., and Harrison, L. C. (1983). Disulfide exchange between insulin and its receptor. J . Eiol. Chem. 258, 11434-1 1438. Crofford, 0. B., and Renold, A. E. (1965a). Glucose uptake by incubated rat epididymal adipose tissue. Rate-limiting steps and site of insulin action. J. Eiol. Chem. 240, 14-21. Crofford, 0. B., and Renold, A. E. (1965b). Glucose uptake by incubated rat epididymal adipose tissue. Characteristics of the glucose transport system and action of insulin. J . B i d . Chem. 240, 3237-3244. Cuatrecasas, O., and Tell, G. P. E. (1973). Insulin-like activity of concanavalin A and wheat germ agglutinin-direct interactions with insulin receptors. Proc. Natl. Acad. Sci. U.S.A. 70, 485489. Cushman, S. W., and Wardzala, L. J. (1980). Potential mechanism of insulin action on glucose transport in the isolated rat adipose cell. J . Biol. Chem. 255, 4758-4762. Cushman, S. W., Karnieli, E . , Foley, J. E., Hissin, P. J., Simpson, I. A., and Salans, L. B. (1982). Mechanism of insulin action on glucose transport in human adipose cells: Translocation of intracellular glucose transport systems to the plasma membrane. Clin. Res. 30, 388A (Abstr.). Cushman, S. W., Wardzala, L. J., Simpson, I. A,, Karnieli, E., Hissin, P. I., Wheeler, T. J., Hinkle, P. C., and Salans, L. B. (1984). Insulin-induced translocation of intracellular glucose transporters in the isolated rat adipose cell. Fed. Proc., Fed. Am. SOC. Exp. Biol. 43, 22512255.

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

497

Czech, M. P. (1976a). Current status of the thiol redox model for the regulation of hexose transport by insulin. J . Cell. Physiol. 89, 66-668. Czech, M. P. (1976b). Cellular basis of insulin insensitivity in large rat adipocytes. J . Clin. Invest. 57, 1523-1532. Czech, M. P. (1980). Insulin action and the regulation of hexose transport. Diabetes 29, 399-409. Czech, M. P., and Fain, J. N . (1972). Antagonism of insulin action on glucose metabolism in white fat cells by dexamethasone. Endocrinology 91, S 18-522. Desbuquois, B., Lopez, S . , and Burlot, H. (1982). Ligand-induced translocation of insulin receptors in intact rat liver. J . Biol. Chem. 257, 10852-10860. Ezaki, O., and Kono, T. (1982). Effects of temperature on basal and insulin-stimulated glucose transport activities in fat cells. J . Biol. Chem. 257, 14306-14310. Fehlmann, M., Carpentier, J.-L., LeCam, A,. Thamm, P., Saunders, D. T., Brandenburg, D., Orci, L., and Freychet, P. (1982). Biochemical and morphological evidence that the insulin receptor is internalized with insulin in hepatocytes. J . Cell B i d . 93, 82-87. Foley, J. E., Cushman, S. W., and Salans, L. B. (1978). Glucose transport in isolated rat adipocytes with measurements of L-arabinose uptake. Am. J . Physiol. 234, El 12-El 19. Foley, J. E., Laursen, A. L., Sonne, 0.. and Gliemann, J. (1980). Insulin binding and hexose transport in rat adipocytes. Diuhetologiu 19, 234-241. Gavin, J. R., 111, Roth, I . , Neville, D. M.. DeMeyts, P., and Buell, D. N. (1974). Insulin-dependent regulation of insulin receptor concentrations: A direct demonstration in cell culture. Proc. Nutl. Acad. Sci. U . S . A . 71, 84-85. Gliemann, I., and Rees, W. D. (1983). The insulin-sensitive hexose transport system in adipocytes. Curr. Top. Membr. Trunsp. 18, 339-379. Goldfine, I. D., and Smith, G. S . (1976). Binding of insulin to isolated nuclei. Proc. Nutl. Acud. Sci. U . S . A . 73, 1427-1431. Goodman, H. M. (1966). Effects of growth hormone on the penetration of L-arabinase into adipose tissue. Endocrinology 78, 819-825. Gorga, F. R., Baldwin, S. A,, and Lienhard, G. E. (1979). The monosaccharide transporter from human erythrocytes is heterogeneously glycosylated. Biochem. Biophvs. Res. Commun. 91, 955-96 I . Gorga, J. C., and Lienhard, G. E. (1984). One transporter per vesicle: Determination of the basis of the insulin effect on glucose transport. Fed. P roc., Fed. Am. Soc. Exp. Biol. 43, 2237-224 I . Green, A . (1983). Glucagon inhibition of insulin-stimulated 2-deoxyglucose uptake by rat adipocytes in the presence of adenosine deaminase. Biochem. J . 212, 189-195. Green, A , , and Olefsky, J. M. (1982). Evidence of insulin-induced internalization and degradation of insulin receptors in rat adipocytes. Proc. Nud. Acad. Sci. U.S.A. 79, 427-431. Hammons, G. T., and Jarett, L. (1980). Lysosomal degradation of receptor-bound 12sI-labeled insulin by rat adipocytes. Diuhetes 29, 475-486. Hiring, H. V.. Kemmler, W . , Renner, R . , and Hepp. K. D. (1978). Initial lagphase in the action of insulin on glucose transport and CAMP levels in fat cells. FEBS Lett. 95, 177-180. Haring, J . U . , Kasuga, M., and Kahn, C. R. (1982). Insulin receptor phosphorylation in intact adipocytes and in a cell-free system. Biochem. Biophys. Res. Commun. 108, 1538-1545. Hedo, J. A , , and Simpson, I. A. (1984). Internalization of insulin receptors in isolated rat adipose cells. Demonstration of vectorial disposition of receptor subunits. J . B i d . Chem. 259, 1108311089. Hedo, J. A,, Harrison, L. C., and Roth, J. (1981a). Binding of insulin receptors to lectins: Evidence for common carbohydrate determinants on several membrane receptors. Biochemistry 20, 33853393. Hedo, J. A,, Kasuga, M . , Van Obberghen, E.. Roth, J., and Kahn, C. R. (1981b). Direct demonstration of glycosylation of insulin receptor subunits by biosynthetic and external labeling: Evidence for heterogeneity. Proc. Nut/. Acud. Sri. U.S.A. 78, 4791-4795. Hissin, P. J . , Foley, J. E., Wardzala. L. J., Karnieli. E., Simpson, 1. A . , Salans, L. B . , and

498

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

Cushman, S. W. (1982a). Mechanism of insulin-resistant glucose transport activity in the enlarged adipose cell of the aged, obese rat. J. Clin. Invest. 70, 780-790. Hissin, P. J., Karnieli, E., Simpson, I. A,, Salans, L. B., and Cushman, S. W. (1982b). A possible mechanism of insulin resistance in the rat adipose cell with high fatllow carbohydrate feeding. Diabetes 31, 589-592. Horuk, R., Rodbell, M., Cushman, S. W., and Simpson, I. A. (1983a). Identification and characterization of the rat adipocyte glucose transporter by photoaffinity crosslinking. FEBS Lerr. 164, 26 1-266. Horuk, R., Rodbell, M., Cushman, S. W., and Wardzala, L. J. (1983b). Proposed mechanism of insulin-resistant glucose transport in the isolated guinea pig adipocyte. Small intracellular pool of glucose transporters. J. Biol. Chem. 258, 7425-7429. Iwamoto, Y.,Maddux, B., and Goldfine, 1. D. (1981). Chloroquine stimulation of insulin binding to IM-9 lymphocytes: Evidence for action at a nonlysosomal site. Biochem. Biophys. Res. Commun. 103, 863-871. Jacobs, S., and Cuatrecasas, P. (1981). Insulin receptor: Structure and function. Endocr. Rev. 203, 251-263. Kahn, B. B., and Cushman, S. W. (1984). Effects of fasting and refeeding on the distribution of glucose transporters in isolated rat adipocytes. Diabetes 33 (Suppl. I), 71A (Abstr.). Kahn, B. B., Horton, E. S., and Cushman, S. W. (1984). Mechanism by which chronic hyperinsulinemia increases insulin-stimulated glucose transport in rat adipocytes: Enlarged intracellular pool of glucose transporters. Clin. Res. 32, 399A (Abstr.). Kahn, C. R. (1983). The insulin receptor and insulin: The lock and key to diabetes, Clin. Res. 31, 326-335. Kahn, C. R., Hedo, J. A., and Kasuga, M. (1983). Structure, biosynthesis and phosphorylation of the insulin receptor. In “Hormone Receptors and Receptor Diseases” (H. Imura and H. Kuzuya, eds.), pp. 3-1 1. Excerpta Medica, Amsterdam. Karnieli, E., Hissin, P. J., Simpson, I. A,, Salans, L. B., and Cushman, S. W. (1981a). A possible mechanism of insulin resistance in the rat adipose cell in streptozotocin-induced diabetes mellitus. J. Clin. Invest. 68, 811-814. Karnieli, E., Zarnowski, M. J., Hissin, P. I., Simpson, I. A,, Salans, L. B., and Cushman, S. W. (I98 1b). Insulin-stimulated translocation of glucose transport systems in the isolated rat adipose cell. Time course, reversal, insulin concentration-dependency and relationship to glucose transport activity. J. Biol. Chem. 256, 4772-4777. Kasahara, M., and Hinkle, P. C. (1976). Reconstitution of D-glucose transport catalyzed by a protein fraction from human erythrocytes in sonicated liposomes. Proc. Natl. Acad. Sci. U.S.A. 73, 396-400. Kasahara, M., and Hinkle, P. C. (1977). Reconstitution and purification of the D-glucose transporter from human erythrocytes. J. Biol. Chem. 252, 7384-7390. Kashiwagi, A., and Foley, J. E. (1982). Opposite effects of a P-adrenergic agonist and a phosphodiesterase inhibitor on glucose transport in isolated human adipocytes: Isoproterenol increases V,,, and IBMX increases Ks. Biochem. Biophys. Res. Commun. 107, 1151-1 157. Kashiwagi, A., Heucksteadt, T. P., and Foley, J. E. (1983). The regulation of glucose transport by CAMP stimulators via three different mechanisms in rat and human adipocytes. J. Biol. Chem. 258, 13685-13692. Kasuga, M., Van Obberghan, E., Nissley, S. P., and Rechler, M. D. (1981). Demonstration of two subtypes of insulin-like growth factor receptors by affinity crosslinking. J. Biol. Chem. 256, 5305-5308. Kasuga, M., Karlson, F. A., and Kahn, C. R. (1982). Insulin stimulates the phosphorylation of the 95,000-dalton subunit of its own receptor. Science 215, 185-187. Kiechle, F. L., Jarett, L., Kotagal, N., and Popp, D. A. (1981). Partial purification from rat

1 1. REGULATION BY INSULIN IN RAT ADIPOSE CELL

499

adipocyte plasma membranes of a chemical mediator which stimulates the action of insulin on pyruvate dehydrogenase. J . Eiol. Chem. 256, 2945-295 I . King, G. L.. Kahn. C. R.. Rechler, M. M., and Nissley, S . P. (1980). Direct demonstration of separate receptors for growth and metabolic activities of insulin and multiplication-stimulating activity (an insulin like growth factor) using antibodies to the insulin receptor. J . Clin. Invest. 66, 130-140. King, G. L., Rechler, M. M., and Kahn, C. R. (1982). Interactions between the receptors for insulin and the insulin-like growth factors on adipocytes. J . Eiol. Chem. 257, 10001-10006. Kirsch, D. M., Kemler, W . , and Haring, H. U. (1983). Cyclic AMP modulates insulin binding and induces post-receptor insulin resistance of glucose transport in adipocytes. Biochem. Eiophys. Res. Commun. 115, 398-405. Kirsch, D. M., Baumgarten, M . , Deufel, I . , Rinninger, F., Kemmler, W., and Haring, H. U . ( 1984). Catecholamine-induced insulin resistance of glucose transport in isolated rat adipocytes. Biochem. J . 217, 737-745. Kono, T., and Barham, F. W. (1971). The relationship between the insulin-binding capacity of fat cells and the cellular response to insulin. J . Biol. Chem. 246, 6210-6216. Kono, T . , Robinson, F. W., and Sarver, J. A. (1975). Insulin-sensitive phosphodiesterase. J . Eiol. Chem. 250, 7826-7835. Kono, T., Robinson, F. W., Sarver, J. A,, Vega, F. V., and Pointer, R. H. (1977). Actions of insulin in fat cells. J . Eiol. Chem. 252, 2226-2233. Kono, T., Suzuki, K., Dansey, L. E., Robinson, F. W . , and Blevins, T. L. (1981). Energydependent and protein synthesis-independent recycling of the insulin-sensitive glucose transport mechanism in fat cells. J . Eiol. Chem. 256, 6400-6407. Kono. T . , Robinson, F. W . , Blevins, T. L., and Ezaki, 0. (1982). Evidence that translocation of the glucose transport activity is the major mechanism of insulin action on glucose transport in fat cells. J . Eiol. Chem. 257, 10942-10947. Kuroda, M., Simpson, I. A , , Honnor, R., Londos, C., and Cushman, S . W. (1984). Counterregulation of insulin-stimulated glucose transport by isoproterenol in rat adipose cells. Fed. Proc. Fed. Am. Soc. Exp. Eiol. 43, 1814 (Abstr.). Lamer, J., Galaski, G., Cheng, K., DePaoli-Roach, A. A., Huang, L., Daggy, P., and Kellog, J. (1979). Generation by insulin of a chemical mediator that controls protein phosphorylation and dephosphorylation. Science 206, 1408-1410. Lienhard, G. E. (1983). Regulation of cellular membrane transport by exocytotic insertion and endocytic retrieval of transporters. Trends Biochern. Sci. 8, 125- 127. Lienhard, G . E . , Kin, H. K., Ransome, K. J., andGorga, J. C . (1982). Immunological identification of an insulin-responsive glucose transporter. Biochern. Eiophys. Res. Commun. 105, I 1501156. Livingston, J. N . , and Lockwood, D. H. (1974). Direct measurements of sugar uptake in small and large adipocytes from young and adult rats. Biochem. Eiophys. Res. Commun. 61, 989-996. Lonnroth, P., and Smith, U . (1983). P-Adrenergic dependent down regulation of insulin binding in rat adipocytes. Biochern. Biophys. Res. Commrtn. 112, 971 -979. Ludvigsen, C., Jarett, L., and McDonald, J . M. (1980). The characterization of catecholamine stimulation of glucose transport by rat adipocytes and isolated plasma membranes. Endocrinology 106, 786-790. McKeel, D. W . , and Jarett, L. (1970). Preparation and characterization of a plasma membrane fraction from isolated fat cells. J . Cell Eiol. 44, 417-432. Maloff, B. L., Levine, S. H., and Lockwood, D. H. (1980). Direct effects of growth hormone on insulin action in rat adipose tissue maintained in vifro. Endocrinology 107, 538-544. Marshall, S. ( 1983). Kinetics of insulin receptor biosynthesis and membrane insertion. Diaberes 32, 319-325.

500

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

Marshall, S., and Olefsky, J. M. (1980). The endocytic-internalization pathway of insulin metabolism: Relationship to insulin degradation and activation of glucose transport. Endocrinology 107, 1937-1945. Marshall, S . , and Olefsky, J. M. (1981). Tris (hydroxymethy1)aminomethanepermits the expression of insulin-induced receptor loss in isolated rat adipocytes. Biochem. Biophys. Res. Commun. 102, 646-653. Massague, J., and Czech, M. P. (1980). Multiple redox forms of the insulin receptor in native liver membranes. Diabetes 29, 945-947. Massague, I., and Czech, M. P. (1982). The subunit structures of two distinct receptors for insulinlike growth factors I and I1 and their relationship to the insulin receptor. J. Biol. Chem. 257, 5038-5045. Mehler, P. S . , Sussman, A. L., Maman, A,, Leitner, J . W., and Sussman, K. E. (1980). Role of insulin secretagogues in the regulation of somatostatin binding by isolated rat islets. J . Clin. Invest. 66, 1334-1338. Munck, A. (1962). Studies on the mode of action of glucocorticoids in rats. Biochim. Biophys. Acta 57, 318-326. Olefsky, J. M. (1975). Effect of dexamethasone on insulin binding, glucose transport, and glucose oxidation of isolated rat adipocytes. J. Clin. Invest. 56, 1499-1508. Olefsky, J. M. (1976). The effects of spontaneous obesity on insulin binding, glucose transport, and glucose oxidation of isolated rat adipocytes. J . Clin. Invest. 57, 842-85 1. Olefsky, J. M., Marshall, S., Berhanu, P., Sarkow, M., Heidenreich, K., and Green, A. (1983). Internalization and intracellular processing of insulin and insulin receptors in adipocytes. Metab. Clin. Exp. 31, 670-690. Oppenheimer, C. L., Pessin, J. E., Massague, J., Gitomer, W., and Czech, M. P. (1983). Insulin action rapidly modulates the apparent affinity of the insulin-like growth factor I1 receptor. J. Biol. Chem. 258, 4824-4830. Pessin, 1. E., Gitomer, W . , Oka, Y., Oppenheimer, C. L., and Czech, M. P. (1983). P-Adrenergic regulation of insulin and epidermal growth factor receptors in rat adipocytes. J. Biol. Chem. 258, 7386-7394. Pessin, I. E., Massague, J . , and Czech, M. P. (1984). Affinity labeling techniques for low abundance membrane components: Peptide hormone receptors and the D-glucose transporter. In “Investigation of Membrane Located Receptors” (E. Reid, G. N. W. Cook, and D. J. Moore, eds.), pp. 295-302. Plenum, New York. Petruzzeli, L. M., Ganguly, S., Smith, C. S., Cobb, M. H., Rubin, C. S., and Rosen, 0. M. (1982). Insulin activates a tryosine-specific protein kinase in extracts of 3T3-LI adipocytes and human placenta. Proc. Nail. Acad. Sci. U.S.A. 79, 6792-6796. Pilch, P. F., and Czech, M. P. (1979). Interaction of cross-linking agents with the insulin effector system of isolated rat cells. J. Biol. Chem. 254, 3375-3380. Pilch, P. F., and Czech, M. P. (1980). The subunit structure of the high affinity insulin receptor. J . Biol. Chem. 255, 1722-2731. Pilch, P. F., Thompson, P. A , , and Czech, M. P. (1980). Coordinate modulation of o-glucose transport activity and bilayer fluidity in plasma membranes derived from control and insulintreated adipocytes. Proc. Natl. Acad. Sci. U.S.A. 77, 915-918. Posner, B. I., Josefsberg, Z., and Bergeron, J. J . M. (1978). Intracellular polypeptide hormone receptors. J. Biol. Chem. 253, 4067-4073. Posner, B. I., Patel, B., Verma, A. L., and Bergeron, J . J . M. (1980). Uptake of insulin by plasmalemma and Golgi subcellular fractions of rat liver. J. Biol. Chem. 255, 735-741. Posner, B. I., Patel, B. A . , Kahn, M. N., and Bergeron, I. J . M. (1982). Effect of chloroquine on the internalization of ~2JI-insulininto subcellular fractions of rat liver. J. Eiol. Chem. 257, 5789-5799.

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

501

Rechler. M. M., Kasuga, M.. Sasaki, N., De Vroede, M. A.. Romanus, J. A., and Nissley, S. P. (1982). Properties of insulin-like growth factor receptor subtypes. I n “Insulin-like Growth Factors/Somatoniedins: Basic Chemistry. Biology, and Clinical Importance” (E. M. Spencer, ed.), pp. 459-490. De Gruyter, New York. Rennie, P., and Gliemann, J. (1981). Rapid down regulation of insulin receptors in adipocytes. Artifact of the incubation buffer. Biochem. Biophys. Res. Commun. 102, 824-83 I , Resh, M. D. (1982). Development of insulin responsiveness of the glucose transporter and ( N a f , K ) adenosine triphosphatase during in litro adipocyte differcntiation. J. B i d . Chem. 257, 6978-6986. Robinson, F. W . , Blevins, T. L., Suzuki, K . , and Kono, T. (1982). An improved method of reconstitution of adipocyte glucose transport activity. Anal. Binchem. 122, 10- 19. Rodbell, M. (1964). Metabolism of isolated fat cells. J. Biol. Chem. 239, 375-380. Roth, R. A , , and Cassels, D. L. (1983). Insulin receptor: Evidence that it is a protein kinase. Science 219, 299-301. Saviolakis, C. A , , Harrison, L. C., and Roth, J . (1981). The binding of 12sI-insulin to specific receptors in IM-9 human lymphocytes. J . Biol. Chem. 256, 4924-4928. Schoenle, E., Zapf, J., and Froesch, E. R. (1976). Binding of nonsuppressible insulin-like activity (NSILA) to isolated fat cells: Evidence for two separate membrane acceptor sites. FEBS Left. 67, 175-179. Schoenle, E., Zapf, J., and Froesch, E. R. (1977). Effects of insulin and NSILA on adipocytes of normal and diabetic rats: Receptor binding, glucose transport and glucose metabolism. Diaberologia 13, 243-249. Schoenle, E., Zapf, J., and Froesch, E. R. (1979a). Effects of insulin on glucose metabolism and glucose transport in fat cells of hormone-treated hypophysectomized rats: Evidence that growth hormone restricts glucose transport. Endocrinologv 105, 1237- 1242. Schoenle, E., Zapf, J., and Froesch, E. R. (1979b). Receptor binding and effects of insulin and NSILA-S on glucose transport and metabolism in adipocytes from hypophysectomized rats. Diabetologia 16, 41-46. Seals, J. R., and Czech, M. P. (1980). Evidence that insulin activates an intrinsic plasma membrane protease in generating a secondary chemical mediator. J. Biol. Chem. 255, 6529-6531, Shanahan, M. F . , and Czech, M. P. (1977). Purification and reconstitution of the adipocyte plasma membrane D-glucose transport system. J. Biol. Chern. 252, 834143343, Shanahan, M. F., Olson, S. A,, Weber. M. J., Lienhard, G. E., and Gorga, J. C. (1982). Photolabeling of glucose-sensitive cytochalasin B binding proteins in erythrocyte, fibroblast and adipocyte membranes. Biochem. Biophys. Res. Commun. 107, 38-43. Simpson, I . A,, and Hedo, J. A. (1984). Insulin-induced phosphorylation of its own receptor may not be a prerequisite for acute insulin action. Science 223, 1301-1304. Simpson, 1. A,, Martin, M. L., and Cushnian, S . W. (1982). Effect of TRIS on insulin’s stimulatory action on glucose transport in the isolated rat adipose cell. Diaberes 31(Suppl. 2). 2A (Abstr.). Simpson, 1. A., Smith, U., and Cushman, S . W. (1983a). Counterregulation of insulin-stimulated glucose transport by isoproterenol in rat adipose cells: Modulation of intrinsic glucose transporter activity. Diabetes 66, 1334- 1338. Simpson, I. A , , Yver, D. R., Hissin, P. J., Wardzala, L. J., Kamieli, E., Salans, L. B., and Cushman, S. W. ( 1983b). Insulin-stimulated translocation of glucose transporters in the isolated rat adipose cell: Characterization of subcellular fractions. Biochim. Biophys. Acta 763, 393407. Simpson, I. A,, Zamowski, M. J., and Cushman. S . W. ( 1 9 8 3 ~ )Mechanism . of insulin’s stimulatory action on glucose transport: Effects of low temperature. Fed. Proc., Fed. Am. SOC. Exp. Biol. 42, 1790 (Abstr.). Simpson, 1. A,, Hedo, J. A,, and Cushman, S. W. (1984a). Insulin-induced internalization of the +

502

IAN A. SIMPSON AND SAMUEL W. CUSHMAN

insulin receptor in the isolated rat adipose cell: Detection of both major receptor subunits following their biosynthetic labeling in culture. Diabetes 33, 13- 18. Simpson, I. A,, Karnieli, E., Hissin, P. I., Smith, U., and Cushman, S. W. (l984b). Mechanism of insulin’s stimulatory action on glucose transport in the isolated rat adipose cell. Proc. Gen. Physiol. Soc. U.S.A. 39, 43-55. Smith, R. M., and Jarett, L. (1982a). A simplified method of producing biologically active monomeric ferritin-insulin for use as a high resolution ultrastructural marker for occupied insulin receptors. J. Histochem. Cytochem. 30, 650-656. Smith, R. M., and Jarett, L. (1982b). Ultrastructural basis for chloroquine-induced increase in intracellular insulin in adipocytes: Alteration of lysosomal function. Proc. Natl. Acad. Sci. U.S.A. 79, 7302-7306. Smith, U. (1974). Influence of insulin and medium glucose concentration on cellular metabolism. J. Clin. Invest. 53, 91-98. Sogin, D. C . , and Hinkle, P. C. (1980). Immunological identification of the human erythrocyte glucose transporter. Proc. Natl. Acad. Sci. U.S.A. 77, 5725-5729. Sonne, 0.. and Simpson, I. A. (1984). Internalization of insulin and its receptor in the isolated rat adipose cell: Time course and insulin concentration dependency. Biochern. Biophys. Acta 804, 404-413. Suzuki, K., and Kono, T. (1980). Evidence that insulin causes translocation of glucose transport activity to the plasma membrane from an intracellular storage site. Proc. Narl. Acad. Sci. U.S.A. 77, 2542-2545. Taylor, W. M., Mak, M., and Halperin, M. L. (1976). Effect of 3’:5’-cyclic AMP on glucose transport in rat adipocytes. Proc. Natl. Acad. Sci. U.S.A. 73, 4359-4363. Van Obberghan, E., and Kahn, C. R. (1981). Autoantibodies to insulin receptors. Mol. Cell. Endocrinol. 22, 277-293. Van Obberghen, E., Kasuga, M., LeCam, A., Hedo, I. A., Roth, I., and Kahn, C. R. (1981). Biosynthetic labeling of insulin receptors: Studies of subunits in cultured human IM-9 lymphocytes. Proc. Natl. Acad. Sci. U.S.A. 78, 1052-1056. Vaughan, M. (1961). The metabolism of adipose tissue in virro. J. Lipid Res. 12, 293-316. Vinten, J., Gliemann, J . , and Osterlind, K. (1976). Exchange of 3-0-methyl-glucose in isolated fat cells. Concentration dependence and effect of insulin. J. Biol. Chem. 251, 794-800. Wang, C.-C., Hedo, J. A., Kahn, C. R., Saunders, D. T., Thamm, P., and Brandenburg, D. (1982). Photoreactive insulin derivatives. Comparison of biologic activity and labeling properties of three analogues in isolated rat adipocytes. Diabetes 31, 1068- 1076. Wang, C.-C., Sonne, O., Hedo, J. A., Cushman, S. W., and Simpson, I. A. (1983). Insulin-induced internalization of the insulin receptor in the isolated rat adipose cell. J. B i d . Chem. 258, 51295134. Wardzala, L. J. (1979). Identification of the glucose transport system in purified rat adipose cell plasma membranes using a cytochalasin B binding assay: Effects of insulin and altered physiological states. Ph.D. dissertation, Dartmouth College, Hanover, N.H. Wardzala, L. J., and Jeanrenaud, B. (1981). Potential mechanism of insulin action on glucose transport in isolated rat diaphragm. Apparent translocation of intracellular transport systems to the plasma membrane. 1. Biol. Chem. 256, 7090-7093. Wardzala, L. J., and Jeanrenaud, B. (1983). Identification of the D-glucose inhibitable cytochalasin B binding site as the glucose transporter in rat diaphragm plasma and microsomal membranes. Biochim. Biophys. Acta 730, 49-56. Wardzala, L. J., Cushman, S. W., and Salans, L. B. (1978). Mechanism of insulin action on glucose transport in the isolated rat adipose cell. Enhancement of the number of functional transport systems. J. Biol. Chem. 253, 8002-8005. Wardzala, L. J . , Simpson, I. A,, Rechler, M. M., and Cushman, S. W. (1984). Mechanism of the stimulatory action of insulin on insulin-like growth factor I1 binding to the isolated rat adipose

11. REGULATION BY INSULIN IN RAT ADIPOSE CELL

503

cell. Apparent redistribution of receptors cycling between a large intracellular pool and the plasma membrane. J . Biol. Chem. 259, 8378-8383. Wheeler, T. J . , Simpson, 1. A . , Sogin, D. C., Hinkle, P. C., and Cushman, S. W. (1982). Detection of the rat adipose cell glucose transporter with antibody against the human red cell glucose transporter. Biochem. Biophys. Res. Commun. 105, 89-95. Winegrad, A. I . , and Renold, A. E. (1958). Studies on rat adipose tissue in viiro. J . Biol. Chem. 233, 267-272. Yip, C. C . , Yeung, C. W. T., and Moule, M. L. (1978). Photoaffinity labeling of insulin receptor of rat adipocyte plasma membrane. J . B i d . Chem. 253, 1743-1745. Zapf, J . , Schoenle, E., and Froesch, E. R. (1978). Insulin-like growth factors I and 11: Some biological actions and receptor binding characteristics of two purified constituents of nonsuppressible insulin-like activity of human serum. Eur. J . Biochem. 87, 285-296.

This Page Intentionally Left Blank

Index

A

N-Acetylglucosaminyl lipid, in protein glycosylation. 193 Addressing signals, see Sorting signals Adenosine deaminase effect on insulin-stimulated glucose uptake, 474-476 isoproterenol addition and, 474-476 Adipocytes insulin-like growth factor I1 binding, 485487 insulin-stimulated glucose uptake, 459-462 regulation, 474-482 membrane subcellular fractions cytochalasin B binding, 463-466 glucose transporters distribution, 466-468 in reconstituted liposomes, 465-466 yield and recovery, 469-470 marker enzyme activities, 462, 464 preparation, 462 Adrenocorticotropin, secretion pathway, 276 Aging, insulin-resistant glucose uptake, 47948 1 Alkaline phosphatase signal peptide mutants, 83 amino acid sequences, 85 (table) Amino acid sequences eukaryotic transmembrane segments, 164165 (table) internal signal peptides, 22 (table) N-terminal of imported mitochondrial proteins, 324 (table) signal peptides at N-terminus eukaryotic proteins, 8-9 (table)

LamB protein mutants, 123 (table) maltose-binding protein mutants, 117, 129 (tables) prokaryotic, 9- 10, 68-70 (tables) coniparison with eukaryotic, 73 mutations, 84-85, 87 (tables) Apocytochrome c, extramitochondrial binding to mitochondrial receptor, 299-300 comparison with cytochrome c conformation, 298 structure, 317 ATPase Ca2+-, in sarcoplasmic reticulum during development, 345-347 of fast and slow muscles, 360-361 synthesis Ca2+-dependent. 358-359 in muscle cell culture, 354-355 pathway, 356-358 in RER, 352 Ca2+ ,Mg2 -, in sacroplasmic reticulum during development, 348 location, 340-341 synthesis in muscle cell culture, 356 F,-, P-subunit precursor isolation and purification, 3 17-3 I8 mitochondrial import, 302-303 posttranslational, 3 I4 processing by protease. 319 Mg2+ -, in developing sarcoplasmic reticulum, 346 in developing sarcoplasmic rcNa+ ,K ticulum, 347 ATPase synthase, subunit 9 N-terminus, 323-325 processing by mitochondrial protease, 3 19320 +

+

505

-.

506

INDEX

B

Bacteriophage M 13, coat protein insertion into E. coli membrane, 171 free energy profile, 171 posttranslational, mechanism of, 77 transmembrane segments, structure, 168 Bacteriorhodopsin membrane-spanning segments helical wheel plot, 168 a-helices, 167 structure, 39 translocation, mechanism of, 39

C

Calcium ion protein synthesis in sarcoplasmic reticulum and, 358-360 uptake by developing sarcoplasmic reticulum, 345-346 in fast and slow muscles, 360-361 Calsequestrin, in sarcoplasmic reticulum Ca2 binding, 34 1 location, 341-342 synthesis in muscle cell culture, 355-356 pathway, 356-358 in RER, 352 Chitobiosylpyrophosphoryl lipid metabolism pathways, 196-197 in RER, orientation, 221-222 synthesis, 195-196 Choline phosphotransferase, in developing sarcoplasmic reticulum, 349 Chorionic gonadotropin, human, interaction with receptor, 230-23 I Clathrin in coated pits of membranes, 385 receptor attachment to, 394 Coated pits, membrane clathrin role, 385, 394 endocytosis and, 385-386 lipid composition, 395 receptor association with, 388-396, 424 mechanism of, 393-396

Cytochalasin B binding to glucose transporter, 463-464 insulin effect, 465 in reconstituted liposomes, 465-466 Cytochrome b2, mitochondrial in intermembrane space, 313 maturation from precursor, 307-309 Cytochrome b2 precursor extramitochondrial synthesis, 306 mitochondrial import and maturation, 306309 Cytochrome bg, embedding into membrane, 32-33 Cytochrome c comparison with apocytochrome c conformation, 298 structure, 317 synthesis from apocytochrome c, 298-299 Cytochrome c peroxidase in mitochondrial intermembrane space, 305 N-terminus, 323-326 Cytochrome c , , mitochondrial maturation, 307-308 Cytochrome P-450 reductase, translocation mechanism, 33-34 Cytosolic soluble factor, protein import into mitochondria and, 320

+

D

Dexamethasone, insulin-stimulated glucose uptake and, 476-477 Diabetes, streptozotocin-induced insulin-resistant glucose uptake, 481 Diet, low fat/low carbohydrate insulin-resistant glucose uptake, 477-478 Docking protein, see SRP-receptor Dolichol interconvertible forms, 188- 189 phosphate removal from, 189 protein glycosylation and, 185- 186 subcellular distribution, liver, 190 synthesis from acetate, 188-189 from mevalonic acid, 187- 188 Dolichol kinase protein glycosylation and, 2 18-2 19 in RER, orientation, 221

INDEX

Dolichyl phosphate conversion-to dolichol, 188- 189 subcellular distribution, liver, 190- I9 I

E EGF receptor, see Epidermal growth factor receptor Endocytosis adsorptive non-receptor-mediated. 41 7 receptor-mediated description, 4 16-4 I7 of LDL, 416-417, 440-441 mechanism of, 440-45 I in membrane glycoprotein sorting, 266-268 of transferrin, 416-417, 432-433 coated pits in membranes and, 385-386 description and role, 413-414 fluid-phase, 416 membrane constituent recycling and, 424430 membrane protein lateral mobility and, 38639 1 phagocytosis, 415 pinocytosis, 415 Endoglycosidase H, oligosaccharide sensitivity to, 208-210 Endoplasmic reticulum glycosyl transferases, location, 220 intrinsic transmembrane glycoproteins, 259260 oligosaccharide transferase, location, 220 rough, see Rough endoplasmic reticulum

(RW sarcoplasmic reticulum development and, 352-354 secretory protein sorting, 259-260 P-Endorphin, secretion pathway, 276 Epidermal growth factor receptor (EGF receptor) association with coated pits conformational change, 392-393 endocytic down-regulation, 392 degradation and occupancy, 404 internalization rate defects, 396

lateral diffusion coefficient and, 390 technique, 383 recycling, 399, 442-443 Escherichiu coli ( E . coli) extracytoplasmic compartments, 105- 107 extragenic suppressor mutants LamB protein export, 138-142 maltose-binding protein export, 140- 142 gene fusion, 107, 109 internal signal peptides in proteins, 21, 23 intragenic malEA12-18 suppressor strains, 127-130 maltose-binding protein alteration, 128130 lamB-lacZ protein fusion strains hybrid proteins, 120- 122 LamB protein mutants, 122-124. 133I35 malE-lacZ protein fusion strains hybrid proteins characteristics, 108- 110 processing, I 19- I20 signal protein alterations, 116- I19 maltose-sensitive phenotype (Mals), 110I13 mutants defective in maltose-binding protein export, 113-116, 127-133, 136- 138, 140- 142 protein secretion, 25 coupling with protein synthesis, 142-145 defects in mutants, 96-97 signal peptidases, 95-96 SRP-like factor, 17-18 Ethanolamine phosphotransferase, in developing sarcoplasmic reticulum, 349

F

Familial hypercholesterolemia, LDL-receptor mutants, 271-272, 393 Fasting, insulin-resistant glucose uptake, 48 1 Fibroblasts interaction with anti-PM IgG binding and uptake, 418-419 plasma membrane area recycling, 424. 428-429 subcellular distribution, 4 18-420

508

INDEX

control IgG, fluorescein-labeled binding and uptake, 420 degradation product release, 420-42 1 lysosomal accumulation, 421-424 mechanism of, 424-428 LDL receptor recycling, 443 transfernin endocytic pathway, 432-437

G

Galactosylphosphoryl lipid, in protein glycosylation, 194- 195 Galactosyl transferase, in adipocyte membrane fractions, 462, 464 Genes fusion, E. coli lumB-lad, 120-124 loci encoding LamB export, 138- 142 loci encoding maltose-binding protein export, 136-138, 140-142 malE-lucZ, 108-120 technology, 107- 109 for membrane-bound IgM in lymphocytes, 272 for mitochondria1 proteins, 320-327 molecular cloning, yeast immunological screening, 322 mutant complementation, 321-322 RNA hybridization, 323 screening with synthetic oligonucleotide probe, 321 Genetic engineering protein fusion strain construction, E. coli. 107-109 viral envelope glycoprotein sorting and, 277-279 Glucose oligosaccharide transfer to protein and, 21 1212 starvation, protein glycosylation and, 223225 Glucose transporters insulin effects on number, 460 cytochalasin B binding assay, 463-465 in reconstituted liposomes, 465-466 subcellular distribution, 466-468 regulation, 474-475, 478-482

yield and recovery, 469-470 structure heterogeneity, 484 purification methods, 482-484 Glucose uptake, adipocytes insulin resistance aging, obesity and, 479-481 fasting and, 481 high fat/low carbohydrate diet and, 477478 hyperinsulinemia and, 48 1-482 streptozotocin-induced diabetes and, 48 1 insulin-stimulated, 459-462 regulation by adenosine deaminase with isoproterenol, 474-476 dexamethasone, 476-477 growth hormone, 476 in subcellular fractions cytochalasin B binding and, 463-466 in reconstituted liposomes, 465-466 insulin effect mechanism of, 470-474 reversal, 473-476 time course, 468-469 Glycosylphosphoryl lipid, in protein glycosyla. tion, 193 Glucosylphosphoryl lipid synthetase, in RER, orientation, 223 Glucuronosyl-N-acet ylglucosaminylpyrophosphoryl lipid, in protein glycosylation, 193-194 Glycerophosphate acyltransferase, in developing sarcoplasmic reticulum, 349 Glycine, in signal peptides, 71 mutations, 88-89 Glycoproteins asparagine-linked oligosaccharides, see Oligosaccharides, asparagine-linked carbohydrate moiety role in compartmentalization, 23 1-232 enzymatic functions, 230 interactive functions, 230-23 I solubility, 228 structure, 229 turnover, 229-230 membrane, see Membrane glycoproteins in sarcoplasmic reticulum location, 341 synthesis pathways, 357 sorting in secretory mutants, yeast, 270-271

INDEX

509

Glycosyl transferases, in endoplasmic reticulum, 220-221 Golgi apparatus enzymatic activities, 26 1 secretory protein sorting, 260-262 in cell-free system, 270 transmembrane glycoproteins, 26 1 Growth hormone, insulin-stimulated glucose uptake and, 476 GTP, protein glycosylation and, 220

H

Hemoglobin- haptoglobin intravenously injected, fate in liver, 44645 I transport in hepatocytes, 445-446 Hepatocytes transferrin endocytic pathway, 436, 438 transport of galactosylated serum albumin, 445-446 hemoglobin-haptoglobin, 445-446 polymeric IgA, 445-446 High affinity CaZ+-binding protein location in sarcoplasmic reticulum, 342 synthesis in muscle cell culture, 355 Hybrid proteins, E . coli in laml-lucZ protein fusion strains, 120124, 133-135 in malE-kucZ fusion strains, 108-120 Hydrophobicity concept, 152-153 scales empirical, 153- 154 membrane proteins and, 156-157 statistical, 153, 155-156 Hyperinsulinemia, insulin-resistant glucose u p take, 481-482

I a-L-Iduronidase, carbohydrate role in activity, 230 Immunoglobulins IgA, polymeric intravenously injected, fate in liver, 44645 1

transport in hepatocytes, 445-446 IgG fluorescein-labeled, interaction with fibroblasts association with lysosomes. 42 1-424 binding and uptake, 420 degradation product release, 420-42 1 mechanism of, 424-428 against plasma membrane antigens (antiPM Ig), interaction with fibroblasts mechanism of, 424, 428-429 slow binding to antigen, 418-419 subcellular localization, 418-420 IgM, membrane-bound, genes for, lymphocytes, 272 Insulin glucose uptake stimulation mechanism of, 470-474 regulation, 474-475, 478-482 in subcellular fractions cytochalasin B binding and, 463-466 glucose transporters and, 469-470 reversal, 469, 473-476 time course, 468-469 internalization from complex with ferritin, 489 during endocytosis, 488-489 Insulin-like growth factor I1 binding to adipocytes insulin-stimulated, 485-487 KCN-inhibited, 486-487 Insulin-like growth factor I1 receptor recycling, inhibition by KCN, 486-487 Insulin receptor degradation and occupancy, 404 internalization down-regulation, 489-490 subcellular distribution and, 490-492 subunit radiolabeling assay, 492-494 recycling evidence for, 398 mechanism of, 494-495 structure, 488 tyrosine kinase activity, 488 Integral membrane proteins incorporation embedding, 32-34 models for, 31-32 signal peptide-initiated, 34-35 SRP and, 35-36 insertion, sequential, models of, 42-46

INDEX

internal signal sequences, 35-37 lack of precursors, 36 membrane-spanning domains amino acid sequences, 22, 38-39 signal function, 41 stop-transfer function, 40-41, 43 structure, 39-40 Internal signal peptides amino acid sequences, 8-10 (table) in E . coli proteins, 2 1, 23 membrane protein translocation and, 35-37 in ovalbumin, 20-21 in prolipoprotein, loop model, 80-81 Iron metabolism and reutilization, 430-431 in transferrin intracellular uptake, 434-436 models of, 438-439 Isoproterenol, combined with adenosine deaminase insulin-stimulated glucose uptake and, 474416

dissociation from receptor. 443-446 transfer to lysosomes, 442 Lipids, subcellular distribution, 258 Liver, subcellular fractions, ligand distribution, 441-45 I Low-density lipoproteins (LDL) endocytosis non-receptor-mediated, 417 receptor-mediated, 416-417, 440-441 Lysolecithin acyltransferase, in developing sarcoplasmic reticulum, 349 Lysosomal enzymes, mislocation in mucolipidosis I1 and 111, 272 Lysosomal proteins mannose 6-phosphate group of, 256, 262 sorting, 262 Lysosomes IgG accumulation, 42 1-424 ligand transfer from receptor to, 442 in membrane fragment recycling phagocytosis, 428 pinocytosis, 424-426, 428-430 transferrin accumulation, 434

L M Lactoperoxidase phagocytosis by macrophages, 428 LamB protein, E . coli hybrids in lumB-lacZ fusion strains characterization, 120- 122 export-defective mutants, 122- 124 mature portion, export information, 133-135 signal peptide sequences, 123 (table) signal peptide mutants, 81-82 amino acid sequences, 84-85 (table) LDL, see Low-density lipoproteins LDL-fenitin conjugate, receptor-mediated endocytosis, 441 LDL receptor association with coated pits, 271, 392-393 endocytic down-regulation, 392 at steady state, 393 mutants in familial hypercholesterolemia, 271-272, 393 properties, 440-442 recycling, 398, 442-443 Ligands binding to receptor, 440-441

a2-Macroglobulin receptor, recycling, 398 Macrophages, lactoperoxidase receptor, recycling, 428 Maltose-binding protein, E . coli export-defective, OmpA protein export and, 131-132 hybrids in rnalE-IacZ fusion strains characteristics, 108-1 10 export-defective in mutants, 113-1 16 signal peptide sequences, 116-1 19 maltose sensitivity and, 110-1 13 processing, 119-120 in rnulEA12-18 mutant characterization, 127- 128 signal peptide alterations, 129- 130 mature portion, export initiation, 130- 133 signal peptide mutants, 82-83 amino acid sequences, 85 (table) synthesis and export coupling, 143- 145 Maltose sensitivity, E . coli in malE-lacZ strain, PB72-47, 110-1 13 defects in hybrid protein export and, I I2I13

INDEX

P-galactosidase high activity and, 1 1 I 112 Mannosylphosphoryl lipid, in protein glycosylation, 192- 193 Mannosylphosphoryl lipid synthetase, in RER, orientation, 222 Membrane fragments recycling in phagocytosis, macrophages, 428 in pinocytosis, fibroblasts, 424-430 Membrane glycoproteins characterization, 264-265 (table) in endoplasmic reticulum, 259-260 externalization, constitutive pathway, 276 in Golgi apparatus, 261 molecular sorting, 263-266 orientation, 263-265 receptor-mediated endocytosis, 266-268 recycling, 266-267 synthesis and processing, 262-263 viral envelope, see Viruses, enveloped Membrane proteins hydrophobicity scales and, 156- 157 intrinsic, transmembrane segments amino acid sequences, eukaryotic, 163, 164-165 (table) conformation within membrane, 167- 169 functions, 173 a-helices, 167-168 hydrophobicity, 166, 168 lateral mobility, endocytosis and, 386-391 localization, 1-5 signals for, 2-5 orientation, 2 (table) receptors as model, 369-370 in RER, 2. 4, 6-7 terminology, 5 turnover, definition, 370-372 Membrane trigger hypothesis bacteriophage M I 3 procoat protein insertion, 77 protein posttranslational translocation, 170171 Microsomes in developing sarcoplasmic reticulum, 34935 I heterogeneity, 35 1-353 low-density, insulin receptor detection, 490492 protein sorting in cell-free system, 269

51 1 Mitochondria biogenesis, 295-297 imported proteins assembly, 312-313 distribution, 313 inner membrane, protein transport, 301-304 intermembrane space protein import to, 305-306 protein processing, 307-3 I0 outer membrane biogenesis, 310-312 receptors, 299-301 solubilization and reconstitution, 3 I8 protein precursors, see Mitochondrial proteins, precursors ribosomes bound to, 313-316 Mitochondria1 proteins assembly, 312 insertion into outer membrane, 310-3 12 maturation from precursors, 307-3 10 precursors binding to receptors, 299-301 distinction from mature proteins, 297-298 isolation and characterization, 3 17-3 18 N-terminal regions, 323-327 proteolytic processing, 304-305 transport conserved among different species, 328-329 cotranslational, 313-314 energy-dependent across inner mrmbrane, 301-304 to intermembrane space, 305-306 posttranslational, 3 14-316 submitochondrial distribution, 313 Mucolipidosis, I1 and 111, lysosomal enzyme mislocation, 272 Mutants LamB protein export-defective, E . coli, 122- I24 export information in mature protein, 133- I35 signal peptide alteration, 123 (table) interpretation, 126-127 maltose-binding protein export-defective, E . coli

export information in mature proteins, 130-133 intragenic suppressor mulEAfZ-f8, 127130

512

INDEX

maltose-resistant malE-lacZ strains, 113116 signal peptide alterations, 116-1 19 interpretation, 116-1 19 of prolipoprotein signal peptides, E. coli export-defective, 8 1-86 site-specific, oligonucleotide-directed, 8691 secretory, protein sorting, yeast, 270-271

N

NADH-cytochrome c reductase, in adipocyte membrane fractions, 464 5'-Nucleotidase, in adipocyte membrane fractions, 464

0

Obesity, insulin-resistant glucose uptake, 47948 I Oligonucleotides, mutagenic site-specific mutants in signal peptides, 8688 Oligosaccharides, asparagine-linked biosynthesis, 184 in glucose absence, 224 lipid intermediates, 185-191 structure, 182-183 transfer from lipid to protein cotranslational, 215-216 endoglycosidase H and, 208-210 glucose effect, 211-212 mechanism of, 215-216 posttranslational, 21 6 Oligosaccharide lipids containing N-acetylglucosamine and mannose, 197-199 containing N-acetylglucosamine, mannose, and glucose, 199-200 endogenous in eukaryotic cells, 207-208 formation from trisaccharides, 201-208 glycosylation, 205-207 mannosyl residue transfer, 202-205

Oligosaccharide transferase peptide substrates, 213-216 purification, 213 solubilization, 212-213 subcellular location, 213 in endoplasmic reticulum, 220 OmpA protein, E. coli export, effect of maltose-binding protein mutants, 131-132 Ornithine transcarbamylase, mitochondrial conformation, 3 13 maturation from precursor, 309-3 10 Ovalbumin internal signal sequences, 20-21

P

Phagocytosis, 415 membrane fragment recycling and, 428 Phosphatases, dolichol-metabolizing subcellular distribution, liver, 191 Pinocytosis, 415 membrane fragment recycling and, 424430 Porin, mitochondrial insertion into outer membrane, 310-31 1 properties, 3 1 I synthesis on free polysomes, 310 Proline in signal peptides, 71 mutations, 88-89 Prolipoprotein secretion steps maturation and assembly, 94-95 membrane association, 92-93 signal peptide cleavage, 94 signal peptide modification, 93-94 translocation, 93 signal peptide mutants, 86-87 amino acid sequences, 87 (table) translocation, E. coli internal signal peptide and, 80 orientation, 80 signal peptide, loop model, 79-81 Proteases, mitochondrial imported protein processing, 319-320 isolation from matrix, 3 18-3 I 9

513

INDEX

Proteins, see also sperifj:r proteins glycosylation dolichyl phosphate level and, 217-219 of enzymes phospholipid effects, 225-226 in RER, orientation, 220-223 glucose starvation and, 223-225 of intermediates, orientation in RER, 22 1-222 protein synthesis inhibition and, 2 19-220 GTP role, 220 integral membrane, see Integral membrane proteins lysosomal, see Lysosomal proteins membrane, see Membrane proteins mitochondrial, see Mitochondria1 proteins secretory, see Secretory proteins sorting constitutive and regulated pathways, 276 genetic engineering and, 277-279 in genetic systems, mutants human, 171-172 yeast, 270-271 in reconstitution systems Golgi membranes, 270 microsomal membranes, 268-269 translocation cotranslational, 153- 156 fusion-fission interaction between intracellular compartments and, 257 lateral mobility restriction and, 257-258 posttranslational, 255-257 Protein synthesis coupling with export, E . coli. 142-145 inhibition, protein glycosylation and, 219220 in sarcoplasmic reticulum, Ca2 effect, 358-360 sites of, 253 Proteolipids, low-molecular-weight, in sar. coplasmic reticulum, 341 +

R

Receptors association with coated pits, 388-396, 440441, 424

mechanism of, 393-396 biosynthesis-degradation rate, 403-404 dissociation from ligand, 443-446 lysosome role, 444-445 down-regulation, terminology, 371 endocytosis mediation, 441 -442 internalization rate apparent in cultured cells, 388-389 lateral diffusion coefficient and, 389-391 ligand binding computer programs for, 378 equilibrium at low temperature, 376-377 fluorimetry, 373 kinetic analysis, 373-376 rnorphometry, 372 radiolabeling, 378 steady-state analysis, 379-384 endocytic down-regulation, 38 1-382 internalization rate, 383-384 visualization, 372 recycling evidence for, 397-399, 425-430, 442443 ligand properties and, 442-443 models of, 400-403 Ribophorins, on RER, ribosome binding, 24 Ribosomes mitochondria-bound, protein transport and, 3 13-316 RER interaction with through ribophorins I and 11, 24 through SRP, 24-25, 75-76 secretory protein translocation and, 6- 13 RNA, messenger (mRNA) secretory protein translocation and, 6-7, 11-14 Rough endoplasmic reticulum (RER) Ca2+-ATPase synthesis, 352 calsequestrin synthesis, 352 docking protein, 23-24 glycosylation intermediates and enzymes. orientation, 220-223 membrane proteins and, 2, 4, 6-7 secretory protein translocation across, 12, 14-15, 28-31 energy for, 26 signal peptidase. 27-28 signal peptide receptor, see Signal recognition particle virus budding from, 275-276

514

INDEX

S

Saccharomyces cerevisiae mitochondria1 protein genes, cloning, 321323 secretory mutants, 270-271 glycoprotein transport, 27 1 secretory enzyme synthesis, 271 Sarcoplasmic reticulum biogenesis models, 35 1-354 endoplasmic reticulum and, 352-354 in differentiating muscle cells function and composition, 345-348 microsome fractions, 349-35 1 phospholipid-synthesizing enzymes, 348349 ultrastructure, 343-345 transverse tubules, 344 triad formation, 344-345 of fast and slow muscles, 360-361 proteins composition, 339-342 synthesis, Ca2+ effect, 3.58-360 structure, 338-339 Secretory proteins molecular sorting in endoplasmic reticulum, 266-268 Golgi apparatus, 260-262 lysosomes, 262 prokaryotic posttranslational translocation, 76-78 protein required for secretion, 96-97 signal peptides, 65-98 synthesis and processing, 259 SRP-receptor and, 23-24 translocation, 5- I 1 energy for, 25-26 in E. coli. 25 internal signal sequences and, see Internal signal peptides mechanism of, 28-30 ribosomes and, 24-25 signal sequences and, see Signal peptides Semliki Forest virus, proteins signal peptides, 22, 41 synthesis and membrane insertion, 36-37 Serine in signal peptides, 71 mutations, 89-90

Serum albumin, galactosylated intravenous:y injected, fate in liver, 44645 1 transport in hepatocytes, 445-446 Signal hypothesis direct transfer model, 170, 171-173, 253 formulation, 7 , 11, 151-152, 157, 252, 255 history, 5-7 linear model, 78 loop model, 21, 78-79, 170 evidence for, 79-81 revision, SRP and SRP-receptor discovery, 75-76 Signal peptidase eukaryotic from RER, 27-28 prokaryotic from E. coli in cell-free system, 79-80 SPase I, characteristics, 28, 95 SPase 11, characteristics, 28, 95-96 Signal peptides cleavage protein transport and, 26-27, 79-80 signal peptidase and, 27-28, 79-80 site of, 158-159 experimental changes, 162- 163 functions, 157 hydrophobic core, 159- 162 N-terminal region, 160-161 recognition by SRP, 18-20 internal, see Internal signal peptides recognition by membranes receptor nature, 14- 18 ubiquitous mechanism, 13-14 secondary structure, 161-162 secretory protein translocation and, 11- 13, 30, 34-35, 66 Signal pedtides, eukaryotic amino acid sequences, 8-9 comparison with prokaryotic, 73 hydrophobicity, 161 Signal peptides, prokaryotic amino acid sequences, 9, 66-71 in mutants, 84-85, 87 amino-terminal region, 67, 71 in mutants, 86-88 cleavage site, 72 in mutants, 90-91 comparison with eukaryotic, 73 in different bacteria, 68-70, 72-73

INDEX

hydrophobic domain, 7 I Sucrase-isomaltase, embedding into memin mutants, 88-90 brane, 33 of LamB protein mutants, 122-124 amino acid sequences, 123 (table) functions, hydrophobic axis length and, T 124-127 of maltose-binding protein in export-defective mutants, I 16- I19 Temperature effects amino acid sequences, 117 (table) on insulin receptor internalization, 491 functions, hydrophobic axis length and, on insulin-stimulated glucose uptake, 473124- 127 474 in intragenic malEAI2-18 suppressor on ligand binding to receptors, 376-377 strains, 129- I30 Threonine, in signal peptides, 71 mutants mutations, 89-90 export-defective, 8 1-86 site-specific, oligonucleotide-directed, 86- Transfemn degradation product release, 434-437 91 iron uptake from, 434-436 in prolipoprotein models of, 438-439 P-turn structure, 80 lysosomal accumulation, 434 cleaved, location, 79-80 receptor-mediated endocytosis, 432-433 during secretion, 92-95 Trisaccharide lipids. conversion to oligosacsecondary structure, 74-75 charides, 201-208 a-helical conformation, 74 Tyrosine kinase, insulin receptor activity. 48X Signal peptide receptor, see Signal recognition particle (SRP) Signal recognition particle (SRP) binding by docking protein, 23-24 V evidence for, 14-15 isolation from RER, 15 -like factor, E . coli. 17-18 Viruses, enveloped membrane protein translocation and, 35-36 glycoprotein sorting in infected cells, 263properties, 15-17 266, 273-275 ribosome interaction with, 24-25, 76 budding from internal cell membranes secretory protein translocation and, 75-76 and, 275-276 signal sequence recognition, 18-20 genetic engineering and, 277-279 Sorting signals polarized budding and, 273-275 cotranslational translocation and, 254-255 Vitellogenin receptor, recycling, Xenopus decoding mechanism, 256 oocytes, 397-398 for endocytosed ligands, 267 genetic engineering study, 277-279 posttranslational translocation and, 254-255 primary, functions, 253-255 X secondary, functions, 253-256 SRP, see Signal recognition particle SRP-receptor, in RER membranes, 23-24 Xylosylphosphorylpolyisoprenol.in protein protein translocation and, 35, 75 glycosylation, 194

This Page Intentionally Left Blank

Contents of Recent Volumes Volume 11 Cell Surface Glycoproteins: Structure, Biosynthesis, and Blological Functions The Cell Membrane-A Short Historical Perspective ASbR ROTHSTEIN The Structure and Biosynthesis of Membrane GI ycoproteins JENNIFER STURCESS, AND MARIOMOSCARELLO, HARRY SCHACHTER Techniques for the Analysis of Membrane Glycoproteins R. L. JULIANO Glycoprotein Membrane Enzymes AND JOHNR. RIORDAN GORDONG. FORSTNER Membrane Glycoproteins of Enveloped Viruses RICHARD W. COMPANSA N D MAURICE c. KEMP Erythrocyte Glycoproteins MICHAEL J. A. TANNER Biochemical Determinants of Cell Adhesion LLOYDA. CULP Proteolytic Modification of Cell Surface Macromolecules: Mode of Action in Stimulating Cell Growth KENNETH D. NOONAN Glycoprotein Antigens of Murine Lymphocytes MICHELLELETARTE Subject Index

Volume 12 Carrlers and Membrane Transport Proteins Isolation of Integral Membrane Proteins and Criteria for Identifying Camer Proteins MICHAEL J. A. TANNER

The Carrier Mechanism S . B. HLADKY The Light-Driven Proton Pump of Halobucrerium hulobium: Mechanism and Function AND MICHAEL EISENBACH S. ROY CAPLAN Erythrocyte Anion Exchange and the Band 3 Protein: Transport Kinetics and Molecular Structure PHILIP A . KNAIJP The Use of Fusion Methods for the Microinjection of Animal Cells R. G . KULKAA N D A. LOYTER Subject Index

Volume 13 Cellular Uechanlsms of Renal Tubular Ion Transport PART I: ION ACTIVITY AND ELEMENTAL COMPOSITION OF INTRAEPITHELIAL COMPARTMENTS Intracellular pH Regulation WALTERF. BORON Reversal of the pH,-Regulating System in a Snail Neuron R . C. THOMAS How to Make and Use Double-Barreled IonSelective Microelectrodes THOMASZUETHEN The Direct Measurement of K. CI. Na, and H Ions in Bullfrog Tubule Cells MAMORUFUJIMOTO, KUNIHIKO KOTERA,A N D YUTAKAMATSUMURA Intracellular Potassium Activity Measurements in Single Proximal Tubules of Necrurus Kidney TAKAHIRO KUBOTA,BRUCEBIAGI,A N D GERHARD GIEBISCH 517

518 lntracellular Ion Activity Measurements in Kidney Tubules RAJAN. KHURI Intracellular Chemical Activity of Potassium in Toad Urinary Bladder JOELDELONGA N D MORTIMER M. CIVAN Quantitative Determination of Electrolyte Concentrations in Epithelial Tissues by Electron Microprobe Analysis ROGERRICK,ADOLFDORGE, RICHARD BAUER,FRANZBECK, JUNEMASON,CHRISTIANE ROLOFF, A N D KLAUSTHURAU PART 11: PROPERTIES OF INTRAEPITHELIAL MEMBRANE BARRIERS IN THE KIDNEY Hormonal Modulation of Epithelial Structure JAMESB. WADE Changes in Cell Membrane Surfaces Associated with Alterations of Transepithelial Ion Movement MICHAEL KASHGARIAN The Dimensions of Membrane Barriers in Transepithelial Flow Pathways LARRYW. WELLINGA N D DANJ . WELLING Electrical Analysis of lntraepithelial Barriers EMILEL. BOULPAEP AND HENRYSACKIN Membrane Selectivity and Ion Activities of Mammalian Tight Epithelia SIMONA. LEWIS.NANCYK. WILLS, A N D DOUGLAS C. EATON Ion Conductances and Electrochemical Potential Differences across Membranes of Gallbladder Epithelium Luis REUSS A Kinetic Model for Ion Fluxes in the Isolated Perfused Tubule BRUCEBIAGI, ERNESTOGONZALEZ, A N D GERHARD GIEBISCH The Effects of Voltage Clamping on Ion Transport Pathways in Tight Epithelia ARTHURL. FINN AND PAULA R ~ G E N E S Tubular Permeability to Buffer Components as a Determinant of Net H Ion Fluxes

CONTENTS OF RECENT VOLUMES

G . MALNIC, v. L. COSTA S I L V A , s. s. CAMPIGLIA, M. DE MELLOAIRES,A N D G. GIEBISCH tonic Conductance of the Cell Membranes and Shunts of Necrurus Proximal Tubule GENJIRO KIMURAA N D KENNETH R. SPRING Luminal Sodium Phosphate Cotransport as the Site of Regulation for Tubular Phosphate Reabsorption: Studies with Isolated Membrane Vesicles HEIN1 MURER,REINHARD STOLL, CARLAEVERS,ROLFKINNE, A N D JEAN-PHILIPPE BONJOLJR, AND HERBERT FLEISCH The Mechanism of Coupling between Glucose Transport and Electrical Potential in the Proximal Tubule: A Study of Potential-Dependent Phlorizin Binding to Isolated Renal Microvillua Membranes PETERS . ARONSON Electrogenic and Electroneutral Na GradientDependent Transport Systems in the Renal Brush Border Membrane Vesicle BERTR4M SACKTOR PART 111: INTRAMEMBRANE CARRIERS AND ENZYMES IN TRANSEPITHELIAL TRANSPORT Sodium Cotransport Systems in the Proximal Tubule: Current Developments R. KINNE,M. BARAC,A N D H. MLJRER ATPases and Salt Transport in the Kidney Tubule MARCARITA PEREZ-GONZALEZ DE LA MANNA,FULGENCIO PROVERRIO, AND GUILLERMO WHITEMRURY Further Studies on the Potential Role of an Anion-Stimulated Mg-ATPase in Rat Proximal Tubule Proton Transport E. KINNE-SAFFRAN A N D R. KINNE Renal Naf -K+-ATPase: Localization and Quantitation by Means of Its K+-Dependent Phosphatase Activity REINIER BEEUWKES Ill A N D SEYMOUR ROSEN Relationship between Localization of N + -K+-ATPase, Cellular Fine Structure,

CONTENTS OF RECENT VOLUMES

and Reabsorptive and Secretory Electrolyte Transport A . ERNST, STEPHEN CLARAV. RIDDLE,A N D KARL.J . KARNAKY. JR. Relevance of the Distribution of Na Pump Sites to Models of Fluid Transport across Epithelia JOHNW. MILLSA N D DONALDR. DIBONA Cyclic AMP in Regulation of Renal Transport: Some Basic Unsolved Questions THOMASP. DOLJSA Distribution of Adenylate Cyclase Activity in the Nephron F. MOREL,D. C H A B A R D ~ S . A N D M. IMBEKT-TEBOUL +

Subject Index

Volume 14 Carriers and Membrane Transport Proteins Interface between Two Immiscible Liquids as a Tool for Studying Membrane Enzyme Systems L. I. B(XUSLAVSKY Criteria for the Reconstitution of Ion Transport Systems ADII. E. SHAMOO AND WILLIAM F. TIVOL The Role of Lipids in the Functioning of a Membrane Protein: The Sarcoplasmic Reticulum Calcium Pump J. P. BENNET.K . A. MCGILL,A N D G . B. WARREN The Asymmetry of the Hexose Transfer System in the Human Red Cell Membrane W. F. WIDDAS Permeation of Nucleosides, Nucleic Acid Bases, and Nucleotides in Animal Cells PETERG. W. PLAGEMANN AND ROBERT M. WOHLHUETER Transmembrane Transport of Small Peptides D. M. MATTHEWSA N D J . W. PAYNE Characteristics of Epithelial Transport in Insect Malpighian Tubules S. H. P. MADDRELL Subject Index

Volume 15 Molecular Mechanisms of Photoreceptor Transduction PART I: THE ROD PHYSIOLOGICAL RESPONSE The Photocurrent and Dark Current of Retinal Rod5 G . MATHEWS A N D D. A. BAYLOR Spread of Excitation and Background Adaptation in the Rod Outer Segment K.-W. YAU,T. D. LAMB,A N D P. A. MCNAUGHTON Ionic Studies of Vertebrate Rods W. GEOFFREY OWENA N D V I N U ~ NTORRE T Photoreceptor Coupling: Its Mechanism and Consequences GEOH-REYH. GOLD PART 11: THE CYCLIC NUCLEOTIDE ENZYMATIC CASCADE AND CALCIUM ION First Stage of Amplification in the CyclicNucleotide Cascade of Vision LUBERTSTRYLR.JAMESB . HLIRLEY, AND BERNARD K.-K. FUNG Rod Guanylate Cyclase Located in Axonemes DARRELL FLEISCHMAN Light Control of Cyclic-Nucleotide Concentration in the Retina THOMASG. EBREY,PAULKILBRIDE, JAMESB. HURLEY,ROGERCALHOON, A N D MOTOYUKI TSUDA Cyclic-GMP Phosphodiesterase and Calmodulin in Early-Onset Inherited Retinal Degenerations G . J . CHADER,Y. P. LIU, R. T. FLETCHER. G . AGLJIRRE, A N D M. T'so R. SANTOS-ANDERSON. Control of Rod Disk Membrane Phosphodiesterase and a Model for Visual Transduct ion P. A. LIEBMAN AND E. N. PUGH,JR. Interactions of Rod Cell Proteins with the Disk Membrane: Influence of Light, Ionic Strength, and Nucleotides HERMANN KUHN

CONTENTS OF RECENT VOLUMES

520 Biochemical Pathways Regulating Transduction in Frog Photoreceptor Membranes M. DERICBOWNDS The Use of Incubated Retinas in Investigating the Effects of Calcium and Other Ions on Cyclic-Nucleotide Levels in Photoreceptors ADOLPH1. COHEN Cyclic A M P Enrichment in Retinal Cones DEBORAB. FARBER Cyclic-Nucleotide Metabolism in Vertebrate Photoreceptors: A Remarkable Analogy and an Unraveling Enigma M. w. BITENSKY, G . L. WHEELER, A. YAMAZAKI, M. M. RASENICK, AND P. J . STEIN Guanosine Nucleotide Metabolism in the Bovine Rod Outer Segment: Distribution of Enzymes and a Role of GTP HITOSHISHICHI Calcium Tracer Exchange in the Rods of Excised Retinas ETE Z. SZUTS The Regulation of Calcium in the Intact Retinal Rod: A Study of Light-Induced Calcium Release by the Outer Segment GEOFFREY H. GOLD A N D JUANI. KORENBROT Modulation of Sodium Conductance in Photoreceptor Membranes by Calcium Ions and cGMP ROBERTT. SORBI PART 111: CALCIUM, CYCLIC NUCLEOTIDES, AND THE MEMBRANE POTENTIAL

SANFORD E. OSTROY, EDWARD P. MEYERTHOLEN, PETERJ. STEIN, ROBERTAA. SVOBODA, A N D MEEGAN J . WILSON [Ca2+Ii Modulation of Membrane Sodium Conductance in Rod Outer Segments BURKSOAKLEYI1 A N D LAWRENCE H. PINTO Cyclic-GMP-Induced Depolarization and Increased Response Latency of Rods: Antagonism by Light WILLIAM H. MILLERA N D GRANTD. NicoL PART IV: AN EDITORIAL OVERVIEW Caz+ and cGMP WILLIAM H. MILLER Index

Volume 16 Electrogenlc Ion Pumps PART I. DEMONSTRATION OF PUMP ELECTROGENICITY IN EUKARYOTIC CELLS Electrophysiology of the Sodium Pump in a Snail Neuron R. C. THOMAS Hyperpolarization of Frog Skeletal Muscle Fibers and of Canine Purkinje Fibers during Enhanced Na+-K + Exchange: Extracellular K Depletion or Increased Pump Current? DAVIDC. GADSBY The Electrogenic Pump in the Plasma Membrane of Nitella ROGERM. SPANSWICK Control of Electrogenesis by ATP, Mg2 , H , and Light in Perfused Cells of Chara MASAHITAZAWAA N D TERUOSHIMMEN +

Calcium and the Mechanism of Light Adaptation in Rods BRUCEL. BASTIANAND GORDONL. FAIN Effects of Cyclic Nucleotides and Calcium Ions on Bufo Rods JOEL E. BROWNAND GERALDINE WALOGA The Relation between Ca2 and Cyclic GMP in Rod Photoreceptors STUARTA. LIPTONAND JOHNE. DOWLING Limits on the Role of Rhodopsin and cGMP in the Functioning of the Vertebrate Photoreceptor

+

+

+

PART 11. THE EVIDENCE IN EPITHELIAL MEMBRANES An Electrogenic Sodium Pump in a Mammalian Tight Epithelium S. A. LEWISAND N. K. WILLS

521

CONTENTS OF RECENT VOLUMES

A Coupled Electrogenic Na+-K+ Pump for Mediating Transepithelial Sodium Transport in Frog Skin ROBERTNIELSEN Transepithelial Potassium Transport in Insect Midgut by an Electrogenic Alkali Metal Ion Pump MICHAELG. WOLFERSBERGER, WILLIAM R. HARVEY,A N D MOIRACIOFPE The ATP-Dependent Component of Gastric Acid Secretion G. SACHS,B. WALLMARK. G . SACCOMANI, E. RABON. H. B. STEWART.0. R. DIBONA,A N D T. BERGLINUH PART 111. REVERSIBILITY: ATP SYNTHESIS DRIVEN BY ELECTRIC FIELDS Effect of Electrochemical Gradients on Active H Transport in an Epithelium QAISAL-AWQATIAND TROYE. DIXON Coupling between H + Entry and ATP Synthesis in Bacteria PETERC. MALONEY Net ATP Synthesis by H -ATPase Reconstituted into Liposomes YASUOKACAWA Phosphorylation in Chloroplasts: ATP Synthesis Driven by A$ and by ApH of Artificial or Light-Generated Origin PETERGRABER +

+

PART IV. SOME THEORETICAL QUESTIONS Response of the Proton Motive Force to the Pulse of an Electrogenic Proron Pump ERICHHEINZ Reaction Kinetic Analysis of Current-Voltage Relationships for Electrogenic Pumps in Neurospora and Acetabularia DIETRICHGRADMANN, ULF-PETERHANSEN,A N D CLIFFORDL. SLAYMAN Some Physics of Ion Transport HAROLDJ . MOROWITZ

PART V. MOLECULAR MECHANISMS OF CHARGE SEPARATION

An H -ATP Synthetase: A Substrate Translocation Concept 1. A. KOZLOVA N D V. P. SKULACHEV Proton Translocation by Cytochrome Oxidase MARTENW I K S T R ~ M Electrogenic Reactions of the Photochemical Reaction Center and the UbiquinoneCytochrome b/r2 Oxidoreductase P. LESLIEDUITON, PAULMUELLER. DANIELP. O’KEEFE, NIGELK. PACKHAM, ROGERC. PRINCE,A N D DAVIDM. TIEDE Proton-Membrane Interactions in Chloroplast Bioenergetics R. A. DILLEY.L. J . PROCHASKA, 0. M. BAKER,N. E. TANDY,AND P. A. MILLNER Photochemical Charge Separation and Active Transport in the Purple Membrane BARRYHONIG Mitochondrial Transhydrogenase: General Principles of Functioning I. A. KOZLOV Membrane Vesicles, Electrochemical Ion Gradients, and Active Transport H. R. KABACK +

PART VI. BIOLOGICAL SIGNIFICANCE OF ELECTROGENIC ION PUMPS

The Role of Electrogenic Proton Translocation in Mitochondria1 Oxidative Phosphorylation JANNA P. WEHRLE Electrogenic Reactions and Proton Pumping in Green Plant Photosynthesis WOLFGANG JUNGE The Role of the Electrogenic Sodium Pump in Controlling Excitability in Nerve and Cardiac Fibers MARIOVASSALLE Pumps and Currents: A Biological Perspective FRANKLIN M. HAROLD Index

522 Volume 17 Membrane Liplds of Prokaryotes Lipids of Prokaryotes-Structure and Distribution HOWARDGOLDFINE Lipids of Bacteria Living in Extreme Environments THOMASA. LANGWORTHY Lipopolysaccharides of Gram-Negative Bacteria OTTO L~JDERIZ, MARINAA. FREUDENBERG, CHRISGALANOS, VOLKERLEHMANN, ERNSTTH. RIETSCHEL,A N D DEREKH. SHAW Prokaryotic Polyterpenes: Phylogenetic Precursors of Sterols GUY OURISSON AND MICHELROHMER Sterols in Mycoplasma Membranes SHMUELRAZIN Regulation of Bacterial Membrane Lipid Synthesis CHARLES0. ROCK AND JOHNE. CRONAN, JR. Transbilayer Distribution of Lipids in Microbial Membranes SHLOMOROTTEM Lipid Phase Transitions and Regulation of Membrane Fluidity in Prokaryotes DONALDL. MELCHIOR Effects of Membrane Lipids on Transport and Enzymic Activities RONALDN. MCELHANEY Index

Volume 18 Membrane Receptors PART I. ADENYLATE CYCLASE-RELATED RECEPTORS Hormone Receptors and the Adenylate Cyclase System: Historical Overview B. RICHARD MARTIN The Elucidation of Some Aspects of Receptor Function by the Use of a Kinetic Approach A. M. TOLKOVSKY The P-Adrenergic Receptor: Ligand Binding

CONTENTS OF RECENT VOLUMES

Studies Illuminate the Mechanism of Receptor- Adenylate Cyclase Coupling JEFFREY M. STADEL A N D ROBERT J. LEFKOWITZ Receptor-Mediated Stimulation and Inhibition of Adenylate Cyclase DERMOTM. F. COOPER Desensitization of the Response of Adenylate Cyclase to Catecholamines JOHN P. PERKINS Hormone-Sensitive Adenylate Cyclase: Identity, Function, and Regulation of the Protein Components ELLIOTT M. Ross, STEENE. PEDERSEN. AND VINCENTA. FLORIO The Regulation of Adenylate Cyclase by Glycoprotein Hormones BRIANA. COOKE The Activity of Adenylate Cyclase Is Regulated by the Nature of Its Lipid Environment MILESD. HOUSLAYA N D LARRYM. GORDON The Analysis of Interactions between Hormone Receptors and Adenylate Cyclase by Target Size Determinations Using Irradiation Inactivation B. RICHARD MARTIN PART 11. RECEPTORS NOT INVOLVING ADENYLATE CYCLASE

Vasopressin Isoreceptors in Mammals: Relation to Cyclic AMP-Dependent and Cyclic AMP-Independent Transduct ion Mechanisms SERGEJARD Induction of Hormone Receptors and Responsiveness during Cellular Differentiation L. MICHAELC. LIN AND SUZANNE BECKNER Receptors for Lysosomal Enzymes and Glycoproteins VIRGINIA SHEPHERD, PAUL SCHLESINGER, A N D PHILIPSTAHL The Insulin-Sensitive Hexose Transport System in Adipocytes 3. GLIEMANN AND W. D. REES Epidermal Growth Factor Receptor and Mechanisms for Animal Cell Division MANJUSRI DAS

CONTENTS OF RECENT VOLUMES

523

Ultrastructure of Na,K-ATPase in Plasma Membranes Vesicles ELISABETH SKRIVER. ARVIDB. MAUNSBACH, A N D PETERLETH JORGENSEN Electron Microscope Analysis of Two-Dimensional Crystals of Membrane-Bound Na,KATPase ARVIDB. MAUNSBACH, ELISABETH SKRIVER. HANSHEBERT,A N D PETER LETH JORGENSEN Organization of the Transmembrane Segments Volume 19 of Na,K-ATPase. Labeling of Lipid Embedded and Surface Domains of the a-Subunit and Its Structure, Mechanism, and Function of Tryptic Fragments with [ 12sl]Iodothe Na/K Pump naphthylazide, (3ZPJATP,and Photolabeled Ouabain PART I . THERMODYNAMIC ASPECTS OF PETERLETH JeRCENSEN, STEVENJ. D. MEMBRANE TRANSPORT KARLISH,A N D CARLOSGITLER Structural Studies on Lamb Kidney Na,KWhat is a Coupled Vectorial Process'? ATPase WII.I.IAMP. JENCKS J. H. COLLINS,BLISSFORBUSH111, L. The Membrane Equilibrium with Chemical K . LANE.E. LING,ARNOLD SCHWARTZ, Reactions A N D A. (REEVES) ZOT F R l t D R I C H A. SAUER Two Slightly Different a-Subunit Components of Kidney Na,K-ATPase Induced by Heat PART 11. STRUCTURAL ANALYSIS OF Treatment Na. K- ATPase T. OHTA,M. KAWAMURA, T. A N D K. HASECAWA, H. ISHIKURA, Structural Aspects of Na,K-ATPase NAGANO ROBtRT L. POST Radiation Inactivation Analysis of Na,KDetergent Solubilization of Na,K-ATPase ATPase MIKAEL. ESMANN J. CLlVE ELLORY, PAUL OTTOLENGHI, Methods for the Cleavage of the Large Subunit A N D ROGERA. KLEIN of Na,K-ATPase and the Resolution of the Stoichiometrical Binding of Ligands to Less Peptidea Produced than 160 Kilodaltons of Na,K-ATPase HENRYRODRIGUEZ, RICHARD HARKINS. H. MATSUI,Y. HAYASHI. A N D JACK KYTE H. HOMAREDA, A N D M. TACUCHI The Active Site Structure of Na,K-ATPase: Selective Purification of Na.K-ATPase and Ca'+ ,MgZ -ATPase from Eel Electroplax Location of a Specific Fluorescein IsothiocyaL. M. AMENDE,S. P. CHOCK,A N D nate-Reactive Site R. W. ALHERS CYNTHIA T. CARILLI. ROBERTA. High-Performance Gel Chromatography of FARLEY,A N D LEWISC. CANTLEY Horse Kidney Na,K-ATPase Subunit Distribution of Sulfhydryl Groups and MAKOTONAKAO,TOSHIKONAKAO, Disulfide Bonds in Renal Na.K-ATPase TOMOKOOHNO,YOSHIHIRO FUKUSHIMA, M . KAWAMURA, T. OHTA,A N D K . YUKICHI HARA,A N D MASAKOARAI NAGANO Native Membranes from Dog Kidney Outer Lipid Regions of Na,K-ATPase Examined with Medulla, Enriched in Na,K-ATPase, and Ve. Fluorescent Lipid Probes sicular in Nature KIMBERLY A. MUCZYNSKI, WARDE. BLISSFORBUSH111 HARRIS,A N D WILLIAML. STAHL The Linkage between Ligand Occupation and Response of the Nicotinic Acetylcholine Receptor PALMERTAYLOR. ROBERTDALEBROWN. A N D DAVID A. JOHNSON The Interaction of Cholera Toxin with Gangliosides and the Cell Membrane S I M O NVANHEYNINGEN Subject Index

+

524 Role of Cholesterol and Other Neutral Lipids in Na,K-ATPase J. J . H . H . M . D E P o N T , W . H . M . PETERS,AND S. L. BONTING PART 111. LIGAND INTERACTIONS: CARDIAC GLYCOSIDES AND IONS Cardiotonic Steroid Binding to Na,K-ATPase BLISSFORBUSH111 Binding of Monovalent Cations to the Na,KATPase M. YAMAGUCHI, J. SAKAMOTO, A N D Y. TONOMURA Half-of-the-Sites Reactivity of Na,K-ATPase Examined by the Accessibility of Vanadate and ATP into E n z y m e a u a b a i n Complexes OTTO HANSEN Binding of Rb+ and ADP to a Potassium-Like Form of Na,K-ATPase J0RGEN JENSEN A N D PAUL OTTOLENGHI Side-Dependent Ion Effects on the Rate of Ouabain Binding to Reconstituted Human Red Cell Ghosts H. H. BODEMANN, T. J. CALLAHAN, H. REICHMANN, A N D J. F. HOFFMAN Intracellular Sodium Enhancement of Ouabain Binding to Na,K-ATPase and the Development of Glycoside Actions TAI AKERA,KYOSUKE TEMMA, AND SATOSHIYAMAMOTO Lithium-Catalyzed Ouabain Binding to Canine Kidney Na,K-ATPase GEORGER. HENDERSON Ouabain Binding and Na,K-ATPase in Resealed Human Red Cell Ghosts D. G. SHOEMAKER A N D P. K. LAUF Stereoelectronic Interaction between Cardiotonic Steroids and Na,K-ATPase: Molecular Mechanism of Digitalis Action F. DITTRICH,P. BERLIN,K. KOPKE,A N D K. R. H. REPKE Use of Prophet and MMS-X Computer Graphics in the Study of the Cardiac Steroid Receptor Site of Na,K-ATPase DWIGHTS. FULLERTON, DOUGLASC. ROHRER,KHALILAHMED,ARTHURH. L. FROM,EITAROKITATSUJI,AND TAMBOUEDEWO

CONTENTS OF RECENT VOLUMES

Photoaffinity Labeling of the Ouabain Binding Site of Na,K-ATPase CLIFFORDC. HALLAND ARNOLDE. RUOHO New Ouabain Derivatives to Covalently Label the Digitalis Binding Site BERNARD Rossi, MAURICE GOELDNER, GILLESPONZIO,CHRISTIAN HIRTH,A N D MICHELLAZDUNSKI Ouabain Sensitivity: Diversity and Disparities JOHNS. WILLISA N D J. CLIVEELLORY PART IV: LIGAND INTERACTIONS: NUCLEOTIDES, VANADATE, AND PHOSPHORYLATION Ligand Interactions with the Substrate Site of Na,K-ATPase: Nucleotides, Vanadate, and Phosphorylation Jens G. Nerby Conformational Changes of Na,K-ATPase Necessary for Transport LEWISC. CANTLEY, CYNTHIA T. CARILLI, RODERIC L. SMITH,A N D D A V I D PERLMAN On the Mechanism behind the Ability of Na,K-ATPase LO Discriminate between Na+ and K + JENSCHR. SKOU Characteristics of the Electric Eel Na,K-ATPase Phosphoprotein ATSUNOBUYODAAND SHIZUKO YODA Sulfhydryl Groups of Na,K-ATPase: Effects of N-Ethylmaleimide on Phosphorylation from ATP in the Presence of Na+ + Mg2+ MIKAELESMANNAND IRENAKLODOS Alternative Pathways of Phosphorylation of Na,K-ATPase Regulated by Na+ Ions on Both Sides of the Plasma Membrane HORSTWALTER Structurally Different Nucleotide Binding Sites in Na,K-ATPase HERMANN KOEPSELLA N D DORISOLLIC Study of Na,K-ATPase with ATP Analogs WILHELMSCHONER, HARTMUTPAULS, ENGINH. SERPERSU, GEROLD REMPETERS, ROSEMARIE PATZELTWENCZLER,AND MARIONHASSELBERG Affinity Labeling Studies of the ATP Binding Site of Canine Kidney Na,K-ATPase

CONTENTS OF RECENT VOLUMES

JAMES B. COOPER,CARLJOHNSON.A N D CHARLESG . WINTER 3’P[’MOO]NMR Kinetic Analysis of 180Exchange Reaction between Pi and H 2 0 Catalyzed by Na,K-ATPase A. STEPHENDAHMSA N D JOELLEE. MIARA PART V . CONFORMATIONAL CHANGES, STRUCTURE/FUNCTION, AND ACTIVE SITE PROBES Principal Conformations of the a-Subunit and Ion Translocation PETER L. JORGENSEN Magnesium-Induced Conformational Changes in Na,K-ATPase S . L. BONTING.H. G. P. SWARTS,W. H. M. PETERS,F. M. A. H. SCHUURMANS STEKHOVEN, A N D I. J. H. H. M. DE PONT Rubidium Movements in Vesicles Reconstituted with Na,K-ATPase, Measured in the Absence of ATP and Pi. in the Presence of Either Ligand, and in the Presence of Both Ligands: Role of the “Occluded State” in Allowing for the Control of the Direction of Ion Movements S. J . D. KARLISHA N D W. D. STEIN Eosin: A Fluorescent Probe of ATP Binding to Na, K- ATPase J. C . SKOUA N D MIKAELESMANN Interaction of Divalent Cations with Fluorescein-Labeled Na,K-ATPase MARCIASTEINBERG, JAMESG. KAPAKOS,A N D PARIMAL C. SEN Cation Activation of Na,K-ATPase after Treatment with Thimerosal MANISHA D. MONEA N D JACKH. KAPLAN Alteration of Conformational Equilibria in Na,K-ATPdse by Glutaraldehyde Treatment DAVIDM. CHIPMAN, E. ELHANANY, R. BERGER,A N D A. LEV Conformational Transition between ADP-Sensitive Phosphoenzyme and Potassium-Sensitive Phosphoenzyme KAZUYATANIGUCHI, KUNIAKISUZUKI. AND SHOICHI IIDA

525 Relation between Red Cell Membrane Na,KATPase and Band 3 ERICT. FOSSEL A N D A. K. SOLOMON PART VI. REACTION MECHANISM AND KINETIC ANALYSIS Kinetic Analyses and the Reaction Mechanism of the Na,K-ATPase JOSEPHD. ROBINSON Evidence for Parallel Pathways of Phosphoenzyme Formation in the Mechanism of ATP Hydrolysis by Electrophorus Na,KATPase JEFFREYP. FROEHLICH, ANNS. HOBBS, A N D R. WAYNEALBERS Evaluation of the Reaction Mechanism of the Sodium Pump by Steady-State Kinetics JOHNR. SACHS Kinetic Evidence in Favor of a Consecutive Model of the Sodium Pump D. A. EISNERA N D D. E. RICHARDS Kinetic Models of Na-Dependent Phosphorylation of Na,K-ATPase from Rat Brain DONALDM. FOSTER,STANLEY 1. RUSSELL,A N D KHALILAHMED Reinvestigation of the Sequence of Sensitivity of Phosphoenzyme of Na,K-ATPase to ADP and K + during the Presteady State of the Phosphorylation by ATP Y. FUKUSHIMA AND M. NAKAO Interaction of N a + , K + , and ATP with Na,KATPase P. I. GARRAHAN, R. Rossi, A N D A. F. REGA Sodium Ion Discharge from Pig Kidney Na,KATPase YUKICHI HARAA N D MAKOTONAKAO ADP Sensitivity of the Native and Oligomycin-Treated Na,K-ATPase ANNS. HOBBS,R. WAYNEALBERS,AND JEFFREY P. FROEHLICH Three (at Least) Consecutive Phosphointermediates of Na-ATPase I . KLODOS,J. G. NORBY, AND N. 0. CHRISTIANSEN Aspects of the Presteady State Hydrolysis of ATP by Na,K-ATPase A. G. LOWE AND L. A. REEVE Identity of the Na Activation Sites in ATPase

526 with the K Activation Sites in p-Nitrophenylphosphatase L. A. PARODI,J. F. PINCUS,L. JOSEPHSON, D. J. SORCE,AND S. R. SIMON On the Existence of Two Distinct Hydrolysis Cycles for Na,K-ATPase with Only One Active Substrate Site IGORW. PLESNER Kinetic Analysis of the Effects of Na+ and K + on Na,K-ATPase LISELOTTE PLESNERAND ICOR W. PLESNER Divalent Cations and Conformational States of Na,K-ATPase JOSEPH D. ROBINSON PART VI1. ION TRANSLOCATION AND REACTION MECHANISM Na.K-ATPase: Reaction Mechanisms and Ion Translocating Steps PAULDE WEER Existence and Role of Occluded-Ion Forms of Na,K-ATPase I. M. GLYNNAND D. E. RICHARDS Na and K Fluxes Mediated by ATP-Free and ATP-Activated Na,K-ATPase in Liposomes BEATRICE M. ANNER Sidedness of Cations and ATP Interactions with the Sodium Pump L. BEAU& A N D R. DIPOLO Sidedness of Sodium Interactions with the Sodium Pump in the Absence of K + RHODABLOSTEIN Magnesium Dependence of Sodium Pump-Mediated Sodium Transport in Intact Human Red Cells P. W. FLATMAN A N D V. L. LEW K+-Independent Active Transport of Na+ by Na,K-ATPase MICHAELFORGACAND GILBERTCHIN ADP-ATP Exchange in Internally Dialyzed Squid Giant Axons PAULDE WEER,GERDAE. BREITWIESER, BRIANG. KENNEDY,AND H. GILBERTSMITH Sodium Pump-Catalyzed ATP-ADP Exchange in Red Blood Cells: The Effects of Intracellular and Extracellular Na and K Ions JACKH. KAPLAN

CONTENTS OF RECENT VOLUMES

Ouabain-Sensitive ATP-ADP Exchange and Na-ATPase of Resealed Red Cell Ghosts J. D. CAVIERES Effect of Internal Adenine Nucleotides on Sodium Pump-Catalyzed Na-Na and Na-K Exchanges BRIANG. KENNEDY, CORMLUNN,A N D JOSEPH F. HOFFMAN NalK Pump in Inside-Out Vesicles Utilizing ATP Synthesized at the Membrane ROBERTW. MERCER,BEVERLEY E. FARQUHARSON, A N D PHILIP B. DUNHAM Anion-Coupled Na Efflux Mediated by the Na/K Pump in Human Red Blood Cells S. DISSINGA N D J. F. HOFFMAN Effect of Trypsin Digestion on the Kinetic Behavior of the Na/K Pump in Intact Erythrocytes DONNAL. KROPP Sodium Movement and ATP Hydrolysis in Basolateral Plasma Membrane Vesicles from Proximal Tubular Cells of Rat Kidney F. PROVERBIO, T. PROVERBIO, AND R . MAR" Stoichiometry of the Electrogenic Na Pump in Barnacle Muscle: Simultaneous Measurement of Na Efflux and Membrane Current M. T. NELSONA N D W. J. LEDERER PART VIII. BIOSYNTHESIS, MULTIPLE FORMS. AND IMMUNOLOGY Regulation of Na,K-ATPase by Its Biosynthesis and Turnover NORMANJ. KARINA N D JOHN S . COOK Biosynthesis of Na,K-ATPase in MDCK Cells J. SHERMAN, T. MORIMOTO, A N D D. D. SABATINI Possible Functional Differences between the Two Na,K-ATPases of the Brain KATHLEEN J. SWEADNER Antigenic Properties of the a, p, and y Subunits of Na,K-ATPase WILLIAMBALL, JR., JOHNH. COLLINS, L. K. LANE,AND ARNOLDSCHWARTZ Antibodies to Na,K-ATPase: Characterization and Use in Cell-Free Synthesis Studies ALICIA MCDONOUGH, ANDREW HIATT, A N D ISIDORE EDELMAN Immunoreactivity of the a- and a(+)-Subunits

CONTENTS OF RECENT VOLUMES

of Na,K-ATPase in Different Organs and Species GERARD D. SCHELLENBERG, IRENE V. PECH,A N D WILLIAM L. STAHL Role of Na and CaZ Fluxes in Terminal Differentiation of Murine Erythroleukemia Cells I. G. MACARA,R. D. SMITH,A N D LEWIS C. CANILEY NaiK Pumps and Passive K + Transport in Large and Small Reticulocytes of Anemic Low- and High-Potassium Sheep P. K . LAW A N D G . VALET Enhancement of Biosynthesis of Na,K-ATPase in the Toad Urinary Bladder by Aldosterone But Not T3 K. GEERING. M. GIRARDET, C. BRON. A N D B. C. ROSSIER J.-P. KRAEHENBL~HL, NaKATPase Activity in Rat Nephron Segments: Effect of Low-Potassium Diet and Thyroid Deficiency LALC. GARGA N D C. CRAKTISHER Axonal Transport of Na,K-ATPase in Optic Nerve of Hamster SUSANC. SPECHT +

+

PART IX. Na,K-ATPase AND POSITIVE INOTROPY; ENDOGENOUS GLYCOSIDES Positive Inotropic Action of Digitalis and Endogenous Factors: Na,K-ATPase and Positive Inotropy; “Endogenous Glycosides” ARNOLDSCHWARTZ Endogenous Glycoside-Like Substances GARNtR T. HAUPERT, JR. Monovalent Cation Transport and Mechanisms of Digitalis-Induced lnotropy THOMASW . SMITHA N D WILLIAM H. BARRY Effects of Sodium Pump Inhibition on Contraction in Sheep Cardiac Purkinje Fibers AND D. A. EISNER,W. J. LEDERER, R. D. VAUGHAN-JONES Quantitative Evaluation of L3H]Ouabain Binding to Contracting Heart Muscle, Positive Inotropy, Na,K-ATPase Inhibition, and s6Rb+ Uptake in Several Species ERLANDERDMANN, LINDSAY BROWN, KARLWERDAN,A N D WOLFGANG KRAWIETZ

Contractile Force Effects of Low Concentrations of Ouabain in Isolated Guinea Pig, Rabbit, Cat, and Rat Atria and Ventricles GUNTER GRUPP,INGRID L. GRUPP.J. GHYSEL-BURTON, T. GODbRAlND. A. DE POVER,A N D ARNOLDSCHWARTZ Difference of Digitalis Binding to Na.KATPase and Sarcolemrna Membranes I . KUROBANE,D. L. NANDI,A N D G. T. OKITA Pharmacological and Biochemical Studies on the Digitalis Receptor: A Two-Site Hypothesis for Positive Inotropic Action ARNOLDSCHWARTZ, INGRID L. GRUPP, ROBERT J. ADAMS,TREVORPOWELL, GUNTER GRUPP,A N D E. T. WALLICK Hypothesis for the Mechanism of Stimulation of the Na/K Pump by Cardiac GlycosidesRole of Endogenous Digitalis-Like Factor T. GODFRAIND, G. CASTANEDAHERNANDEZ, J. GHYSEL-BURTON, AND A . DE POVER Immunochemical Approaches to the Isolation of an Endogenous Digoxin-like Factor KENNETHA. GRUBER, JANICEM. WHITAKER, A N D VARDAMAN M. BUCKALEW, JR. Demonstration of a Humoral Na/K Pump Inhibitor in Experimental Low-Renin Hypertension MOTILALPAMNANI, STEPHENHUOT. DAVIDCLOUGH,JAMESBUGGY,A N D FRANCIS J. HADDY Absence of Ouabain-Like Activity of the Na,K-ATPase Inhibitor in Guinea Pig Brain Extract GEORGER. KRACKE Brain Na,K-ATPase: Regulation by Norepinephrine and an Endogenous Inhibitor ALANC. SWANN Inhibitory and Stimulatory Effects of Vanadate on Sodium Pump of Cultured Heart Cells from Different Species KARLWERDAN,GERHARD BAURIEDEL, A N D ERLAND WOLFGANGKRAWIETZ, ERDMANN Endogenous Inhibitor of Na,K-ATPase: “Endodigin” K. R. WHITMER,D. EPPS, A N D ARNOLD SCHWARTZ

CONTENTS OF RECENT VOLUMES

PART X. PHYSIOLOGY AND PATHOPHYSIOLOGY OF THE Na/K PUMP Disorders in Molecular Assemblies for Na Transport in Essential Hypertension MITZYL. CANESSA, NORMAC. ADRAGNA, ISABELBIZE, HAROLD AND DANIEL C. TOSTESON SOLOMON, The Na-K Cotransport System in Essential Hypertension R. P. GARAY,C. NAZARET,A N D P. HANNAERT Loss of Na,K-ATPase Activity during Cataract Formation in Lens PARIMAL C. SEN A N D DOUGLAS R. PFEIFFER Na/K Pump: Effect of Obesity and Nutritional State M. DELUISE,P. USHER,A N D J. FLIER Decreased Na,K-ATPase Activity in Erythrocyte Membranes and Intact Erythrocytes from Obese Man DAVIDM. M o n , IWAR KLIMES,AND RANDILL. CLARK Functionally Abnormal Na/K Pump in Erythrocytes from a Morbidly Obese Subject J . FLIER,P. USHER,A N D M. DELUISE Specific Insulin Binding to Purified Na,KATPase Associated with Rapid Activation of the Enzyme JULIEE. M. MCGEOCH Mechanism for Cholinergic Stimulation of Sodium Pump in Rat Submandibular Gland DAVIDJ . STEWARTAND AMARKK. SEN Evidence for an Aldosterone-Mediated, NaDependent Activation of Na,K-ATPase in the Cortical Collecting Tubule KEVINJ. PETTY,JUHAP. KOKKO,AND DIANAMARVER Vanadate and Somatostatin Having Divergent Effects on Pancreatic Islet Na,K-ATPase KENJIlKEJIRl AND SEYMOUR R. LEVIN Phosphorylation of a Kidney Preparation of Na,K-ATPase by the Catalytic Subunit of CAMP-Dependent Protein Kinase SVENMARDH Modulation of Na,K-ATPase Activity in Rat Brain by Adenosine 3’5’-Monophosphate A N D AMARK. RUSSELLB. LINGHAM SEN Stimulation and Inhibition by Plasma of Oua-

bain-Sensitive Sodium Efflux in Human Red Blood Cells A. R. CHIPPERFIELD lnhibition of the Na Pump by Cytoplasmic Calcium in Intact Red Cells A. M. BROWNA N D V. L. LEW Involvement of Calmodulin in the Inhibition of Na,K-ATPase by Ouabain LIONELG. LLLIEVRE, M. T . PIASCIK, J . D. POTTER,E. T. WALLICK, AND ARNOLDSCHWARTZ Index

Volume 20 Molecular Approaches to Epithelial Transport PART 1. FREQUENCY DOMAIN ANALYSIS OF ION TRANSPORT Fluctuation Analysis of Apical Sodium Transport T. HOSHIKO Impedance Analysis of Necturus Gallbladder Epithelium Using Extra- and lntracellular Microelectrodes J . J. LIM, G . KOTTRA, L. KAMPMANN, A N D E. FROMTER Membrane Area Changes Associated with Proton Secretion in Turtle Urinary Bladder Studied Using Impedance Analysis Techniques CHRISCLAUSENAND TROYE. DIXON Mechanisms of Ion Transport by the Mammalian Colon Revealed by Frequency Domain Analysis Techniques N. K. WILLS Analysis of Ion Transport Using Frequency Domain Measurements SIMONA. LEWISAND WILLIAMP. ALLES Use of Potassium Depolarization to Study Apical Transport Properties in Epithelia G. PALMER LAWRENCE PART 11. USE OF ANTIBODIES TO EPITHELIAL MEMBRANE PROTEINS Biosynthesis of Na+ ,K+-ATPase in Amphibian Epithelial Cells B. C. ROSIER

529

CONTENTS OF RECENT VOLUMES

Use of Antibodies in the Study of N a + , K + ATPase Biosynthesis and Structure ALICIAM. MCDONOUGH Encounters with Monoclonal Antibodies to Na+ .K -ATPase MICHAEL KASHGARIAN, DANIEL BIEMESDERPER, A N D BLISSFORBUSH 111 Monoclonal Antibodies as Probes of Epithelial Cell Polarity GEORGEK. OJAKIAN AND DORISA. HERZLINGER Immunolabeling of Frozen Thin Sections and Its Application to the Study of the Biogenesis of Epithelial Cell Plasma Membranes IVANEMANUILOV IVANOV. HEIDE PLESKEN, DAVIDD. SABATINI, A N D J . RINDLER MICHAEL Development of Antibodies to Apical Membrane Constituents Associated with the Action of Vasopressin JAMES B. WADE, VICTORIA GUCKIAN, A N D INGEBORG KOEPPEN Molecular Modification of Renal Brush Border Maltase with Age: Monoclonal Antibody-Specific Forms of the Enzyme BtRTRAM SACKTOR A N D UZI RE~SS +

PART 111. BIOCHEMICAL CHARACTERIZATION OF TRANSPORT PROTEINS Sodium-o-Glucose Cotransport System: Biochemical Analysis of Active Sites R. KINNE,M. E. M. DA CRUZ,A N D I. T. LIN Probing Molecular Characteristics of Ion Transport Proteins DARRELL D. FANESTIL, RALPHI. KESSLER,A N D CHUNSIKPARK Aldosterone-Induced Proteins in Renal Epithelia MALCOLM COX AND MICHAELGEHEB Development of an Isolation Procedure for Brush Border Membrane of an Electrically Tight Epithelium: Rabbit Distal Colon MICHAELc. GUSTIN AND D A V I D B. P. GOODMAN Index

Volume 21 Ion Channels: Molecular and Physlologlcal Aspects Ionic Selectivity of Channels at the End Plate PETERH. BARRYAND PETERW. GAGE Gating of Channels in Nerve and Muscle: A Stochastic Approach RICHARD HORN The Potassium Channel of Sarcoplasmic Reticulum CHRISTOPHER MILLER,JOAN E. BELL, AND ANAMARIAGARCIA Measuring the Properties of Single Channels in Cell Membranes H.-A. KOLB Kinetics of Movement in Narrow Channels DAVIDG . LEVITT Structure and Selectivity of Porin Channels R. BENZ Channels in the Junctions between Cells WERNERR. LOEWENSTEIN Channels across Epithelial Cell Layers SIMONA. LEWIS,JOHN w.HANRAHAN, AND W. VAN DRIESSCHE Water Movement through Membrane Channels ALANFINKELSTEIN Channels with Multiple Conformational States: Interrelations with Carriers and Pumps P. LAUCER Ion Movements in Gramicidin Channels S. B . HLADKYAND D. A. HAYWN Index

Volume 22 The Squld Axon PART I. STRUCTURE Squid Axon Ultrastructure GLORIAM. VILLEGAS AND RAIMUNWVILLEGAS The Structure of Axoplasm RAYMOND I. LASEK PART 11. REGULATION OF THE AXOPLASMIC ENVIRONMENT Biochemistry and Metabolism of the Squid Giant Axon

CONTENTS OF RECENT VOLUMES

HAROLDGAINER,PAULE. GALLANT, ROBERTCOULD,A N D HARISHC. PANT Transport of Sugars and Amino Acids P. F. BAKERAND A. CARRUTHERS Sodium Pump in Squid Axons Luis BEAU& Chloride in the Squid Giant Axon JOHNM. RUSSELL Axonal Calcium and Magnesium Homeostasis P. F. BAKERAND R. DIPOLO Regulation of Axonal pH WALTERF. BORON Hormone-Sensitive Cyclic Nucleotide Metabolism in Giant Axons of Loligo P. F. BAKERA N D A. CARRUTHERS PART 111. EXCITABILITY Hodgkin-Huxley: Thirty Years After H. MEVES Sequential Models of Sodium Channel Gating AND CLAYM. ARMSTRONG DONALDR. MATTESON Multi-Ion Nature of Potassium Channels in Squid Axons TED BEGENISICH A N D CATHERINE SMITH Noise Analysis and Single-Channel Recordings FRANCOCONTI Membrane Surface Charge A N D GERALD DANIELL. GILBERT EHRENSTEIN Optical Signals: Changes in Membrane Structure, Recording of Membrane Potential, and Measurement of Calcium LAWRENCE B. COHEN,DAVIDLANDOWNE, LESLIEM. LOEW,AND BRIANM. SALZBERG Effects of Anesthetics on the Squid Giant Axon

D. A. HAYDON, J. R. ELLIOTT, AND B. M. HENDRY Pharmacology of Nerve Membrane Sodium Channels TOSHIONARAHASHI PART 1V. INTERACTION BETWEEN GIANT AXON AND NEIGHBORING CELLS The Squid Giant Synapse RODOLFOR. LLINAS

Axon-Schwann Cell Relationship JORGEVILLEGAS Index

Volume 23 Genes and Membranes: Transport Proteins and Receptors PART I: RECEPTORS AND RECOGNITION PROTEINS

Sensory Transduction in Bacteria MELVINI. SIMON,ALEXANDRA KRIKOS, NORIHIRO MUTOH,AND ALANBOYD Mutational Analysis of the Structure and Function of the Influenza Virus Hemagglutinin MARY-JANE GETHING,CAROLYN DOYLE, MICHAELROTH, AND JOE SAMBROOK PART 11: CHANNELS Ca2+ Channels of Paramecium: A Multidisciplinary Study CHINGKUNGAND YOSHIROSAIMI Studies of Shaker Mutations Affecting a K + Channel in Drosophila LILYYEH JAN, SANDRA BARBEL, LESLIETIMPE,CHERYLLAFPTR, LAWRENCE SALKOFF, PATRICKO'FARRELL, AND YUHNUNGJAN Sodium Channels in Neural Cells: Molecular Properties and Analysis of Mutants WILLIAMA. CAITERALL,TOHRUGONOI, AND MARIACOSTA PART 111: TRANSPORT SYSTEMS The Histidine Transport System of Salmonella ryphimurium GIOVANNA FERRO-LUZZIAMES A Study of Mutants of the Lactose Transport System of Escherichia coli T. HASTINGSWILSON, DONNASETO-YOUNG, SYLVIEBEDU, RESHAM. h T Z R A T H , AND BENNOMOLLER-HILL The Proton-ATPase of Escherichia coli A. E. SENIOR

531

CONTENTS OF RECENT VOLUMES

The Kdp System: A Bacterial K + Transport ATPase WOLFGANGEPSTEIN Molecular Cloning and Characterization of a Mouse Ouabain Resistance Gene: A Genetic

Approach to the Analysis of the Na ,K -ATPase ROBERTLEVENSON +

Index

+

This Page Intentionally Left Blank