Current Topics in Membranes and Transport. Volume 11: Cell Surface Glycoproteins: Structure, Biosynthesis, and Biological Functions (v. 11)

Current Topics in Membranes and Transport Volume 11 Ce llSutfaee Glycoproteins Structure, Biosynthesis, and Biological Functions

Advisory Board

I . S . Edelman Alvin Essig Franklin M . Harold James D. Jamieson Anthony Martonosi Shmuel Raxin Martin Rodbell Aser Rothstein Stanley G . Schultz Contributom

Richard W. Compans Lloyd A . Culp Gordon G . Forstner R. L. Juliano Maurice C . Kemp Michelle Letarte Mario Moscarello Kenneth D. Noonan John R. Riordan Aser Rothstein Harry Schachter Jennifer Sturgess Michael j . A. Tanner

Current Topics in Membranes and Transport Edited by Felix Bronner

Amort Kleinzeller

Department of Oral Biology University of Connecticut Health Center Farmington, Connecticut

Department of Physiology University of Pennsylvania School of Medicine Philadelphia, Pennsyloania

VOLUME 11 Cell Surface Giycqproteinx Structure, Biosynthesis, and Biological Functions Guest Editors R. 1. Juliano

h e r Rothrtein

Research Institute The Hospital for Sick Children Toronto, Ontario, Canada

Research Institute The Hospital for Sick Children Toronto, Ontario, Canada

1978

Academic Prom

New Yo&

San Fmncisco

London

A Subsidiary of Harcourt Brace jovanovich, Publishers

COPYRIGHT @ 1978,BY ACADEMIC PRESS,INC. ALL RIGHTS RESERVED. NO PART OF THIS PUBLICATTON 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. 111 Fifth Avenue, New

York, New York 10003

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- 11709 1 ISBN 0-1 2-15331 1-5 PHINTED IN THE UNITED STATES OF AMERICA

79808182

9 8 7 6 5 4 3 2

List of Contributors, ix Preface, xi Contents of Previous Volumes, xiii The Cell Membrane-A

Short Historical Perspective

ASER ROTHSTEIN Text, 1 References, 12

The Structure and Biosynthesis of Membrane Glycoproteins JENNIFER STURGESS, MARIO MOSCARELLO, AND HARRY SCHACHTER I. 11. 111. IV. V. VI.

Carbohydrate Asymmetry Across the Membrane Bilayer, 16 Glycoprotein Structure, 19 Model Membrane Systems, 34 The Glycosylation Reaction, 37 Subcellular Sites of Glycosylation, 61 Membrane Biogenesis, 67 References, 85

Techniques for the Analysis of Membrane Glycoproteins

H. L. JULIANO I. 11. 111. IV. V.

Introduction, 107 Identification of Cell Surface Glycoproteins, 108 Fractionation of Membrane Glycoproteins, 117 Chemical Analysis of Membrane Glycoproteins, 129 Genetic Analysis of Membrane Glycoproteins, 133 References. 134

Glycoprotein Membrane Enzymes JOHN R. RIORDAN AND GORDON G. FORSTNER I. Introduction, 146 11. Specific Enzymes, 147 111. Membrane Association, 188 IV. Structure, 1% V

vi

CONTENTS

V. Functional Interrelationships, 201 VI. Biosynthetic and Developmental Aspects, 206 VII. Are All Ectoenzymes Glycoproteins?, 210 References, 210

Membrane Glycoproteins of Enveloped Viruses RICHARD W.COMPANS AND MAURICE C. KEMP I. 11. 111. IV. V.

Membranes of Lipid-Containing Viruses, 233 Components of Viral Membranes, 236 Arrangement of Viral Envelope Components, 240 Structure and Function of Viral Glycoproteins, 242 Assembly of Viral Membranes, 260 References, 268

Erythrocyte Glycoproteins MICHAEL J. A. TANNER I. 11. 111. IV. V. VI.

Introduction, 279 The Origin and Turnover of the Erythrocyte, 280 The Glycoproteins of the Erythrocyte Membrane, 281 Organization of the Glycoproteins in the Erythrocyte Membrane, 285 Structure of the Glycoproteins, 288 Functions of Glycoproteins, 304 References, 316

Biochemical Determinants of Cell Adhesion LLOYD A. CULP I. 11. 111. IV.

Introduction, 327 Cell Substrate Adhesion, 328 Cell-Cell Adhesion, 356 Conclusion, 381 References, 383

Proteolytic Modification of Cell Surface Macromolecules: Mode of Action in Stimulating Cell Growth KENNETH D. NOONAN I. 11. 111. IV. V. VI. VII.

Introduction, 398 The Erythrocyte Membrane, 399 Stimulation of Cell Division in a Resting Lymphocyte Population, 404 Protease Induction of Cell Division in Fibroblasts, 412 Effects of Proteases on Fibroblast Surface Structure, 421 Protease-Induced Transmembrane Events, 432 Limited Autolysis as a Mechanism for Inducing Cell Division, 437

CONTENTS

VIII. Role Media Components Play in Protease-Stimulated Cell Division, 444 IX. Summary, 450 References, 452 Note Added in Proof, 461 Glycoprotein Antigens of Murine lymphocytes

MICHELLE LETARTE I. Introduction, 464 11. Methods of Analysis of Cell Surface Antigens, 467 111. Isolation and Characterization of H-2 Antigens, 477

IV. V. VI. VII. VIII.

Isolation and Characterization of IA Antigens, 487 Isolation and Characterization of Thy-1 Antigen, 493 Preliminary Characterization of Tla Antigens, 500 Preliminary Characterization of Ly-2,3 Antigens, 502 Conclusion, 503 References, 505

Subject Index, 513

vii

This Page Intentionally Left Blank

Numbers in parentheses indicate the pages on which the authors’ contributions begin. Richard W. Companr, Department of Microbiology, University of Alabama at Bir-

mingham, The Medical Center, Birmingham, Alabama 35294 (233) Lloyd A. Culp, Department of Microbiology, School of Medicine, Case Western Reserve

University, Cleveland, Ohio 44106 (327) Gordon 0. Forrtner, The Research Institute, The Hospital for Sick Children, and Depart-

ment of Physiology, University ofToronto, Toronto M5G 1x8, Ontario, Canada (145) R. 1. Juliano, The Research Institute, The Hospital for Sick Children, Toronto M5G 1x8,

Ontario, Canada (107) Maurice C. Kemp, Department of Microbiology, University of Alabama at Birmingham,

The Medical Center, Birmingham, Alabama 35294 (233) Michelle Letarta, Department of Biological Research, The Ontario Cancer Institute,

Toronto, Ontario, Canada (463) Department of Biochemistry, The Research Institute, The Hospital for Sick Children, Toronto M5G 1x8, Ontario, Canada (15)

Mario Morcarello,

of Biochemistry and Molecular Biology, JHM Health Center, University of Florida, Gainesville, Florida 32610 (397)

Kenneth D. Noonan, Department

John R. Riordan, The Research Institute, The Hospital for Sick Children, and Depart-

ment of Clinical Biochemistry, University of Toronto, Toronto M5G 1x8, Ontario, Canada (1) Aser Rothrtein, The Research Institute,The Hospital for Sick Children, TorontoM5G 1x8,

Ontario, Canada (1) Harry Schaehter, Departments of Biochemistry and Pathology, The Research Institute,

The Hospital for Sick Children, Toronto M5G 1x8, Ontario, Canada (15) Jennifer Sturgerr, Department of Pathology, The Research Institute, The Hospital for

Sick Children, Toronto M5G 1x8, Ontario, Canada (15) Michael J. A. Tanner, Department of Biochemistry, University of Bristol, Bristol BS8

lTD, United Kingdom (279)

ix

This Page Intentionally Left Blank

Due to the rapid pace at which our field is expanding and its increasing reach, the Editors, the Advisory Board, and the Publishers have decided to include in the continuing broad coverage of our series topic-centered volumes which will explore in depth emerging research areas of particular significance. Cell surface glycoproteins are such a topic. We were fortunate that our colleagues, R. L. Juliano and A. Rothstein, were willing to serve as Guest Editors to bring together authors whose work has dealt with various aspects of this topical subject and to structure these contributions into a special volume of Current Topics in Membranes and Trunsport. The introductory chapter presents an historical review and a general discussion of the cell membrane, while the other chapters deal with structure, biosynthesis, and function of the glycoproteins. For our part we are particularly pleased that the first of our special, topic-centered issues deals with a membrane component that appears to function in cellular recognition and assembly. This volume may therefore contribute to the molecular understanding of cellular differentiation, one of biology’s major puzzles. We hope our readers will share our excitement at this eleventh volume of this serial publication.

FELIXBRONNER

ARNOST KLEINZELLER

xi

This Page Intentionally Left Blank

Contents of Previous Volumes Volume 1 Some Considerations about the Structure of Cellular Membranes MAYNARD M. DEWEYAND LLOYDBARR The Transport of Sugars across Isolated Bacterial Membranes H. R. KABACK Galactoside Permease of Escherichia coli ADAM KEPES Sulfhydryl Groups in Membrane Structure and Function ASER ROTHSTEIN Molecular Architecture of the Mitochondrion DAVID H. MACLENNAN Author Index-Subject Index

Volume 2 The Molecular Basis of Simple Diffusion within Biological Membranes W. R. LIEB AND W. D. STEIN The Transport of Water in Erythrocytes ROBERT E. FORSTER Ion-Translocation in Energy-Conserving Membrane Systems B. CHANCE AND M. MONTAL Structure and Biosynthesis of the Membrane Adenosine Triphosphatase of Mitochondria

ALEXANDER TZAGOLOFF Mitochondria1 Compartments: A Comparison of Two Models HENRYTEDESCHI Author Index-Subject Index

Volume 3 The Na+, K+-ATPase Membrane Transport System: Importance in Cellular Function ARNOLD SCHWARTZ, GEORGEE. LINDENMAYER, AND JULIUS C. ALLEN Biochemical and Clinical Aspects of Sarcoplasmic Reticulum Function ANTHONY MARTONOSI The Role of Periaxonal and Perineuronal Spaces in Modifying Ionic Flow across Neural Membranes W. J. ADELMAN,JR. AND Y. PALTI Properties of the Isolated Nerve Endings GEORGINARODdGUEZ DE LORES ARNAIZ AND

EDUARDODE ROBERTIS Transport and Discharge of Exportable Proteins in Pancreatic Exocrine Cells: In Vitro Studies J. D. JAMIESON

xiii

xiv

The Movement of Water across Vasopressin-Sensitive Epithelia RICHARD M. HAYS Active Transport of Potassium and Other Alkali Metals by the Isolated Midgut of the Silkworm WILLIAM R. HARVEYAND KARL ZERAHN Author lndex-Subject lndex

Volume 4 The Genetic Control of Membrane Transport CAROLYNW. SLAYMAN Enzymic Hydrolysis of Various Components in Biomembranes and Related Systems MAHENDRA KUMAR JAIN Regulation of Sugar Transport in Eukaryotic Cells HOWARDE. MORGANAND CAROLF. WHITFIELD Secretory Events in Gastric Mucosa RICHARD P. DURBIN Author Index-Subject lndex

CONTENTS

OF PREVIOUS VOLUMES

A Macromolecular Approach to Nerve Excitation ICHIJI TASAKIAND EMILIO CARBONE Subject Index

Volume 6 Role of Cholesterol in Biomembranes and Related Systems MAHENDRAKUMAR JAIN Ionic Activities in Cells A. A. LEV AND W. McD. ARMSTRONG Active Calcium Transport and Caz+Activated ATPase in Human Red Cells H. J. SCHATZMANN The Effect of Insulin on Glucose Transport in Muscle Cells TORBENCLAUSEN Recognition Sites for Material Transport and Information Transfer HALVORN. CHRISTENSEN Subject lndex

Volume 7 Volume 5 Cation Transport in Bacteria: K+, Na+, and H+ FRANKLIN M. HAROLDAND KARLHEINZ ALTENWRF Pro and Contra Carrier Proteins; Sugar Transport via the Periplasmic GalactoseBinding Protein WINFRIEDBoos Coupling and Energy Transfer in Active Amino Acid Transport ERICH HEINZ The Means of Distinguishing between Hydrogen Secretion and Bicarbonate Reabsorption: Theory and Applications to the Reptilian Bladder and Mammalian Kidney WILLIAM A. BRODSKYAND THEODOREP. SCHILB Sodium and Chloride Transport across Isolated Rabbit Ileum STANLEYG. SCHULTZ AND PETER F. CURRAN

Ion Transport in Plant Cells E. A. C. MACROBBIE H+ Ion Transport and Energy Transduction in Chloroplasts RICHARD A. DILLEYAND ROBERTT. GIAQUINTA The Present State of the Carrier Hypothesis PAUL G. LEFEVRE Ion Transport and Short-circuit Technique WARRENS. REHM Subject lndex

Volume 8 Chemical and Physical Properties of Myelin Proteins M. A. MOSCARELLO The Distinction between Sequential and Simultaneous Models for Sodium and Potassium Transport AND R. P. GARAY P. J. GARRAHAN

xv

CONTENTS OF PREVIOUS VOLUMES

Soluble and Membrane ATPases of Mitochondria, Chloroplasts, and Bacteria: Molecular Stnicture, Enzymatic Properties, and Functions RIVKAPANETAND D. RAO SANADI Competition, Saturation, and Inhibition-Ionic Interactions Shown by Membrane Ionic Currents in Nerve, Muscle, and Bilayer Systems ROBERTJ. FRENCH AND WILLIAMJ. ADELMAN,JR. Properties of the Glucose Transport System in the Renal Brush Border Membrane R. KINNE Subject Index

Volume 9 The State of Water and Alkali Cations within the Intracellular Fluids: The Contribution of NMR Spectroscopy MORDECHAISHPORER AND MORTIMER M. CIVAN Electrostatic Potentials at MembraneSolution Interfaces STUARTMCLAUGHLIN A Thermodynamic Treatment of Active Sodium Transport S. ROY CAPLANAND ALVIN ESSIG Anaerobic Electron Transfer and Active Transport in Bacteria WIL N. KONINGS AND JOHANNESBOONSTRA Protein Kinases and Membrane

Phosphorylation M. MARLENEHOSEYAND MARIANO TAO Mechanism and Physiological Significance of Calcium Transport across Mammalian Mitochondria1 Membranes LEENA MELA Thyroidal Regulation of Active Sodium Transport F. ISMAIL-BEIGI Subject Index

Volume 10 Mechanochemical Properties of Membranes E. A. EVANSAND R. M. HOCHMUTH Receptor-Mediated Protein Transport into Cells. Entry Mechanisms for Toxins, Hormones, Antibodies, Viruses, Lysosoma1 Hydrolases, Asialoglycoproteins, and Carrier Proteins DAVIDM. NEVILLE,JR. AND TA-MINCHANG The Regulation of Intracellular Calcium ERNESTOCARAFOLI AND MARTIN CROMPTON Calcium Transport and the Properties of a Calcium-Sensitive Potassium Channel in Red Cell Membranes VIRGIL10 L. LEW AND HUGOG. FERREIRA Proton-Dependent Solute Transport in Microorganisms A. A. EDDY Subject Index

This Page Intentionally Left Blank

Current Topics in Membranes and Transport volume 11

Cell Surface Glycoproteins: Structure, Biosynthesis, and Biological Functions

This Page Intentionally Left Blank

CURRENT TOPICS IN MEMBRANES AND TRANSPORT, VOLUME

The Cell Membrane-A

11

Short Historical Perspective

ASER ROTHSTEIN Research Institute The Hospital for Sick Children Toronto, Cunudu

This volume reviews several selected topics concerning membrane glycoproteins. Although it is a relatively large volume, it does not offer a complete and comprehensive coverage of existing knowledge on this subject. The incomplete coverage reflects the intense and diverse interests in glycoproteins of large segments of the biological research community. Although soluble glycoproteins have been studied for many years, interest in membrane-bound glycoproteins is of relatively recent origin. For example, in a general review on glycoproteins published in 1970 (Spiro, 1970), only 2 pages out of 39, and 15 out of 274 references, were devoted to membrane glycoproteins. Today, however, membrane glycoproteins (and membrane proteins in general) are a central theme of membrane research. A major shift in interest has occurred that can be illustrated by the following observations. In 1968, when the membrane literature was reviewed for the previous year in theAnnua1 Reuiew of Physiology (Rothstein, 1968), about 400 references were used, but an estimated total of about 600 membrane papers had been published. A breakdown of the topics covered (Table I) indicated that the large majority of papers were concerned with membrane transport activities (65%). Proteins and glycoproteins, per se, were not a recognizable category of membrane study, with the exception of papers concerned with membrane ATPases. The membrane, at that time, was largely the domain of the physiologist, the biophysicist, and the electron microscopist. Table I indicates that by 1976, however, biochemically oriented studies constituted the major thrust of membrane research (42% of the total), with a large proportion of such studies devoted to glycoproteins or to membrane receptors known or implied to be glycoproteins. Such studies now constitute a large interdisciplinary area of research, encompassing major compo1 Copyright 0 1978 by Academic Press. Inc. All rights of reproduction in any form reserved. ISBN 0-12-153311-5

2

ASER ROTHSTEIN

TABLE I ANALYSISOF NUMBERS AND CONTENT OF PAPERS ON MEMBRANES PUBLISHED I N 1968 AND 1976 1968"

1976b

Topic

Number

Percent

Number

Percent

Structure Fractionation Models (real and theoretical) Physical properties Transport Membrane biosynthesis Proteins and glycoproteins Enzymes Receptors (hormones, antibodies, and carriers) Lipids and glycolipids Miscellaneous Total

11 10 36

3 2 9

336 120 324 131 996 144 300 516c 420

9 3 9 3 27 4 8 14 12

156 156 35w

4 4

-

-

265 6

65 1 11 9

47 36

411d

From Table 1 in Rothstein (1971);papers cited in Rothstein (1968). Averages from three randomly selected issues of Biological Membrane Abstracts, Information Retrieval Ltd., London. About one-half of the papers on enzymes involve ATPase, and one-quarter CAMP. Only two-thirds of the papers collected were cited, thus the total is about 600.

nents of biochemistry, immunology, endocrinology, embryology, cell biology, and physiology. The change in emphasis in membrane research is reflected not only in the shift toward a biochemical-cell biological approach, but also in a massive increase in the amount of research effort. It is estimated that, from 1948 to 1968, the number of publications in the membrane field increased at a uniform rate of about 8% per year based on the cumulative number of references cited in the biennial reviews in the Annual Review of Physiology. (During that period, the reviews tended to more or less cover the field of membranes in general.) In view of the fact that this was a period of rapid expansion of biological science, growth of the membrane field was, in all probability, not quite keeping pace with that of other biological fields. Between 1968 and 1976 a major discontinuity in the trend had occurred; the annual rate of production of papers jumped from about 600 to about 3600. Much of the increase took place in the 5-year period between 1968

CELL MEMBRANE-SHORT

HISTORICAL PERSPECTIVE

3

and 1973, because by 1973 the annual rate was already over 2500 (based on the number of abstracts appearing annually in Biological Membrane Abstracts, Information Retrieval Ltd., London). In the field of transport alone, Andreoli and Schaffer (1976) estimated that 3000 papers appeared between 1973 and 1976. It is an astounding fact that, in 1977 alone, almost as many papers on membranes were published as in the 20-year period 1948-1968. The increase in productivity is also reflected in a variety of ways: more journals publishing papers; new journals largely devoted to original papers on membranes or to biomembrane reviews; numerous books on membranes; numerous membrane symposia, workshops, and training courses; and so on. The growth reflects not only an expanded interest in the more classic kinds of membrane studies (transport and membrane structure), but the attraction into the field of investigators from other disciplines (biochemistry, immunology, endocrinology, genetics, and cell biology). The membrane field has become attractive to many new investigators willing to invest their efforts, to journals willing to devote their pages, and to funding agencies willing to donate their money. As an introduction to this volume, we try to analyze the reasons for the rapid growth of interest in membranes and also for the shift in emphasis toward studies relating to proteins and glycoproteins. Because current trends have their origins in the past, we try to place the analysis in historical perspective by briefly outlining past achievements, with appropriate attention paid to both conceptual and technological development. The membrane field, like many fields of biology, is often technology-limited, so that perhaps as many major advances have resulted from technical as from conceptual breakthroughs. The membrane field is relatively old as far as experimental biology is concerned, dating back to the middle of the nineteenth century. Simple observations with the microscope on the extent of shrinking (plasmolysis) of plant cells in osmotic gradients led to the realization that the cells obeyed the laws of osmosis. It was therefore concluded that the cells must be covered by an invisible, semipermeable membrane that allowed water, but not solutes, to pass during the period of observation. Thus the biophysical approach was born. Later in the century, it became clear that osmotically shrunken cells in some solutes slowly recovered their normal volume, and that normal cells swelled (and hemolysis occurred in the case of erythrocytes), leading to the conclusion that some solutes could also penetrate the membrane. By the end of the nineteenth century, Overton and others had systematically determined the rates of penetration of a variety of nonelec-

4

ASER ROTHSTEIN

trolytes and reported a strong corrleation between the measured permeabilities of solutes and their lipid solubilities (partition coefficients). He concluded that the hypothetical, invisible cell membrane must be largely lipid. He also observed that solutes of small molecular size penetrated faster than would be predicted from their lipid solubilities. It was therefore proposed that the lipid membrane might have aqueous membrane “patches” through which small nonelectrolytes could pass. This concept of a mosaic membrane was the progenitor of the modern lipid mosaic membrane theory, formalized by Singer and Nicolson (1972), according to which a continuous lipid matrix is penetrated by transmembrane proteins that may provide permeation pathways for hydrophilic solutes. In quantitative terms, the penetration of nonelectrolytes and the consequent swelling or shrinking of cells were expressed in terms of derivations of the laws of diffusion, with the membrane considered a resistance to diffusion defined in terms of a permeability constant. The concept of permeability in these rigorous terms was refined and developed during the period 1900-1940. A finding that was to lead to the next major advance in membranology was made in the 1920s. Gorter and Grendel(l925) reported that the red blood cell contained just sufficient lipid to form a continuous bilayer over its surface. This observation formed the starting point for the general concept of a continuous lipid bilayer membrane as proposed by Danielli (1943).This model elegantly explained a large variety of permeation data on the basis of lipid partition coefficients and on the basis of the energy barriers to solute flow that existed at the water-lipid interface and within the membrane interior (viscosity). At this time, the membrane was a hypothetical lipid structure invented to explain permeability and osmotic behavior of cells, largely in relation to nonphysiological nonelectrolytes. No functional role was assigned to proteins, although they were presumed to be associated in some way with the surfaces (particularly the inner surface) of the bilayer. It was recognized that physiological substrates such as sugars could be taken up and used by cells or absorbed in the intestine and kidney, but membrane activities relating to metabolites were not extensively explored because meaningful permeability coefficients could not be assigned. By the operational criteria used, the cells were considered relatively impermeable to electrolytes. The next 20 years (1940-1960) represented a revolutionary period in membrane work. A major stimulus was a technical breakthrough, the production of radioactive isotopes and of counting devices which allowed the study of movements of ions, particularly Na+ and K+,

CELL MEMBRANE-SHORT

HISTORICAL PERSPECTIVE

5

Ca’+, and anions, and of substrates such as sugars and amino acids. The entire emphasis shifted from nonfunctional nonelectrolytes to the functionally important ions and metabolites. It became clear that cells could maintain nonequilibrium distributions of cations despite the fact that they were permeable to these ions. Most of the clarification of ion transport mechanisms came from a rigorous analysis, in thermodynamic terms, of ion flows and distributions across membranes, with appropriate consideration given to the role of the membrane potential as well as to the concentration gradients as driving forces. It was noted that, in addition to the expected fluxes of ions down their electrochemical gradients, fluxes could also take place in the uphill direction against electrochemical gradients. The energy to drive such movements was obviously derived from cell metabolism. The process became known as active transport. An important technical breakthrough was the short-circuit technique developed by Ussing (1954) for use with frog skin (and later with other tissues). With this technique, all external driving forces (electrical and chemical) for ion flow across the membrane could be eliminated, so that only active transport generated by metabolic events within the skin resulted in net movements. During the same period, the thermodynamic approach was used to develop the foundations of modern electrophysiology. In nerve (and muscle), it became clear that membrane potentials and electrical conduction were determined largely by cation gradients (Na+ and K+) maintained by active transport systems, and by cation permeabilities. Technical advances, such as the development of microelectrodes, and the voZtage clamp procedure (similar in concept to the short-circuit technique) were used to describe in detail the cation permeability changes occurring during conduction (Hodgkin, 1958). While flows and forces were being defined in thermodynamic terms, the mechanisms of permeation and transport were being determined by the application of enzyme kinetics. As in the case of enzyme activity, many transport activities reached limiting (maximal) rates as solute concentrations were increased. Furthermore, pairs of chemically related substrates appeared to compete for transport. It was therefore proposed that transport must involve a reversible interaction of the transported molecule with a hypothetical membrane site called a carrier. The substrate-carrier complex was assumed to move across the membrane and was therefore considered a mobile carrier (Wilbrandt and Rosenberg, 1961). On the structural side, another major technical breakthrough was the development of the electron microscope. With its use, an abstract

6

ASER ROTHSTEIN

membrane conceptualized from permeability and transport data became a real structure that could be seen as a “railroad track” on a photographic print (Robertston, 1959). The spacings of the images obtained with metal stains were consistent with the Danielli (1943) continuous bilayer model. The bilayer structure was also confirmed by x-ray diffraction analysis (Finean, 1961). On the biochemical side, progress was slow. It was recognized that membrane entities (carriers) must be involved in the many kinds of transport and that metabolic reactions must provide the energy for active transport, but there seemed to be no way to identify these hypothetical substances and reactions. The membrane was therefore accepted as a “black box,” albeit a lipid one, across which certain substances could flow in a characteristic manner and which could transform metabolic energy into uphill flows of certain solutes. Although it was generally accepted that the basic structure of the membrane was a lipid bilayer, studies on water permeability suggested that flow through “pores” might be involved (Paganelli and Solomon, 1957) (this was a revival of the mosaic membrane concept and another preview of the modern lipid mosaic model). No one knew as yet what to do about membrane proteins, although it was clearly demonstrated during this period that several enzymes were localized on the membrane outer surface (Rothstein, 1954). In the period 1960-1970, the momentum of membrane research continued to increase along lines established during the previous 20 years, but with intensified interest in membrane proteins, and with the beginning of an interest in membrane glycoproteins. The kinetic analysis of transport resulted in more complex and sophisticated models. Asymmetric behaviors were identified, and the substrate-carrier binding was found to be influenced by interactions at other sites (regulator or modifier sites). Thus the structure and behavior of proposed carriers became more and more complicated. On the thermodynamic side, a major conceptual advance was the application of irreversible thermodynamics, which formalized and quantitated the interactions between the flow of energy, solutes, and solvent and which allowed a more rigorous definition of transport (Kedem and Katchalsky, 1961). It became clear that the membrane could be considered an energy transducer in which the flow of a solute could be coupled to the flow of energy (active transport), to the flow of a solvent (water), or to the flow of another solute. Such coupling could account, for example, for the relationship of Na+ and fluid transport in epithelia, for the dependence of sugar and amino acid transport on Na+, and for the dependence of solute transport on proton gradients.

CELL MEMBRANE-SHORT

HISTORICAL PERSPECTIVE

7

In addition to the fuller exploitation of thermodynamic and kinetic concepts of transport, this period saw development of the foundations of the molecular and biochemical approach to membrane structure and function. Increasing attention was paid to the biochemistry of membrane lipids and to the behavior of lipid models (monolayers, bilayers, and liposomes). Investigations on membrane proteins were stimulated by several observations:

1. Membrane ATPases were identified as components of the Na+-

K+ transport system (Skou, 1965). 2. Sugar transport systems (permeases) in bacteria were found to be under genetic control, presumably through mutations in membrane protein structures; proteins associated with two forms of sugar transport in bacteria, i.e., M protein (Fox and Kennedy, 1965)and the phosphotransferase system (Roseman, 1972),were directly identified. 3. Many kinds of permeation were found to be perturbed by agents that react with proteins (sulfhydryl reactive agents) (Rothstein, 1970).

4. Hormone-binding sites on the cell surface were identified as proteinaceous (glycoproteins) (Kahn, 1976), and CAMP, part of the hormone response system, was identified (Robison et al., 1970). Much of the work on membrane proteins had become possible because of technological success in preparing, identifying, and purifying plasma membrane fractions. Gross analytical determinations indicated that at least half of the total membrane content was protein, that many of the proteins contained substantial quantities of carbohydrate, and that many metabolic enzymes were associated with the membrane. These findings raised many questions about the architectural arrangements of proteins in the membrane and about their functional roles. It was established that membrane functions included not only the interchange of materials between the cell and its environment, but also the recognition of and response to signals from the environment, such as hormones, antibodies, and other cells. The membrane therefore became the focus of attention not only for physiologists and biophysicists interested in transport function, for anatomists interested in structure, and for biochemists interested in membrane lipids, enzymes, and other proteins, but also for endocrinologists, immunologists, embryologists, and cell biologists. This growing interest was matched at the same time by an unusual number of important technical advances that allowed startling progress to be made in the 1970s. They included: (1) improved procedures for marking, isolating, and purifying membranes; (2) sophisticated use of detergents for “dissolv-

8

ASER ROTHSTEIN

ing” membrane proteins; (3) development of covalent, nonpenetrating probes for determining the “sidedness” of membrane components; (4) development of specific probes to “mark” functional proteins and functional sites; ( 5 )use of proteolytic enzymes as probes of membrane-bound proteins; (6) development of subcellular vesicle systems derived from specific membranes; (7) reconstitution of functional proteins in model membranes; (8)development of cross-linking agents to investigate “near neighbors” among components; (9)sophisticated technologies such .as microcalorimetry, nuclear magnetic resonance, electron spin resonance, infrared spectroscopy, circular dichroism, and fluorescence analysis for evaluation of the physical state of the membrane; (10)acrylamide gel electrophoresis for separating and identifying membrane proteins; (11)gel filtration procedures for fractionating vesicles and macromolecular membrane components; (12) the freeze-fracture technique of electron microscopy; and (13)use of purified lectins to explore the sugar arrays of surface glycoproteins. On the conceptual side, research was also stimulated by formulation of the lipid mosaic model of membrane structure, which described the molecular arrangements of proteins in the membrane with respect to the lipids (Singer and Nicolson, 1972). It was proposed that the lipid bilayer forms a continuous phase that is interrupted or partially interrupted by a class of relatively hydrophobic (intrinsic) proteins inserted into the bilayer and held tenaciously by hydrophobic interactions. Such proteins can be removed from the membrane only by the use of detergents. They can be visualized as “particles” within the bilayer by the use of the freeze-fracture technique of electron microscopy. Other proteins (extrinsic) are associated largely with the inner face of the membrane by ionic or hydrogen bonding. They can be extracted by alterations in ionic strength or pH, by the use of cation chelating agents, or by protein perturbants. One of the key features of the membrane is its asymmetry. The lipid compositiion of the two halves of the bilayer is not the same, so that the preponderance of positively charged head groups faces inward (in red blood cells). The asymmetry with respect to proteins is especially pronounced. All the neutral sugar moieties and all the sialic acid groups of the glycoproteins (and glycolipids) are exposed on the outer face. The polypeptides bearing them are intrinsic proteins embedded in the lipid. In contrast, the extrinsic proteins are largely or entirely located on the inner face of the membrane. At least some of the intrinsic proteins pass all the way through the bilayer (span the membrane), so that they are exposed to both the external and internal environments. Because the internal segments of such intrinsic proteins may

CELL MEMBRANE-SHORT

HISTORICAL PERSPECTIVE

9

be closely associated with some of the extrinsic proteins at the cytoplasmic face of the membrane, the membrane-spanning proteins may play an essential role in providing connections and communications across the bilayer that are important in transport phenomena and in the responses of cells to external stimuli. Much recent membrane work has involved the elucidation of detailed molecular architecture, the relationships of lipids to membrane proteins, interactions between membrane proteins, the identification and characterization of specific proteins, and the molecular mechanisms underlying functional activities. Some interest has also developed in the ways in which cell membranes are assembled, and how their structure is regulated so that they respond appropriately to external signals, growth, and changes in physiological state. From these manifold but interrelated investigations, patterns are beginning to emerge which we briefly outline, with emphasis on the glycoproteins which may play a key role in some membrane phenomena. Only highlights are considered. Detailed discussions of certain aspects are found in the following chapters. The glycoproteins of the cell membrane are largely intrinsic proteins, anchored by hydrophobic interactions in the lipid bilayer with the carbohydrate groups facing toward the outside. Conversely, many intrinsic plasma membrane proteins are glycosylated. The sugar groups form complex arrays with considerable variation from protein to protein. The polypeptide portions may be anchored in the lipid by a sequence of hydrophobic amino acid residues in a-helical form that may traverse the bilayer, but in some cases several polypeptides may collectively form a complex that traverses the membrane. The important point, from a functional point of view, is that protein continua span the bilayer so that they are in contact with both the external and cytoplasmic environments. They can thus provide a means of communication across the bilayer either for flow of solutes and water, or for signals” in response to external “messengers” such as hormones, antibodies, or other cells. Because under normal in vivo conditions the lipid bilayer is fluid, the intrinsic proteins can be considered to “float” in it. Although the proteins can migrate laterally under certain conditions, they may also be relatively fixed with respect to each other because of their interactions with a matrix of extrinsic proteins on the inner surface (and perhaps with each other). Thus their arrangement is not only relatively stable but can be unique. The particular arrays of glycoproteins on the surface may provide for structural features of the cell, and they may also provide specific loci for interaction with external factors. The surfaces of cells contain several exoenzymes, such as phospha6‘

10

ASER ROTHSTEIN

tases, polysaccharide hydrolases, cholinesterase, and aminopeptidase. These enzymes seem to have the simple function of digesting certain external substrates. Although the enzymes may be glycosylated, there is no evidence that the sugar moieties are involved in the recognition and binding of the substrate, or in the enzyme process. Such enzymes are held in the lipid bilayer by a hydrophobic segment of the polypeptide, but the insertion seems to serve only to anchor the enzymes in the membrane, playing no role in the catalytic activity. In the case of sucrase in the brush border of intestinal cells, no direct insertion into the bilayer is evident. Sucrase, however, is tightly bound to another enzyme, isomaltase, which has a hydrophobic anchor. The protruding portion containing the substrate binding site can in some cases be cleaved from the anchoring portion by proteolytic enzymes without loss of activity. The transport of solutes and of water is perhaps the most studied membrane function. It is certainly the one that has occupied our interest for the longest time. Although it is generally accepted that proteins are essential components of most transport or permeation systems (those in which a specific carrier seems to be involved), only a few specific transport proteins have as yet been identified, e.g., the cationtransporting ATPases for Na+, K+, and Ca+, the sugar-transporting system of bacteria, and the anion and sugar transport systems of the red blood cell. Characterization of these few identified systems is far from complete. Nevertheless, a general pattern is emerging which allows interesting speculations to be made. At least some polypeptides of each system (more than one polypeptide chain seems to be involved in the systems so far examined at this level) are highly hydrophobic and are inserted into the bilayer so that some protein segments are exposed to the outside and also to the cytoplasmic environment. Thus they form a protein continuum through the lipid through which transport can occur. The transmembrane portion is probably arranged with hydrophobic groups toward the lipid chains and with hydrophilic groups internalized (forming a largely aqueous channel through the bilayer). The arrangement of such protein channels with respect to the inside and outside of the cell is highly asymmetric. Each system contains a carrier site that specifically binds the transported entity and allows its translocation to occur. That site is therefore accessible from either side of the membrane. Translocation probably does not involve a rotational movement of the whole protein, but a small part of the protein containing the carrier site may be mobile, undergoing a conformational change associated with transport. In active transport systems, the movement is associated with enzymic, reactions. Other sites

CELL MEMBRANE-SHORT

HISTORICAL PERSPECTIVE

11

that bind cofactors (such as ATP and magnesium), inhibitory substances, or perhaps regulatory substances are asymmetrically arranged, facing either outside or inside. The outward-facing segments of transport proteins may be glycosylated (so they are technically glycoproteins), but no evidence indicates that sugar moieties contribute to the specificity of binding of substrates or to transport characteristics per se. Not much is known about the molecular details of the transfer of macromolecules across the membrane. One mechanism of uptake involves endocytosis. A first step is the specific binding of the macromolecule. Recognition in this case may involve the sugar moieties of the transported molecule and of the membrane. For example, desialated glycoproteins are recognized by glycoproteins on the surfaces of liver and kidney cells as a prelude to their uptake and destruction by the cells. The set of functions in which glycoproteins seem to play an important and specific role is the response of cells to environmental substances. The response involves three types of interactions associated with three different parts of the glycoprotein molecule. The first is the recognition (specific binding) of an extracellular molecule (hormone, antibody, lectin, soluble glycoprotein, or the glycopeptide portion of another cell). The second involves a triggering reaction (or form of communication) across the membrane-spanning segment. The third involves interactions at the inner face of the membrane between the membrane gl ycoproteins and other proteins. This interaction may involve various responses, e.g., release of CAMP, phosphorylation of protein, release of Ca2+,or modulation of the microtubular network. As a result, the cell may respond in a variety of complex ways involving shifts in metabolic paths, changes in permeability, endocytosis, protein synthesis and growth, movement, and so on. No direct role can presently be given for glycoproteins in membrane movement or flow, or in the maintenance of special forms of membrane architecture, although all these processes may be modulated by external factors through the mediation of surface glycoproteins. In summary, the ubiquity of glycoproteins in cell membranes is well established. Knowledge of their chemical structure and of their mode of synthesis is considerable, as evidenced by the information presented in this volume. The real deficit in our information concerns the specific functional roles of the many glycoproteins. In proteins that function as exoenzymes or play a role in transport, the carbohydrate moiety seems to play no role in the binding of substrates. In a

12

ASER ROTHSTEIN

few cases, the carbohydrate containing segments can even be digested away without loss of function. Why, then, are these proteins glycosylated? The only functions in which the carbohydrate moieties of glycoproteins are directly implicated are those related to recognition phenomena. Even here, hard evidence is relatively sparse and, in the case of cell-cell interactions, limited largely to cells of lower organisms. Presumably this deficit of information will be rectified in the future, for much research activity is evident in this area. The period from 1970 until today has moved membrane studies ahead by a quantum jump. In the next period of time, one can foresee that the functional role of many additional membrane proteins will be established and elucidated, allowing an analysis of the mechanisms of the various functions at a molecular level. More membrane proteins will be characterized in detail, and sequence analysis will also be involved. The relationship of membrane proteins to each other (membrane topology) will become a more popular topic as technology allows, because it is becoming increasingly evident that many functions involve systems of proteins arranged in the membrane in orderly arrays. Perhaps the most interesting studies one can anticipate will be concerned with the assembly of membranes and the control of this assembly in response to growth, differentiation, and physiological state. This kind of information is central to the understanding of embryological development, responses of cells to normal and abnormal conditions, and cell pathologies. It is safe to predict that the study of membranes is still a growth area in biological science. REFERENCES

A few literature citations are listed below that refer to some of the older studies. No references are given to support the many inferences drawn from more recent findings. The chapter covers such a wide array of information that even a selective list of references would be far longer than the text. The tentative conclusions expressed must therefore be considered to b e one man’s unverified opinion. Andreoli, T. E., and Schaffer, J. A. (1976).Mass transport across cell membranes: The effects of antidiuretic hormone on water and solute flows. Annu. Rev. Physiol. 38,

451-500. Danielli, J. F. (1943).The theory of penetration of a thin membrane. I n “The Penneability of Natural Membranes” (by H. Davson and J. F. Danielli), pp. 341-352. Cambridge Univ. Press, London. Finean, J. B. (1961).“Chemical Ultrastructure in Living Tissues.” Thomas, Springfield, Illinois. Fox, C . F., and Kennedy, E. P. (1965).Specific labeling and partial purification of the M protein, a component of the P-galactoside transport system of Escheri>hia coli. Proc. Natl. Acad. Sci. U.S.A. 54,891-899. Gorter, E.,and Grendel, F. (1925).Bimolecular layers of lipoids on chromocytes of blood.]. Erp. Med. 41,439-443.

CELL MEMBRANE-SHORT

HISTORICAL PERSPECTIVE

13

Hodgkin, A. L. (1958). Ionic movements and electrical activity in giant nerve fibers. Proc. R . SOC., Ser. B 148, 1-37. Kahn, C. R. (1976). Membrane receptors for hormones and neurotransmitters. J . Cell Biol. 70,261-286. Kedem, O., and Katchalsky, A. (1961).A physical interpretation of the phenomenological coefficients of membrane permeability. J . Gen. Physiol. 45, 143-179. Paganelli, C. V., and Solomon, A. K. (1957). The rate of exchange of tritiated water across the human red cell membrane.J. Gen. Physiol. 41,259-277. Robertson, J. D. (1959).T h e ultrastructure of cell membranes and their derivatives. Biochem. SOC. Symp. 16,3-43. Robison, G. A., Schmidt, M. J., and Sutherland, E. W. (1970).“Cyclic AMP.” Academic Press, New York. Roseman, S. (1972).Carbohydrate transport in bacterial cells. I n “Metabolic Pathways” (C. Hokin, ed.), 3rd Ed., Vol. 6, pp. 41-89. Academic Press, New York. Rothstein, A. (1954).The enzymology of the cell surface. Protoplamatologia 2, E4. Rothstein, A. (1968).Membrane phenomena. Annu. Reu. Physiol. 30, 15-72. Rothstein, A. (1970).Sulfiydryl groups in membrane structure and function. Curr. Top. Membr. Transp. 1, 135-176. Rothstein, A. (1971). Fashions in Membranology. In “Intestinal Transport of Electrolytes, Amino Acids and Sugars” (W. McD. Armstrong and A. S. Nunn, Jr., eds.), pp. 3-11. Thomas, Springfield, Illinois. Singer, S. J., and Nicolson, G. L. (1972).The fluid mosaic model of the structure of cell membranes. Science 175,720-731. Skou, J. C. (1965). Enzymatic basis for active transport of Na+ and K+ across the cell membrane. Physiol. Reo. 45,596-618. Spiro, R. G. (1970).Glycoproteins. Annu. Reu. Biochem. 39,599-638. Ussing, H. H. (1954).Active transport of inorganic ions. Symp. SOC. E x p . Biol. 8,407-422. Wilbrandt, W., and Rosenberg, T. (1961).The concept of carrier transport and its corollaries in pharmacology. Pharmacol. Rev. 13, 109-183.

This Page Intentionally Left Blank

CURRENT TOPICS IN MEMBRANES AND TRANSPORT. VOLUME

11

The Structure and Biosynthesis of Membrane GIycoproteins JENNZFER STURGESS. MARZO MOSCARELLO. AND HARRY SCHACHTER Departments of Biochemistry and Pathology The Research Institute The Hospital for Sick Children Toronto. Ontario. Canada

I . Carbohydrate Asymmetry across the Membrane Bilayer . . . . . . . I1 . Glycoprotein Structure . . . . . . . . . . . . . . . . A . Glycophorin . . . . . . . . . . . . . . . . . . B Rhodopsin . . . . . . . . . . . . . . . . . . . C . Epiglycanin . . . . . . . . . . . . . . . . . . D . Murine Histocompatibility Antigens . . . . . . . . . . . E Ashwell’s Mammalian Lectin . . . . . . . . . . . . . F. The LETS Glycoprotein . . . . . . . . . . . . . . G . Enveloped Viruses . . . . . . . . . . . . . . . . H . Cytochrome b, and Cytochrome b. Reductase . . . . . . . . I . Glycoproteins of Myelin . . . . . . . . . . . . . . J . Proteolipid Protein Fraction of Myelin . . . . . . . . . . K . General Comments on Membrane Protein Structure . . . . . . I11. Model Membrane Systems . . . . . . . . . . . . . . . IV . T h e Glycosylation Reaction . . . . . . . . . . . . . . . A . Nucleotide Sugar Formation . . . . . . . . . . . . . B . Polyprenol Phosphate Sugar Formation . . . . . . . . . . C . Dolichol Pyrophosphate Oligosaccharides and the Assembly of Asn-GlcNAc Core Oligosaccharide . . . . . . . . . . . D . Elongation of N-Acetyllactosamine-type Oligosaccharides . . . . E . Assembly of Ser(Thr)-GalNAcOligosaccharides . . . . . . . V. Subcellular Sites of Glycosylation . . . . . . . . . . . . . A Autoradiographic Evidence . . . . . . . . . . . . . B. Subcellular Localization of Glycosyltransferases . . . . . . . C . The Role of the Ribosome . . . . . . . . . . . . . . VI . Membrane Biogenesis . . . . . . . . . . . . . . . . A . Biogenesis of Plasma Membranes . . . . . . . . . . . B. Biogenesis of Intracellular Membranes . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . .

.

.

.

16 19 24 25 26 26 27 28 28 29 30 32 33 34 37 37 42 46 53 58 61 61

64 64 67 70 83 85 15

Copyright @ 1WB hy Academic Press. Inc. All rights of reproduction in any form reserved. ISBN 0-12-153311-5

16

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

Glycoproteins are a large and heterogeneous group of macromolecules and serve a variety of different functions. They occur as soluble secretory molecules such as plasma glycoproteins (a,-acid glycoprotein, transferrin, haptoglobin, and so on), protein hormones (gonadotrophins, thyroglobulin), enzymes (RNase, DNase, amylase, and so on), immunoglobulins, mucins, blood group glycoproteins, acid mucopolysaccharides (glycosaminoglycans), collagens, and basement membranes. The current interest in membrane structure has drawn the attention of many researchers to the fact that glycoproteins also occur in an insoluble form as components of cell membranes. The fluid mosaic model of membrane structure (Singer and Nicolson, 1972) emphasized the important role of proteins as both integral and peripheral components of membranes; the structures of some of these membrane proteins have recently been studied, and it has been realized that many of them are glycoproteins. Some membrane glycoproteins have been isolated by extraction of the membrane with dissociating reagents (Table I); the presence of other membrane glycoproteins has been detected by more indirect methods, e.g., the use of lectins (see Noonan, this volume) and of other carbohydrate-detecting reagents has established the presence of glycoproteins on the surfaces of most cells. The receptors for several hormones (insulin, luteinizing hormone, chorionic gonadotrophin, ACTH) are probably glycoproteins. This chapter reviews the structure and biosynthesis of these membrane-bound glycoproteins; since more information is available on secretory glycoproteins than on membrane glycoproteins, it will occasionally be necessary to extrapolate from the former to the latter. I. CARBOHYDRATE ASYMMETRY ACROSS THE MEMBRANE BILAYER

One of the most interesting concepts arising out of the study of membranes is that there is asymmetry across the membrane bilayer (Rothman and Lenard, 1977; Bretscher, 1973; Singer, 1974). This asymmetry is absolute for proteins and carbohydrates, at least at the sensitivity level of the techniques available to study this problem; i.e., every polypeptide and every glycoprotein molecule has the same orientation across the lipid bilayer. Phospholipids show only partial asymmetry in that every phospholipid is present on both sides of the bilayer but in a different amount. The carbohydrate of the cell surface membrane is always oriented toward the external environment; this fact and the analogous carbohydrate asymmetry across intracellular

17

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

TABLE I

MEMBRANEGLYCOPROTEINS

Membrane glycoprotein Intrinsic Erythrocyte membrane glycophorin Erythrocyte membrane coomassie blue band 111 Bovine retina rhodopsin Epigl ycanin Murine histocompatibility antigen (H-2) Human histocompatibility antigen (HL-A) Mammalian liver lectin LETS glycoprotein VSV G protein Cytochrome b5 Cytochrome b, reductase Human myelin lipophilin Mouse liver nucleotide pyrophosphatase Oviduct membrane glycoprotein Infectious mononucleosis heterophile antigen (Paul-Bunnell antigen) Peripheral Sarcoplasmic reticulum calsequestrin I Mitochondria1 Caz+-binding glycoprotein Platelet glycocalicin

Nonpolar residues (%)

35-40

-

MW

31,000 90,000

47

-

35,000 100,000-500,000 43,000-47,000

-

30,000

51 -

-

38

64 40

-

40,000; 48,000 210,000-270,000 67,000 16,700 33,000 25,000-28,000 130,000

References"

1-8 4 9, 10 11 12 13, 14

15 16, 17 18-20 21 22 23,24 25

56

25,000 25,700

26 27

43

46,000

28

25

33,000

29

57

148,000

30

(1)Winder (1972);(2) Marchesi et al. (1972); (3) Segrest et al. (1973); (4) Marchesi et a!. (1976); (5) Tomita and Marchesi (1975);(6) Springer et al. (1966); (7) Cleve et al. (1972); (8) Morawiecki (1964); (9) Plantner and Kean (1976); (10) Heller (1968); (11) Codington et al. (1975a,b); (12) Nathenson and Cullen (1974); (13) Tanigari et al. (1973); (14) Creswell et al. (1973); (15) Kawasaki and Ashwell (1976a,b); (16) Hynes (1976); (17) Hunt and Brown (1975);(18) Morrison and Lodish (1975);(19) Knipe et al. (1977a);(20) Toneguzzo and Ghosh (1975); (21) Ozols (1972); (22) Strittmatter (1971); (23) Moscarelloet al. (1973);(24) Moscarello (1976);(25) Evans et al. (1973);(26) Chen and Lennarz (1976);(27)Merricket al. (1977);(28) MacLennan (1975);(29)Carafoli and Sottocasa (1974); (30) Okumura et al. (1976).

18

J. STURGESS, M. MOSCARELLO, AND

H. SCHACHTER

membranes are essential in understanding membrane glycoprotein biosynthesis. The close association of carbohydrate with the cell membrane has been demonstrated using cytochemical techniques such as the periodicacid-Schiff (PAS)reagent (Leblond, 1950)and shown to be a characteristic feature of all cell surfaces except in the region of junction complexes (Rambourg et al., 1966). The density of PAS-reactive carbohydrates varies; for instance in intestinal epithelial cells the apical surfaces stain heavily and the lateral surfaces stain only moderately. Carbohydrate staining always shows asymmetry in that it is distributed on the external surface but not on the cytoplasmic surface of the plasma membrane (Rambourg and Leblond, 1967). The reactive 1,2glycol groups are presumably due to membrane-associated glycoproteins, since glycolipids are extracted during tissue preparation (Winzler, 1970). Electron-dense markers such as colloidal iron (Gasic and Berwick, 1962; Benedetti and Emmelot, 1967), colloidal thorium (Rambourg and Leblond, 1967), ruthenium red (Luft, 1971), and cationized ferritin (Danon et al., 1972; Hackenbrock and Miller, 1975) have demonstrated the presence of acidic or anionic sites at the cell surface; these sites are usually attributed to the carboxyl groups of protein-bound sialic acid residues. The reactive groups detected by these cytochemical techniques are believed to be on carbohydrate residues attached to the polypeptide portions of integral membrane proteins exposed at the external surface of the cell (Winzler, 1969). Similar cytochemical approaches have been applied to subcellular membrane fractions to study the distribution and intracellular localization of glycoproteins. Among intracellular membranes, staining of carbohydrate residues has implicated the Golgi complex in a central role in glycoprotein biosynthesis. Cytochemical staining of the Golgi complex with periodic acid-silver methanamine and with phosphotungstic acid at low pH has shown that glycoproteins are localized mainly in Golgi membrane saccules, cytoplasmic vesicles, and lysosomes. The reaction product occurs at the inner surface, directed into the intracisternal or intravesicular space (Rambourg et al., 1969). This asymmetry of membranes is the reverse of that observed at the cell surface. Recently, carbohydrate residues have been demonstrated also on the cytoplasmic surface of the membranes; for instance, anionic binding sites have been demonstrated on the outer face of the Golgi complex by the binding of cationized ferritin (Abe et al., 1976). However, the membrane-associated carbohydrates on the cytoplasmic surface are of lower density than those observed on the inner membrane face. In rough endoplasmic reticulum, the binding of conca~

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

19

navalin A (Con A) occurs exclusively on the inner membrane surface, i.e., the surface opposite that occupied by ribosomes (Nicolson and Singer, 1971; Hirano et al., 1972). In the Golgi complex, Con A binds on the inner membrane surface and, to a limited extent, on the cytoplasmic surface of the membranes (Abe et ul., 1977). This indicates that carbohydrate residues are present, and probably assembled, mainly on the inner aspect of the membrane, but that some may be exposed on the cytoplasmic surface. Similarities in cytochemical reactivity between the Golgi complex and the plasma membrane provide evidence for a biogenetic relationship in which the plasma membrane has a complement of Golgi-type membranes that are everted following fusion of membrane vesicles with the cell surface (Fig. 1).

II. GLYCOPROTEIN STRUCTURE

The number of membrane glycoproteins that have been isolated in a form pure enough for structural studies is quite small (Table I). A discussion of membrane glycopratein structure must therefore be limited, and generalizations based on the available data are obviously subject to error. At the present state of knowledge it appears that the oligosaccharide moieties of membrane-bound glycoproteins are structurally similar to the moieties present on secretory glycoproteins. A brief review of glycoprotein structure is therefore presented, although much of this information is based on secretory glycoproteins. It has become customary to discuss glycoprotein structure on the basis of the carbohydrate-amino acid linkage of the oligosaccharide moiety. This approach has been useful because the carbohydrateamino acid linkage dictates the properties of the associated oligosaccharide group and because a single glycoprotein molecule may contain oligosaccharides of more than one linkage type. Animal glycoproteins contain oligosaccharides of four types of carbohydrate-amino acid linkages (Schachter and Roden, 1973): (1) asparagine-N-acetylglucosamine (Asn-GlcNAc), (2) serine(threonine)-N-acetylgalactosamine [Ser(Thr)-GalNAc], (3)hydroxylysine-galactose, and (4)serinexylose. The hydroxylysine-galactose-type linkage occurs only in collagens and basement membranes, while the serine-xylose-type linkage occurs in the chondroitin sulfates and possibly in other acid mucopolysaccharides (Roden, 1970); these types are therefore not discussed further. The Asn-GlcNAc-type linkage is of very wide distribution in animal glycoproteins and is found in many different secretory proteins (plasma glycoproteins, protein hormones, enzymes, immuno-

20

J. STURGESS,

M. MOSCARELLO, AND H . SCHACHTER

hOS RIBOSOMAL SUBUNIT

-mRNA - 605

RIBOSOMAL SUBUNIT

PEPTIOE HYDROPHOBIC SEOUENCE

mRNA

CY TOPLASMIC JFACE

-9 ???? ? ??? - - 111 111A

LIPID BlLAYER OF ROIJGH - ENDOPLASMIC RETICULUM -

9 9 1 P O ? PP- - 11111111- - -

\ INTRAVESICULAR FACE

RIBOSOME BINOING PROTEIN

8 OSYLTRANSFERASE PLASMA MEMBRANE

n

~

EXTRACELLULAR

FIG.1. Schematic illustration of the biosynthesis of membrane glycoproteins and their insertion into the plasma membrane. (1)Synthesis of membrane protein probably begins on free ribosomes. By analogy to the situation for secreted proteins, it is postulated that the N-terminal sequence of newly synthesized peptide carries a hydrophobic signal sequence (Blobel and Dobberstein, 1975a,b; Schechter et al., 1974) which serves to attach the free ribosome to the endoplasmic reticulum membrane. The evidence that all glycoproteins, including membranous glycoproteins, are assembled on membranebound ribosomes is quite convincing (Schachter, 1974a,b; Morrison and Lodish, 1975; Bergeron et al., 1975), but the presence of a signal N-terminal sequence on a nascent membrane glycoprotein has not yet been demonstrated. However, recent work on the biosynthesis of viral membrane glycoproteins (Rothman and Lenard, 1977; Wirth et al., 1977) suggests that data on the question of a signal sequence will soon be forthcoming. (2) The ribosome becomes attached to the endoplasmic reticulum membrane via a ribsome-binding protein which is postulated to assemble into a hydrophilic channel for

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

21

globulins) as well as in membrane glycoproteins. The Ser(Thr)GalNAc-type linkage is found in mucins and blood group glycoproteins and has also been described in the major red cell membrane glycoprotein, i.e., glycophorin. Asn-GlcNAc oligosaccharides always have a core structure containing only GlcNAc and mannose residues; this core often, but not always, has the branched structure shown in Fig. 2 (Montreuil, 1975). The further elongation of the core structure can take one of two gen-

’

the passage of nascent peptide through the membrane into the intravesicular space. (3)As the ribosome moves along the mRNA, translation occurs and more and more of the nascent peptide enters the intravesicular space. (4) It is postulated that intrinsic membrane proteins have a hydrophobic sequence which anchors them into the membrane (Marchesi et al. 1976; Singer and Nicolson, 1972). Presumably, when translation of this sequence occurs, the hydrophilic channel somehow dissipates and the hydrophobic sequence anchors the peptide in the membrane. (5)The disappearance of the hydrophilic channel probably will cause release of the ribosome from the membrane. Completion of the C-terminal sequence can take place, but without further movement of nascent peptide through the membrane. Also, a t some point in this sequence of events, the N-terminal hydrophobic signal sequence may be cleaved by a peptidase or “signalase” (Blobel and Dobberstein, 1975a) present on the inside of the endoplasmic reticulum. Carbohydrate incorporation occurs primarily after the release of nascent peptide from the polyribosome complex, but there is evidence that some carbohydrate may also become attached to the nascent peptide (Schachter, 1974a,b; Kiely et ul., 1976). (6) Peptide synthesis is complete, and the peptide detaches from the polyribosome complex; the ribosome falls off the mRNA. The peptide is now incorporated into the membrane as indicated. Carbohydrate incorporation occurs, catalyzed by a battery of enzymes called a multiglycosyltransferase system (Schachter and Rod&, 1973; Roseman, 1970); these enzymes are firmly attached to the endomembrane system, and there is evidence that the active sites are directed toward the intravesicular space. Since some membrane glycoproteins (e.g., red cell membrane band 111; see Table I and Tanner, this volume) traverse the membrane more than once, it is clear that the N-terminus must be capable, in some cases, of reentering the bilayer from the intravesicular side. Such twisting of the polypeptide chain within the bilayer should not interfere with the asymmetrical incorporation of carbohydrate from the intravesicular side. (7)The membrane glycoprotein somehow migrates through the endomembrane system from the rough-surfaced endoplasmic reticulum to the Golgi apparatus. One theory (the membrane shuttle hypothesis) suggests that migration occurs by the movement of discrete vesicles from one part of the cell to the other (Meldolesi, 1974a,b; Steiner et ul., 1974; Jamieson and Palade, 1971), followed by selectioe lateral diffusion of proteins after membrane fusion (Bergeron et ul., 1973); another theory (the membrane flow hypothesis) suggests the flow of complete membrane domains along continuous membrane channels (Morre et ul., 1974). Carbohydrate incorporation is completed within the Golgi apparatus (Schachter, 1974a,b). (8)Vesicles migrate from the Golgi apparatus to the plasma membrane where fusion occurs. (9) Lateral migration causes the insertion of membrane glycoprotens into the cell surface; unused components of the vesicle membrane return to the cytoplasm for reutilization. Many aspects of this scheme are hypothetical; evidence in support of the model is presented throughout this chapter.

22

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

Gal

I Sialic Acid

OLIMMANNOSIDE TYPE

N-ACETYLLACTOWMNE TYPE

FIG. 2. Schematic representation of a typical Asn-GlcNAc-type oligosaccharide (Montreuil, 1975).The core structure can be elongated to become either an N-acetyllactosamine- or an oligomannoside-type oligosaccharide.

era1 directions. If GlcNAc residues are incorporated into the core, the structure becomes the N-acetyllactosamine type; sialic acid residues are often attached to the galactose residues in this structure (Fig. 2). If, however, further mannose residues are incorporated into the core, the structure becomes the oligomannoside type (Fig. 2). When AsnGlcNAc oligosaccharides contain fucose residues, these are usually found attached to the GlcNAc residue nearest the asparagine residue. In general, the oligomannoside structure does not contain galactose or sialic acid residues, and neither type of Asn-GlcNAc oligosaccharide contains GalNAc residues. The N-acetyllactosamine-type oligosaccharide may contain only two arms, as shown in Fig. 2, or further branching may occur by the attachment of two GlcNAc residues to a single mannose residue of the core; thus a single Asn-GlcNAc oligosaccharide may contain zero, one, two, three, or possibly four residues of sialic acid. A single glycoprotein molecule (e.g., thyroglobulin) may contain both N-acetyllactosamine and oligomannoside oligosaccharides (Arima and Spiro, 1972; Arima et al., 1972). Ser(Thr)-GalNAc oligosaccharides vary greatly in length and complexity. Ovine submaxillary mucin contains mainly Ser(Thr)-GalNAc and Ser(Thr)-GalNAc-sialic acid groupings, whereas the blood group glycoproteins isolated from the fluids of ovarian cysts carry much more complex branched oligosaccharide structures containing GalNAc, GlcNAc, galactose, and fucose residues (Lloyd et al., 1968; Rovis et al., 1973). Ser(Thr)-GalNAc oligosaccharides have never been reported to contain mannose residues but may carry sulfate ester groups.

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

23

Neither Asn-GlcNAc nor Ser(Thr)-GalNAc oligosaccharides have ever been reported to contain glycuronic acids, and the finding of these sugar acids usually indicates the presence of acid mucopolysaccharides. The occurrence of glucose is a more complex problem. The only well-characterized glycoproteins proved to contain glucose are collagens and basement membranes. However, recent studies on the biosynthesis of Asn-GlcNAc oligosaccharides indicate that lipid pyrophosphate oligosaccharides believed to be intermediates in the biosynthetic process contain not only GlcNAc and mannose residues but also glucose residues (Behrens et al., 1973; R. G. Spiro et al., 1976; M. J. Spiro et al., 1976a,b; Herscovics et al., 1977a,b). The role of lipid intermediates in glycoprotein synthesis is discussed in Section IV,C; the important point to be noted at this time is that M. J. Spiro et al. (1976a,b) have reported the presence of glucose residues in lipid pyrophosphate oligosaccharides isolated from calf thyroid, kidney, and thymus, and from hen oviduct, have shown the transfer in thyroid slices of glucose-containing oligosaccharides from lipid intermediates to endogenous protein acceptors, and have found glucose to be a constituent of the glycoprotein fraction of various thyroid membrane preparations. The glucose residues appear to be part of an oligomannoside Asn-GlcNAc oligosaccharide and are present in internal positions between an outer oligomannoside sequence and an internal core structure (R. G. Spiro et al., 1976). It is possible that this glucose-containing structure will prove to be a constituent of the membrane glycoproteins of many tissues, but further structural work is required to establish this hypothesis. The presence of Ser(Thr)-GalNAc oligosaccharides on a glycoprotein can readily be demonstrated by using mild alkali hydrolysis in the presence of borohydride (Carlson, 1968) to effect p-elimination of the oligosaccharide. Asn-GlcNAc oligosaccharides are resistant to the reaction and there is in fact no method available for cleaving the AsnGlcNAc linkage of a glycoprotein without damaging both the oligosaccharide and the polypeptide backbone. Recently, endo-P-N-acetylglucosaminidases have been reported (Koide and Muramatsu, 1974; Tarentino and Maley, 1974; Ito et al., 1975). These can cleave the asparagine-linked GlcNAc-GlcNAc sequence (Fig. 2) but most appear to act efficiently only on glycopeptides and not on the intact glycoprotein. Table I lists the major membrane glycoproteins that have been isolated to date. The integral membrane glycoprotein G present in the membrane of vesicular stomatitis virus (VSV) is included as an example of an enveloped virus glycoprotein; these interesting viruses are

24

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

discussed further by Compans in this volume. The plasma membrane enzyme nucleotide pyrophosphatase is included as an example of a membrane glycoprotein enzyme; further examples are discussed by Riordan and Forstner in this volume. Murine histocompatibility antigens are glycoproteins, as are other antigens on the lymphocyte surface; this topic is discussed by Letarte in this volume. The human red cell membrane contains two major glycoproteins, glycophorin and coomassie blue-staining band 111; these proteins are discussed by Tanner in this volume. The following discussion deals with structural aspects of interest to the problem of biosynthesis. A. Glycophorin

Glycophorin is the major sialic acid-containing glycoprotein of the human red cell membrane. This protein does not stain with coomassie blue after electrophoresis in sodium dodecyl sulfate (SDS) polyacrylamide gels but can be detected with the PAS stain. Both glycophorin and coomassie blue band I11 are glycoproteins which span the lipid bilayer. Glycophorin is an amphipathic polypeptide chain comprised of three sections: (1)a hydrophilic N-terminal sequence which carries all the oligosaccharide groups and which is exposed at the outer side of the plasma membrane, (2) a middle hydrophobic section which interacts with the lipid bilayer, and (3) a hydrophilic C-terminal section which is exposed at the inner surface of the plasma membrane. Glycophorin contains both Ser(Thr)-GalNAc and Asn-GlcNAc oligosaccharides clustered together at the N-terminal end of the molecule (Tomita and Marchesi, 1975). Thomas and Winzler (1969a,b) studied the structures of oligosaccharides released from glycophorin by mild alkaline hydrolysis in the presence of borohydride; the major products were the tetrasaccharide shown in Fig. 3 and smaller chains lacking one or both sialic acid residues or the galactose residue. Ser(Thr)GalNAc oligosaccharides larger than the tetrasaccharide have been isolated from glycophorin and are believed to be responsible for the MN blood group antigenic activities of red cells (Springer and Yang, 1977). The Asn-GlcNAc oligosaccharide present on glycophorin has a typical N-acetyllactosamine structure (Fig. 2; Thomas and Winzler, 1971; Kornfeld and Kornfeld, 1969; Kornfeld and Kornfeld, 1970). Sialic AcidBGalNAc-OH IPl.3

Sialic AcidSGal FIG.3. Structure of a tetrasaccharide isolated from glycophorin by the mild alkaliborohydride elimination reaction (Thomas and Winzler, 1969a,b).

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

25

8. Rhodopsin

Rhodopsin is a glycoprotein found in the outer segment cells of mammalian retinal rods. It is involved in the molecular events of vision, and therefore information on its structure is important. It is an intrinsic membrane protein and is solubilized only by detergents (Osborne et al., 1974). Considerable variation has been found in the reported MWs which vary between 27,000 and 40,000 (Lewis et al., 1974). However, most values are about 30,000, based on amino acid analyses, gel electrophoresis, agarose column chromatography, and analytical ultracentrifugation. The carbohydrate moiety has been studied in various laboratories. Heller and Lawrence (1970) found that bovine rhodopsin contained a covalently linked carbohydrate unit consisting of three GlcNAc and three mannose residues; this was based on the isolation of a single glycopeptide of nine amino acids from a peptic digest. All the carbohydrate was linked through an Asn-GlcNAc linkage. The surface location ofthe oligosaccharide chain was determined b y Con A and wheat germ agglutinin binding (Renthal et hl., 1973). The role of the carbohydrate groups in the chromophoric properties of the protein was studied by periodate oxidation which would be expected to cleave between C-3 and C 4 of GlcNAc and between C-2 and C-3, and C-3 and C-4, of the mannose residues. The 500-nm absorption band characteristic of rhodopsin was not affected by periodate oxidation and was regenerable after bleaching. It was concluded (Renthal et al., 1973) that the intact carbohydrate moiety was not essential for the chromophoric properties of rhodopsin and that the role of the carbohydrate might be to orient rhodopsin in the disk membrane, such that the oligosaccharide protrudes into the hydrophilic environment. Plantner and Kean (1976) reinvestigated the carbohydrate composition of bovine rhodopsin and found 8-9 moles of mannose and 5 moles of GlcNAc per mole of visual pigment. The MW of the rhodopsin apoglycoprotein was calculated to be 38,000 on the basis of amino acid analysis. Bovine rhodopsin carries two oligosaccharide groups per mole at asparagine residues 2 and 15 from the N-terminal end (Hargrave, 1977). Further, like glycophorin, it appears to have a hydrophilic carbohydrate-rich N-terminal sequence exposed at the outer surface of the membrane and a hydrophobic section which interacts with the lipid bilayer (Hargrave, 1977; Saari, 1974; Pober and Stryer, 1975; Worthington, 1973).

26

J. STURGESS,

M. MOSCARELLO, AND H. SCHACHTER

A detailed three-dimensional electron microscope study of the purple membrane of Halobacterium halobium was presented recently by Henderson and Unwin (1975). The purple membrane functions as a light-driven hydrogen ion pump involved in photosynthesis. It has a MW of 26,000, and retinal is covalently linked to each protein molecule in a 1 : 1 ratio, imparting the characteristic purple color. It is often referred to as bacterial rhodopsin. A three-dimensional map of this molecule at a 7 A resolution showed that there were seven rods which were a-helices and extended perpendicularly through most of the width of the membrane. The overall dimensions of the protein were calculated as 25 x 35 x 45 A, with the longest dimension perpendicular to the plane of the membrane. Lipid molecules fill the spaces between the rods, forming a mosaic. Bacterial rhodopsin appears to be an example of an intrinsic membrane protein but is not a glycoprotein. C. Epiglycanin

Codington et al. (1975a,b) have described a high-MW glycoprotein present in the plasma membrane of the murine tumor cell line TA3Ha which they have called epiglycanin. Glycopeptide can be released from epiglycanin by mild proteolysis of intact TA3 cells. The oligosaccharide is of the Ser(Thr)-GalNActype and contains sialic acid, galactose, and GalNAc in the molar ratio 1 : 4 :2. Epiglycanin reacts with Vicia graminea lectin, indicating structural features in common with human blood-group-N substance (Springer et al., 1972). D. Murine Histocompatibility Antigens

Histocompatibility antigens are a complex series of proteins present on the surface of lymphocytes and of probably all nucleated cells (see Letarte, this volume). The genetic regions controlling the expression of histocompatibility antigens are called HL-A (human leukocyte antigens) in humans and H-2 in mice. The products of the H-2 genes are undoubtedly glycoproteins (Nathenson and Muramatsu, 1971), and the products of the HL-A genes are probably also glycoproteins. The oligosaccharide prosthetic groups of the H-2 antigens have been investigated by Nathenson and co-workers (Nathenson and Muramatsu, 1971; Nathenson and Cullen, 1974). A glycopeptide has been isolated from proteolytic digests of purified H-2 antigen, and a partial oligosaccharide structure has been reported on the basis of glycosidase digestion. This structure is of the Asn-GlcNAc N-kcetyllactosamine type

STRUCTURE A N D BlOSYNTHESlS OF M E M B R A N E GLYCOPROTEINS

27

(Fig. 2) and has two or more sialic acid residues per mole and a fucose residue attached to the most internal GlcNAc residue. The position of the fucose residue was determined by digestion of the glycopeptide with an endo-p-N-acetylglucosaminidasefrom Diplococcus pneumoniae. The latter enzyme is of great interest because it apparently can cleave about 80% of the oligosaccharide from intact H-2 glycoprotein (Nathenson and Cullen, 1974); the resulting protein retains all of its H-2 antigenic activity. The current model for the H-2 antigen indicates that it is a protein of about 45,000 MW carrying one or two oligosaccharide units exposed on the external surface of the cell and anchored in the lipid bilayer by a hydrophobic polypeptide of about 3000-6000 MW. E. Ashwell's Mammalian Lectin

In a brilliant series of studies, Ashwell and co-workers (Ashwell and Morell, 1974; Kawasaki and Ashwell, 1976a,b) showed that mammalian hepatocytes carry in their plasma membranes a glycoprotein which acts as a specific binding protein for glycoproteins with exposed terminal galactosyl residues. Thus a variety of sialidase-treated plasma gl ycoproteins are cleared from the circulation by binding to this liver membrane protein followed by endocytosis and proteolysis within the hepatocyte. The liver binding protein has been isolated from rabbit liver and shown to be an intrinsic membrane glycoprotein composed of two subunits with MWs of 48,000 and 40,000, respectively. These subunits tend to aggregate into large complexes in aqueous solution (Kawasaki and Ashwell, 1976a). Pronase digestion of the

GLYCOPEPTIDE I GLYCOPEPTIDE n FIG.4. Oligosaccharides present on a mammalian lectin (asialoglycoprotein-bind-

ing protein) isolated from rabbit liver by Kawasaki and Ashwell (1976b).

28

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

isolated glycoprotein resulted in quantitative recovery of the carbohydrate moieties which were shown to consist of two glycopeptides of varying composition (Fig. 4). Both glycopeptides are of the AsnGlcNAc type, but it should be noted that one of these structures has one less mannose residue than the typical core structure shown in Fig. 2. Whether or not deiriation from this core structure will prove to be typical of membrane glycoproteins cannot be assessed at our present level of knowledge. The mammalian liver binding protein can be considered a lectin, since it recognizes specific carbohydrate structures and since it can agglutinate red celIs and stimulate mitogenesis in desialylated T lymphocytes (Novogrodsky and Ashwell, 1977). It has recently been shown (Pricer and Ashwell, 1976) that the binding protein is not localized to the plasma membrane of liver cells but occurs in Golgi membranes and other intracellular membranes. F. The LETS Glycoprotein

The large external transformation-sensitive (LETS) glycoprotein is a cell surface glycoprotein present on normal fibroblasts and myoblasts, but on the surfaces of transformed cell lines it is reduced in amount or absent (Hynes, 1976). It has not been established that this glycoprotein is an intrinsic membrane protein, and no detailed information on its carbohydrate composition is available. Hunt and Brown (1975)have, however, reported a similar surface glycoprotein which is present on mouse L cells during the GI phase of the cell cycle (Hunt et al., 1975) and which spans the L-cell plasma membrane, indicating it to be an intrinsic membrane glycoprotein. G. Enveloped Viruses

Enveloped viruses (Compans and Kemp, this volume) derive their membrane envelopes from their host cells. The peptide moieties of viral membrane proteins are products of the viral genome, but the oligosaccharide moieties of viral membrane glycoproteins are assembled by the glycosyltransferases of the host cell. The infection of mammalian cell lines by enveloped viruses therefore offers a unique system for studying the biosynthesis of specific membrane glycoproteins. In recognition of this fact, many groups have recently studied the structures of various enveloped virus membrane glycoproteins. VSV has a single intrinsic membrane glycoprotein (G protein); this protein has two oligosaccharide groups per mole, and both of these groups have a

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

29

typical fucose and sialic acid-containing Asn-GlcNAc structure of the N-acetyllactosamine type (Fig. 2) (Etchison and Summers, 1977; Reading et al., 1977). Sindbis virus has two membrane glycoproteins, El and Ef,each of which has two different oligosaccharide groups (Sefton and Keegstra, 1974), one an Asn-GlcNAc group containing only core sugars (mannose and GlcNAc) and the other a fucose- and sialic acid-containing Asn-GlcNAc structure of the N-acetyllactosamine type (Fig. 2). H. Cytochrome b5 and Cytochrome b, Reductase

Fatty acid desaturation takes place on the membranes of the endoplasmic reticulum. The first two components catalyzing the desaturation reaction are NADH-cytochrome b, reductase and cytochrome b,. Both components may be glycoproteins (Ozols, 1972), although this has not been firmly established. Amino acid sequence studies on cytochrome b, revealed the sequence Asn-His-Ser at positions 22 to 24 in the human liver enzyme; this may represent the point of attachment of the oligosaccharide to the polypeptide chain. These studies were done on the enzyme released from liver microsomes by mild proteolysis, which had a MW of 11,000-13,000, A detergent-solubilized preparation was obtained by Spatz and Strittmatter (1971), and the MW of this preparation was estimated at 16,600, considerably higher than that of the preparation released by proteolysis. In the absence of detergent, the water-soluble enzyme aggregated, giving a MW of about 120,000. Trypsin treatment of the detergent-solubilized enzyme yielded several peptides (Spatz and Strittmatter, 1971).One of MW 12,000 was enzymically active and showed no tendency to aggregate in water. Another peptide of apparent MW 5000 represented the major hydrophobic peptide. Amino acid analysis showed that this fragment contained 49% nonpolar residues as compared to 29% for the active fragment. The hydrophobic peptide sequence appears to b e at the C-terminus of the molecule and probably serves to anchor the protein in the membrane. Microsomal cytochrome b, reductase has many properties similar to those described for cytochrome b,. The MW is approximately 33,000, higher than that of cytochrome b, (16,600). Although a quantitative sugar analysis is not available, the purified enzyme stains positively for carbohydrate and is probably a glycoprotein. It appears to have a hydrophobic domain (similar to that of cytochrome b,) b y which it is inserted into the membrane.

30

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

Both cytochrome b, and cytochrome b, reductase have been purified and appear to carry out their catalytic activities in the absence of lipid. However, a lipid requirement for the enzymes is not ruled out. Rogers and Strittmatter (1973) found that electron transport was reduced by more than 80% on the removal of lipid. After restoration of the original phospholipid/protein ratio by incubating the particles in the presence of liposomes (prepared from microsomal lipids), the rate of cytochrome b, reduction was fully restored. More recently, the desaturase system was reconstituted from the isolated components by combining cytochrome b, reductase, cytochrome b,, the desaturase, stearyl-CoA, NADH, oxygen, and lipid, resulting in an active system (Strittmatter et aZ., 1974). Cytochrome b, can be considered an integral membrane protein. The relatively high content of hydrophobic amino acids, especially in the hydrophobic tail of the detergent-solubilized protein (Spatz and Strittmatter, 1971), and its ability to associate with lipids to restore activity, are characteristic of integral membrane proteins. Recent data indicate that cytochrome b, contains two globular domains (Visser et aZ., 1975). One of the domains carries the electron transport site. The function of the other domain is to attach the molecule to the membrane. These two domains are joined by a link region, 30-40 A long, which may be flexible. Therefore the relatively flexible electron transport site is joined to the hydrophobic region anchored in the membrane by a flexible link region. Since these studies were carried out in deoxycholate micelles, one cannot say that cytochrome b, under these conditions is identical with the protein in a natural membrane. Although the complete sequence of cytochrome b, is not known, the C-terminal end (Ozols, 1974) is Glu-Asp-COOH, imparting three negative charges to this end of the molecule; this segment probably cannot lie within the hydrocarbon region of the bilayer, suggesting that the molecule traverses the whole width of the bilayer. However, an alternative model presented recently by Depierre and Dallner (1975) suggests that most of the protein protrudes into the hydrophilic environment on the cytoplasmic side of the membrane, while the hydrophobic portion traverses only half the width of the bilayer.

I. Glycoproteino of Myelin

Myelin is composed of 20-25% protein and 75-80% lipid. The protein composition of myelin is relatively simple (compared to that of other membranes) in that approximately 70% of the total protein is

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

31

made up of two protein fractions, the basic protein and proteolipid.

The basic protein has potent biological activity, eliciting experimental allergic encephalomyelitis at a dose of a few micrograms. The sequence has been established, and considerable data are available on the secondary and tertiary structures. The proteolipid protein fraction consists of several proteins which are soluble in chloroform-methanol. Lipophilin, which constitutes a large proportion of the proteolipid fraction, has been isolated and purified. Basic protein is discussed in this section, although it is not a glycoprotein, because it is a natural acceptor of N-acetyl-D-galactosamine in the presence of a submaxillary gland N-acetylgalactosaminyltransferase.Lipophilin is an intrinsic membrane protein reported to contain carbohydrate. Basic protein was first described by Kies (1965) as the encephalitogenic protein of myelin; i.e., the injection of a small amount (a few micrograms) into a guinea pig elicited a fatal disease characterized by hind limb paralysis, weight loss, and death. Eylar et al. (1971), after several years of study, reported the complete amino acid sequence. Basic protein has a MW of 18,400 (Eylar, 1972) and is characterized by a relatively large number of basic amino acids (lysine plus histidine plus arginine comprise 25% of the. total residues). The basic residues are located at random in the molecule, and there is no obvious hydrophobic region. A proline-rich segment, Pro-Arg-Thr-Pro-Pro-Pro-Ser, surrounds threonine 98, resulting in a sharp bend in the molecule. Although the primary structure is well established, the secondary structure is still in dispute. The proline-rich region mentioned above places severe constraints on the molecule and creates a hairpin turn in the molecule at this site. Early studies (Kies et al., 1965; Palmer and Dawson, 1969; Chao and Einstein, 1970) showed no a-helical or p structures by circular dichroism (CD). In a more recent study, Epand et al. (1974) showed that the protein had a nonrandom structure in solution, despite the fact that it showed no a-helical or p structure. Intrinsic viscosity studies confirmed the axial ratio of 1 : 10 reported earlier (Eylar and Thompson, 1969). The radius of gyration, calculated from low-angle x-ray scattering measurements (Epand et al., 1974), was found to be 39 k 2 A. The model proposed by Epand et al. (1974) was that of a prolate ellipsoid of dimensions 150 x 15 A. Further support for a highly structured molecule was obtained from surface tension measurements (Moscarello et d.,1974), CD and proton magnetic resonance (PMR) changes in myelin basic protein conformation under specific conditions (Liebes et al., 1975), and 13Cnuclear magnetic resonance (NMR) studies (Chapman and Moore, 1976) showing highly structured regions especially near residues 85 to 116. It is interesting

32

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

that the region including residues 81 to 118 has been found constant in myelin basic proteins from humans (Carnegie, 1971), cows (Eylar, 1970), and rats (Dunkley and Carnegie, 1974), suggesting some special function for this region. Threonine 98 has been shown to be the site of glycosylation with UDP-N-acetyl-D-galactosamine and an enzyme from the submaxillary gland (Hagopian and Eylar, 1969b). The only other protein of several tested which was an acceptor in this system was the deglycosylated natural salivary gland protein. In a further in vitro study using UDPN-[’4C]acetyl-D-galactosamineand a purified enzyme from bovine submaxillary gland, over 80% of the radioactivity was recovered on threonine 98 and was shown to be in 0-glycosidic linkage (Hagopian et al., 1971). The significance of this finding is not clear. It has been suggested that the basic protein may be transiently glycosylated during its synthesis and insertion into the membrane. The carbohydrate moiety may play a recognition role, initiating contact between different layers of the myelin sheath and functioning as an “organizer” (Carnegie, 1971; Hughes, 1976). In a tightly compacted structure such as myelin, there would be no room for the oligosaccharide chain, thus it is not surprising that it might subsequently be removed from the protein. J. Proteolipid Protein Fraction of Myelin

A hydrophobic protein has been isolated, purified, and characterized from the chloroform-methanol-soluble fraction of human myelin (Gagnon et al., 1971; Moscarelloet al., 1973). It has several interesting structural features. The hydrophobic residues account for about 6264% of the total residues. Two moles of fatty acid are present per mole of protein (assuming a MW of 28,000); these are esterified directly to the peptide backbone and are not present as a phospholipid or sphingolipid (Folch-Pi and Stoffyn, 1972). Analysis revealed the presence of small amounts of carbohydrate, and therefore it may be a glycoprotein. The protein has been called lipophilin. Attempts to elucidate the primary sequence have met with difficulty, largely because of the insoluble nature of the protein and its cleavage products. Nussbaum et al. (1974) have reported the N-terminal sequence of the performic acid-oxidized rat brain protein to be Gly-Leu-Leu-Glu-Cys SO3-CysSO3-Ala-Arg-CysSO,-Leu -Val-GIy -AlaPro-Phe-Ala-X-Leu-Val-Ala-. Although the N-terminal sequence for the human protein is similar, these data must be interpreted with caution since the sequenator yields were low, often less than 15%.

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

33

In contrast to the basic protein of myelin, which appears to have a fairly rigid secondary structure, lipophilin (formerly called N-2, MOScarello et al., 1973) is conformationally flexible. It can be obtained in water-soluble form in an a-helical or p conformation, depending on the method used to solubilize the protein. The MW of the a-helical form was 86,000, that of the p structure was 500,000, and a monomer MW of 24,000-28,000 was found by equilibrium ultracentrifugation. K. General Comments on Membrane Protein Structure

Although generalizations about membrane protein structure are based on limited data (Table I), a pattern is emerging which supports the concept of a fluid mosaic membrane as detailed by Singer and NicOlson (1972). Integral membrane proteins appear to be amphipathic molecules containing a hydrophobic sequence which interacts strongly with the lipid core of the bilayer. Oligosaccharide groups are clustered near one end of the molecule (probably the N-terminal end), and this hydrophilic portion is exposed on the external side of the plasma membrane or at the inner face (opposite the cytoplasmic face) of intracellular membranes; thus carbohydrate shows an asymmetrical distribution across the membrane bilayer. The oligosaccharide groups studied to date indicate that integral membrane glycoproteins may contain Asn-GlcNAc oligosaccharides of both the oligomannoside and N-acetyllactosamine type (Fig. 2), as well as Ser(Thr)-GalNAc oligosaccharides; these structures do not appear to be significantly different from the oligosaccharides described in secretory glycoproteins, although the possible occurrence of glucose residues in membrane glycoproteins requires further study (Behrens et al., 1973; R. G. Spiro et al., 1976; M. J. Spiro et al., 1976a,b). It should be pointed out that, while Table I lists only membrane glycoproteins, integral membrane proteins have been isolated which do not contain carbohydrate, e.g., the ATPase of sarcoplasmic reticulum (MacLennan, 1975), halophilic bacterial rhodopsin (Oesterhelt and Stoeckenius, 1971), and possibly other proteins. Excluded from the above discussion were acid mucopolysaccharides (glycosaminoglycans),although there is strong evidence for the presence of these molecules at the surface of cultured cells (Kraemer, 1971; Hynes, 1976).These molecules are presently thought to be components of the glycocalyx and therefore more closely associated with peripheral than with integral proteins; however, their topology and function at the cell surface have not been adequately studied.

34

J. STURGESS, M. MOSCARELLO, A N D H . SCHACHTER

111.

MODEL MEMBRANE SYSTEMS

Protein-lipid interactions are difficult to study in natural membranes which possess many different kinds of lipids and proteins. The use of model membranes in which well-characterized isolated proteins are incorporated into lipid vesicles made from well-defined lipids provides us with an experimental model in which to study these protein-lipid interactions. These studies may provide information useful for understanding the incorporation of glycoproteins into membranes in uivo. A large number of proteins has been incorporated into lipid vesicles, and a list of these has been compiled by Tyrell et al. (1976). In addition to membrane glycoproteins, the list includes other proteins such as bovine serum albumin and immunoglobulin (nonmembrane proteins) and synthetic polypeptides such as poly-i-lysine. Some of these are not incorporated into the lipid bilayer but instead interact at the surface of the vesicle. Our discussion here is limited to naturally occurring membrane glycoproteins. Table I lists the properties of some membrane glycoproteins. Two points are worth emphasizing: (1)Where the data are available, the percent of hydrophobic residues (threonine, alanine, proline, tyrosine, valine, methionine, leucine, isoleucine, tryptophan, phenylalanine) is quite high, being 50% or greater in some cases; most soluble glycoproteins contain 20-40% hydrophobic residues, while membrane glycoproteins contain 30-60%; (2) The MWs are relatively low (25,000-50,000). An exception to this rule is the erythrocyte glycoprotein (band 111) of MW 90,000. Although undoubtedly an intrinsic membrane glycoprotein, its relatively high MW must be related to the fact that a considerable portion of it protrudes from both the inner and outer surfaces of the erythrocyte membrane. The membrane portion of this glycoprotein has been estimated to have a MW of 60,000 (Jenkins and Tanner, 1977). Since it is reported to traverse the membrane twice, forming a loop, each arm of this loop has a MW of approximately 30,000. Similarly, the very large size of epiglycanin is due to the fact that it is primarily external to the cell surface. Recent evidence suggests that a polypeptide of about 30,000 MW can span about 60 A of bilayer (Brady et al., 1978). The large percentage of apolar residues in these proteins is no doubt related to the need of the protein to present a large number of hydrophobic groups for interaction with the lipid hydrocarbon chains within the bilayer. An intimate association between protein and lipid described as boundary lipid (Boggs et al., 1976)or lipid annulus (Warren et d . , 1975) has been demonstrated in several cases.

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

35

In some cases, membrane glycoprotein has been shown to possess more than 60% a-helix, e.g., when lipophilin was incorporated into liposomes made of dipalmitoyl phosphatidylcholine (DPPC) (Cockle et al., 1978).X-ray diffraction studies on similar vesicles showed that the thickness of the bilayer was not increased by the presence of the protein, indicating that it was not protruding beyond the surface of the vesicle (Rand et nl., 1976). Surface-labeling studies with a nonpenetrating reagent, 4,4'-diisothiocyanoditritiostilbene disulfonate have been interpreted as indicative that lipophilin traverses the entire width of the bilayer (Wood et nl., 1978). It should be pointed out that proteins lacking carbohydrates can be incorporated into lipid vesicles in vitro (Tyrell et al., 1976) and that there are naturally occurring integral membrane proteins which are not glycoproteins, e.g., the ATPase of sarcoplasmic reticulum (MacLennan, 1975) and the bacterial rhodopsin discussed in the previous section. Thus carbohydrate is probably not required for the stabilization of integral protein within the bilayer. Carbohydrate may be needed during the insertion of protein into the bilayer but may subsequently be removed; there is, however, no evidence for the latter hypothesis. It appears likely that carbohydrate plays its most essential role when it is on the surface of the cell and not during the insertion of glycoprotein into the membrane. Rhodopsin was recombined into egg phosphatidylcholine bilayers, and details of the interactions were studied with electron spin resonance (ESR) and freeze-fracture techniques (Hong and Hubbell, 1972). By incorporating a spin label into bilayers with and without rhodopsin, the effect of the protein was found to inhibit segmental motion of the hydrocarbon chains. The order parameter (SJ, which is a measure of the fluidity of the bilayer, increased as the mole fraction of rhodopsin increased, supporting the contention that the protein was incorporated into the bilayer. Freeze-fracture electron microscopy revealed the presence of numerous particles on the fracture surfaces, suggesting that rhodopsin was buried deep inside the bilayer. More recently, extensive model membrane studies have been carried out with the myelin glycoprotein lipophilin. Because the protein is very hydrophobic, it should be readily incorporated into the lipid bilayer. Freeze-fracture electron microscopy showed that lipophilin was readily incorporated into lipid vesicles (Vail et al., 1974). The possibility of using this model system to study lipid-protein interactions was evident. Several studies were reported by Papahadjopoulos et al. (1975a,b) on the effect of the protein on the ultrastructure, lipid phase transitions and permeability of phospholipid vesicles. Lipophilin was

36

J. STURGESS, M. MOSCARELLO, A N D H. SCHACHTER

found to bind strongly to phospholipids, irrespective of surface charges, the presence of cholesterol, or double bonds in the fatty acyl chains. Large, closed, multilamellar vesicles were formed with a buoyant density intermediate between that of pure lipid and protein. The presence of the protein in the vesicles increased their permeability to zzNa+by two to three orders of magnitude. Differential scanning calorimetry indicated that the presence of the protein had no effect on the lipid phase transition from solid to liquid crystalline. However, the enthalpy of transition decreased as the amount of protein in the vesicles was increased from 0 to 53%, but there was no change in the midpoint temperature. When the concentration of protein was 50% by weight, the enthalpy was approximately one-half that of the pure lipid. Details of the interaction between lipophilin and lipids have been provided in two recent studies. In one study (Boggs et al., 1976), spin labels were used to study the microenvironment of the protein. Lipophilin was inserted into phosphatidylcholine vesicles and studied with fatty acid spin labels. Two distinct components were present in the spectrum. One was immobilized, presumably as a result of the presence of boundary lipid around the protein, and the second was indicative of anisotropic motion, similar to the spectrum for phosphatidylcholine vesicles. Lipophilin was found to increase the order parameter linearly with increasing concentration of protein. The phase transition temperature measured with 'TEMPO (2,2,6,6-tetramethyl piperidine- 1-oxyl)was not changed, in agreement with the differential scanning calorimetry data (Papahadjopoulos et al., 1975a,b). The presence of lipophilin in the lipid bilayer was found to induce lipid phase separation (Boggs et al., 1977). Differential scanning calorimetry was used to study the effect of the protein on the transition temperature of a mixture of phosphatidylserine (PS) and DPPC. PS melts at 8°C and DPPC at 43"C, while the mixture melts at an intermediate temperature depending upon the PS/DPPC ratio. The higher the PS concentration in the mixture, the lower the melting temperature. Addition of the protein to the PS/DPPC mixture at all ratios raised the melting temperature, which could only be explained by the removal of PS from the mixture, resulting in a lipid mixture with a lower proportion of PS. It was concluded that lipophilin bound preferentially to PS, removing it from the mixture and leaving a lipid phase with a higher proportion of DPPC. The binding of certain classes of lipids to the boundary layer of intrinsic membrane proteins may be one way in which asymmetry can be induced and maintained in a membrane. Many membrane glyco-

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

37

proteins have specific enzymic activities. Lipids are known to influence the activity of enzymes. Therefore a specific lipid environment around a membrane-bound enzyme can be maintained by lipid phase separation (discussed above). The existence of different lipid environments surrounding some of the glycosyltransferases of Golgi membranes have been described (Mitranic et al., 1976, 1978) for galactosyl- and sialyltransferases on the basis of a different sensitivity to temperature and drugs such as colchicine. When these glycosyltransferases become available in a pure form, it will be interesting to use them as protein components in model membrane system studies. The principles of lipid-protein interactions derived from the study of model membranes are directly applicable to the more complex systems in naturally occurring membranes. IV.

THE GLYCOSYLATION REACTION

Glycosyltransferases are enzymes which transfer monosaccharides or oligosaccharides from activated derivatives to various acceptors. The activated glycose donors are usually phosphorylated derivatives in which the glycose moiety is connected through the anomeric carbon by either a phosphate or a pyrophosphate linkage to a nucleoside moiety (to form a nucleotide sugar or nucleotide oligosaccharide) or to a polyprenol lipid. A glycosyltransferase can catalyze the formation of O-glycosidic linkages between two sugar moieties or between a sugar and a hydroxyamino acid such as serine or threonine, or it can form an N-glycosidic linkage between N-acetylglucosamine and asparagine. Glycoprotein synthesis depends on a variety of complex factors such as the availability of glycose donors, the substrate specificities of glycosyltransferases, and the arrangement of glycosyltransferases within the cell’s endomembrane system. The following section deals with the synthesis of nucleotide sugars and lipid intermediates and with the utilization of these glycose donors b y glycosyltransferases. Much of the information presented is not specific for membrane glycoproteins but has relevance for all glycoproteins. A. Nueleotide Sugar Formation

The pathways of nucleotide sugar formation from D-ghCOSe are outlined in Fig. 5 (Feingold, 1972; Schachter and Roden, 1973). The following discussion deals only with aspects of interest to the problem of glycoprotein synthesis.

38

J. STURGESS, M. MOSCARELLO, A N D H. SCHACHTER

GL-6-P

I

- -

>..

Glc-1-P

UDP-Glc

Mannose

UDP-Gal

1

Fructose-6-P-Man-6-P

(I-

r -

Mon-1-P

G4cN- 6- P-

Glucosomine

GlcNAC-6-P-

GlcNAc

I

1 Gal-1-P t -Galactose

GDP- Man

-

-GDP;Fuc

I

1

! I-

FuCtl-P Fucose

GlcNAC-1-P

I I uOP-GolNAc f GalNPC- 1-P

UDP-GlcNAc-ManNAc

t

t

Golocrosomtne

ManNAc-6-P

I

N-ocetylneurorntnlc acid-9-P

1

N-acetylneuramlnlc ocld

1

CMP-N-ocelylneuramlnIc acid

FIG.5. Pathways of nucleotide sugar synthesis.

1. UDP-GALACTOSE Glucose can be converted to UDP-galactose as indicated in Fig. 5. The intact animal, the perfused liver, and thyroid slices have been used to study the incorporation of [14Clglucose into glycoprotein (Schachter and Rod&, 1973), and radioactivity was detected in protein-bound galactose and other sugars. When intact liver cells were exposed to [14C]galactose,appreciable radioactivity was found in protein-bound mannose and hexosamine, as well as in protein-bound galactose (Sarcione, 1964; Richmond, 1965; Moscarello et d.,1972). These experiments demonstrate the in vivo interconversion of glucose and galactose and the conversion of exogenous galactose to nucleotide sugars other than UDP-galactose.

2. GDP-MANNOSE GDP-mannose can be formed from D-glUCOSe via fructose 6-phosphate, mannose 6-phosphate, and mannose 1-phosphate (Fig. 5), the final enzyme in the pathway being GDP-mannose pyrophosphorylase. Exogenous D-mannose can also enter the metabolic scheme (Fig. 5 ) via mannose 6-phosphate and give rise to GDP-mannose as well as to other nucleotide sugars. Thus Mitranic and Moscarello (1972) injected D-[2-3H]mannoseinto rats and studied the incorporation of radioactivity into plasma glycoproteins; they recovered over 80% of the total protein-bound radioactivity in sugars other than mannose, namely, ga-

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

39

lactose, fucose, sialic acid, and hexosamine. This extensive conversion of mannose to other monosaccharides in rat liver may not occur in all tissues (Whur et al., 1969).

3. GDP-FUCOSE GDP-fucose can be formed from Dglucose via GDP-mannose or from exogenous L-fucose via fucose kinase and GDP-fucose pyrophosphorylase (Fig. 5 ) . Parenterally administered [14Clfucoseis an excellent precursor of glycoprotein and, unlike monosaccharides such as glucose, galactose, mannose, glucosamine, and galactosamine, fucose is not converted to other monosaccharides (Coffey et al., 1964; Bekesi and Winzler, 1967; Bocci and Winzler, 1969; Sturgess et al., 1973; Shull and Miller, 1960). This finding indicates that the conversion of GDP-mannose to GDP-fucose is not reversible (Fig. 5).

4. UDP-N-ACETYLGLUCOSAMINE AND UDP-N-ACETYLGALACTOSAMINE UDP-N-acetylglucosamine and UDP-N-acetylgalactosamine can both be formed from D-glucose as shown in Fig. 5. The first specific step in hexosamine formation is the conversion of fructose 6-phosphate to glucosamine 6-phosphate7 a process catalyzed by two separate enzymes, ~-glutamine:~-fi-uctose-6-phosphate amidotransferase and glucosamine-6-phosphate deaminase [2-amino-2-deoxy-D-glucose-6-phosphate ketol isomerase (deaminating)]. The amidotransferase reaction is irreversible and is probably the major pathway of hexosamine formation. Inhibition of the amidotransferase by the injection of a glutamine analog (duazomycin A) into rats reduced the level of UDP-N-acetylglucosamine in liver by 85% and caused a 5070% inhibition of glycoprotein synthesis by liver (Bates et ul., 1966; Bates and Handschumacher, 1969); both effects could be prevented by administration of exogenous glucosamine, indicating that duazomycin A did not interfere with the incorporation of glucosamine into glycoprotein but blocked the conversion of glucose to hexosamine via the amidotransferase. The amidotransferase, the first step in hexosamine formation, is subject to feedback control by the end product, UDP-N-acetylglucosamine (Kornfeld et al., 1964; Kornfeld, 1967; Mazlen et d., 1969; Ellis and Sommar, 1971; Miyagi and Tsuiki, 1971; Trujillo and Gan, 1973). UDP-N-acetylglucosamine is formed from N-acetylglucosamine 1phosphate by a pyrophosphorylase, and UDP-N-acetylgalactosamine

40

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

is formed from UDP-N-acetylglucosamine by a 4-epimerase (Fig. 5). N-Acetylglucosamine l-phosphate can be derived from . glucose or from exogenous glucosamine (Fig. 5). Radioactive glucosamine is an excellent precursor for studying glycoprotein synthesis in intact cells; since ~-glutamine:~-fructose-6-phosphate amidotransferase is irreversible, glucosamine is not glycogenic and is not converted to hexoses (McGarrahan and Maley, 1962) although, as expected from the scheme shown in Fig. 5, the administration of radioactive glucosamine to intact cells leads to the appearance of radioactivity in proteinbound sialic acid, as well as in protein-bound glucosamine (Lawford and Schachter, 1966).The in vivo epimerization of UDP-N-acetylglucosamine to UDP-N-acetylgalactosamine in liver is not appreciable (Shetlar et al., 1964; Spiro, 1959; Robinson et al., 1964; MacBeth et al., 1965),presumably because the equilibrium favors formation of the glucosamine derivative. White et al. (1965) injected ~-[l-'~C]galactosamine into rats and recovered most of the plasma protein-bound radioactivity in N-acetylglucosamine and N-acetylgalactosamine; the ratio of labeled glucosamine to labeled galactosamine in serum glycoprotein varied between 6 : l and 19 : l, indicating that most of the exogenous galactosamine underwent epimerization prior to incorporation into protein. The exogenous galactosamine appears to be utilized in rat liver primarily by the same pathway as galactose (Maley et al., 1968), and UDP-N-acetylgalactosamine is synthetized from N-acetylgalactosamine l-phosphate by a pyrophosphorylase (Fig. 5).

5. CMP-SIALICACID The CMP-sialic acids are formed by a series of reactions which originate from N-acetylmannosamine (Fig. 5). The latter compound may be formed from UDP-N-acetylglucosamine b y UDP-N-acetylglucosamine 2-epimerase or from N-acetylglucosamine by N-acetylglucosamine 2-epimerase. N-Acetylmannosamine can be converted to N-acetylneuraminic acid either directly by an aldol condensation with pyruvate (catalyzed by N-acetylneuraminic acid aldolase) or by the series of reactions shown in Fig. 5. The latter route is believed to be the biosynthetic path, while the aldolase is primarily a degradative enzyme in viva Thus N-acetylmannosamine is first phosphorylated to the 6-phosphate which undergoes condensation with phosphoenolpyruvate to form N-acetylneuraminic acid 9-phosphate; cleavage of the phosphate ester results in the formation of N-acetylneuraminic acid. The final step in the activation of sialic acid to CMP-sialic acid is catalyzed by CMP-sialic acid synthetase (CTP:N-acetylneuraminate

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

41

cytidyltransferase). The formation of N-acetylmannosamine from UDP-N-acetylglucosamine is subject to feedback inhibition by the final product, CMP-N-acetylneuraminic acid (Kornfeld et al., 1964; Kikuchi and Tsuiki, 1973). Several studies have appeared indicating that CMP-sialic acid synthetase is a nuclear enzyme (Kean, 1970; Kean and Brunner, 1971; Van Dijk et al., 1973; Van den Eijnden, 1973). Since sialic acid incorporation into glycoproteins and glycolipids appears to be a function of the Golgi apparatus, it is not clear why CMP-sialic acid should be synthesized in the nucleus. Sialic acids are acetyl and glycolyl derivatives of neuraminic acid and are widely distributed in the animal kingdom (mammals, birds, fish, and echinoderms) and in algae and bacteria (Warren, 1963; Cabezas, 1973). The major naturally occurring sialic acids are N-acetylneuraminic acid, N-glycolylneuraminic acid, N-acetyl-4-O-acetylneuraminic acid, N-acetyl-7-O-acetylneuraminic acid, N-acetyl9-0-acetylneuraminic acid, and N-acetyl-7,9-di-O-acetyl-neuraminic acid. Human tissues contain mainly N-acetylneuraminic acid with only trace amounts of other sialic acids, but many tissues of other mammalian species contain appreciable amounts of N-glycolylneuraminic acid, N,O-diacetylneuraminic acid, and triacetylneuraminic acid. Schauer’s group (Schauer et ul., 1974; Buscher et al., 1974; Kamerling et al., 1975; Schauer, 1973; Jancik and Schauer, 1974) has developed gas-liquid and thin-layer chromatographic techniques for resolving the different sialic acids and has undertaken a detailed study of the biosynthesis of the different sialic acids and their incorporation into macromolecules. N-Acetylneuraminic acid is the precursor of N-glycolylneuraminic acid, the N-acetyl-mono-0-acetylneuraminicacids, and t!!e N-acetyloligo-O-acetylneuraminic acids. N-Acetylneuraminate is converted to N-glycolylneuraminate by the enzyme N-acetylneuraminate, ascorbate, or NADPH:oxygen oxidoreductase (N-acetyl-hydroxylating), first described in cell-free preparations of porcine submaxillary gland (Schauer, 1970~).Schauer (1970a,b) also demonstrated the presence in equine submaxillary glands of N-acety1neuraminate:acetyl-CoA4O-acetyltransferase, and in bovine submaxillary glands of N-acetylneuraminate:acetyl-CoA 7- and/or 9-O-acetyltransferase(s) which convert N-acetylneuraminate to the corresponding N-acetyl-mono-0acetylneuraminic acids; the latter enzyme(s) also forms N-acetyl-7,9di-O-acetylneuraminic acid (Schauer and Wember, 1971). These enzymes are membrane-bound and can hydroxylate or O-acetylate either free N-acetylneuraminate before its incorporation into macromole-

42

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

cules or N-acetylneuraminate glycosidically linked to membranebound glycoproteins (Schauer and Wember, 1970, 1971). B. Polyprenol Phosphate Sugar Formation

One of the most important discoveries in the field of oligosaccharide synthesis has been the role of long-chain isoprenyl alcohols as intermediates in glycosyl transfer reactions. It is now well-established (Strominger et al., 1972; Lennarz and Scher, 1972) that phosphorylated derivatives of a C,, polyisoprenoid alcohol are involved in the biosynthesis of microbial cell wall peptidoglycan, lipopolysaccharide, mannans, and other polysaccharides; polyprenols are also involved in the synthesis of plant oligosaccharides. Two different reaction sequences have been described (Fig. 6). In type 1,oligosaccharide is assembled while attached to polyprenol by a pyrophosphate bridge and is then transferred to the growing macromolecule; in this type of mechanism, it is interesting to note that UMP is released. In the type 2 mechanism, polyprenol monophosphate merely acts as an intermediate in the transfer of a single monosaccharide from nucleotide sugar.

1. DOLICHOL PHOSPHATEMONOSACCHARIDES A family of long-chain isoprenyl alcohols, the dolichols, containing 16 to 22 five-carbon isoprene units, with the first unit saturated, has been described in animal tissues (Burgos et al., 1963; Butterworth and TYPE 1 MECHANISM UW-A

UMP

Lipid-P

UDP

Lipid-P-P-A

x

Lipid-P-P-A-B . .....etc - .....

UDP-B

....2

Liptd-P-P-A-B

____._

....Z

R-A-B

x

R-H

Li0id-P

P

TYPE 2 MECHANISM UDP-A

UOP

Lipid-P

Lipid-P-A

11': 1

Lipid-P-B

or

UDP-B .. ...etc. - . ..

R-A-B

FIG.6. Mechanisms by which polyisoprenol lipids are involved in complex carbohydrate assembly. A,B . . Z, monosaccharides; P, a monophosphate group; R-H, the acceptor (e.g., a peptide) to which the oligosaccharide becomes attached.

.

STRUCTURE A N D BIOSYNTHESIS OF M E M B R A N E G L Y C O P R O T E I N S

43

Hemming, 1968; Gough and Hemming, 1970), but until recently the function of dolichols in anim'al tissues was unknown. The work of Leloir's group (Behrens and Leloir, 1970; Behrens et al., 1971a,b, 1973; Parodi et al., 1972a,b) has shown that various dolichol monophosphate monosaccharides are synthesized in mammalian tissues and has indicated that these compounds are intermediates in polysaccharide and possibly glycoprotein assembly. Many other laboratories have followed this lead and are now engaged in mammalian polyprenol research. The first strong evidence for the formation of glycosylated polyprenol phosphates in mammalian tissues came from Leloir's group. In their first report (Behrens and Leloir, 1970), lipid was prepared from pork liver and tentatively identified as dolichol monophosphate by comparison with a standard preparation of this compound; rat liver microsomes were shown to catalyze the transfer of glucose from UDP glucose to this lipid to form a compound with the properties to be expected for dolichol monophosphate glucose (type 2 mechanism, Fig. 6). Further, dolichol monophosphate prepared by the chemical phosphorylation of highly purified dolichol also acted as a glucose acceptor in the enzyme reaction. The transferase required Mg2+for activity and was stimulated by Triton X-100 or deoxycholate. In a later paper, Behrens et al. (1971b) showed that microsomes from rat liver and brain catalyzed the formation not only of dolichol phosphate glucose from UDP-glucose but also of dolichol phosphate N-acetylglucosamine from UDP-N-acetylglucosamine and dolichol phosphate mannose from GDP-mannose; the nature of the phosphate bridge (monophosphate or pyrophosphate, Fig. 6) was not established. Both natural dolichol monophosphate and synthetic dolichol monophosphate served as exogenous acceptors in these reactions. No glycosyl transfer to dolichol monophosphate was observed with UDP-galactose, UDPN-acetylgalactosamine, or ADP-glucose. Ghalambor et al. (1974) showed that calf pancreas microsomes catalyzed the incorporation of N-acetylglucosamine from UDP-N-acetylglucosamine into endogenous lipid to form a product identified as dolichyl pyrophosphate N acetylglucosamine (P1-2-acetamido-2-deoxy-D-glucosyl P2-dolichyl pyrophosphate). The enzymic reaction required Mn2+or Mg2+,was inhibited by EDTA and by Triton X-100, but was stimulated by the chaotropic agent potassium thiocyanate. The product was identified by chromatographic and chemical comparisons with synthetic dolichyl pyrophosphate N-acetylglucosamine; both compounds released N-acetylglucosamine on acid hydrolysis (0.1 M hydrochloric acid at 100°C for 10 minutes) and N-acetylglucosamine l-phosphate on alka-

44

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

line hydrolysis (0.2M sodium hydroxide at 68°C for 1 hour) or on treatment with phosphodiesterase. The formation of lipid product was stimulated by the addition of exogenous dolichyl monophosphate. The evidence supported an earlier contention of Molnar et al. (1971) that a type 1 mechanism (Fig. 6) was involved in the transfer of N-acetylglucosamine to endogenous lipid. Molnar (1974) showed that the formation of lipid pyrophosphate N-acetylglucosamine by rat liver rough-surfaced microsomes is a reversible reaction; UDP-N-acetylglucosamine can be generated by the addition of UMP to glycosylated lipid, and UMP inhibits the formation of glycosylated lipid. 2. DOLICHOLPYROPHOSPHATE N,N'-DIACETYLCHITOBIOSE As indicated in Fig. 6, a type 1 mechanism in bacterial systems has usually been associated with the assembly of an oligosaccharide attached to a polyprenol by a pyrophosphate linkage; this also appears to be the case in mammalian systems. Thus Leloir et al. (1973) found that incubation of rat liver microsomes with exogenous dolichyl monophosphate and UDP N-[14Clacetylglucosamineresulted not only in the formation of radioactive dolichyl pyrophosphate N-acetylglucosamine but also of radioactive dolichyl pyrophosphate N,N'-diacetylchitobiose [P1-2-acetamido-2-deoxy-O-~-~glucopyranosyl-( 1-+ 4)-2acetamido-2-deoxy-~-glucosylPz-dolichylpyrophosphate 11. Mannose residues can be incorporated into dolichyl pyrophosphate N,N'-diacetylchitobiose (Hsu et al., 1974; Lucas et al., 1975; Lucas and Waechter, 1974) and a structure is assembled analogous to the mannose and N-acetylglucosamine-containingcores of Asn-GlcNAc-type prosthetic groups.

3. DOLICHOLD-MANNOPYRANOSYL PHOSPHATE

The most extensive investigations in this area have been carried out with the mannose incorporation systems, and the role of mannosylated polyprenol phosphates as intermediates in mammalian glycoprotein assembly appears well established. The pig liver system has been studied in detail by Hemming's group (Richards et al., 1971, 1972; Richards and Hemming, 1972; Hemming, 1973).Pig liver microsomes catalyze the transfer of mannose from GDP-mannose to endogenous lipid acceptor, and this reaction is stimulated by adding exogenous dolichol phosphate. A large-scale incubation using endogenous lipid acceptor was carried out with microsomes from 10 kg pig liver (Evans and Hemming, 1973), and highly purified l4C-.1abeledmannolipid was

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

45

isolated by a combination of mild alkaline saponification to destroy glycerolipids, chromatography on silicic acid, DEAE-cellulose and Kieselguhr columns, and preparative thin-layer chromatography. This mannolipid was compared with synthetic dolichyl a-D-mannopyranosyl phosphate (Warren and Jeanloz, 1973a,b), and the two compounds showed identical infrared (IR) and NMR spectra, chromatographic mobilities on thin-layer chromatography in two solvent systems, lability to mild acid, stability to mild alkali, and resistance to catalytic hydrogenation; the last-mentioned finding indicates that the endogenous lipid has a saturated terminal isoprene unit, like dolichol, and is not an allylic alcohol phosphate like the bacterial C,, polyprenol phosphate. Mass spectrometry could not be carried out on either dolichol phosphate or dolichol phosphate mannose, presumably because these compounds are not sufficiently volatile; dolichol gives a good mass spectrum, as do allylic polyprenol phosphates which dephosphorylate readily in the spectrometer (Evans and Hemming, 1973). Similar studies have been carried out with microsomes from myeloma tumor and bovine liver (Baynes et al., 1973), microsomes from hen oviduct and bovine thyroid (Waechter et al., 1973), and microsomes from calf pancreas (Tkacz et al., 1973, 1974). In all cases, divalent cation-dependent mannose transfer from GDP-mannose to an endogenous lipid acceptor was observed. The mannolipid released mannose on mild acid hydrolysis, was stable to saponification with mild alkali, but released varying amounts of mannose l-phosphate on strong alkaline hydrolysis (0.1N sodium hydroxide in 90% ethanol at 80"-100°C for 10 minutes). Baynes et al. (1973) identified their lipid component as a dolichol by mass spectrometry after strong alkaline hydrolysis; the other mannolipid preparations were identified as dolichol monophosphate mannose by chromatographic comparisons with synthetic dolichyl a-D-mannopyranosyl phosphate. Since the bridge between mannose and dolichol appears to be monophosphate in all these systems, it can be concluded that the transfer reaction has a type 2 mechanism (Fig. 6). Further proof of this is the finding that GDP strongly inhibits mannolipid formation when added prior to the start of the reaction and causes loss of mannose from mannolipid when added after initiation of the reaction; GMP has no such effect. The reversibility of the reaction is also indicated by the rapid and extensive transfer of mannose from mannolipid to GDP to form GDP-mannose (Richards et al., 1972; Richards and Hemming, 1972; Baynes et al., 1973; Waechter et al., 1973). Exogenous dolichol phosphate stimulates mannose transfer to lipid in these systems. Other polyprenol phosphates were tested as exogenous acceptors in the calf

46

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

pancreas system (Tkacz et al., 1973, 1974), and it was found that ficaprenyl phosphate (an allylic polyprenol with 11 isoprene units) and solanesyl phosphate (an allylic poIyprenol with 9 isoprene units) stimulated mannose transfer almost as effectively as dolichyl phosphate (a nonallylic polyprenol with 19 isoprene units). Farnesyl phosphate and citronellyl phosphate (3 and 2 isoprene units, respectively) had no stimulatory effect; retinyl phosphate (4 isoprene units) had a slight stimulatory effect. It was shown that the phosphates of ficaprenol, solanesol, and retinol acted as acceptors for mannose and did not act by stimulating incorporation into endogenous dolichol phosphate. Although there is overwhelming evidence implicating the dolichols as the major endogenous polyprenols involved in mammalian glycosyl transfer reactions, several reports have appeared suggesting a role for retinyl phosphate (DeLuca et al., 1973; Barr and DeLuca, 1974). C. Dolichol Pyrophorphate Oligosaccharides and the Assembly of Asn-GlcNAc Core Oligosaccharide

Initiation of the synthesis of an Asn-GlcNAc-type prosthetic group requires the attachment of an N-acetylglucosamine residue to an asparagine (or aspartic acid) residue in a polypeptide chain. RNase from bovine pancreas exists in at least four forms with identical amino acid sequences; RNase A is carbohydrate-free, whereas RNase B has a single polysaccharide prosthetic group attached to residue 34 (Tarentino et al., 1970).Since this residue is asparagine in RNase A, it is believed that asparagine rather than aspartic acid is the acceptor for N acetylglucosamine. A similar conclusion was drawn from studies comparing the incorporation of L3H]asparticacid and [3H]asparagine into rat serum glycoproteins (Kohno and Yamashina, 1973). There are presently two hypotheses for initiation of the AsnGlcNAc prosthetic group; i.e., either N-acetylglucosamine is incorporated directly into the polypeptide backbone from UDP-N-acetylglucosamine (Marshall, 1974; Khalkhali and Marshall, 1975) or the oligosaccharide core is preassembled while attached to dolichol by a pyrophosphate bridge and is subsequently transferred to the polypeptide (type 1 mechanism, Fig. 6). A large body of amino acid sequence data is now available on glycoproteins carrying Asn-GlcNAc-type prosthetic groups (Hunt and Dayhoff, 1970; Marshall, 1972, 1974).The sequence at the linkage region is always Asn-X-Ser(Thr), where X can be almost any amino acid. However, this tripeptide sequence does not invariably result in glycosylation, since many proteins with this sequence remain unglycosylated even in organs capable of making the Asn-GlcNAc linkage, e.g.,

STRUCTURE A N D BIOSYNTHESIS OF M E M B R A N E G L Y C O P R O T E I N S

47

RNase A in bovine pancreas. The factors which control oligosaccharide initiation are not known, but conformational constraints are undoubtedly involved. 1. DOLICHOLPYROPHOSPHATE OLIGOSACCHAFUDE

In the earlier work from Leloir's laboratory (Behrens and Leloir, 1970; Behrens et al., 1971b), it was reported that dolichol monophosphate glucose could transfer glucose to an endogenous acceptor to form a trichloroacetic acid-insoluble product which was not extracted with chloroform-methanol(2 : 1, v/v). This was at first assumed to be a glycoprotein but, in a most important development (Behrens et al., 1971a; Parodi et al., 1972b), it was subsequently shown that the product was in fact a dolichol pyrophosphate oligosaccharide; this latter material could be solubilized by chloroform-methanol-water (1: 1: 0.3, v/v). This experimental device was quickly applied by other laboratories, and it is now established that mannose also undergoes a similar series of reactions, namely, transfer from GDP-mannose to dolichol monophosphate to form dolichol monophosphate mannose, and transfer from this intermediate to an oligosaccharide attached to lipid (probably dolichol) by a pyrophosphate bridge (Behrens et al., 1973; Waechter et al., 1973; Hsu et al., 1974). Dolichol pyrophosphate oligosaccharides have solubility properties which allow their separation from both dolichol monophosphate monosaccharides and glycoprotein. They are insoluble in water, in aqueous ethanol, in chloroform-methanol(2 : 1, v/v), and in trichloroacetic acid, and fractionate with protein in the insoluble interphase obtained after extraction by the procedure of Folch et al. (1957); they are soluble in dimethyl sulfoxide, 6 M pyridine acetate (pH 4.4), chloroformmethanol-water (1: 1:0.3, v/v), and aqueous detergent solutions. The compounds can be purified by chromatography on DEAE-cellulose with chloroform-methanol-water (1: 1:0.3, v/v) as solvent, using a linear gradient of ammonium formate; dolichol monophosphate monosaccharides are eluted from this column well before lipid pyrophosphate monosaccharides and lipid pyrophosphate oligosaccharides (Behrens et al., 1971a). Mild acid methanolysis (0.1 N hydrochloric acid in methanol at 30°C for 60 minutes) released uncharged oligosaccharides from both [''C]glucose- and [14C]mannose-labeledlipid pyrophosphate oligosaccharides (Behrens et al., 1971a, 1973); the MWs indicated that the mannose-labeled oligosaccharides had from 5 to 18 monosaccharide units, and that the glucose-labeled oligosaccharides were somewhat larger (about 20 monosaccharide units). Mild acid hydrolysis or methanolysis released dolichol monophosphate from a

48

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

large-scale preparation of glucose-containing lipid pyrophosphate oligosaccharide from rat liver (Parodi et al., 19728); a compound believed to be dolichol pyrophosphate was also released by these treatments. Lipid pyrophosphate oligosaccharide is stable to saponification with mild alkali but can be decomposed by strong alkali treatment; treatment with 10% aqueous ammonia at 100°C for 3 hours released an oligosaccharide with a negative charge due to a phosphate ester (Behrens et al., 1971a). Alkaline hydrolysis of the products of acid methanolysis of lipid pyrophosphate oligosaccharides led to the appearance of positively charged substances believed to be due to the deacetylation of N-acetylhexosamine residues (Parodi et al., 1973; Behrens et al., 1973). ASSEMBLY 2. INITIATION OF Asn-GlcNAc OLIGOSACCHARIDE

The major unresolved question in this problem concerns the role that dolichol phosphate monosaccharides and dolichol pyrophosphate oligosaccharides play in glycoprotein synthesis. It appears to be reasonably well-established in several systems that radioactivity can be transferred either directly or indirectly from GDP-[l4C1mannose,dolichol monophosphate [14C]mannose,dolichol pyrophosphate [l4C1glucose-oligosaccharide and dolichol pyrophosphate [14Clmannose-oligosaccharide to endogenous acceptors to form glycoproteins (Parodi et al., 1972a; Behrens et al., 1973; Hsu et al., 1974; Waechter et al., 1973; Baynes et ul., 1973). For example, Behrens et al. (1973) have reported that rat liver microsomes catalyze the transfer of oligosaccharide from dolichol pyrophosphate [14C]mann~~e-oligosaccharides to endogenous protein acceptors; this product released radioactive glycopeptides on proteolysis or after alkaline hydrolysis. Behrens et al. (1973) have therefore proposed a scheme for the biosynthesis of the core of Asn-GlcNAc-type prosthetic groups: UDP-GlcNAc + dolichol-P 4 UMP + dolichol-P-P-GlcNAc UDP-GlcNAc + dolichol-P-P-GlcNAc + UDP + dolichol-P-P-GlcNAc-GlcNAc (GDP-Man and/or dolichol-P-Man), + dolichol-P-P-(GlcNAc), (GDP and/or dolichol-P), + dolichol-P-P-(GlcNAc),-(Man), (dolichol-P-Glc), + dolichol-P-P-(GlcNAc),-(Man), + (dolichol-P), + dolichol-P-P-(GlcNAc),-(Man),-(Glc), --f

It has been shown (Parodi et al., 1972a; Behrens et al., 1973) that both

dolichol-P-P-(GlcNAc),-(Man), and dolichol-P-P-(GlcNAc),-(Man),(Glc), can transfer their oligosaccharides to endogenous protein acceptors (see type 1 mechanism, Fig. 6).

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

49

Although this work most elegantly proved the role of dolichol intermediates in glycoprotein synthesis, it is not clear whether the above mechanism is applicable to all Asn-GlcNAc-type prosthetic groups. Several laboratories are accordingly engaged in studying the role of dolichols in the synthesis of specific glycoproteins such as immunoglobulin light chain (Baynes et al., 1973; Hsu et al., 1974) and hen oviduct glycoproteins (Waechter et nl., 1973). CELLSYSTEM 3. PLASMA Baynes et al. (1973) and Hsu et ul. (1974) carried out studies with the plasma cell tumor MOPC46B which secretes a K-type immunoglobulin light chain with a single Asn-GlcNAc-type oligosaccharide prosthetic group containing four mannose, three N-acetylglucosamine, four galactose, two fucose, and two sialic acid residues. Dolichol monophosphate mannose (mannolipid) synthesis can be distinguished from the transfer of mannose from mannolipid to glycoproteins by the use of EDTA; the former reaction requires a divalent cation, whereas the latter does not. Since EDTA inhibits transfer of mannose from GDP-mannose to glycoprotein, but not transfer of mannose from mannolipid to glycoprotein, it can be concluded that the latter reaction proceeds directly and not through conversion of mannolipid to GDP-mannose. Dolichol pyrophosphate oligosaccharide has been isolated from the plasmacytoma MOPC-46B (Hsu et al., 1974). A Con A-Sepharose column was used to fractionate dolichol monophosphate mannose from dolichol pyrophosphate oligosaccharide; the former compound passed through unretarded, whereas the latter compound absorbed and was eluted with a-methylmannoside. The structure of this compound has been tentatively characterized as dolichol-P-P-(GlcNAc),-(Man),. Kinetic evidence indicated formation of the dolichol pyrophosphate oligosaccharide from dolichol monophosphate mannose with subsequent transfer of the entire oligosaccharide to endogenous protein. The nature of the glycoprotein product formed in plasmacytoma is not known; the protein is membrane-bound and is solubilized by detergent. However, it is interesting that 10-20% of the plasmacytoma glycoprotein product reacted with antiserum to MOPC-46B light chain.

4. HENOVIDUCTSYSTEM Lennarz and co-workers (Waechter et al., 1973; Lucas et al., 1975; Chen et al., 1975; Pless and Lennarz, 1975) have done extensive stud-

50

J. STURGESS, M. MOSCARELLO, A N D H. SCHACHTER

ies on the role of dolichol intermediates in glycoprotein synthesis by hen oviduct. This tissue also carries out the series of reactions indicated above for rat liver and plasmacytoma. Hen oviduct synthesizes a dolichol pyrophosphate oligosaccharide with the structure (a-Man),Manal-4GlcNAc~l-4GlcNAc-pyrophosphate-dolichol. Direct transfer of radioactive oligosaccharide from this compound to endogenous glycoprotein has been demonstrated (Lucas et al., 1975).The final glycoprotein product could be digePced with proteases; however, the protein did not react with antiovalbumin antiserum and was strongly membrane-bound. The membrane-bound glycoprotein product was shown to contain the same oligosaccharide structure, (a-Man),ManPl4GlcNAca14GlcNAc, as the dolichol pyrophosphate oligosaccharide precursor. Studies have now been carried out on various other tissues; the isolation of dolichol pyrophosphate oligosaccharides from these tissues has provided strong evidence for existence of the biosynthetic pathway outlined in Fig. 7 (see reviews by Lennarz, 1975; Behrens, 1974; Heath et al., 1974; Waechter and Lennarz, 1976). However, this scheme is still tentative, because all the reactions have not been demonstrated in enzyme preparations from a single tissue source; further, none of the enzymes have been isolated and purified. 5. ANOMERICCONFIGURATION

It should be pointed out that every glycosylation reaction in the assembly of dolichol pyrophosphate oligosaccharide is believed to proceed with an inversion of configuration (Fig. 7). The nucleotide sugars required for glycosylations are all a-linked, except for GDP-L-fucose

8 ManeGlcNPcLGLcNPcdP-P-~l~~M~n~=~an~~lcNAc~G,~~~eP-P-~l

S ~ - R

&%%+Dd-p-p

I I

(Ma~=ManLGkNAc+GLcNA&ASN

FIG. 7. The role of dolichol intermediates in glycoprotein biosynthesis. Do], Dolichol; P, a monophosphate group.

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

51

which is p-linked. Herscovics et al. (1974) and Tkacz and Herscovics (1975) showed that calf pancreas microsomes and human lymphocyte homogenates catalyze the transfer of mannose from GDP-a-mannose to dolichol monophosphate to form dolichol P-Dmannopyranosyl phosphate. The transfer of mannose from GDP-a-mannose to dolichol pyrophosphate N,N’-diacetylchitobiose has been demonstrated with hen oviduct and rat liver microsomes (Levy et al., 1974); the product was tentatively identified as dolichol pyrophosphate P-mannosylN,N’-diacetylchitobiose. Dolichol monophosphate P-mannose could not serve as mannose donor for the synthesis of this dolichol pyrophosphate trisaccharide but could serve as a mannose donor for the further addition of a-linked mannose residues to form larger dolichol pyrophosphate oligosaccharides (Fig. 7). Similar findings were reported for human lymphocyte membranes (Wedgwood et al., 1974). Chen and Lennarz (1976) have reported the synthesis of dolichol pyrophosphate P-mannosyl-N,N‘-diacetylchitobiose by hen oviduct membrane and the transfer of trisaccharide from this lipid to endogenous protein acceptors. Herscovics et al. (1977a) showed that calf pancreas microsomes can effect the transfer of a-linked mannose residues from dolichyl P-D-[14C]mann~pyran~syl phosphate to dolichol pyrophosphate oligosaccharide without prior conversion to GDP-mannose. Thus it appears that GDP-a-mannose is the direct precursor of Plinked mannose residues and dolichyl P-D-mannopyranosyl phosphate is the direct precursor of a-linked mannose residues (Fig. 7). ARE ASSEMBLED BY THE 6. WHAT GLYCOPROTEINS DOLICHOLPATHWAY?

The transfer of oligosaccharide from dolichoI pyrophosphate oligosaccharide to endogenous protein acceptors has been demonstrated in several systems. The linkage of oligosaccharide to protein is stable to mild alkali, indicating that the linkage is probably of the Asn-GlcNAc type, but this has not been established (Waechter and Lennarz, 1976). The nature of the endogenous protein acceptors is not known. In all cases, the protein products are tightly bound to membrane. Further, the protein products could not be identified with the major secretory products of the tissues under study (e.g., immunoglobulin in the case of plasmacytoma, and ovalbumin in the case of hen oviduct). Finally, glucose has consistently been found in dolichol pyrophosphate oligosaccharides synthesized by rat liver (Parodi et al., 1972a; Behrens et al., 1973), calf thyroid (M. J. Spiro et al., 1976a; R. G. Spiro et al., 1976), calf kidney cortex, calf thymus and hen oviduct (M. J. Spiro et

52

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

al., 1976b), and calf pancreas (Herscovics et al., 1977a,b). Slices from kidney, oviduct, thymus, and thyroid synthesize these glucose-containing, lipid-bound oligosaccharides, while cell-free systems from plasmacytoma and oviduct make smaller lipid-bound oligosaccharides free of glucose; cell-free systems from rat liver (Behrens et al., 1973) and calf pancreas (Herscovics et al., 1977a,b) have been reported to synthesize glucose-containing lipid oligosaccharides. These various findings have indicated that the dolichol pathway (Fig. 7) may be involved in the biosynthesis of membrane-bound glycoproteins but not of secretory glycoproteins. The suggestion that novel glucose-containing glycoproteins may occur in membranes (M. J. Spiro et al., 1976a) has added weight to this hypothesis. Recent data, however, have indicated that the dolichol pathway may apply to both secreted and membrane-bound glycoproteins of the Asn-GlcNAc type. In fact, the biosynthetic scheme shown in Fig. 1 probably applies to both secreted and membrane-bound glycoproteins (Schachter, 1974a,b); the secreted protein, lacking a hydrophobic central section, presumably passes through the endoplasmic reticulum membrane into the intravesicular space, while the membrane protein becomes attached to membrane as indicated in Fig. 1. The dolichol pathway probably becomes operative either while the peptide is still attached to the ribosome or shortly after its release from the ribosome (see Section V). Earlier studies failed to demonstrate the transfer of oligosaccharide from dolichol pyrophosphate oligosaccharide to exogenous protein acceptors derived from secreted proteins such as ovalbumin; however, recent reports have indicated that such transfer can be demonstrated provided the exogenous acceptor is first unfolded by disulfide bond cleavage (Pless, 1976; Struck et al., 1977). Further evidence that the dolichol pathway is involved in ovalbumin synthesis was provided by Struck and Lennarz (1977), who showed that the antibiotic tunicamycin, which inhibits the formation of dolichol pyrophosphate N-acetylglucosamine (Tkacz and Lampen, 1975; Takatsuki et al., 1975), allows normal synthesis by hen oviduct slices of carbohydrate-free ovalbumin but inhibits the incorporation of all carbohydrate into ovalbumin. It is not clear what controls transfer of the oligosaccharide from a lipid oligosaccharide to protein; Chen and Lennarz (1976) showed that a unit as small as a trisaccharide can be transferred to protein. Since the oligosaccharide moieties of the larger dolichol pyrophosphate oligosaccharides resemble the oligomannoside structure (Fig. 2), it is reasonable to suggest that the latter are completely preassembled on lipid prior to incorporation into protein. Adamany and Spiro

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

53

(1975a,b), however, showed that al-2-linked mannose residues can be incorporated, one at a time, into exogenous glycopeptide acceptors using dolichol phosphate P-mannose as a donor (type 2 mechanism, Fig. 6). This suggests that further elongation of oligomannoside-type oligosaccharides can occur after their transfer from lipid intermediates to protein. While it appears likely that the core of N-acetyllactosamine-type oligosaccharides (Fig. 2) is assembled by the dolichol pathway (Fig. 7), elongation by the addition of N-acetylglucosamine, fucose, galactose, and sialic acid residues occurs by the addition of one sugar at a time to the glycoprotein (see Section IV,D); this process occurs primarily in the Golgi complex (see Section V). D. Elongation of N -AcetyIlactosamine-type 01igosaccharides

Four sugars are involved in the elongation of N-acetyllactosaminetype oligosaccharides, i.e., N-acetylglucosamine, fucose, galactose, and sialic acid (Fig. 2). The glycosyltransferases involved in these reactions have been reviewed (Schachter and Roden, 1973; Schachter, 1974a,b; Spiro et al., 1974); the following sections review more recent work on these enzymes.

1. N-ACETYLGLUCOSAMINYLTWSFERASES A glycoprotein N-acetylglucosaminyltransferasehas been described in goat colostrum (Johnston et al., 1966, 1973), rat liver (Johnston et al., 1973; Schachter et al., 1970),various other rat tissues (Johnston et al., 1973), guinea pig liver (Bosmann, 1970), pig liver (Hudgin and Schachter, 1 9 7 1 ~and ) ~ rat, human, and pig serum (Mookerjea et al., 1971,1972; Hudgin and Schachter, 1971~). The enzyme catalyzes the transfer of N-acetylglucosamine from UDP-N-acetylglucosamine to various glycoprotein acceptors containing prosthetic groups with WDmannose residues at the nonreducing ends, e.g., a,-acid glycoprotein and fetuin pretreated with sialidase, P-galactosidase and P-N-acetylglucosaminidase, and native ovalbumin and RNase B. There is a large variation in the V,,, achieved with these various acceptors (Johnston et al., 1973), suggesting that more than one N-acetylglucosaminyltransferase is present in the crude enzyme preparations that have been studied. The liver enzyme is strongly membrane-bound; solubilization was achieved with Triton X-100 or acetone treatment (Bosmann, 1970; ) ~ subsequent purification has so far Hudgin and Schachter, 1 9 7 1 ~but

54

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

not been possible. The serum and colostrum enzymes are nonsedimentable even in the absence of detergent, but only a 200-fold purification of the colostrum enzyme has been achieved (Johnston et al., 1973). The enzyme requires Mn2+for activity; other divalent cations (Mg+, C$+) are not as effective as stimulants (Bosmann, 1970; Hudgin and Schachter, 1971~). Recent work (Stanley et al., 1975; Narasimhan et al., 1977) has shown that this N-acetylglucosaminyltransferase activity is due to at least two separate enzymes, GlcNAc transferases I and 11. GlcNAc transferase I attaches GlcNAc in Pl-2 linkage to the Manal-S(Man-al,6-)-Man~1-4GlcNAc~1-4GlcNAc-Asn core of Asn-GlcNAc glycopeptides. This enzyme is deleted in a lectin-resistant mutant of Chinese hamster ovary cells. GlcNAc transferase I1 attaches GlcNAc in Pl-2 linkage to the product of GlcNAc transferase I (Fig. 8); it is fully active in the GlcNAc transferase I-deficient Chinese hamster ovary cell line. GlcNAc transferase I action is essential for elongation; if this enzyme is absent, GlcNAc addition cannot occur and, consequently, fucose, galactose, and sialic acid addition cannot occur (Fig. 8).Thus the GlcNAc transferase I-deficient Chinese hamster ovary cell line cannot elongate the Asn-GlcNAc core, thereby explaining the inability of these cells to bind several lectins.

,

P Dol

Gn-F -A:"-

I

G: Gn-F -Ak-

FIG. 8. Elongation of an N-acetyllactosamine-type oligosaccharide. The (Man),(GlcNAc), core strucure is presumably transferred from dolichol pyrophosphate oligosaccharide to a peptide acceptor and subsequently elongated by stepwise addition of sugars in the Golgi complex. The enzymes catalyzing these reactions are N-acetylglucosaminyltransferases (GlcNAc-Tr), fucosyltransferase (Fuc-Tr), galactosyltransferase (Gal-Tr),and sialyltransferases (Sialyl-Tr). See text for additional comments.

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

55

GlcNAc transferase I can act on Mana1-3Man/31-4GlcNAc and larger oligosaccharides, glycopeptides, and glycoproteins containing this trisaccharide sequence. GlcNAc transferase I1 acts only on the branched structure Man-(GlcNAc-Man)-Man-GlcNAc-R, where R is H or a glycopeptide or glycoprotein. Low-MW compounds such as methyl a-mannopyranoside are ineffective acceptors. 2. FUCOSYLTRANSFERASE Fucose has been found at only one position in Asn-GlcNAc glycoproteins, i.e., attached to the most internal GlcNAc residue (Fig. 2). Fucose has not been conclusively shown attached to the galactose residue of Asn-GlcNAc glycopeptides, although amino acid-free AsnGlcNAc-type oligosaccharides with a Fucal-2Gal sequence have been isolated from the tissues of a fucosidosis patient (Tsay et al., 1976). Two glycoprotein fucosyltransferase activities have been described in pig liver (Jabbal and Schachter, 1971) and human serum (Munro and Schachter, 1973) which transfer fucose from GDP-fucose to sialidase-treated a,-acid glycoprotein and sialidase-, P-galactosidase-treated a,-acid glycoprotein, respectively. The former activity attaches fucose to a terminal P-galactoside residue, and the human serum enzyme is the blood group H-dependent a-2-fucosyltransferase. The transferase acting on sialidase-, P-galactosidase-treated alacid glycoprotein has recently been shown to transfer fucose to the asparagine-linked GlcNAc residue (Fig. 8) (Wilson et al., 1976). The transfer of fucose to the most internal GlcNAc residue requires the prior incorporation of at least one GlcNAc residue in Pl-2 linkage to a mannose residue at the core oligosaccharide (Fig. 8).

3. GALACTOSYLTRANSFERASES Galactose appears in only a single location in Asn-GlcNAc-type oligosaccharides, namely, linked to N-acetylglucosamine either as the terminal or penultimate (to sialic acid) residue at the nonreducing end (Fig. 2). Many tissues are capable of transferring galactose from UDPgalactose to either free N-acetylglucosamine to form N-acetyllactosamine or to gl ycoproteins with a P-N-acetylglucosaminide nonreducing terminus, e.g., @,-acidglycoprotein or fetuin pretreated with sialidase and 0-galactosidase. The linkages synthesized by the glycoprotein galactosyltransferases have not been characterized and, since G a l p l + 4GlcNAc and Galj31+ 6GlcNAc linkages have been described in Asn-GlcNAc-type oligosaccharides, more than a single ga-

56

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

lactosyltransferase may contribute to the incorporation of galactose into glycoprotein acceptors. Asn-GlcNAc-type glycoprotein galactosyltransferase is tightly bound to membrane within the cell but exists in a soluble form in milk, colostrum, and serum (see Schachter and Rod&, 1973, for references). The milk galactosyltransferase is in fact equivalent to the A protein, one of the two components of lactose synthetase (Brew et al., 1968). The A protein by itself has a very low affinity for glucose, although it can synthesize lactose at very high glucose concentrations; in the presence of the B protein (a-lactalbumin),however, the affinity for glucose is greatly increased and lactose synthesis occurs readily. Since a-lactalbumin occurs only in mammary gland, lactose synthesis is confined to this organ; the function of the A protein in other tissues is believed to be in the biosynthesis of Asn-GlcNAc-type glycoproteins. It is interesting that the galactosyltransferase in liver, serum, and other tissues can make lactose provided exogenous a-lactalbumin is provided (Hudgin and Schachter, 1971b; Fitzgerald et al., 1971); it is not, however, certain that the milk A protein and the various membrane-bound galactosyltransferases are identical proteins. Bovine and human milk A protein has been purified to homogeneity by classic methods (Fitzgerald et al., 1970) and by affinity chromatography with an a-lactalbumin-Sepharose column (Trayer and Hill, 1971; Andrews, 1970; Khatra et al., 1974). The purified enzyme catalyzes the incorporation of galactose into N-acetylglucosamine as well as into oligosaccharides and glycoproteins with a nonreducing p-Nacetylglucosaminide terminus, indicating that these activities are catalyzed by a single enzyme. Although a-lactalbumin under some conditions inhibits transfer to N-acetylglucosamine (Kitchen and Andrews, 1972), there is no effect on galactose incorporation into oligomers of N-acetylglucosamine and into glycoproteins (Schanbacher and Ebner, 1970). The pure enzyme requires Mn2+for activity. Detailed kinetic studies have been carried out on the milk transferase in the presence and absence of a-lactalbumin (Morrison and Ebner, 1971a,b; Khatra et al., 1974), and orders of binding to enzyme have been suggested for monosaccharide, UDP galactose, Mn2+,UDP, disaccharide product, and a-lactalbumin. This galactosyltransferase has been demonstrated in goat colostrum, rat liver, and other rat tissues (McGuire et al., 1965; Schachter et al., 1970; Carlson et al., 1973a), pig liver and serum (Hudgin and Schachter, 1971b), rat serum (Wagner and Cynkin, 1971), human serum (Kim et al., 1972a,b), amniotic fluid (Nelson et al., 1974), cerebrospinal fluid (Den et al., 1970), mouse mastocytoma (Helting and Erbing, 1973), thyroid (Spiro and Spiro, 1968b), human saliva (Naku-

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

57

mura et al., 1974; Nakumura and Tsunemitsu, 1975a,b), and human urine (Chester, 1974). The A protein is a glycoprotein (Lehman et al., 1975) which exists in milk in at least two forms (MWs 58,000 and 42,000); the smaller proteins are proteolytic products, and a single protein (MW 51,000) can be obtained from bovine colostrum which contains protease inhibitors (Powell and Brew, 1974). The enzyme requires a sulfhydryl group (Magee and Ebner, 1974; Kitchen and Andrews, 1974). Its interaction with a-lactalbumin has been studied fluorimetrically (Prieels and Barel, 1975), and its interaction with UDP-galactose by CD (Geren et al., 1975). 4. SIALYLTWSFERASES Like galactose, sialic acid also occurs in only a single position in Asn-GlcNAc-type prosthetic groups, namely, attached to galactose at the nonreducing terminus (Fig. 2). However, the linkages to galactose may be a 2 + 2, a2 + 3, a2 + 4, and a 2 4 6, and a family of several different sialyltransferases is probably involved in the synthesis of Asn-GlcNAc-type glycoproteins. These sialyltransferases are assayed by measuring the incorporation of sialic acid from CMP-sialic acid into low-MW acceptors such as lactose or N-acetyllactosamine and into gl ycoproteins with P-galactoside termini such as fetuin or a,-acid glycoprotein pretreated with sialidase; these activities have been described in rat mammary gland (Carlson et al., 1973b), goat, bovine, and human colostrum (Bartholomew et al., 1973), rat liver (Schachter et al., 1970; Bernacki and Bosmann, 1973), pig liver and serum (Hudgin and Schachter, 1971a),human serum (Kim et al., 1971; Mookerjea et al., 1972), and thyroid (Spiro and Spiro, 1968a). The same enzyme probably acts on both low- and high-MW acceptors (Hudgin and Schachter, 1971a; Bartholomew et al., 1973). Rat mammary gland sialyltransferase is strongly membrane-bound and could not be solubilized with detergents. There was no requirement for metal. The crude microsomal enzyme utilized either CMPN-acetylneuraminic acid or CMP-N-glycolylneuraminic acid as sialic acid donor with lactose as acceptor. Only compounds with P-galactoside termini were effective acceptors; a large number of other compounds, including several a-galactosides, were ineffective. The finding that the mammary gland enzyme was relatively inactive toward high-MW acceptors (Carlson et al., 197313) is probably due to the fact that no detergent was present in the assay; detergent is known to be required for optimal sialyltransferase activity with glycoprotein acceptors (Schachter et al., 1970). Goat, bovine, and human colostrum contain a soluble sialyltrans-

58

J. STURGESS, M. MOSCARELLO, A N D H. SCHACHTER

ferase which transferred sialic acid from CMP sialic acid to low- and high-MW /3-galactosides (Bartholomew et al., 1973). The bovine colostrum enzyme has recently been purified 440,000-fold (Paulson et al., 1977a) by the use of affinity chromatography using CDP as a ligand. The colostrum enzyme differs from the mammary gland enzyme in two important respects: (1)the rat mammary gland enzyme showed approximately equal activity with the following acceptors: Galpl + 4GlcNAc, GalP1 + SGlcNAc, Gal/31+ GGlcNAc, and Gal/3l+ 4Glc; the colostrum enzyme (and the pig liver enzyme, Hudgin and Schachter, 1971a) showed a marked preference for Gal/31+.4GlcNAc (the requirement for Galpl + 4GlcNAc was absolute in the case of the pure enzyme, Paulson et al., 1977b); (2) the mammary gland enzyme made predominantly the a2 + 3 linkage, while the pure bovine colostrum enzyme synthesized only the a2 + 6 linkage. The first observation illustrates the important point that, although the terminal sugar of the acceptor is the major factor controlling transferase specificity, internal sugars can also influence transferase activities. The second observation shows that different positional isomers of the same two sugars are synthesized by separate transferases; similar conclusions were drawn from a study of sialyltransferases during embryological development (Hudgin and Schachter, 1972). Considerations of this type are the basis of the one linkage-one enzyme hypothesis (Schachter and Roden, 1973). E. Assembly of Ser(1hr)-GalNAc Oligosaccharides

Ser(Thr)-GalNAc oligosaccharide assembly probably does not involve lipid intermediates and occurs by the sequential addition of one sugar at a time to the growing glycoprotein. The process is illustrated using salivary gland mucins as examples. Ovine and porcine submaxillary mucins (OSM and PSM) have been thoroughly characterized, and all the glycosyltransferases required for assembly of their oligosaccharide prosthetic groups have been described (Schachter and Roden, 1973; McGuire, 1970; Schachter et al., 1971; Carlson et al., 1973c; McGuire and Roseman, 1967). Figure 9 summarizes the biosynthetic path for the major oligosaccharide prosthetic groups of OSM and PSM and indicates that five different glycosyltransferases are involved. Since these glycosyltransferases have not been purified to homogeneity, the possibility of a lipid intermediate in monosaccharide transfer from nucleotide sugar to glycoprotein (type 2 mechanism, Fig. 6) has not been completely ruled out.

59

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

“

R - G GolNAvSA

R-O-GolNAc I

GP-Fur

R-O-GalNAc

Gbi

-

“cr._NcI Got-Fuc

R- O- G o i N k I 001- Fuc

GoINAc

,,

I?

I R-O-GolNAc-SA

Gbi-Fuc (Kil -,PSM

,

(1,

R ~ Ga:NAc. o ~

I

9

Gai-Fuc GolNAc IA.1 -PSM

FIG. 9. Biosynthesis of OSM and PSM. Question marks designate pathways not directly tested but which could exist because of the presence of the above oligosaccharides in PSM (Carlson, 1968). (A-)-PSM and (A+)-PSM refer to mucins lacking or carrying the human blood-group-A determinant, i.e., GalNAca l-B[Fucu 1-2]Gal-.

The first step in the assembly process is the attachment of N-acetylgalactosamine to an hydroxyamino acid in the peptide backbone (McCuire and Roseman, 1967; McCuire, 1970). The polypeptide N acetylgalactosaminyltransferase is present in particulate form in mammalian submaxillary glands and has been partly purified. The polypeptide acceptor is prepared by treating OSM sequentially with sialidase and a-N-acetylgalactosaminidaseto remove the disaccharide prosthetic groups. The enzyme has a high specificity for this acceptor; a large number of other compounds was ineffective, and pronase digestion of the carbohydrate-free OSM destroyed its acceptor activity. This high degree of specificity for a particular high-MW polypeptide is shown by all the known transferases involved in synthesizing linkages between monosaccharides and amino acids. Bovine submaxillary gland has a similar N-acetylgalactosaminyltransferase(Hagopian and Eylar, 1968a,b, 1969a,b) which transfers N-acetylgalactosamine to carbohydrate-free bovine submaxillary mucin; this enzyme is also very specific for its polypeptide acceptor. Hagopian et al. (1971) showed, however, that the enzyme can transfer N-acetylgalactosamine to a specific threonine residue in a basic protein isolated from bovine myelin; this myelin protein is not a glycoprotein but presumably accepts N acetylgalactosamine because of an amino acid sequence similar to the polypeptide core of submaxillary mucins. Comparison of amino acid sequences near Ser(Thr)-GalNAc linkage regions (Marshall, 1972; Hill, 1976; Hill et al., 1977a,b) has not clarified the specificity requirements for the transferase.

60

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

There is a branch point in the biosynthetic pathway (Fig. 9) after the first N-acetylgalactosamine is incorporated. If galactose is incorporated before sialic acid, the pathway proceeds toward the synthesis of complex oligosaccharide side chains (the major components of PSM); if sialic acid is incorporated before galactose, galactose cannot be incorporated and assembly stops at the disaccharide stage (the major component of OSM). The sialyl- and galactosyltransferases responsible for these two reactions are present in both ovine and porcine submaxillary glands, but there is relatively little galactosyltransferase activity in ovine glands (McGuire, 1970; Schachter et al., 1971; Carlson et al., 1973~). Ovine glands therefore make predominantly the disaccharide side chain, whereas porcine glands make both disaccharide and larger oligosaccharide side chains. The key enzyme in this control process is the galactosyltransferase (Schachter et al., 1971). This enzyme is strongly bound to membrane, is activated by Triton X-100, has an absolute requirement for Mn2+, and incorporates galactose into linkage with the terminal N-acetylgalactosamine of sialidase-treated OSM probably in pl 3 linkage. The galactosyltransferase will not act if the terminal N-acetylgalactosamine of the acceptor is substituted with a sialic acid residue (Fig. 9). Both galactosyl- and sialyltransferases compete for the same substrate (Fig. 9), namely, N-acetylgalactosamine attached to the polypeptide core of the mucin. Sialyltransferases capable of transferring sialic acid from CMP-sialic acid to sialidase-treated ovine, bovine, and porcine submaxillary mucins have been described in ovine, bovine, and porcine submaxillary glands (Carlson et al., 1973~). The product formed by the sheep enzyme with CMP-N-acetylneuraminic acid as donor was shown to be N-acetylneuraminyl-(2 -+6)-N-acetylgalactosamine peptide, the same linkage found in naturally occurring OSM. A variety of low-MW compounds either with or without terminal N acetylgalactosaminide residues were ineffective as acceptors. Carlson et al. (1973~) found that the sheep enzyme could use either CMP-Nacetylneuraminic acid or CMP-N-glycolylneuraminic acid as sialic acid donors. Schauer and Wember (1973) examined the latter point more carefully and found that the sialyltransferases of bovine, porcine, and equine submaxillary glands transferred N-acetyl-, N-glycolyl, N-acetyl-7(or g)-O-acetyl-, and N-acetyl-4-O-acetylneuraminic acids from their respective CMP-glycosides to endogenous acceptors at similar rates; all four nucleotide sugars gave the same pH optima and K, values, and there was competition between nucleotide sugars for a common enzyme active site. Thus the mucin sialyltransferases are not specific for the acyl groups attached to neuraminic acid; the

STRUCTURE A N D BIOSYNTHESIS OF M E M B R A N E GLYCOPROTEINS

61

different ratios of various sialic acids found in salivary mucus glycoproteins from different species apparently depend neither on CMPsialic acid synthetase nor on sialyltransferases, but on the activities of the oxidoreductase and acetyltransferases which modify N-acetylneuraminic acid. If galactose is incorporated into protein-bound N-acetylgalactosamine before sialic acid (Fig. 9), further growth of the oligosaccharide can occur. A fucosyltransferase is present in porcine submaxillary gland capable of transferring fucose from GDP-fucose to both high and low-MW acceptors with terminal galactose residues, e.g., sialidasetreated a,-acid glycoprotein, lactose, and the P l + 3, /3l +. 4, and P 1 4 6 isomers of galactopyranosyl-N-acetylglucosamine(McGuire, 1970); the enzyme can also transfer fucose to the terminal galactose residue of the product of the galactosyltransferase reaction (Fig. 9) obtained either from a large-scale galactosyltransferase incubation or by treating blood group A-negative PSM with 1 N hydrochloric acid at 70°C for 2 hours to remove terminal fucose and sialic acid residues. The final enzyme in the synthetic scheme (Fig. 9) occurs only in pigs genetically capable of making an antigen similar to the human blood-group-A antigen. This porcine submaxillary gland N-acetylgalactosaminyltransferase converts blood group A-negative PSM to blood group A-positive PSM (McGuire, 1970; Schwyzer and Hill, 1977a,b) and has the same substrate specificities as the blood group A-dependent N-acetylgalactosaminyltransferase present in human tissues; the enzyme will incorporate N-acetylgalactosamine in a1 -+ 3 linkage to a terminal P-galactoside residue of both low- and high-MW acceptors, provided the terminal galactose has a fucose residue attached in a1 + 2 linkage. The blood group enzymes are not discussed in this chapter. V.

SUBCELLULAR SITES OF GLYCOSYLATION

Figure 1 illustrates a model for membrane glycoprotein biogenesis. Data on the subcellular localization of the various glycosylation reactions support this scheme (Schachter and Rod&, 1973; Schachter, 1974a,b), and some of this evidence is reviewed in this section. A. Autoradiographic Evidence

With the use of radioactively labeled carbohydrate precursors, considerable information has been gained about the site of glycosylation

62

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

of macromolecules and the turnover of glycoproteins. Monitoring the fate of radioactivity has involved biochemical studies on subcellular membrane fractions (Schachter, 1974a,b) and electron microscope autoradiographic studies to localize the site of radioactivity. The latter results must be interpreted with caution, since the technique lacks specificity for one type of macromolecule, since the precursor may be metabolized prior to incorporation, and since incorporation may not occur at one specific site of the macromolecule. Early autoradiographic studies showed that radioactive carbohydrate precursors produced grains initially in the cytoplasm and later at the cell surface; this was direct evidence that each cell is the source of its own surface material (Ito, 1969). Since that time, autoradiographic techniques have been used extensively to study the sites of assembly of secretory glycoproteins and of membrane-associated glycoproteins (Neutra and Leblond, 1966; Bennett et d.,1974). Labeling of cell surface glycoproteins has been demonstrated using radioactive carbohydrates as precursors and, in general, the pattern of labeling is similar in several cell types. For instance, in the duodenum, incorporation of mannose is observed diffusely in the cytoplasm at 10 minutes and at the cell surface after approximately 5 hours. Fucose appears within 2 minutes in the Golgi complex and within 20 minutes at the apical and lateral cell surfaces, where it remains for up to 30 hours. Galactose appears within 10 minutes in the Golgi complex and at 30 minutes at the apical cell surface (Leblond and Bennett, 1974). From such data, it appears that the elaboration of cell surface glycoprotein follows a pathway similar to that of soluble secretory glycoproteins and also follows a similar time sequence. It should be noted that these techniques do not distinguish membrane glycoproteins from secretory products which remain associated with the cell surface. The incorporation of glucosamine and mannose residues occurs at the level of the rough endoplasmic reticulum and also within the Golgi complex (Sturgess et d.,1972; Moscarello et d.,1972). The interpretation of studies with mannose is subject to some dispute, since it is readily metabolized to other glycosyl residues (Mitranic and Moscarello, 1972). The addition of galactose, fucose, and sialic acid residues occurs exclusively within the Golgi complex. These findings support the idea that the mannose- and glucosamine-containing core (Fig. 2) is assembled in the rough-surfaced endoplasmic reticulum, while elongation (the addition of N-acetylglucosamine, galactose, fucose, and sialic acid) occurs almost entirely in the Golgi’apparatus. The cisternae and tubular network of the hepatocyte Golgi complex (Fig. 10)are the earliest sites of assembly of carbohydrates into macromolecules, and labeled macromolecules are then transferred to se-

STRUCTURE A N D BIOSYNTHESIS OF M E M B R A N E GLYCOPROTEINS

63

FIG. 10. Transmission electron micrograph of rat liver showing the characteristic cisternae (C), vesicles (V) representing the tubular network, and larger secretory vesicles of the Golgi complex. The Golgi complex is bounded by microtubules (M) and polarized so that its secretory or trans face (SF)is adjacent to the plasma membrane (PM). The tubular network is presumed to be the site of giycosylation of macromolecules and is a site which may show extensive proliferation following perturbation of the Golgi complex functions. x 56,000.

cretory vesicles which migrate to the cell surface (Sturgess et al., 1973);radioactivity which remains associated with the hepatocyte cell surface for prolonged time intervals has been attributed to the labeling of integral membrane glycoproteins. Few studies of glycoprotein biosynthesis have focused on the labeling of specific membrane glycoproteins, an approach essential in characterizing the biosynthetic site and turnover of these molecules. With rhodopsin, specific labeling experiments have been carried out, and these indicate that assembly occurs in the Golgi complex, with subsequent transfer to plasma membrane (Bok et al., 1974; Papermaster et al., 1975). Rhodopsin offers many advantages in studying this problem. It forms 80-90% of the protein of rod outer segments and is readily purified. The biosyn-

64

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

thesis of rhodopsin has been studied in frog retina using radioactive amino acids (Papermaster et d.,1975). Incorporation of label was rapid, and radioactivity was transferred from retinal subcellular fractions to rod outer segments after 2 hours. Immunochemical analysis showed that the newly synthesized protein was membrane-bound upon completion of synthesis. A soluble form was not detected. Darkadapted retinas were incubated in vitro in the presence of [3Hlglucosamine (Bok et al., 1974). Radioactivity appeared early over the ribosomes, in the Golgi apparatus after 20 minutes and, after 2 hours, in the base of the outer segment where it was inserted into the disk membrane. It was concluded that rhodopsin was fully glycosylated prior to its assembly into the disk membrane. On the basis of x-ray diffraction analysis, Worthington (1973) concluded that the nonpolar parts of rhodopsin reside in the low-density hydrocarbon chain region and that the polar parts may protrude a short distance from the membrane surface. Support for the conclusion that rhodopsin is probably buried deeply in the membrane comes from studies on model membranes (see Section 111). B. Subcellular localization of Glycosyltrunsferases

The subcellular localization of glycosyltransferases has been reviewed (Schachter, 1974a,b). The evidence is extensive that the Golgi complex is the major site of the transferases involved in elongation of N-acetyllactosamine-type oligosaccharides, i.e., the N-acetylglucosaminyl-, fucosyl-, galactosyl-, and sialyltransferases discussed in Section IV,D. Table I1 shows supportive data obtained from rat liver (Munro et al., 1975). The extensive literature on the cell surface localization of glycosyltransferases (Shur and Roth, 1975; Culp, this volume) has been obtained primarily with cultured cells; further, the relative contributions of the Golgi complex and plasma membrane to cellular glycosyltransferase activity were usually not assessed in these studies. Rat liver plasma membrane is essentially devoid of glycosyltransferase activity (Munro et al., 1975). Thus, evidence suggests that the Golgi apparatus is the main site of oligosaccharide elongation. The role of cell surface glycosyltransferases either in glycosylation reactions or in cell-cell interactions remains controversial. C. The Role of the Ribosome

Initial stages of peptide assembly occur at the ribosomal level, and there is considerable evidence that membrane glycoproteins, des-

66

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

tined for the cell surface, are synthesized on bound ribosomes (Siekevitz, 1972; Morrison and Lodish, 1975; Bergeron et al., 1975). In contrast, it has been suggested that nonglycosylated membrane proteins, destined for the cytoplasmic face of the endoplasmic reticulum or plasmalemma, are synthesized by free ribosomes (Autuori et al., 1975a,b; Elhammer et al., 1975; Svensson et al., 1976; Lodish, 1973; Ito and Sato, 1969). The membrane glycoproteins of the endoplasmic reticulum present a special biosynthetic problem, since glycosylation is completed in the Golgi complex, and there must be a mechanism for transporting these molecules from the Golgi apparatus back to the endoplasmic reticulum. To explain the transfer of proteins across membranes, Blobel and Dobberstein (1975a,b) have proposed the “signal” hypothesis for which supporting data have been reviewed previously (Blobel and Sabatini, 1971). The hypothesis is based on the presence of a sequence of signal codons on mRNAs whose translation products are destined for transfer across a membrane. Translation of the signal codons results in a unique sequence of amino acids at the N-terminal end of the nascent peptide chain. The assembly of peptides begins on free ribosomes, at which stage the membrane can recognize the N-terminal end with its signal sequence of hydrophobic amino acids 10 to 40 amino acid residues long (Fig. 1).Attachment of the ribosomes to membrane receptor protein occurs only if the signal is present, whereas other peptides remain free in the cytoplasm. As a result of membrane-associated proteolysis, the N-terminal end of the peptide is removed (Milstein et al., 1972; Blobel and Dobberstein, 1975a). The protein may then become an integral part of the membrane or be released into the “soluble” compartment of the endoplasmic reticulum as a secretory protein. The signal hypothesis is based primarily on work with secretory proteins such as pancreatic enzymes and immunoglobulins (Blobel and Dobberstein, 1975a,b; Schechter et al., 1974). It explains why secretory proteins are translated on membrane-bound ribosomes and how a soluble hydrophilic protein crosses the lipid bilayer from the cytoplasmic side of the endoplasmic reticulum membrane to the intravesicular side. It seems highly probable that intrinsic membrane glycoproteins, which must also traverse the lipid bilayer (at least partly) during translation, are also translated on membrane-bound ribosomes and that the nascent peptides of these proteins have signal sequences at their N-terminal ends (Fig. 1).There is strong evidence that all glycoproteins are translated on membrane-bound ribosomes (see Schachter, 1974a,b, for reviews), but evidence suggesting a signal sequence for membrane glycoproteins has only recently become avail-

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

67

able. It has been shown that the mRNAs coding for the membrane glycoproteins of enveloped animal viruses (such as VSV and Sindbis virus) are translated solely on membrane-bound ribosomes (Morrison and Lodish, 1975; Wirth et al., 1977; Toneguzzo and Ghosh, 1975); nonglycoprotein components of these viruses are 'translated on free ribosomes. Sindbis virus has three structural proteins, two envelope glycoproteins (E, and E,), and a nonglycosylated core protein. El and E, are inserted into the plasma membrane of infected cells as integral membrane proteins and become part of the viral envelope as the virus buds through the plasma membrane. All three structural proteins are translated from a single mRNA using a single initiation site. Wirth et al. (1977) found that all three proteins were translated on membranebound ribosomes; however, El and E, are vectorially transported through the endoplasmic reticulum membrane (they sediment with the membrane and become unavailable to proteases), while the core protein is released to the cytoplasmic side of the membrane. These findings show that neither the mRNA nor the ribosome can determine the segregation of nascent protein to either the cytoplasmic or intravesicular side of the membrane. Rather, the peptide itself, presumably by means of some sort of signal sequence, directs the binding of the ribosome to the membrane and subsequent vectorial discharge through the lipid bilayer (Fig. 1). In rat liver and in plasmacytomas, a small amount of N-acetylglucosamine has been shown to be incorporated into nascent peptide still attached to ribosomes (Lawford and Schachter, 1966; Molnar and Sy, 1967; Sherr and Uhr, 1969; Cowan and Robinson, 1970). In a recent reinvestigation of this point, Kiely et al. (1976) showed that both glucosamine and mannose were present on nascent ovalbumin chains still bound to ribosomes by tRNA. It is likely that this incorporation represents only a small fraction of the total and that most glycosylation is a postribosomal event (Schachter, 1974a,b; Jamieson, 1977).

VI.

MEMBRANE BlOGENESlS

On the basis of present evidence, it is assumed that the biosynthetic mechanisms for secretory glycoproteins and structural glycoproteins of cell membranes are very similar. Common features are found in the soluble, secreted proteins and the membrane glycoproteins in terms of their overall composition and the assembly of oligosaccharide sequences. There is no evidence to date indicating that the mode of biosynthesis is different. The role of the rough- and smooth-surfaced endoplasmic reticulum

68

J. STURGESS, M. MOSCARELLO, A N D H. SCHACHTER

and Golgi complex in the biosynthesis and intracellular transport of secretory glycoproteins makes it difficult to distinguish glycoproteins destined for secretion from those which are to become an integral part of the membrane, presumably by incorporation into the lipid bilayer through specialized hydrophobic peptide sequences (Fig. 1).At the present time, it is assumed that membrane glycoproteins are synthesized on the rough endoplasmic reticulum and then transported through the endomembrane system to the Golgi complex and then to the final membrane site (Figs. 1 and 11). Other membrane components, presumably lacking carbohydrate, may be made on free ribosomes in the cytoplasm and may migrate through the cytosol as soluble components until they are incorporated into a membrane structure (Bretscher, 1973; Bretscher and Raff, 1975). Glycosyltransferase enzymes are tightly bound to membranes of the endoplasmic reticulum and the Golgi complex (Schachter, 1974a,b), and the glycosylations catalyzed by these enzymes take place predominantly at the internal surface of the cisternae and within the Golgi vesicles (Nicolson and Singer, 1971; Hirano et al., 1972). Thus glycosylation is a membranebound phenomenon, and the available evidence indicates that all glycoproteins must pass through the endomembrane system (Fig. 1). Bretscher (1973) has suggested that glycosylation may occur at the plasma membrane to "lock" into the membrane proteins that migrate to it through the cytosol. This seems unlikely for at least two reasons: (1) Cells such as liver lack plasma membrane glycosyltransferases (Munro et al., 1975); (2) model membrane studies (Section 111) indicate that carbohydrate is not essential for the insertion of a hydrophobic protein into the membrane. Rather, nonglycosylated proteins that migrate to the plasma membrane through the cytosol probably bind to sites on the membranes to become peripheral membrane proteins at the cytoplasmic face (Rothman and Lenard, 1977). The difficulty in distinguishing secretory and membrane proteins at early stages of their biosynthesis suggests that it is best to study membrane glycoprotein synthesis in a nonsecretory tissue (e.g., retinal rod cells which make rhodopsin). Rat liver microsomes contain membrane-associated glycoproteins which can be released only by detergents, but these glycoproteins contain fewer sugars than glycoproteins discharged into the lumen of the microsomal vesicles (Redman and Cherian, 1972); it is therefore not clear whether these membrane glycoproteins are destined to be membrane glycoproteins 'firmly integrated into the membrane by hydrophobic interactions or are unfinished polypeptide moieties of future soluble secretory glycoproteins which are still attached to the membrane. Similarly, in the Golgi com-

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

69

FIG. 11. Transmission electron micrograph of rat liver showing the Golgi complex and endoplasmic reticulum involved in membrane biogenesis. The Golgi complex (G) is polarized with one side associated with the formation of secretory products. This surface is commonly directed toward the plasma membrane. In membrane biogenesis, the flow of membrane is thought to occur from the nuclear envelope (N), to the rough endoplasmic reticulum (RER), and by transition vesicles (arrows) to the Golgi complex; at this site, membrane differentiation occurs, and membrane is transferred to the cell surface (PM) via small and large secretory vesicles (SV). x 19,000.

plex of rat liver, glycoproteins destined for secretion remain firmly bound to the membrane during early stages of assembly (Moscarello et al., 1972), so that it is difficult to dissociate precursors of soluble glycoproteins from integral membrane glycoproteins. This problem is a complicating factor in many of the studies that

70

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

bear on the intracellular migration of membrane glycoproteins from the endoplasmic reticulum to the plasma membrane and to other intracellular membrane sites. A. Biogenerir of Plasma Membranes

The transport of secretory (and presumably membrane) proteins from their site of assembly in the rough endoplasmic reticulum to the Golgi complex is an energy-dependent process that is blocked by respiratory inhibitors (Jamieson and Palade, 1968).The transfer of membrane has been characterized morphologically by the appearance of transition vesicles between the rough endoplasmic reticulum and the periphery of the Golgi complex (Fig. 11).These vesicles may arise by degranulation of rough endoplasmic reticulum (Morre et al., 1970), possibly accompanied by insertion of new protein, or by lateral diffusion of membrane protein. An alternative mechanism may involve the growth of membrane from a restricted area of the rough endoplasmic reticulum (Claude, 1970), in which case the individual membrane components may be synthesized and assembled at this site. The transitional elements (smooth-surfaced vesicles) present between the rough endoplasmic reticulum and Golgi complex appear to have distinctive properties. Beadlike structures have been identified at the base of the transition vesicles by staining with bismuth. These structures, which are 10-12 nm in diameter and are probably protein in nature (Locke and Huie, 1976), are exclusive to this intracellular site and may function in membrane biogenesis. 1. THEROLE OF THE GOLGICOMPLEX

The Golgi complex has been assigned a role in the differentiation and distribution of membrane as well as of secretory products. While some phospholipid assembly occurs in the Golgi complex (Chang et al., 1977), it is dependent on the rough endoplasmic reticulum for its protein. Following glycosylation, secretory glycoproteins are transferred from the Golgi complex to the cell surface primarily by the release of secretory vesicles and their fusion with the plasma membrane (Jamieson and Palade, 1971); the same vesicles probably also carry out the transport of membrane glycoproteins. Membrane fusion is followed by restricted transfer and/or recycling of membrane components (Fig. 1).Alternatively, membrane glycoproteins may be transferred to the plasma membrane (or to intracellular membranes) in

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

71

individual cytoplasmic glycoprotein-lipid complexes which are wholly assimilated into the membrane bilayer. The Golgi complex is highly developed in secretory cells, where it is involved in the biosynthesis and secretion of macromolecules. In monolayer cultures, where cell surfaces are very extensive, a welldeveloped Golgi complex is observed even when secretory activity is low (Fig. 12).In hepatocyte cultures (Odashimaet al., 1976),the Golgi complex shows particularly elaborate networks of tubules (Fig. 13)far more extensive than those observed in isolated hepatocytes. Although these cells continue to secrete plasma glycoproteins in culture, the levels are low, indicating a possible role of the Golgi complex in the differentiation of membrane destined for the cell surface. At the cell surface, the carbohydrate portion of the glycoprotein molecule is exposed at the outer surface of the plasma membrane. In contrast, the carbohydrate portions of glycoproteins membrane-bound in the Golgi complex are predominantly situated at the intracisternal surface of the membrane. Palade and his co-workers demonstrated that, although enzymically and chemically Golgi membranes showed distinct differences from plasma membranes, there was a gradation of

FIG. 12. Transmission electron micrograph of human skin fibroblast showing characteristic cisternae (C) of the Golgi complex and distinctive tubular network (arrows). Secretory vesicles and vacuoles appear to arise from this network and from the segregation of products within the cistemae. x 19,OOO.

72

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

FIG.13. Transmission electron micrograph of rat hepatocyte in monolayer culture showing elaborate Golgi complex (arrows) characteristic of these rapidly proliferating cells. The Golgi complex exhibits a large membrane surface with numerous cisternae (C) and elaborate tubular networks (arrows). In these cells which secrete serum glycoproteins in small amounts, the extensive Golgi network suggests a direct relationship between the Golgi complex and the extensive cell surfaces. x 3000.

composition. For example, 5'-nucleotidase is concentrated at both the external surface of the liver plasma membrane and at the inner surface of the Golgi secretory vesicle (Farquhar et al., 1974). Similarly, insulin-binding glycoprotein (Bergeron et al., 1973) and polypeptide hormone receptors (Bergeron and Posner, 1975) are present in rat liver Golgi membranes and plasma membranes, a relationship presumed to be biogenetic rather than functional. Current hypotheses indicate that vesicles arising from the Golgi complex are transported to the cell surface, fuse with the plasma membrane, and thus transfer structural proteins into the cell surface. Carbohydrate residues originally intracisternal in the Golgi complex would then be externalized (Fig. 1). With the continuous contribution of membrane components to the cell surface, a balancing mechanism must exist to maintain its normal area. Selected proteins and glycoproteins may congregate by lateral diffusion followed by internalization of membrane vesicles by endocytosis; or internalization of individual proteins may occur in a lipid

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

73

complex. Degradation and shedding of membrane glycoproteins on the external surface of the cell is also a feature of this mechanism. 2. “MEMBRANEFLOW”AND “MEMBRANE SHUTTLE”

In secretory systems, exportable proteins are transported through the cell along a discontinuous pathway, including a number of membrane-bound compartments (rough endoplasmic reticulum, Golgi complex, secretion granules). The products are segregated within granules and discharged into the extracellular space following fusion of the granule membrane with the plasma membrane. There may be no permanent connection of the membrane compartments, but there remains a functional connection via a fusion-fission system, with specific or exclusive membrane interactions. These interactions are controlled, since the peculiar structure and composition of the different membrane systems are preserved throughout the process. This fact of chemically discrete intracellular membrane compartments is important in understanding the two main hypotheses proposed to explain the synthesis and circulation of membrane in the cell-the membrane flow hypothesis and the membrane shuttle hypothesis. In the membrane flow model (Morre et aZ., 1974), sequential and unidirectional flow of membranes and their contents from the nuclear envelope to the cell surface is coupled with irreversible membrane differentiation in both membrane composition and organization. This model indicates that secretions and membrane move together. In invertebrate cells, morphological studies suggest that the membrane flow hypothesis may be an adequate explanation of membrane biogenesis (Morr6 et aZ., 1974). However, in vertebrate cells, the system is more complex, and the membrane flow hypothesis does not account for certain membrane characteristics. First, the turnover of microsoma1 and plasma membrane components is slower than that of intravesicular materials; e.g., in the rat pancreas, the turnover of secretory protein ranges between 10 and 20 hours, whereas the half-life of specific membrane proteins varies from 3 to 28 days (Meldolesi, 1974a,b). Second, the membrane flow hypothesis does not account for the heterogeneity of turnover rates among specific membrane proteins. The rates of synthesis and assembly of membrane proteins vary and can be selectively modified by pharmacological agents (Schimke, 1974). For example, phenobarbital and 3-methylcholanthrene both cause the synthesis of selected endoplasmic reticulum components; each therefore results in a different pattern of induction of membrane protein. The induction of enzymes in the endoplasmic reticulum is associated

74

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

with changes in membrane morphology and proliferation (Conney, 1967; Schimke, 1973). Thus the transport of secretory proteins and membrane biogenesis are regulated on different time scales. I n contrast, membrane flow implies simultaneous movement, relocation, and degradation of entire membrane domains. However, membrane flow may occur to a limited extent, provided that it is slower than the flow of secretion and that it is coupled with extensive modification of the transported membrane in its new compartment. The shuttle hypothesis provides an alternative mechanism for membrane biogenesis in which the intracellular transport of membrane components occurs by a process of fusion, fission, and recycling of membranes. In this process the identity of the membrane types is preserved. It is not clear whether complete membrane patches or specific macromolecules migrate from one compartment to another, since the shuttle vehicles have not been characterized. The theory implies as yet unexplained membrane recognition phenomena during fusion and fission, to regulate the specific interactions of membranes in a precise and controlled manner. 3. NONRANDOMTOPOGRAPHY OF INTEGRAL PROTEINS In formulating a model for membrane biogenesis, allowances must be made for the nonrandom topography of integral membrane proteins and glycoproteins. I n most epithelial cells, the membrane glycoproteins vary in distribution in apical, basal, and lateral plasma membranes. For example, the apical membrane of the pancreas closely resembles zymogen granule membrane and not other regions of the plasma membrane (DeCamilli et al., 1976), and membrane receptors of the hepatocyte are confined to specific regions of the plasma membrane. This variation in distribution of membrane macromolecules at the cell surface indicates that a mechanism must exist to segregate the integral proteins and to restrict their lateral movement or free diffusion (Figs. 14 and 15).The segregation of membrane proteins may be determined by specific recognition phenomena during the fusion of vesicles and plasma membrane such that selected membrane components become integrated into specific regions of the plasma membrane. Barriers to the movement of membrane glycoproteins, may include tight junctions, since the dissociation of junction complexes in epithelial cells is accompanied by lateral diffusion of surface carbohydrates (Pisam and Repoche, 1976). Recent evidence indicates that a similar segregation of membrane

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

75

FIG. 14. Freeze-fracture replica of plasma membranes from rat liver. The fracture plane traverses the hydrophobic interior of the membrane, exposing the protoplasmic face (PF)which represents the hydrophobic aspect of the inner leaflet, and the exoplasmic face (EF) which represents the hydrophobic aspect of the outer leaflet. Numerous particles, presumably intramembrane proteins, are observed on the PF, and fewer on the EF. The distribution is random except for specialized regions of the membrane, such as gap junctions, where the particles have an ordered distribution. X49,OOO.

components may occur within the Golgi complex (Fig. 16), since there is significant variation in the distribution of membrane-bound enzymes and glycoproteins. In the liver Golgi complex, for example, cytochemical studies indicate that the distribution of enzymes such as 5’-nucleotidase is nonrandom, being restricted to certain classes of secretory vesicles (Farquhar et al., 1974). The specific activity of 5’-nucleotidase in Golgi fractions is intermediate between those of the endoplasmic reticulum and plasma membrane (Bergeron et d.,1975),

76

J. STURGESS, M. MOSCARELLO, A N D H. SCHACHTER

FIG.15. Electron micrographs showing the distribution of anionic binding sites on the plasma membrane, using cationic fenitin as a marker (arrows). Cationic ferritin binds randomly to the inner cytoplasmic surface of the plasma membrane and to associated material. The close proximity of the ferritin particles to the inner membrane surfaces is observed particularly at sites of tight junctions. x 41,000.

but localization of the enzyme varies within the organelle. Reaction product from the enzyme appears on the inside of the very low density lipoprotein-filled secretory granules and on the outside of the cisternal elements. Functional specialization within the Golgi complex is also reflected in the morphological heterogeneity of this organelle and by the variation in distribution of anionic sites on the membrane surfaces (Abe et al., 1976). Based on the binding of cationic ferritin (Fig. 17), the density of anionic sites (presumably sialic acid residues) is greatest on the tubular network. The cisternae show few anionic sites, but those that are present are asymmetrically distributed with a high density on convex surfaces and a low density on concave surfaces (Fig. 17). The restriction on density of binding sites indicates membrane differentiation within the Golgi complex. The localization of specific sites indicates that, despite the fluidity of membrane components, the integral proteins are not free to diffuse within the plane of the membrane. The distribution of sites for binding the lectin Con A to the Golgi complex also varies, with most binding sites restricted to localized areas of the tubular network (Abe et al., 1977). These observations

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

77

FIG. 16. Freeze-fracture replica of the Golgi complex showing the hydrophobic aspects of the membranes. The intramembrane particles presumably include the membrane-bound enzymes of the glycosyltransferase system. The cistemae (C) and tubular networks (T) of the Golgi complex show mainly a random distribution of particles, with some linear arrays. The number of particles varies on successive cistemae in the Golgi complex, and they are more numerous on the PF than on the E F face (see Fig. 14). The sites of formation of tubules and vesicles (V) are sites for the contribution of membranes to the cell surface. x 115,000.

provide additional evidence that the segregation of membrane glycoproteins occurs in the hepatocyte Golgi complex and that there may be functional specialization among its component structures (Sturgess et al., 1974).Membrane specialization may not necessarily be accompanied b y the segregation of different secretory macromolecules in different vesicles, since Kraehenbuhl et al. (1977) used immunocytochemical localization of different pancreatic secretory proteins to show that all Golgi vesicles appeared to contain the same complement of secretory proteins. Segregation of components at the cell surface may result from selec-

78

J. STURGESS, M. MOSCARELLO, A N D H. SCHACHTER

FIG.17. Electron micrograph showing distribution of anionic binding sites on membranes of a Golgi complex isolated from rat liver. Cationic ferritin binds mainly to small vesicles and to the tubular network and, with increasing concentration, binding is observed on convex surfaces of the cisternae (C, arrows). Binding is not observed on the concave surfaces, indicating the polarity of the membrane system within this organelle. X84,OOO.

tive incorporation of membrane components. For instance, the fusion of membranes may involve a specific recognition phenomenon. In Tetrahyrnena the fusion of mucocyst granules with the plasma membrane is preceded by reorganization of intramemhrane particles into specific patterns (Satir et al., 1973).The transfer of membrane protein following fusion may not be random. Evidence for this has been presented by freeze-fracture studies on the pancreas, where fusion of secretory granules with the plasmalemma does not result in a random mixing of membrane components (DeCamilli et al., 1976).

STRUCTURE A N D BIOSYNTHESIS OF M E M B R A N E GLYCOPROTEINS

79

4. ROLE OF THE CYTOSKELETON The control of integral membrane protein and glycoprotein distribution and mobility may depend on a cytoskeletal structure, involving linked microfilaments and microtubules (Nicolson, 1976). Evidence has been presented which indicates that microfilaments and microtubules are linked to integral membrane proteins (Poste et al., 1975), that the movement of intramembrane particles, observed by freezefracture, is influenced by the agents cytochalasin and colchicine, and that perturbation of the membrane bilayer with local anesthetics can be attributed to a reversible disruption of membrane-associated microfilaments and microtubules (Nicolson, 1976). Capping phenomena involving lateral movement of membrane proteins often occur at the cell surface over the region of the Golgi complex (De Petris, 1975); this may be related to the process of membrane turnover, with interaction of intracellular and plasma membranes through the cytoskeletal system. High-resolution studies with the scanning electron microscope provide a new approach to examining the topography of membrane surfaces. Recent work with plasma membrane and Golgi fractions isolated from rat liver (Sturgess and Moscarello, 1976) demonstrate the asymmetry of the plasma membrane (Fig. 18). The outer surface is apparently smooth, with occasional arrays of globular units. The inner surface has a reticulate appearance and is associated with numerous fine filaments and smooth-surfaced vesicles (Fig. 19). The Golgi membranes are characterized by stacks of superimposed plates which form a dome-shaped complex with cisternae and tubular networks (Fig. 20). Individual Golgi complexes are interconnected by smooth-surfaced tubules. 5.

KINETICS OF VIRAL GLYCOPROTEIN MIGRATION THROUGH THE HOSTCELL

To establish the concepts of membrane biogenesis outlined above, it would be most desirable to follow the kinetics of migration of specific and well-characterized integral membrane glycoprotein precursors through the endomembrane system of the cell. There are two general approaches to this type of experiment. Radioactive precursors can be administered, and the kinetics of incorporation and subsequent migration of the tracer can be followed either by autoradiography (Section V,A) or by subcellular fractionation and subsequent biochemical analysis. These approaches have previously been used primarily on secretory glycoproteins (Schachter, 1974a,b; Schachter and Roden,

80

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

FIG.18. Scanning electron micrograph of plasma membrane isolated from rat liver, freeze-dried, and gold-coated. The plasma membrane shows two distinct surfaces: The outer surface (E) appears relatively smooth except for occasional linear arrays of large globular particles (arrows) which may be related to specialized regions of the cell surface; and the inner surface (I) adjacent to the cytoplasm has a rougher, reticulate structure. X37,OOO.

1973). Suitable systems for the study of membrane glycoprotein biosynthesis by these methods are limited, but the incorporation of rhodopsin into retinal rod outer segment membranes or of glycoproteins into viral envelopes offers excellent opportunities for experimentation. As mentioned earlier, enveloped viruses such as VSV obtain their membrane envelopes by budding through the plasma membrane of the infected cell. VSV RNA codes for a single glycoprotein polypeptide; glycosylation of this peptide is carried out by the infected cell’s endomembrane machinery. The VSV glycoprotein (G protein) is incorporated into the infected cell’s plasma membrane and eventually becomes a component of the virus envelope. Atkinson et al. (1976) using radioactive fucose, and Knipe et al. (1977a) using radioactive methionine, followed the kinetics of VSV G-protein synthesis and showed a lag of 20 minutes between the completion of glycosylation and the appearance of glycoprotein at the cell surface; nonglycosyl-

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

81

FIG. 19. Detail of the inner surface of the plasma membrane showing the reticulate appearance. Long, filamentous structures (arrows)and smooth membranes are observed in association with this surface. x 72,000.

ated viral membrane protein (M protein) appears at the plasma membrane within 5 minutes. The extra time is presumably required for the processing of G protein in the host cell’s endomembrane system, while nonglycosylated M protein is quickly transferred from free ribosomes through the cytosol to the plasma membrane (Knipe et al., 1977b), where it rapidly becomes incorporated into the viral envelope; M protein is believed to be a peripheral membrane protein on the cytoplasmic side of the membrane. The study of enveloped viruses offers yet another advantage, namely, the availability of temperature-sensitive viral mutants defecOne such tive in the synthesis of viral proteins (Knipe et al., 1977~). mutant can make G polypeptide, can incorporate certain sugars into this peptide, but cannot add sialic acid (and possibly other sugars); the defect is in the G-protein peptide sequence, since the presence of normal virus in the infected cells cannot correct the defective synthesis. The partially glycosylated G-protein precursor accumulates in the

82

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

FIG.20. Scanning electron micrograph of Golgi complexes isolated from rat liver, freeze-dried, and gold-coated. The Golgi complex appears as a stack of platelike structures, each with a central saccule or cisterna (C) continuous with a network of fine tubules (arrows).Individual Golgi complexes appear to be interconnected through long, smooth-surfaced tubules. x 37,500.

rough endoplasmic reticulum (Knipe et al., 1977d). There are at least two interpretations of this interesting observation. The defect in the peptide may prevent proper initial glycosylation such that elongation by the addition of external sugars (Fig. 8) is not possible; this hypothesis requires that defective initial glycosylation prevent movement from the rough endoplasmic reticulum to the Golgi apparatus. A more likely theory is that the primary defect is inability to move to the Golgi apparatus for elongation. The two theories could be distinguished by testing the ability of isolated Golgi apparatus to elongate in vitro G precursor purified from cells infected with mutant virus; such experiments have not been carried out. The inability of abnormal G-protein precursor to migrate to the Golgi apparatus and plasma membrane in a normal endomembrane system seems to argue against the membrane flow hypothesis of membrane biogenesis; obviously passive transport

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

83

of membrane G protein by membrane movement does not occur in cells infected with mutant virus, and some property of the G protein must be required for it to be transported in the endomembrane system. B. Biogenesis of lntmcellular Membranes

There is considerable evidence that glycoproteins are integral components of intracellular membranes. Analyzing the content of sialic acid in ceIl membranes, Click et (12. (1971) demonstrated high levels in plasma membranes, but low levels were present in nuclei and mitochondria. In fact, a low sialic acid content is characteristic of intracelM a r membranes except for those of lysosomes, where the level approvhes that of the plasma membrane. Intracellular membranes also contain protein-bound mannose, galactose, and glucosamine but lack galactosamine (Bergman and Dallner, 1976). 1. ENDOPLASMIC RETICULUM Glycoproteins of the endoplasmic reticulum are stable membrane components and can be distinguished from secretory glycoproteins. It has been suggested that the protein framework of the endoplasmic reticulum is more rigid than that of the plasma membrane, since it retains its form after lipid extraction, and lipid components can be reintroduced to restore enzyme activity (Trump et al., 1970), suggesting that the membrane organization may vary from that at the cell surface. Current evidence suggests that the final stages of membrane glycoprotein assembly occur primarily in the Golgi complex, so that either the complete glycoprotein or an oligosaccharide chain has to be transported from the Golgi complex to the endoplasmic reticulum. The possibility that some glycosylation may occur in other subcellular membranes has not been eliminated. The transport of membrane glycoproteins between intracellular compartments may be mediated by membrane vesicles which fuse to permit the lateral diffusion of integral proteins. An alternative theory is that an unstable cytoplasmic complex is formed between lipid and membrane glycoprotein to create a hydrophobic environment for the protein component; glycoprotein molecules can thus be incorporated into the endoplasmic reticulum membrane as a lipoprotein complex (Svennson et al., 1976).According to Autuori et al. (1975a,b), cytoplasmic sialoproteins, labeled with [3Hlgluc~~amine and [14Clleucine,can

84

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

be incorporated into microsomal membrane when incubated in vitro with rough microsomes; they cannot be removed by subsequent washing with detergent. These investigators conclude that membrane glycoproteins are transferred from the Golgi complex via cytoplasmic pools into microsomes.

2. MITOCHONDRIA Mitochondria have some capacity to synthesize and glycosylate proteins. Incorporation of glucose, mannose, and galactose occurs in rat liver and brain mitochondria (Bosmann, 1971) and, when radioactively labeled sugar nucleotides are incubated with intact mitochondria, radioactivity is incorporated into acid-precipitable glycoproteins. Autonomous mitochondria1 protein synthesis assembles a total of 4 glycoproteins out of a complete set of at least 15 proteins and 8 gycoproteins. The glycosylation reactions carried out by mitochondria utilize protein moieties made on mitochondrial ribosomes. In vivo, other glycosylation reactions may be carried out by mitochondria on proteins made elsewhere in the cell. It is possible that the activation of nucleotide sugars can take place in the mitochondrial matrix. The incorporation of sugar into mitochondrial glycoproteins is associated with the inner membrane rather than the outer one. Other evidence has shown that the proteins of the outer membranes are synthesized in the cytoplasm, while mitochondrial protein synthesis is concerned with the components of the inner membrane and the cristae. Inhibitors such as chloramphenicol block the incorporation of amino acids and sugars into mitochondrial gl ycoproteins, whereas cycloheximide has no effect.

3. NUCLEI The origin of the nuclear envelope glycoproteins is unknown. It has been assumed that glycoproteins of the nuclear membrane result from polypeptide synthesis and glycosylation in the endoplasmic reticulum and Golgi complex, respectively, with a subsequent exchange between these compartments. The possibility has been raised, however, that CMP-sialic acid synthesis may occur within the nucleus (Van Dijk et al., 1973). In the membrane flow hypothesis of membrane biogenesis, the nuclear envelope releases vesicles which fuse with rough endoplasmic reticulum. If this is the case, then it would be of interest to confirm whether glycosylation mechanisms are significant in the nucleus and

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

a5

whether they contribute to the biosynthesis of integral membrane glycoprotein in the nuclear envelope and in the endoplasmic reticulum.

4. SECRETORYVESICLES The membranes of the secretory vesicles originate from segregation of membrane from specialized regions of the Golgi complex. They may have a protein composition different from that of the other smooth membranes of the Golgi complex (Palade, 1975) and similar to that of the plasma membrane (Hodson and Brenchley, 1976). Major glycoproteins of the zymogen granule membranes are similar to those of the apical cell membrane of the pancreatic acinar cell but are absent or present in trace amounts only in lysosomal and mitochondria1 membranes. REFERENCES Abe, H., Moscarello, M. A., and Sturgess, J. M. (1976). The distribution of anionic sites on the surface of the Golgi complex. J. Cell Biol. 71,973-979. Abe, H., Young, M. E. M., Moscarello, M. A., and Sturgess, J. M. (1977). Electron microscopic studies of concanavalin A binding to membranes of the Colgi complex. Cyto-

bios 17, 7-15. Adamany, A. M., and Spiro, R. G. (1975a). Glycoprotein biosynthesis: Studies on thyroid mannosyltransferases. I. Action on glycopeptides and simple glycosides. j .

Biol. Chem. 250,2830-2841. Adamany, A. M., and Spiro, R. G. (1975b). Glycoprotein biosynthesis: Studies on thyroid mannosyltransferases. 11. Characterization of a polyisoprenyl mannosyl phosphate and evaluation of its intermediary role in the glycosylation of exogenous acceptors. J . Biol. Chem. 250,2842-2854. Andrews, P. (1970). Purification of lactose synthetase A protein from human milk and demonstration of its interaction with a-lactalbumin. FEBS Lett. 9,297-300. Arima, T., and Spiro, R. G. (1972). Studies on the carbohydrate units of thyroglobulin: Structure of the carbohydrate units of thyroglobulin. Structure of the mannose Nacetylglucosamine containing unit (unit A) of the human and calf pancreasj. Biol.

Chem. 247, 1836-1848. Arima, T., Spiro, M. J., and Spiro, R. G. (1972). Studies on the carbohydrate units of thyroglobulin: Evaluation of their microheterogeneity in the human and calf proteins. ]. Biol. Chem. 247, 1825-1835. Ashwell, G., and Morel], A. C. (1974). The role of surface carbohydrates in the hepatic recognition and transport of circulating glycoproteins. Adv. Enzymol. Relat. Areus M o l . Biol. 41, 99-128. Atkinson, P. H., Moyer, S. A., and Summers, D. F. (1976).Assembly of vesicular stomatitis virus glycoprotein and matrix protein into Hela cell plasma membranes.J. Mol.

Biol. 102,613-631. Autuori, F., Svensson, H., and Dallner, G. (1975a). Biogenesis of microsomal membrane glycoproteins in rat liver. I. Presence of glycoproteins in microsomes and cytosol. J .

Cell Biol. 67,687-699. Autuori, F., Svensson, H., and Dallner, G. (1975b). Biogenesis of microsomal mem-

86

J. STURGESS, M. MOSCARELLO, A N D H. SCHACHTER

brane glycoproteins in rat liver. 11. Purification of soluble glycoproteins and their incorporation into microsomal membranes. J . Cell Biol. 67,700-714. Barr, R. M., and DeLuca, L. M. (1974). The in vivo incorporation of mannose, retinol and mevalonic acid into phospholipids of hamster liver. Biochem. Biophys. Res. Commun. 60,355-363. Bartholomew, B. A., Jourdian, G. W., and Roseman, S. (1973). The sialic acids, XV. Transfer of sialic acid to glycoproteins by a sialyltransferase from co1ostrum.J. Biol. Chem. 248,5751-5762. Bates, C. J., and Handschumacher, R. E. (1969). Inactivation and resynthesis of glucosamine-6-phosphate synthetase after treatment with glutamine analogues. Adv. Enzyme Regul. 7, 183-204. Bates, C. J., Adams, W. R., and Handschumacher, R. E. (1966). Control of the formation of UDP-N-acetylhexosamine and glycoprotein synthesis in rat liver,J. Biol. Chem. 241,1705-1712. Baynes, J. W., Hsu, A. G., and Heath, E. C. (1973). The role of mannosyl-phosphoryl-dihydropolyisoprenol in the synthesis of mammalian glycoproteins. J . Biol. Chem. 248,5693-5704. Behrens, N. H. (1974). Polyprenol sugars and glycoprotein synthesis. In “Biology and Chemistry of Eucaryotic Cell Surfaces” (E. Y. C. Lee and E. E. Smith, eds.), pp. 159-180. Academic Press, New York. Behrens, N. H., and Leloir, L. F. (1970). Dolichol monophosphate glucose: An intermediate in glucose transfer in liver. Proc. Natl. Acad. Sci. U.S.A. 66, 153-159. Behrens, N. H., Parodi, A. J., and Leloir, L. L. (1971a). Glucose transfer from dolichol monophosphate glucose: The product formed with endogenous microsomal acceptor. Proc. Natl. Acad. Sci. U.S.A. 68,2857-2860. Behrens, N. H., Parodi, A. J., Leloir, L. F., and Krisman, C. R. (1971b). 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 oligosaccharides containing mannose: Their role in glycoprotein synthesis. Proc. Natl. Acad. Sci. U.S.A. 70,3390-3394. Bekesi, J. G., and Winder, R. J. (1967).The metabolism of plasma glycoproteins: Studies on the incorporation of ~-fucose-l-’~C into tissue and serum in the normal rat.J. Biol. Chem. 242,3837-3879. Benedetti, E. L., and Emmelot, P. (1967). Studies on plasma membranes. IV. Ultrastructurd localisation and content of sialic acid in plasma membranes isolated from rat liver and hepatoma. J . Cell Sci. 2,499-512. Bennett, G., Leblond, C. P., and Haddad, A. (1974). Migration of glycoprotein from the Golgi apparatus to the surface of various cell types as shown by radioautography after labelled fucose injection into rats.]. Cell Biol. 60,258-284. Bergeron, J. J., and Posner, B. I. (1975). Polypeptide hormone receptors in the Golgi apparatus of rat hepatocytes. J . Cell Biol. 67, 59a. Bergeron, J. J., Evans, W. H., and Geschwind, I. I. (1973).Insulin binding to rat liver Golgi fracti0ns.J. Cell Biol. 59, 771-6. Bergeron, J. J., Berridge, M. V., and Evans, W. H. (1975).Biogenesis of plasmalemmal glycoproteins: Intracelluiar site of synthesis of mouse liver plasmalemmal 5’-nucleotidase as determined by the subcellular location of messenger RNA coding for 5‘-nucleotidase. Biochim. Biophys. Acta 407,325-337. Bergman, A., and Dallner, G. (1976). Distribution of protein-bound sugar residues in microsomal subfractions and Golgi membranes. Biochim. Biophys. Acta 433,496508.

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

87

Bernacki, R. J., and Bosmann, H. B. (1973). Activity of a rat liver glycoprotein sialyltransferase utilizing desialyzed prothrombin as an acceptor in normal and hypothrombinemic animals. Eur. J . Biochem. 33,49-58. Blobel, G., and Dobberstein, B. (1975a).Transfer of proteins across membranes. I. Presence of proteolytically processed and unprocessed nascent immunoglobulin light chains on membrane-bound ribosomes of murine myeloma. J . Cell Biol. 67,835851. Blobel, G., and Dobberstein, B. (1975b). Transfer of proteins across membranes. 2. Reconstitution of functional rough microsomes from heterologous comp0nents.J. Cell Biol. 67,852-862. Blobel, G., and Sabatini, D. D. (1971). Ribosome-membrane interaction in eukaryotic cells. Biomembranes 2, 193-195. Bocci, V., and Winzler, R. J. (1969). Metabolism of ~ - f u c o s e - l - * and ~ C of fucose glycoproteins in the rat. Am. J . Physiol. 216, 1337-1342. Boggs, J. M., Vail, W. J., and Moscarello, M. A. (1976). Preparation and properties of vesicles of a purified myelin hydrophobic protein and phospholipid: A spin label study. Biochim. Biophys. Acta 448,517-530. Boggs, J. M., Wood, D. D., Moscarello, M. A., and Papahadjopoulos, D. (1977). Lipid phase separation induced by a hydrophobic protein in phosphatidylserine-phosphatidylcholine vesicles. Biochemistry 16,2325-2329. Bok, D., Basinger, S. F., and Hall, M. 0. (1974).Autoradiographic and radiobiochemical studies on the incorporation of [6-3HlGlucosamine into frog rhodopsin. Exp. Eye Res. 18, 225-240. Bosmann, H. B. (1970). Glycoprotein biosynthesis: Purification and properties of glycoprotein N-acetylglucosaminyltransferasesfrom guinea pig liver utilizing endogenous and exogenous acceptors. Eur. J . Biochem. 1 4 , 3 3 4 0 . Bosmann, H. B. (1971). Identification of the products of autonomous mitochondrial protein and glycoprotein synthesis. Nature (London),New Biol. 234,54-56. Brady, G , Papahadjopoulos, D., Moscarello M. A. (1978). In press. Bretscher, M. S. (1973). Membrane structure: Some general principles. Membranes are asymmetric lipid bilayers in which cytoplasmically synthesized proteins are dissolved. Science 181,622-629. Bretscher, M. S., and Raff, M. C. (1975). Mammalian plasma membranes. Nature (London) 2 5 8 , 4 3 4 9 . Brew, K., Vanaman, T. C., and Hill, R. L. (1968). T h e role of a-lactaalbumin and the A protein in lactose synthetase: A unique mechanism for the control of a biological reaction. Proc. Natl. Acad. Sci. U S A . 59, 491-497. Brostoff, S., and Eylar, E. H. (1971). Localization of methylated arginine in the A 1 Protein from myelin. Proc. Natl. Acad. Sci. U.S.A. 68,765. Burgos, J., Hemming, F. W., Pennock, J. F., and Morton, R. A. (1963). Dolichol: A naturally-occurring C,,, isoprenoid alcohol. Biochem. J . 88,470-482. Buscher, H. P., Casals-Stenzel, J., and Schauer, R. (1974).New sialic acids: Identificaacids and N-acety1-O-glycoloylneuraminic tion of N-glycoloyl-0-acetylneuraminic acids by improved methods for detection of N-acyl and 0-acyl groups and by gas-liquid chromatography. Eur. J . Biochem. 50, 71-82. Butterworth, P. H. W., and Hemming, F. W. (1968).Intracellular distribution of the free and esterified forms of dolichol in pig liver. Arch. Biochem. Bioph!/s. 128,503-508. Cabezas, J. A. (1973). The type of naturally occurring sialic acids. Reu. Esp. Fisiol. 29, 307-322. Carafoli, E., and Sottocasa, G. (1974).The Ca++transport system of the mitochondrial

88

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

membrane and the problem of the Ca++carrier. lnt. Congr. Biochem.,Proc. Honour Britton Chance, 9th. Carlson, D. M. (1968).Structures and immunochemical properties of oligosaccharides isolated from submaxillary mucins. J . Biol. Chem. 243,616-626. Carlson, D. M., David, J., and Rutter, W. J. (1973a). Galactosyltransferase activities in pancreas, liver and gut of developing rat. Arch. Biochem. Biophys. 157,605-612. Carlson, D. M., Jourdian, G. W., and Roseman, S. (197313).The sialic acids. XIV. Synthesis of sialyl-lactose by a sialyltransferase from rat mammary g1and.J. Biol. Chem. 248,5742-5750. Carlson, D. M., McGuire, E. J., Jourdian, G. W., and Roseman, S. (1973~).The sialic acids. XVI. Isolation of a mucin sialyltransferase from sheep submaxillary gland.]. Biol. Chem. 248,5763-5773. Carnegie, P. R. (1971).Amino acid sequence of the encephalitogenic basic protein from human myelin. Biochem. J . 123,5747. Chang, P., Moscarello, M. A., and Sturgess, J. M. (1977). Biosynthesis of phosphatidylerthanolamine by the Golgi complex of rat liver in uitro. Experienta 33, 11361137. Chao, L. P., and Einstein, E. R. (1970). Physical properties of the bovine encephalitogenic protein: Molecular weight and conformation. J . Neurochem. 17, 1121-1 132. Chapman, B. E., and Moore, W. J. (1976).Conformation of myelin basic protein in aqueous solution from nuclear magnetic resonance spectroscopy. Biochem. Biophys. Res. Commun. 73,758-766. Chen, W. W., and Lennarz, W. J. (1976).Participation of a trisaccharide lipid in glycosylation of oviduct membrane glycoprotein. J . Biol. Chem. 251,7802-7809. 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. Chester, M. A. (1974). The occurrence of a P-galactosyltransferase in normal human urine. FEBS Lett. 46, 59-62. Claude, A. (1970).Growth and differentiation of cytoplasmic membranes in the course of lipoprotein granule synthesis in the hepatic cell. I. Elaboration of elements of the Golgi complex. J . Cell Biol. 47,745-766. Cleve, H., Hamaguchi, H., and Hutteroth, T. (1972).Preparation and further characterization of the MN glycoprotein of human erythrocyte membranes./. E r p . Med. 136, 1140- 1156. Cockle, S . , Epand, R. M., Boggs, J. M., and Moscarello, M. A. (1978). Circular dichroism studies on lipid-protein complexes of a hydrophobic myelin protein. Biochemistry 17,624-629. Codington, J. F., Linsley, K. B., Jeanloz, R. W., Irimura, T., and Osawa, T. (1975a). Immunochemical and chemical investigations of the structure of glycoprotein fragments obtained from epiglycanin, a glycoprotein at the surface of the TA3-Ha cancer cell. Carbohyilr. Res. 40, 171. Codington, J. F., Cooper, A. G., Brown, M. C., and Jeanloz, R. W. (1975b).Evidence that the major cell surface glycoprotein of the TA3-Ha carcinoma contains the Vicia graminea receptor sites. Biochemistry 14,855-859. Coffey, J. W., Miller, 0. N., and Sellinger, 0. Z. (1964). The metabolism of L-fucose in the rat. J . Biol. Chem. 239,4011-4017. Conney, A. (1967). Pharmacological implications of microsomal enzyme induction. Pharmacol. Reu. 19,317-366. Cowan, N. J.,and Robinson, G. B. (1970). The ribosomal incorporation of hexosamine into glycoprotein in a mouse myeloma. FEBS Lett. 8,6-8.

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

89

Creswell, P., Turner, M. J., and Strominger, J. L. (1973). Papain solubilized HL-A antigens from cultured human lymphocytes contain two peptide fragments. Proc. Natl. Acad. Sci. U.S.A. 70, 1693-1707. Danon, D. L.,Addstein, V., Marikovsky, Y., and Skutelsky, E. (1972).Use of cationised ferritin as a level of negative charges on cell surfaces.J. Ultrustruct. Res. 38,500-

510. DeCamilli, P., Peluchetti, D., and Meldolesi, J. (1976).Dynamic changes of the hminal plasmalemma in stimulated parotid acinar cells: A freeze fracture study. J . Cell Biol. 70,59-74. DeLuca, L., Maestri, N., Rosso, G., and Wolf, G. (1973).Retinol glycolipids. J. Biol. Chem. 248,641-648. Den, H., Kaufman, B., and Roseman, S. (1970).Properties of some glycosyltransferases in embryonic chicken brain. J. Biol. Chem. 245,6607-6615. D e Petris, S. (1975). Concanavalin A receptors, immunoglobulins, and theta antigen of the lymphocyte surface: Interactions with concanavalin A and with cytoplasmic structures.]. Cell Biol. 65,123-146. Depierre, J. W., and Dallner, G. (1975).Structural aspects of the membrane of the endoplasmic reticulum. Biochim. Biophys. Acta 415,411-472. Dunkley, P. R., and Carnegie, P. R. (1974).Amino acid sequence of the smaller basic protein from rat brain myelin. Biochem. J. 141,243-261. Elhammer, A,, Svensson, H., Autuori, F., and Dallner, G. (1975).Biogenesis of microsoma1 membrane glycoproteins in rat liver. 111. Release of glycoproteins from the Golgi fraction and their transfer to microsomal membranes. J. Cell Biol. 67,715-

724. Ellis, D.B., and Sommar, K. M. (1971).Regulation of ~-glutamine:~-fructose-6-phosphate aminotransferase activity in bovine trachea. Biochim. Biophys. Acta 230,531

-534. Epand, R. M., Moscarello, M. A., Zirenberg, B., and Vail, W. J. (1974).T h e folded conformation of the encephalitogenic protein of the human brain. Biochemistry 13, 1264-1267. Etchison, J. R., and Summers, D. F. (1977).Oligosaccharide structure of the membrane glycoprotein of vesicular stomatitis virus. J. Supramol. Struct., Suppl. 1, 6.New York, N.Y. 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. Evans, W. H.,Hood, D. O., and Gurd, J. W. (1973).Purification and properties of a mouse liver plasma membrane glycoprotein hydro1ysing nucleotide pyrophosphate and phosphodiester bonds. Biochem. J . 135,819-826. Eylar, E. H. (1970).Amino acid sequence of the basic protein of the myelin membrane. Proc. Natl. Acad. Sci. U.S.A. 67, 1425. Eylar, E . H. (1972).The structure and immunologic properties of basic proteins of myelin. Ann. N.Y. Acad. Sci. 195,481-491. Eylar, E. H., and Thompson, M. (1969).Allergic encephalomyelitis: The physico-chemical properties of the basic protein encephalitogen from bovine spinal cord. Arch. Biochem. Biophys. 129,468-479. Eylar, E. H., Brostoff, S., Hashim., Caccam, J., and Burnet, P. (1971).T h e basic A, protein of the myelin membrane: The complete amino acid sequence.J. Biol. Chem.

246,5770-5784. Farquhar, M. G., Bergeron, J. J., and Palade, G. E. (1974).Cytochemistry of Golgi fractions prepared from rat liver. J. Cell Biol. 60,8-25.

90

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

Feingold, D. S. (1972). The role of pyridine nucleotides in interconversions of nucleoside diphosphate sugars. In “Biochemistry of the Glycosidic Linkage” (H. Pontis and R. Piras, eds.), pp. 79-112. Academic Press, New York. Fitzgerald, D. K., Brodbeck, U., Kiyosawa, I., Mawal, R.,Colvin, B., and Ebner, K. E. (1970).a-Lactalbumin and the lactose synthetase reaction.]. Biol. Chem. 245,2103 -2108. Fitzgerald, D. K., McKenzie, L., and Ebner, K. E. (1971).Galactosyl transferase activity in a variety of sources. Biochim. Biophys. Acta 235,425-428. Folch-Pi, J., and Stoffyn, P. J. (1972). Proteolipids from membrane systems. Ann. N.Y. Acad. Sci. 195,86-107. Folch, P. J., Lees, M., and Sloane-Stanley, G. H. (1957). A simple method forthe isolation and purification of total lipids from animal tissues. J . Biol. Chem. 226,497-509. Gagnon, J., Finch, P. R., Wood, D. D., and Moscarello, M. A. (1971).Isolation of a highly purified myelin protein. Biochemistry 10,4756-4763. Gasic, G., and Berwick, L. (1962). Hale stain for acid containing mucins: Adaptation to electron microscopy. J. Cell Biol. 19,223-228. Geren, C. R., Magee, S. C., and Ebner, K. E. (1975). Circular dichroism changes in galactosyltransferase upon substrate binding. Biochemist y 14, 1461-1463. Ghalambor, M. A., Warren, C. D., and Jeanloz, R. W. (1974). Biosynthesis of aPi-2-acetamido-2-deoxy-~glucosylP2-polyisoprenyl pyrophosphate by calf pancreas microsomes. Biochem. Biophys. Res. Commun. 56,407-414. Glick, M. C., Comstock, C. A., Cohen, M. A,, and Warren, L. (1971). Membranes of animal cells. VIII. Distribution of sialic acid hexosamines and sialidase in the L cell. Biochim. Biophys. Acta 233,247-257. Gough, D. P., and Hemming, F. W. (1970).The characterization and stereochemistry of biosynthesis of dolichols in rat liver. Biochem. J. 118, 163-166. Hackenbrock, C. R., and Miller, K. J. (1975). The distribution of anionic sites on the surface of mitochondria1 membranes. J. Cell Biol. 65,615-630. Hagopian, A., and Eylar, E. H. (1968a). Glycoprotein biosynthesis: The basic protein encephalitogen from bovine spinal cord, a receptor for the polypeptidyl: N-acetylgalactosaminyl transferase from bovine submaxillary glands. Arch. Biochem. Biophys. 126,785-794. Hagopian, A., and Eylar, E. H. (1968b). Glycoprotein biosynthesis: Studies on the receptor specificity of the po1ypeptidyl.N-acetylgalactosaminyltransferase from bovine submaxillary glands. Arch. Biochem. Biophys. 128,422-433. Hagopian, A., and Eylar, E. H. (1969a). Glycoprotein biosynthesis: The solubilization of glycosyl transferases from membranes of HeLa cells and bovine submaxillary glycoproteins. Arch. Biochem. Biophys. 129,447-455. Hagopian, A,, and Eylar, E. H. (1969b). Glycoprotein biosynthesis: The purification and characterization of a polypeptide W-acetylgalactosaminyl transferase from bovine submaxillary glands. Arch. Biochem. Biophys. 129,515-524. Hagopian, A., Westall, F. C., Whitehead, J. S., and Eylar, E. H. (1971). Glycosylation of the A1 protein from myelin by a polypeptide N-acetylgalactosaminyl transferase: Identification of the receptor sequence. J . Biol. Chem. 246,2519-2523. Hargrave, P. A. (1977). Molecular properties of rhodopsin. J . Supramol. Struct., Suppl. 1, 3. Heath, E. C., Baynes, J. W., and Hsu, A. F. (1974). The role of mannosyl phosphoryl dihydropolyisoprenol in the synthesis of mammalian glycoproteins. In “Biology and Chemistry of Eucaryotic Cell Surfaces” (E. Y. C. Lee and E. E. Smith, eds.), pp. 181-211. Academic Press, New York.

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

91

Heller, J. (1968). Structure of visual pigments. I. Purification, molecular weights and composition of bovine visual pigments. Biochemistry 7,2906-2913. Heller, J., and Lawrence, M. A. (1970). Structure of the glycopeptide from bovine visual pigment 500. Biochemistry 9,864-869. Helting, T., and Erbing, B. (1973). Galactosyl transfer in mouse mastocytoma: Purification and properties ofN-acetyllactosamine synthetase. Biochim. Biophys. Acta 293, 94-104. Hemming, F. W. (1973). The formation of sugar derivatives of dolichol phosphate in mammalian systems. Biochem. Soc. Trans. 1, 1029-1033. Henderson, R., and Unwin, P. N. T. (1975). Three-dimensional model of purple membrane obtained by electron microscopy. Nature (London)257,28-32. Herscovics, A., Warren, C. D., Jeanloz, R. W., Wedgwood, J. F., Liu, I. Y., and Strominger, J. L. (1974). Occurrence of a p-D-mannopyranosyl phosphate residue in the polyprenyl mannosyl phosphate formed in calf pancreas microsomes and in human lymphocytes. FEBS Lett. 45,312-317. Herscovics, A., Golovtchenko, A. M., Warren, C. D., Bugge, B., and Jeanloz, R. W. (19774. Mannosyltransferase activity in calf pancreas microsomes. Formation of I4C-labeled lipid-linked oiigosaccharides from GDP-D-[14C]-mannose and panphosphate. J . B i d . Chem. 252, 224creatic dolichyl ~-~-['~C]-mannopyranosyl 234. Herscovics, A., Bugge, B., and Jeanloz, R. W. (19771~).Clucosyltransferase activity in calf pancreas microsomes: Formation of dolichyl D-[14c]-glucosylphosphate and "C-labeled lipid-linked oligosaccharides from U DP-~-[~~C]-glucose.J. Biol. Chem, 252,2271-2277. Hill, H. D., Jr. (1976). Structural characterization of ovine submaxillary mucin. Fed. Proc., Fed. Am. SOC.E x p . Biol. 35, 1443. Hill, H. D., Jr., Reynolds, J. A., and Hill, R. L. (1977a). Purification, composition, molecular weight, and subunit structure of ovine submaxillary mucin.J. Biol. Chem. 252, 3791-3798. Hill, H. D., Jr., Schwyzer, M., Steinman, H. M., and Hill, R. L. (197713).Ovine submaxillary mucin: Primary structure and peptide substrates of UDP-N-acetylgalactosamine:mucin transferase. J . Biol. Chem. 252,3799-3804. Hirano, H., Parkhouse, B., Nicolson, G. L., Lennox, E.S., and Singer, S, J. (1972). Distribution of saccharide residues on membrane fragments from a myeloma-cell homogenate: Its implications for membrane biogenesis. Proc. Natl. Acad. Sci. U S A . 69,2945-2949. Hodson, S., and Brenchley, C . (1976). Similarities of the Golgi apparatus membrane and the plasma membrane in rat liver cel1s.J. Cell Sci. 20, 167-182. Hong, K., and Hubbell, W. L. (1972). Preparation and properties of phospholipid bilayers containing rhodopsin. Proc. Natl. Acad. Sci. U.S.A. 69,2617-2621. Hsu, A. F., Baynes, J. W., and Heath, E. C. (1974).The role ofa dolichol-oligosaccharide as an intermediate in glycoprotein biosynthesis. Proc. Natl. Acad. Sci. U.S.A. 71, 2391-2395. Hudgin, R. L., and Schachter, H. (1971a). Porcine sugar nuc1eotide:glycoprotein glycosyltransferases. I. Blood serum and liver sialyltransferases. Can. J . Biochem. 49, 829-837. Hudgin, R. L., and Schachter, H. (1971b). Porcine sugar nuc1eotide:glycoprotein glycosyltransferases. 11. Blood serum and liver galactosyltransferase. Can. J . Biochem. 49,838-846. Hudgin, R. L., and Schachter, H. (1971~). Porcine sugar nuc1eotide:glycoprotein glyco-

92

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

syltransferases. 111. Blood serum and liver N-acetylglucosaminyltransferase. Can.J . Biochem. 49,847-852. Hudgin, R. L., and Schachter, H. (1972). Evidence for two CMP-N-acetylneuraminic acid:lactose sialyltransferases in rat, porcine, bovine and human liver. Can.J . Biochem. 50,1024-1028. Hughes, R. C. (1976).“Membrane Glycoproteins, A Review of Structure and Function.” Butterworth, London. Hunt, L. T., and Dayhoff, M. 0. (1970). The occurrence in proteins of the tripeptides Asn-X-Ser and Asn-X-Thr and of bound carbohydrate. Biochem. Biophys. Res. Commun. 39,757-765. Hunt, R. C., and Brown, J. C. (1975). Identification of a high molecular weight transmembrane protein in mouse L cells. J . Mol. Biol. 97,413-422. Hunt, R. C., Gold, E., and Brown, J. C. (1975).Cell cycle dependent exposure of a high molecular weight protein on the surface of mouse L cells. Biochim. Biophys. Actu 413,453-458. Hynes, R. 0. (1976). Cell surface proteins and malignant transformation. Biochim. Biophys. Acta 458, 73-107. Ito, A., and Sato, R. (1969). Proteolytic microdissection of smooth-surfaced vesicles of liver microsomes.J. Cell Biol. 40, 179-189. Ito, S. (1969). Structure and function of the glycocalyx. Fed. Proc., Fed. Am. SOC. Exp. B i d . 28, 12-25. Ito, S., Muramatsu, T., and Kobata, A. (1975). Endo-P-N-acetylglucosaminidasesacting on carbohydrate moieties of glycoproteins: Purification and properties of the two enzymes with different specificities from Clostridium pegringens. Arch. Biochem. Biophys. 171,78-86. Jabbal, I., and Schachter, H. (1971). Pork liver guanosine diphosphate-L-fucose glycoprotein fucosyltransferases. J . Biol. Chem. 246,5154-5161. Jamieson, J. C. (1977). Studies on the site of addition of sialic acid glucosamine to rat a,-acid glycoprotein. Can.J. Biochem. 55,408-414. Jamieson, J. D., and Palade, G. E. (1968). Intracellular transport of secretory proteins in the pancreatic exocrine cell. IV. Metabolic requirements. J . Cell Biol. 39, 589603. Jamieson, J. D., and Palade, G. E. (1971). Synthesis, intracellular transport and discharge of secretory proteins in stimulated pancreatic exocrine ce1ls.J. Cell Biol. 50, 135-158. Jancik, J., and Schauer, R. (1974). Sialic acid-A determinant of the life-time of rabbit erythrocytes. Hoppe-Seyler’s 2. Physiol. Chem. 355,395-400. Jenkins, R. E., and Tanner, M. J. A. (1977). The structure of the major protein of the human erythrocyte membrane. Biochem. J . 161, 139-147. Johnston, I. R., McGuire, E. J., Jourdian, G. W., and Roseman, S. (1966). Incorporation of N-acetyl-D-glucosamine into glycoproteins. J . Biol. Chem. 241, 5735-5737. Johnston, I. R., McGuire, E. J., and Roseman, S. (1973). Sialic acids. XVII. A uridine diphosphate N-acetylg1ucosamine:glycoprotein N-acetylglucosaminyltransferase from goat co1ostrum.J. Biol. Chem. 248,7281-7288. Kamerling, J. P., Vliegenthart. J. F. G.,Versluis, C., and Schauer, R. (1975).Identification of 0-acetylated N-acylneuraminic acids by mass spectrometry. Carbohydr. Res. 41, 7-17. Kawasaki, T., and Ashwell, G. (1976a). Chemical and physical properties of an hepatic membrane protein that specifically binds asialoglycoproteins. J . Biol. Chem. 251, 1296- 1302. Kawasaki, T., and Ashwell, G. (1976b). Carbohydrate stxycture of glycopeptides iso-

STRUCTURE A N D BIOSYNTHESIS OF M E M B R A N E GLYCOPROTEINS

93

lated from an hepatic membrane-binding protein specific for asialoglycoproteins.]. Biol. Chem. 251,5292-5299. Kean, E. L. (1970). Nuclear cytidine 5'-monophosphosialic acid synthetase. J. Biol. Chem. 245,2301-2308. Kean, E. L., and Bruner, W. E. (1971). Cytidine 5'-monophosphosialic acid synthetase: Activity and localization in the nuclear fragments of the unfertilized sea urchin egg. Exp. Cell Res. 69,384-392. Khalkhali, Z., and Marshall, R. D. (1975). Glycosylation of rihonuclease A catalysed by rabbit liver extracts. Biochem. 1. 146,299-307. Khatra, B. S., Herries, D. G., and Brew, K. (1974). Some kinetic properties of human milk galactosyl transferase. Eur. J . Biochem. 44, 537-560. Kiely, M. L., McKnight, G. S., and Schimke, R. T. (1976). Studies on the attachment of carbohydrate to ovalbumin nascent chains in hen 0viduct.J. Biol. Chem. 251,54905495. Kies, M. (1965). Chemical studies on an encephalitogenic protein from guinea pig brain. Ann. N.Y. Acad. Sci. 122, 161-170. Kies, M., Thompson, G. E., and Alvord, E. C. (1965).The relationship of myelin proteins to experimental allergic encephalomyelitis. Ann. N.Y. Acad. Sci. 122, 148160. Kikuchi, K., and Tsuiki, S. (1973). Purification and properties of UDP-N-acetylglucosamine 2'-epimerase from rat liver. Biochim. Biophys. Acta 327, 193-206. Kim, Y. S., Perdomo, J., Bella, A., Jr., and Nordberg, J. (1971). Properties of a CMP-Nacetylneuraminic acid:glycoprotein sialyltransferase in human serum and erythrocyte membranes. Biochim. Biophys. Actu 244,505-512. Kim, Y. S., Perdomo, J., and Whitehead, J. S. (1972a). Glycosyltransferases in human blood. I. Galactosyltransferase in human serum and erythrocyte membranes. J . Clin. Invest. 51,2024-2032. Kim, Y. S., Perdomo, J., Whitehead, J. S., and Curtis, K. J. (1972b). Glycosyltransferases in human blood. 11. Study of serum galactosyltransferase andN-acetylgalactosaminyltransferase in patients with liver diseases. J . Clin. Inoest. 51,2033-2039. Kitchen, B. J., and Andrews, P. (1972). The effect of temperature on a reaction catalyzed by lactose synthetase A protein. FEBS Lett. 26,333-335. Kitchen, B. J.,and Andrews, P. (1974). Studies on the thiol group of lactose synthetase A protein from human milk and on the binding of uridine diphosphate galactose to the enzyme. Biochem. J . 141,173-178. Knipe, D. M., Lodish, H. F., and Baltimore, D. (1977a). Localization of two cellular forms of the vesicular stomatitis viral glycoprotein. J. Virol. 21, 1121-1127. Knipe, D. M., Baltimore, D., and Lodish, H. F. (1977h). Separate pathways of maturation of the major structural proteins of vesicular stomatitis virus.1. Virol. 21, 11281139. Analysis of the defects of temperaKnipe, D., Lodish, H. F., and Baltimore, D. (1977~). ture-sensitive mutants of vesicular stomatitis virus: Intracellular degradation of specific viral proteins. J . Virol. 21, 1140-1148. Knipe, D. M., Baltimore, D., and Lodish, H. F. (1977d). Maturation of viral proteins in cells infected with temperature-sensitive mutants of vesicular stomatitis virus. J. Virol. 21, 1149-1158. Kohno, M., and Yamashina, I. (1973). Identification of asparagine as the site of glycosylation in glycoprotein biosynthesis. J. Biochem. (Tokyo)73, 1089-1094, Koide, N., and Muramatsu, T. (1974). Endo-p-N-acetylglucosaminidaseacting on carhohydrate moieties of glycoproteins: Purification and properties of the enzyme from Diplococcus pneumoniae. 1.B i d . Chem. 249,4897-4904.

94

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

Kornfeld, R. (1967).Studies on L-glutamine D-fructose 6-phosphate amidotransferase. I. Feedback inhibition by UDP-N-acetylglucosamine. J . Biol. Chem. 242,3135. Kornfeld, R., and Kornfeld, S. (1970).The structure of a phytohemagglutinin receptor site from human erythrocytes. J . Biol. Chem. 245,2536-2545. Kornfeld, S., and Kornfeld, R. (1969). Solubilization and partial characterization of a phytohemagglutinin receptor site from human erythrocytes. Proc. Natl. Acad. Sci. U.S.A. 63, 1439-1446. Kornfeld, S., Kornfeld, R., Ndufeld, E. F., and O’Brien, P. J. (1964).The feedback control of sugar nucleotide biosynthesis in liver. Proc. Natl. Acad. Sci. U.S.A.52,371379. Kraehenbuhl, J. P., Racine, L., and Jamieson, J. D. (1977).Immunocytochemical localization of secretory proteins in bovine pancreatic exocrine cells. J . Cell Biol. 72, 406-423. Kraemer, P. M. (1971).Heparan sulfates of cultured cells. I. Membrane-associated and cell sap species in Chinese hamster cells. Biochemistry 10, 1437-1445. Lawford, G. R., and Schachter, H. (1966).Biosynthesis of glycoprotein by liver. The incorporation in oiuo of ‘4C-glucosamine into protein-bound hexosamine and sialic acid of rat liver suhcellular fractions. J . Biol. Chem. 241,5408-5418. Leblond, C. P. (1950).Distribution of periodic acid-reactive carbohydrates in the adult rat. Am. J . Anat. 86, 1-25. Leblond, C. P., and Bennett, G. (1974).Elaboration and turnover of cell goat glycoproteins. In “The Cell Surface in Development” (A. A. Moscona, ed.), p. 29. Wiley, New York. Lehman, E. D., Hudson, B. G., and Ebner, K. E. (1975). Studies on the carbohydrate structure of bovine milk galactosyltransferase. FEBS Lett. 54,65-69. Leloir, L. F., Staneloni, R. J., Carminatti, H., and Behrens, N. H. (1973).The biosynthesis of a N,”-diacetylchitobiose containing lipid by liver microsomes: A probable dolichol pyrophosphate derivative. Biochem. Biophys. Res. Commun. 52, 12851292. Lennarz, W. J. (1975).Lipid linked sugars in glycoprotein synthesis. Science 188,986991. Lennarz, W. J., and Scher, M. G. (1972). Metabolism and function of polyisoprenol sugar intermediates in membrane associated reactions. Biochim. Biophys. Acta 265,417-441. Levy, J. A., Carminatti, H., Cantarella, A. I., Behrens, N. H., Leloir, L. F., and Tabora, E. (1974). Mannose transfer to lipid linked di-N-acetylchitobiose. Biochem. Biophys. Res. Commun. 60, 118-125. Lewis, M. S., Krieg, L. C., and Kirk,W. D. (1974).The molecular weight and detergent binding of bovine rhodopsin. E x p . Eye Res. 18,29-40. Liebes, L. F., Zand, R., and Phillips, W. D. (1975). Solution behaviour, circular dichroism and 220 MHz PMR studies of the bovine myelin basic protein. Biochim. Biophys. Acta 405,27-39. Lloyd, K. O., Kabat, E. A., and Licerio, E. (1968).Immunochemical studies on blood groups. XXXVIII. Structures and activities of oligosaccharides produced by alkaline degradation of blood-group Lewis substance. Proposed structure of the carbohydrate chains of human blood-group A, B, H, Lea, and Leb substances. Biochemistry 7,2976-2990. Locke, M., and Huie, P. (1976).The beads in the Golgi complex-endoplasmic reticulum region. J. Cell Biol. 70,384-394. Lodish, H. F. (1973). Biosynthesis of reticulocyte membrane proteins by membranefree polyribosomes. Proc. Natl. Acad. S c i . U.S.A. 70, 1526-1530.

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

95

Lucas, J. J., and Waechter, C. J. (1974). Structure of an oligosaccharide-lipid intermediate. Fed. Proc., Fed. Am. SOC. E x p . Biol. 33, 1368. Lucas, J. J., Waechter, C. J., and Lennarz, W. J. (1975). The participation of lipid-linked oligosaccharide in synthesis of membrane g1ycoproteins.J. Biol. Chem. 250,19922002. Luft, J. H. (1971). Ruthenium red and violet. 11. Fine structural localisation in animal tissues. Anat. Rec. 171,369-416. MacBeth, R. A. L., Bekesi, J. C., Sugden, E., and Bice, S. (1965).The metabolism of plasma glycoproteins. I. Studies on the rate of incorporation of g l u c ~ s a m i n e - l - ~ ~ C into protein-bound hexosamine and N-acetylneuraminic acid in the normal rat. J . Biol. Chem. 240,3707-3713. MacLennan, D. H. (1975). Resolution of the calcium transport system of sarcoplasmic reticulum. Can. J . Biochem. 53,251-261. McCarrahan, J. F., and Maley, F. (1962). Hexosamine metabolism. I. The metabolism in uioo and oitro of D-glucosamine-l-Cl4 and N-acetyl-D-ghcosamine-l-Cl4in rat liver. J. Biol. Chem. 237, 2458-2465. McGuire, E. J. (1970). Biosynthesis of submaxillary mucins. In “Blood and Tissue Antigens” (D. Aminoff, ed.), pp. 461-478. Academic Press, New York. McGuire, E. J., and Roseman, S. (1967). Enzymatic synthesis of the protein-hexosamine linkage in sheep submaxillary mucin. J. Biol. Chem. 242,3745-3747. McGuire, E. J.. Jourdian, G. W., Carlson, D. M. and Roseman, S. (1965)J. Biol. Chem. 240, PC 4113. Magee, S. C., and Ebner, K. E. (1974). Inactivation of soluble bovine milk galactosyltransferase (lactose synthetase) by sulhydryl reagents and trypsin. J . Biol. Chem. 249,6992-6998. Maley, F., Tarentino, A. L.,McGarrahan, J. F., and DelCiacco, R. (1968). The metabolism of D-galactosamine and N-acetyl-D-galactosaminein rat liver. Biochem. J . 107, 637 -644. Marchesi, V. T., Furthmayr, H., and Tomita, M. (1976). The red cell membrane. Annu. Reo. Biochem. 45,667-698. Marchesi, V. T., Tillack, T. W., Jackson, R. L., Segrest, J. P., and Scott, R. E. (1972). Chemical characterization and surface orientation of the major glycoprotein of the human erythrocyte membrane. Proc. Natl. Acad. Sci. U S A . 69, 1445-1449. Marshall, R. D. (1972). Glycoproteins. Annu. Rev. Biochem. 41,673-702. Marshall, R. D. (1974). The nature and metabolism of the carbohydrate-peptide linkages of glycoproteins. Biochem. SOC. Symp. 40, 17-26. Mazlen, R. G., Muellenberg, C. G., and OBrien, P. J . (1969). The control of L-glutamine:D-fmctose-6-phosphateamidotransferase in bovine retina. Biochim. Biophys. Acts 171,352-354. Meldolesi, J. (1974a). Membranes and membrane surfaces: Dynamics of cytoplasmic membranes in pancreatic acinar cells. Phil. Trans. R. SOC. London, Ser. B 268,3953. Meldolesi, J. (1974h). Secretory mechanisms in pancreatic acinar cells: Role of the cytoplasmic membranes. Ado. Cytophamacol. 2, 71-85. Merrick, J. M., Schifferle, R., Zadarlik, K., and Milgrom, F. (1977). Isolation and partial characterization of the infectious mononucleosis heterophile antigen from bovine erythrocytes. Fed. Proc., Fed. Am. SOC. E x p . Biol. 36,702. Milstein, C., Brownlee, G . G., Harrison, T. M., and Mathews, M. B. (1972). A possible precursor of immunoglobulin light chains. Nature (London), New Biol. 239, 117120. H proteinMitranic, M., and Moscarello, M. A. (1972). The conversion of r n a n n o ~ e - ~to

96

J. STURGESS,

M. MOSCARELLO, AND H. SCHACHTER

bound sialic acid, fucose, galactose and hexosamines in rat serum. Biochem. Biophys. Res. Commun. 47,74-80. Mitranic, M., Sturgess, J. M., and Moscarello, M. A. (1976).The effect of temperature on the galactosyl- and sialyltransferases and on the ultrastructure of Golgi membranes. Biochim. Biophys. Acta 443, 190-197. Mitranic, M., Boggs, J. M., Sturgess, J. M., and Moscarello, M. A. (1978). The effect of colchicine on galactosyl- and sialyltransferases of rat liver Golgi and N-acetylgalactosaminyltransferase from porcine submaxillary gland. Submitted for publication. Miyagi, T., and Tsuiki, S. (1971). Effect of phosphoglucose isomerase and glucose-6phosphate on UDP-N-acetylglucosamine inhibition of L-glutamine:D-fructose-6phosphate aminotransferase. Biochim. Biophys. Acta 250,51-62. Molnar, J. (1974). A proposed pathway of plasma glycoprotein synthesis in the liver. Colloq. Int. CNRS No. 221, 921-935. Molnar, J., and Sy, D. (1967).Attachment of glucosamine to protein at the ribosomal site of rat liver. Biochemistril 6, 1941-1947. Molnar, J., Chao, H., and Ikehara, Y. (1971). Phosphoryl-N-acetylglucosaminetransfer to a lipid acceptor of liver microsomal preparations. Biochim. Biophys. Acta 239, 401-4 10. Montreuil, J. (1975). Recent data on the structure of the carbohydrate moiety of glycoproteins: Metabolic and biological implications. Pure Appl. Chem. 42,431-477. Mookejea, S . , Chow, A., and Hudgin, R. L. (1971). Occurrence of UDP-N-acetylglucosamine:glycoprotein N-acetylglucosaminyltransferase activity in human and rat sera. Can. J . Biochem. 49,297-299. Mookerjea, S., Michaels, M. A., Hudgin, R. L., Moscarello, M. A., Chow, A., and Schachter, H. (1972).The levels of nucleotide-sugar:glycoproteinsialyl- and N-acetylglucosaminyltransferases in normal and pathological human sera. Can. J . Biochem. 50,738-740. Morawiecki, A. (1964). Dissociation of M- and N-group mucoprotein into subunits in detergent solution. Biochim. Biophys. Acta 83,339-347. MorrB, D. J., Mollenhauer, H. H., and Bracker, C. E. (1970). Origin and continuity of Golgi apparatus. In “Results and Problems in Cell Differentiation. Vol. 2: Origin and Continuity of Cell Organelles” (J. Reinert and H. Ursprung, eds.), pp. 82-126. Springer-Verlag, Berlin and New York. Morre, D. J., Keenan, T. W., and Huang, C. M. (1974).Membrane flow and differentiation: Origin of Golgi apparatus membranes from endoplasmic reticulum. Adu. Cytophamacol. 2, 107-125. Morrison, J. F., and Ebner, K. E. (1971a). Studies on galactosyltransferase: Kinetic investigations with N-acetylglucosamine as the galactosyl group acceptor. J . Biol. Chem. 246,3977-3984. Morrison, J. F., and Ebner, K. E. (1971b). Studies on galactosyltransferase: Kinetic investigations with glucose as the galactosyl group accept0r.J. Biol. Chem. 246,39853991. Morrison, T. G., and Lodish, H. F. (1975). Site of synthesis of membrane and nonmembrane proteins of vesicular stomatitis virus. J . Biol. Chem. 250,6955-6962. Moscarello, M. A. (1976). Chemical and physical properties of myelin proteins. Curr. Top. Membr. Transp. 8, 1-28. Moscarello, M. A., Kashuba, L., and Sturgess, J. M. (1972). The incorporation of l4C-Dgalactose and ’H-D-mannose into Golgi fractions of rat liver and into serum. FEBS Lett. 26,87-91.

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

97

Moscarello, M. A., Cagnon, J., Wood, D. D., Anthony, J., and Epand, R. M. (1973).Conformational flexibility of a-myelin protein. Biochemistry 12,3402-3406. Moscarello, M. A., Katona, E., Neumann, W., and Epand, R. M. (1974).The ordered structure of the encephalitogenic protein from normal human myelin. B i o p h p . Chem. 2,290-295. Munro, J. R., and Schachter, H. (1973). The presence of two GDP-L-fucose:glycoprotein fucosyltransferases in human serum. Arch. Biochem. Biophys. 156,534-542. Munro, J. R., Narasimhan, S., Wetmore, S., Riordan, J. R., and Schachter, H. (1975).Intracellular localization of GDP-L-fucose:glycoprotein and CMP-sialic acid:apolipoprotein glycosyltransferases in rat and pork livers. Arch. Biochem. Biophys. 169, 269-277. Nakamura, R., and Tsunemitsu, A. (19754.Presence of cytidine 5‘-monophospho sialic acid:glycoprotein sialyltransferase in human saliva. J. Dent. Res. 54, 188. Nakamura, R., and Tsunemitsu, A. (1975b).Presence of guanosine diphosphofucose:glycoprotein fucosyltransferase in human saliva. J. Dent. Res. 54,417. Nakamura, R., Watanabe, T., Yasutake, A., and Tsunemitsu, A. (1974).Presence of urigalactosyltransferase in human dine 5’-diphosphogalactose:N-acetylglucosamine saliva. J. Dent. Res. 53,937. Narasimhan, S., Stanley, P., and Schachter, H. (1977).Control of glycoprotein synthesis: Lectin-resistant mutant containing only one of two distinct N-acetylglucosaminyltransferase activities present in wild type Chinese hamster ovary cells. J. Biol. Chem. 252,3926-3933. Nathenson, S. G., and Cullen, S. E . (1974).Biochemical properties and immunochemical-genetic relationships of mouse €I-2alloantigens. Biochim. Biophys. Acta 344, 1-25. Nathenson, S. G., and Muramatsu, T. (1971).Properties of the carbohydrate portion of mouse H2 alloantigens. In “Glycoproteins of Blood Cells and Plasma” ( G . A. Janiieson and T. J. Greenwalt, eds.), pp. 245-262.Lippincott, Philadelphia, Pennsylvania. The occurrence and propNelson, J. D., Jato-Rodriguez, J. J., and Mookejea, S. (1974). erties of soluble UDP-ga1actose:glycoprotein galactosyltransferase in human amniotic fluid. Can. ]. Biochem. 52,42-50. Neutra, M., and Leblond, C. P. (1966).Radioautographic comparison of the uptake of galactose-H3 and glucose-H3 in the Golgi region of various cells secreting glycoproteins or mucopolysaccharides. J. Cell Biol. 30, 137-150. Nicolson, G.L. (1976).Transmembrane control of the receptors on normal and tumour cells. I. Cytoplasmic influence over cell surface components. Biochim. Biophys. Acta 457,57-108. Nicolson, G. L., and Singer, S . J. (1971).Ferritin conjugated plant agglutinins as specific saccharide stains for electron microscopy: Application to saccharides bound to cell membranes. Proc. Natl. Acad. Sci. U.S.A. 68, 942-945. Novogrodsky, A., and Ashwell, G. (1977).Lymphocyte mitogenesis induced by a mammalian liver protein that specifically binds desialyated glycoproteins. Proc. Natl. Acud. Sci. U.S.A.74, 676-678. Nussbaum, J. L., Rouayreng, J. F., Mandel, P., Jolles, J., and Jolles, P. (1974).Isolation and terminal sequence determination of the major rat brain myelin proteolipid P7 apoprotein. Biochem. Biophys. Res. Commun. 57, 1240-1247. Odashima, S., Sturgess, J. M., and Rothstein, A. (1976).Characteristics of adult rat liver cells in culture. I. Morphological and biochemical studies. Cell Tissue Res. 169, 167.

98

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

Oesterhelt, D., and Stoeckenius, W. (1971). Rhodopsin-like protein from the purple membrane ofHalobacterium halobium. Nature (London),New Biol. 233,149-152. Okumura, T., Lombart, C., and Jamieson, G. A. (1976). Platelet glycococalicin. 11. Purification and characterization. J . Biol. Chem. 251,5950-5955. Osborne, Beverley, H., Sardet, C., and Helenius, A. (1974). Bovine rhodopsin: Characterization of the complex formed with Triton X-100. Eur. J. Biochem. 44,383-390. Ozols, J. (1972). Cytochrome b, from normal human liver: Isolation and partial amino acid sequence. J . Biol. Chem. 247,2242-2245. Ozols, J. (1974). Cytochrome b, from microsomal membranes of equine, bovine, and porcine livers: Isolation and properties of preparations containing the membranous segment. Biochemistry 13,426-434. Palade, G. E. (1975). Intracellular aspects of the process of protein secretion. Science 189,347-349. Palmer, F. B., and Dawson, R. M. C. (1969).The isolation and properties of experimental allergic encephalitogenic protein. Biochem. J . 111,629-636. Papahadjopoulos, D., Vail, W. J., and Moscarello, M. A. (1975a).Interaction of a purified hydrophobic protein from myelin with phospholipid membranes: Studies on ultrastructure, phase transitions and permeability. J . Mernbr. Biol. 22, 143-164. Papahadjopoulos, D., Moscarello, M., Eylar, E. H., and Isac, T. (1975b). The effects of proteins on thermotropic phase transitions of phospholipid membranes. Biochim. Biophys. Acta 401,317-335. Papermaster, D. S., Converse, C. A., and Sill, J. (1975). Membrane biosynthesis in frog retina: Opsin transport in the photoreceptor cell. Biochemistry 14, 1343-1352. Parodi, A. J., Behrens, N. H., Leloir, L. F., and Carminatti, H. (1972a).The role of polyprenol-bound saccharides as intermediates in glycoprotein synthesis in liver. Proc. Natl. Acad. Sci. U.S.A.69, 3268-3272. Parodi, A. J., Behrens, N. H., Leloir, L. F., and Dankert, M. (1972b). Glucose transfer from dolichol monophosphate glucose: The lipid moiety of the endogenous acceptor. Biochim. Biophys. Acta 270,529-536. Parodi, A. J., Staneloni, R., Cantarella, A. I., Leloir, L. F., Behrens, N. H., Carminatti, H., and Levy, J. A. (1973). Further studies on a glycolipid formed from dolichyl-Dglucosyl monophosphate. Carbohydr. Res. 26,393-400. Paulson, J. C., Beranek, W. E., and Hill, R. L. (1977a).Purification ofa sialyltransferase from bovine colostrum by affinity chromatography on CDP-agarose.J . Biol. Chem. 252,2356-2362. Paulson, J. C., Rearick, J. I., and Hill, R. L. (1977b). Enzymatic propeities of p-D-galactoside a-26 sialyltransferase from bovine co1ostrum.J. Biol. Chem. 252,2363-2371. Pisam, M., and Repoche, P. (1976). Renouvellement des glycoproteins de surface: ktude radioautographique de I'epithelium de la vessie de grenouille. J . Microsc. (Paris) 17,261. Plantner, J. J., and Kean, E. L. (1976).The carbohydrate composition of bovine rhodopsin. J. Biol. Chem. 251, 1548-1552. Pless, D. D. (1976). Oligosaccharide-lipid as substrate for glycosylation of exogenous proteins. Fed. Proc., Fed. Am. SOC.E x p . Biol. 35, 1539. Pless, D. D., and Lennarz, W. J. (1975). A lipid-linked oligosaccharide intermediate in glycoprotein synthesis: Characterization of [Man-14C] glycoproteins labeled from [Man-14C]oligosaccharide-lipid and GDP-["C] Man. J. Biol. Chem. 250, 70147019. Pober, J. S., and Stryer, L. (1975). Light dissociates enzymatically cleaved rhodopsin into two different fragments. J . Mol. Biol. 95,477-481.

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

99

Poste, G., Papahadjopoulos, D., and Jacobson, K. (1975). Local anesthetics increase susceptibility of untransformed cells to agglutination of concanavalin A. Nature (London) 253,552-554. Powell, J. T., and Brew, K. (1974).The preparation and characterization of two forms of bovine galactosyl transferase. Eur. J. Biochem. 48,217-228. Pricer, W. E., Jr., and Ashwell, G. (1976). Subcellular distribution of a mammalian hepatic binding protein specific for asialoglycoproteins. J. Biol. Chem. 251, 75397544. Prieels, J. P., and Barel, A. 0. (1975).A fluorimetric study of the interactions of insolubilized human a-lactalbumin with galactosyl transferase (A protein) and with anti-alactalbumin antibodies. Biochim. Biophys. Acta 393,496-504. Rambourg, A., and Leblond, C. P. (1967). Electron microscope observations on the carbohydrate-rich cell coat present at the surface of cells in the rat. J. Cell Biol. 32, 27-53. Rambourg, A., Neutra, M. M., and Leblond, C. P. (1966). Presence of a “cell coat” rich in carbohydrate at the surface of cells in the rat. Anat. Rec. 154,41-71. Rambourg, A., Hernandez, W., and Leblond, C. P. (1969). Detection of complex carbohydrates in the Colgi apparatus of rat ce1ls.J. Cell Biol. 40,395-414. Rand, P., Papahadjopoulos, D., and Moscarello, M. A. (1976). On the interaction of lecithins both with a purified hydrophobic protein from myelin and with fatty acids and lysolecithin. Biophys J. 16, 192. Reading, C. L., Penhoet, E. E., and Ballou, C. E. (1977). Carbohydrate structure of vesicular stomatitis virus g1ycoprotein.J. Supramol. Struct., Suppl. 1, 7. Redman, C. M., and Cherian, M. G. (1972).The secretory pathways of rat serum glycoproteins and albumin: Localisation of newly formed proteins within the endoplasmic reticu1um.J. Cell Biol. 52, 231-245. Renthal, R., Steinmemann, A., and Stryer, L. (1973).The carbohydrate moiety of rhodopsin: Lecting binding, chemical modification and fluorescence studies. E x p . Eye Res. 17,511-515. 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. Richards, J. B., Evans, P. J., and Hemming, F. W. (1971).Dolichol phosphates as acceptors of mannose from guanosine diphosphate mannose in liver systems. Bi0chem.J. 124,957-959. Richards, J. B., Evans, P. J., and Hemming, F. W. (1972).Dolichol phosphate and mannose transferase reactions in animal systems. I n “Biochemistry of the Glycosidic Linkage” (R. Piras and H. G. Pontis, eds.), pp. 207-220. Academic Press, New York. Richmond, J. E. (1965). Role of sugar nucleotides in the incorporation of sugars into glycoproteins. Biochemistry 4, 1834-1838. Robinson, G. B., Molnar, J., and Winzler, R. J. (1964). Biosynthesis of glycoproteins. I. Incorporation of gl~cosamine-’~C into liver and plasma proteins of the rat. J. Biol. Chem. 239, 1134-1141. Roden, L. (1970). Biosynthesis of acidic glycosaminoglycans (mucopolysaccharides). In “Metabolic Conjugation and Metabolic Hydrolysis” (W. H. Fishman, ed.), Vol. 2, pp. 345-442. Academic Press, New York. Rogers, M. J., and Strittmatter, P. (1973).Lipid-protein interactions in the reconstitution of the microsomal reduced nicotinamide adenine dinucleotide-cytochrome bs reductase system. ]. Biol. Chem. 248,800-806.

100

J. STURGESS, M. MOSCARELLO, A N D H. SCHACHTER

Roseman, S. (1970). The synthesis of complex carbohydrates by multiglycosyltransferase systems and their potential function in intercellular adhesion. Chem. Phys. Lipids 5,270-297. Rothman, J. E., and Lenard, J. (1977). Membrane asymmetry. Science 195,743-753. Rovis, L., Anderson, B., Kabat, E. A., Gruezo, F., and Liao, J. (1973). Immunochemical studies on blood groups: Structures of oligosaccharides produced by base-borohydride degradation of human ovarian cyst blood group H, heb and Lea active glycoproteins. Biochemistry 12,5340-5354. Saari, J. C. (1974). The accessibility of bovine rhodopsin in photoreceptor membranes. J. Cell Biol. 63,480-491. Sarcione, E. J. (1964). The initial subcellular site of incorporation of hexoses into liver protein. J. Biol. Chem. 239, 1686-1689. Satir, B. C., Schooley, C., and Satir, P. (1973). Membrane fusion in a model system. J . Cell Biol. 59, 153-176. Schachter, H. (1974a). The subcellular sites of glycosylation. Biochem. SOC. Symp. 40, 57-71. Schachter, H. (1974b). Glycosylation of glycoproteins during intracellular transport of secretory products. Ado. Cytophannacol. 2,207-218. Schachter, H., and Roden, L. (1973). The biosynthesis of animal glycoproteins. I n “Metabolic Conjugation and Metabolic Hydrolysis” (W. H. Fishman, ed.), Vol. 3, pp. 1149. Academic Press, New York. Schachter, H., Jabbal, I., Hudgin, R. L., Pinteric, L., McGuire, E. J., and Roseman, S. (1970). Intracellular localization of liver sugar nucleotide glycoprotein glycosyltransferases in a Golgi-rich fraction. J . Biol. Chem. 245, 1090-1100. Schachter, H., McGuire, E. J., and Roseman, S. (1971). Sialic acids. XIII. A uridine diphosphate D-ga1actose:mucin galactosyltransferase from porcine submaxillary gland. J. Biol. Chem. 246, 5321-5328. Schanbacher, F. L., and Ebner, K. E. (1970). Galactosyltransferase acceptor specificity of the lactose synthetase A protein. J . Biol. Chem. 245, 5057-5061. Schauer, R. (1970a). Biosynthesis of N-acetyl-0-acetylneuraminicacid I. Incorporation of I4C-acetate into slices of the submaxillary salivary glands of ox and horse. HoppeSeyler’s 2.Physiol. Chem. 351,595-602. Schauer, R. (1970b). Biosynthesis of N-acetyl-0-acetylneuraminicacid. 11. Substrate and intracellular localization of the bovine acetyl-coenzyme AN-acetylneuraminate-7- and 8-0-acetyltransferase. Hoppe-Seyler’s Z. Physiol. Chem. 351,783-791. Schauer, R. (19704. Biosynthesis of N-glycoloylneuraminic acid by an ascorbate or NADPH-dependent of N-acetyl hydroxylating “N-acety1neuraminate:oxygenoxidoreductase in homogenates of porcine submaxillary gland. Hoppe-Seyler’s 2.Physiol. Chem. 351,783-791. Schauer, R. (1973).Chemistry and biology of acylneuraminic acids. Angew. Chem. 85, 128- 140. Schauer, R., and Wember, M. (1970). Studies on the subcellular site of the biosynthesis of N-glycoloylneuraminic acid in porcine submaxillary gland. Hoppe-Seyler’s 2. Physiol. Chem. 351,1353-1358. Schauer, R., and Wember, M. (1971). Hydroxylation and 0-acetylneuraminic acid bound to glycoproteins of isolated subcellular membranes from porcine and bovine submaxillary glands. Hoppe-Seyler’s 2.Physiol. Chem. 352, 1282-1290. Schauer, R., and Wember, M. (1973). Studies on the substrate specificity of acylneuraminate cytidyltransferase and sialyltransferase of submandibular glands from cow, pig and horse. Hoppe-Seyler’s 2.Physiol. Chem. 354, 1405-1414.

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

101

Schauer, R., Buscher, H. P., and Casals-Stenzel, J. (1974). Sialic acids: Their analysis and enzymic modification in relation to the synthesis of submandibular gland glycoproteins. Biochem. SOC.Symp. 40,87-116. Schechter, I., McKean, D. J., Guyer, R., and Terry, W. (1974). Partial amino acid sequence of the precursor of immunoglobulin light chain programmed by messenger RNA in uitro. Science 188, 160-162. Schimke, R. T. (1973). Control of enzyme levels in mammalian tissues. Adu. Enzymol. Relat. Areas Mol. Biol. 37, 135-187. Schimke, R. T. (1974). The synthesis and degradation of membrane proteins. Adu. Cytopharmacol. 2,63-69. Schwyzer, M., and Hill, R. L. (1977a). Porcine A blood group-specific N-acetylgalactosaminyltransferase. I. Purification from porcine submaxillary glands. J. B i d . Chem. 252,2338-2345. Schwyzer, M., and Hill, R. L. (1977b). Porcine A blood group-specific N-acetylgalactosaminyltransferase. 11. Enzymatic properties. J. B i d . Chem. 252,2346-2355. Sefton, B., and Keegstra, K. (1974). Glycoproteins of Sindbis virus: Preliminary characterization of the oligosaccharides. J. Virol. 14, 522-530. Segrest, J. P., Kahne, I., Jackson, R. L., and Marchesi, V. T. (1973). Major glycoprotein of the human erythrocyte membrane: evidence for an amphipathic molecular structure. Arch. Biochem. Biophys. 155, 167-183. Sherr, C. J., and Uhr, J. W. (1969). Immunoglobulin synthesis and secretion. 111. Incorporation of glucosamine into immunoglobulin on polyribosomes. Proc. Natl. Acad. Sci. U.S.A. 64,381-387. Shetlar, M. R., Capps, J. C., and Hem, D. L. (1964). Incorporation of radioactive glucosamine into the serum proteins of intact rats and rabbits. Biochim. Biuphys. Acta 83,93-101. Shull, K . H., and Miller, 0. N. (1960).Formation in uiuo of glycogen by certain intermediates of the lactate-propranediol pathway. J . B i d . Chem. 235, 551-553. Shur, B. D., and Roth, S. (1975). Cell surface glycosyltransferases. Biochim. Biophys. Acta 415,473-512. Siekevitz, P. (1972). The turnover of proteins and the usage of information. J . Theor. B i d . 37,321-333. Singer, S . J. (1974). The molecular organization of membranes. Annu. Reu. Biochem. 43, 805-833. Singer, S . J., and Nicolson, G. L. (1972). The fluid mosaic model of the structure of cell membranes. Science 175,720-731. Spatz, L., and Strithnatter, P. (1971). A form of cytochrome b, that contains an additional hydrophobic sequence of 40 amino acid residues. Proc. Natl. Acad. Sci. U.S.A. 68, 1042-1046. Spiro, R. G. (1959). Studies on the biosynthesis of glucosamine in the intact rat. J . B i d . Chem. 234,742-748. Spiro, M. J., and Spiro, R. G. (1968a). Glycoprotein biosynthesis: Studies on thyroglobulin. Thyroid sialyltransferase. J. B i d . Chem. 243,6520-6528. Spiro, M. J., and Spiro, R. G. (1968b). Glycoprotein biosynthesis: Studies on thyroglobulin. Thyroid galactosyltransferase. J . B i d . Chem. 243,6529-6537. 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. 251,64006408.

Spiro, M. J., Spiro, R. G., and Bhoyroo, V. D. (197613).Lipid-saccharide intermediates in

102

J. STURGESS, M. MOSCARELLO, AND H. SCHACHTER

glycoprotein biosynthcsis. 111. Comparison of oligosaccharide-lipids formed by slices from several tissues.]. Biol. Chem. 251,6420-6425. Spiro, R. G., Spiro, M. J., and Adamany, A. M. (1974).Enzymic assembly of the carbohydrate units of glycoproteins and the role of lipid intermediates. Biochem. S O C . Symp. 40,37-56. Spiro, R. G., Spiro, M. J., and Bhoyroo, V. D. (1976). Lipid-saccharide intermediates in glycoproteinhiosynthesis. 11. Studies on the structure of an oligosaccharide-lipid from thyroid.]. Biol. Chem. 251,6409-6419. Springer, G . F., and Yang, H. J. (1977). Isolation and partial characterization of blood group M- and N-specific glycopeptides and oligosaccharides from human erythrocytes. lmmunochemistry 14,497-502. Springer, G. F., Nagai, Y., and Tegtmeyer, H. (1966). Isolation and properties of human blood group NN and meconium-Vg antigens. Biochemistry 5,3254-3272. Springer, G . F., Codington, J. F., and Jeanloz, R. W. (1972). Surface glycoprotein from a mouse tumour cell as specific inhibitor of antihuman blood group or agglutinin. J . Natl. Cancer Inst. 49, 1469-1470. Stanley, P., Narasimhan, S., Siminovitch, L., and Schachter, H. (1975). Chinese hamster ovary cells selected for resistance to the cytotoxicity of phytohemagglutinin are deficient in a UDP-N-acetylglucosamine-glycoproteinN-acetylglucosaminyltransferase activity. Proc. Natl. Acad. Sci. U.S.A. 72,3323-3327. Steiner, D. F., Kemmler, W., Tager, H. S., and Rubenstein, A. H. (1974). Molecular events taking place during intracellular transport of exportable proteins: The conversion of peptide hormone precursors. Ado. Cytopharmacol. 2, 195-205. Strittmatter, P. (1971). The characterization of interconversion of two conformational states of cytochrome b, reduct0se.J. Biol. Chem. 246, 1017-1074. Strittmatter, P., Spatz, L., Corcoran, D., Rogers, M. J., Setlow, B., and Redline, R. (1974). Purification and properties of rat liver microsomd stearyl coenzyme A desaturase. Proc. Natl. Acad. Sci. U S A . 71,4565-4569. Strominger, J. L., Higashi, Y., Sandermann, H., Stone, K. J., and Willoughby, E. (1972). The role of polyisoprenyl alcohols in the biosynthesis of the peptidoglycan of bacterial cell walls and other complex polysaccharides. In “Biochemistry of the Glycosidic Linkage” (R. Piras and H. C. Pontis, eds.), pp. 135-154. Academic Press, New York. Struck, D. K., and Lennarz, W. J. (1977). Evidence for the participation of saccharidelipids in the synthesis of the oligosaccharide chain of ovalbumin. J . Biol. Chem. 252, 1007-1013. Struck, D. K., Chen, W. W., and Lennarz, W. J. (1977). Enzymatic incorporation of saccharides from saccharide-lipids into endogenous and exogenous protein acceptors. Fed. Proc., Fed. Am. Soc. E x p . Biol. 36,745. Sturgess, J. M., and Moscarello, M. A. (1976). Surface ultrastructure of the Golgi complex. Scanning Electron Microsc. symp., IITtU/1976 7, Part 2, 146. Sturgess, J. M., Mitranic, M. M., and Moscarello, M. A. (1972).The incorporation of Dglucosamine-3H into the Golgi complex from rat liver and into serum glycoproteins. Biochem. Biophys. Res. Commun. 46, 1270-1277. Sturgess, J. M., Minaker, E., Mitranic, M. M., and Moscarello, M. A. (1973). The incorporation of L-fucose into glycoproteins in the Golgi apparatus of rat liver and in serum. Biochim. Biophys. Acta 320, 123-132. Sturgess, J. M., Moscarello, M. A., and Vail, W. J. (1974). The fine structure of membranes of the Golgi complex. Proc. Int. Congr. Electron Microsc., 8th 11, 194. Svensson, H., Elhammer, A,, Autuori, F., and Dallner, G. (1976).Biogenesis of microsoma1 membrane glycoproteins in rat liver. IV. Characteristics of a cytoplasmic lipo-

STRUCTURE AND BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

103

protein having properties of a membrane precursor. Biochim. Biophys. Actu 455, 383-398.

Takatsuki, A., Kohno, K., and Tamura, G . (1975). Inhibition of hiosynthesis of polyisoprenol sugars in chick embryo microsomes by tuniram-r-i- A-w:- Biol. Chem. 39, 2089-2091.

Tanigari, N., Nakamuro, K., Apella, E., Poulik, M. d.,- - - A :';asman, D. (1973). Identity of the HL-A common portion fragment and human p,-microglobulin. Biochem. Biophys. Res. Commun. 55, 1234-1239. Tarentino, A. L., and Maley, F. (1974). Purification and properties of an endo-@-N-acetylglucosaminidase from Streptomyces griseus. J. Biol. Chem. 249,811-817. Tarentino, A., Plummer, T. H., Jr., and Maley, F. (1970). Studies on the oligosaccharide sequence of ribonuclease B. J. B i d . Chem. 245,4150-4157. Thomas, D. B., and Winzler, R. J. (1969a). Oligosaccharides from M-active sialoglycopeptide of human erythrocytes. Biochem. Biophys. Res. Commun. 35,811-818. Thomas, D. B., and Winzler, R. J. (1969b).Structural studies on human erythrocyte glycoproteins: Alkali-labile oligosaccharides. J. B i d . Chem. 244,5413-5946. Thomas, D. B., and Winzler, R. J. (1971). Structure of glycoproteins of human erythrocytes: Alkali-stable oligosaccharides. Biochem. J. 124, 55-59. Tkacz, J. S., and Herscovics, A. (1975). Ozonolytic cleavage of authentic and pancreatic dolichyl mannopyranosyl phosphates: Determination of sugar configuration in the fragments with a-and p-mannosidases. Biochem. Biophys. Res. Comrnun. 64,1009 -1017.

Tkacz, J . S., and Lampen, J. 0. (1975). Tunicamycin inhibition of polyisoprenyl N-acetylglucosaminyl pyrophosphate formation in calf liver microsomes. Biochem. Biophys. Res. Commun. 65,248-257. Tkacz, J. S., Herscovics, A., Warren, C. D., and Jeanloz, R. W. (1973). Characterization of mannolipids formed in a calf pancreatic microsomal fraction from guanosine diphosphate [I4C]mannose and either endogenous lipid or exogenous polyisoprenyl phosphates. Biochem. SOC. Truns. 1, 1174-1175. Tkacz, J. S., Herscovics, A., Warren, C. D., and Jeanloz, R. W. (1974). Mannosyltransferase activity in calf pancreas microsomes: Formation from guanosine diphosphate-~-['~C mannose ] of a I4C-labeled mannolipid with properties of dolichyl mannopyranosyl phosphate. J. Biol. Chem. 249,6372-6381. Tomita, M., and Marchesi, V. T. (1975). Amino acid sequence and oligosaccharide attachment sites of human erythrocyte glycophorin. Proc. Natl. Acad. Sci. U.S.A. 72, 2964 -2968.

Toneguzzo, F., and Ghosh, H. P. (1975). Cell-free synthesis of vesicular stomatitis virus proteins: Translation of membrane-bound polyribosomal mRNAs. FEBS Lett.50, 369-373.

Trayer, I. P., and Hill, R. L. (1971). The purification and properties of the A protein of lactose synthetase. J. Biol. Chem. 246,6666-6675. Trujillo, J. L., and Can, J. C. (1973). Gycoprotein biosynthesis. VI. Regulation of UDPN-acetyl-D-glucosamine metabolism in bovine thyroid gland slices. Biochim. Biophys. Actu 304,32-41. Trump, B. F., Sultera, S. M., and Byrne, W. L. (1970). Membrane structure: Lipid-protein interactions in microsomal membranes. Proc. Nutl. Acud. Sci. U S A . 66,433440.

Tsay, G. C., Dawson, G., and Sung, S. S. J. (1976). Structure of the accumulating oligosaccharide in fucosidosis. J. Biol. Chem. 251,5852-5859. Tyrell, D. A., Heath, T. D., Colley, C. M., and Ryman, B. E. (1976). New aspects of liposomes. Biochim. Biophys. Actu 457,259-302.

104

’

J. STURGESS, M. MOSCARELLO, A N D H. SCHACHTER

Vail, W. J,, Papahadjopoulos, D., and Moscarello, M. A. (1974). Interaction of a hydrophobic protein with liposomes: Evidence for particles seen on freeze fracture as being proteins. Biochim. Biophys. Actu 345,463-467. Van den Eijnden, D. H. (1973). Subcellular localization of cytidine 5‘-monophospho-Nacetylneuraminic acid synthetase in calf brain. J . Neurochem. 21,949-958. Van Dijk, W., Ferwerda, W., and Van den Eijnden, D. H. (1973). Subcellular and regional distribution of CMP-N-acetyl neuraminic acid synthetase in calf kidney. Biochim. Biophys. Acta 315,162-175. Visser, L., Robinson, N. C., and Tanford, C. (1975). The two-domain structure of cytochrorne b, in deoxycholate solution. Biochemistry 14, 1194-1199. Waechter, C. J,, and Lennan, W. J. (1976). The role of polyprenol-linked sugars in glycoprotein synthesis. Annu. Rev. Biochem. 45,95-112. Waechter, C. J., Lucas, J. J., and Lennan, W. J. (1973).Membrane glycoproteins. I. Enzymatic synthesis of mannosyl phosphoryl polyisoprenol and its role as a mannosyl donor in glycoprotein synthesis. J . Biol. Chem. 248,7570-7579. Wagner, R. R., and Cynkin, M. A. (1971).Glycoprotein metabolism and UDP-galactose:glycoprotein galactosyltransferase of rat serum. Biochem. Biophys. Res. Commun. 45,57-62. Warren, C. D., and Jeanloz, R. W. (1973a). The characterization of glycolipids derived from long-chain polyprenols: Chemical synthesis of a-Dmannopyranosyl dolichyl phosphate. FEBS Lett. 31,332-334. Warren, C. D., and Jeanloz, R. W. (197313). Chemical synthesis of dolichyl a-Dmannopyranosyl phosphate and citronellyl a-D-mannopyranosyl phosphate. Biochemistry 12,5038-5045. Warren, G. B., Houslay, M. D., Metcalfe, J. C., and Birrall, N. J. M. (1975). Cholesterol is excluded from the phospholipid annulus surrounded by an active calcium transport protein. Nature (London)225,684-687. Warren, L. (1963). The distribution of sialic acids in nature. Comp. Biochem. Physiol. 10, 153-157. Wedgwood, J. F., Warren, C. D., Jeanloz, R. W., and Strominger, J. L. (1974). Enzymatic P*-dolichyl pyrophosphate and its chemical utilization of PI-di-N-acetylchitobiosyl synthesis. Proc. Natl. Acad. Sci. U.S.A. 71, 5022-5026. White, B. N., Shetlar, M. R., Shurley, H. M., and Schilling, J. A. (1965). Incorporation of ~ - [ l - ’ ~ Cgalactosamine l into serum proteins and tissues of the rat. Biochim. Biophys. Actu 101,259-266. Whur, P., Herscovics, A., and Leblond, C. P. (1969).Radioautographic visualization of the incorporation of galactose-3H by rat thyroids in uitro in relation to the stages of thyroglobulin synthesis. J . Cell Biol. 43,289-311. 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 L-fucose to the asparagine-linked N-acetylglucosamine residue of glycopeptide from a,-acid glycoprotein. Biochem. Biophys. Res. Commun. 72,909-916. Winzler, R. J. (1969).A glycoprotein in human erythrocyte membranes. In “Red Cell Membrane” (G. A. Jamieson and T. J. Greenwalt, eds.), pp. 157-171. Lippincott, Philadelphia, Pennsylvania. Winzler, R. J. (1970). Carbohydrates in cell surfaces. Znt. Reu. Cytol. 29,77-125. Winzler, R. J. (1972). Glycoproteins of plasma membranes: Chemistry and function. In “Glycoproteins. Their Composition, Structure and Function,” Part B (A. Gottschalk, ed.), pp. 1268-1293. Elsevier, Amsterdam. Wirth, D. F., Katz, F., Small, B., and Lodish, H. F. (1977).How a single Sindbis virus

STRUCTURE A N D BIOSYNTHESIS OF MEMBRANE GLYCOPROTEINS

105

mRNA directs the synthesis of one soluble protein and two integral membrane glycoproteins. Cell 10,253-263. Wood, D. D., Boggs, J., and Moscarello, M. A. (1978). Labelling of lipophilin in phosphatidylcholine vesicles. In preparation. Worthington, C. R. (1973). X-ray analysis of retinal photoreceptor structure. E x p . Eye Res. 17,487-501.

This Page Intentionally Left Blank

CURRENT TOPICS IN MEMBRANES AND TRANSPORT. VOLUME

11

Techniques for the Analysis of Membrane Glycoproteins R . L. JULZANO Research Institute The Hospital for Sick Children Toronto, Onturio

I. Introduction. . . . . . . . . . . . . . . . . . 11. Identification of Cell Surface Glycoproteins . . . . . A. Surface Labeling . . . . . . . . . . . . . . B. Membrane Isolation . . . . . . . . . . . . . C. MetabolicLabelingof Membrane Glycoproteins . . . . . 111. Fractionation of Membrane Glycoproteins . . . . . . . . A. Detergents. . . . . . . . . . . . . . . . B. Denaturing Solvents. . . . . . . . . . . . . . C. Fractionation of Membrane Glycoproteins by Column and Polyacrylamide Gel Techniques . . . . . . . . . . D. Lectin Affinity Techniques . . . . . . . . . . . E. Other Approaches to Membrane Glycoprotein Fractionation . IV. Chemical Analysis of Membrane Glycoproteins . . . . . . . A. Preparation of Glycoproteins and Oligosaccharides . . . . B. Analysis of the Polypeptide Portion of Glycoproteins . C. Analysis of Individual Sugars . . . . . . . . . . . D. Determination of the Carbohydrate Structure of Glycoproteins E. MWs of Membrane Glycoproteins . . . . . . . . . V. Genetic Analysis of Membrane Glycoproteins . . . . . . . References . . . . . . . . . . . . . . . . . .

.

.

.

. .

1.

. .

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

. .

. . . .

. . . . . . . . . .

107 108 108

115 116 117 117 119 120 122 128 129 129 130 131 131 132 133 134

INTRODUCTION

This article is intended as a guide to current approaches for the identification, solubilization, fractionation, and chemical analysis of plasma membrane glycoproteins. The material reviewed is confined to studies of glycoproteins emanating from mammalian and avian cells, tumors, and tissues, and does not deal with prokaryotic or lower eukaryotic organisms. Certain highly specialized surface glycoproteins, such as hormone receptors and surface immunoglobulins, have been reviewed elsewhere (Kahn, 1976; Vitetta and Uhr, 1975)and are Copyright @ 19711 by Academic Press. lnc. All rights ofreprodiiction in any form reserved. ISBN 0- 12- 1533I 1-5

107

108

R. 1. JULIANO

not considered here. This article deals primarily with the more recent literature dating from about 1973; for surveys of the earlier literature, the reader is directed to several excellent reviews and treatises on the biochemistry of glycoproteins (Marshall, 1972; Spiro, 197313; Gottschalk, 1972; Cook and Stoddart, 1973; Ginsberg, 1972; Sharon, 1975; Hughes, 1976). Much of the methodology for the analysis of individual sugars and of oligosaccharide sequences is dealt with in the treatises edited by Gottschalk and by Ginsberg in greater detail than is possible here. To reiterate, this article is a guide to current approaches for membrane glycoprotein analysis and reflects the author’s own interests and experience in this area; it is not intended as a comprehensive survey of all techniques which have been applied to the biochemistry of glycoproteins. II. IDENTIFICATION OF CELL SURFACE GLYCOPROTEINS A. Surface Labeling

A nontrivial problem in the analysis of membrane glycoproteins is to discriminate molecules residing on the outer surface of the cell from other glycoprotein species associated with internal elements such as the endoplasmic reticulum, mitochondria, and lysosomes. Since the overall glycoprotein pattern of animal cells is so complex (Wray and Perdue, 1974), there has been widespread use of a very powerful approach to the problem of the cellular localization of proteins and glygoproteins, namely, the employment of surface-labeling reagents. The term “surface label” connotes a type of reagent which reacts covalently with proteins or glycoproteins, but whose size or solubility properties render it unable to enter cells, thus restricting its sites of reaction to the cell periphery. The surface label approach to the analysis of membrane organization has been reviewed previously (Juliano, 1973; Carraway, 1975); nonetheless, it seems appropriate to discuss this technique and its advantages and liabilities within the confines of this article. Basically, two types of surface label reagents are currently in use: (1)low-MW, impermeant molecules which react directly with accessible surface groups, and (2) enzymes which catalyze or promote the transfer of a radiolabel to accessible sites on the cell surface. Carraway (1975) lists seven low-MW membrane-labeling reagents which have been used on animal cells; most of these reagents are bulky anions, moieties known to penetrate cell membranes only very slowly (Roth-

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

109

stein et al., 1976). In most instances these reagents react primarily with exposed amino groups; this is true in the cases or isothiocyanostilbene disulfonic acids (SITS and DIDS), diazotized sulfanilic acid (DASA), formylmethionyl methyl phosphate (FMMP), and trinitrobenzene sulfonate (TNBS), while the photoreactive compound N (4-azido-2-nitrophenyl)-2-amino ethanesulfonate (NAP taurine) has a much broader range of reactivity. Most of these reagents, including DIDS (Cabantchik and Rothstein, 1974), FMMP (Bretscher, 1971), and NAP taurine (Staros and Richards, 1974; Cabantchik et al., 1976), have been used primarily to investigate molecular organization in erythrocytes. Another amino-reactive probe, namely, pyridoxal phosphate-[3H]borohydride, has also been used on enveloped viruses (Rifkin et al., 1972) and on cultured mammalian cells (Juliano and Behar-Bannelier, 1975b; Hunt and Brown, 1974). Some of the problems associated with the use of low-MW probes in studies on complex nucleated cells are considered in the following discussion. 1. LACTOPEROXIDASE

The most widely used surface-labeling reagent is the enzyme lactoperoxidase, which catalyzes the iodination of accessible tyrosine and possibly histidine residues when used in the presence of hydrogen peroxide. The reaction proceeds mainly via an enzyme-bound activated iodine intermediate and requires close contact between the enzyme and its protein substrate (Morrison et d.,1971). The application of the lactoperoxidase technique to the analysis of erythrocyte membrane organization has been reviewed elsewhere (Juliano, 1973; Carraway, 1975). Lactoperoxidase iodination has also been applied to the analysis of surface membrane proteins in lymphoid cells (Marchalonis et al., 1971; Vitetta et al., 1971; Vitetta and Uhr, 1975), normal and transformed fibroblasts (Hynes, 1973; Hogg, 1974; Yamada and Weston, 1974; Teng and Chen, 1975, 1976), platelets (Phillips, 1972; Tanner et al., 1974), adipocytes (Trosper and Levy, 1974), and a variety of cultured and ascitic tumor cells (Shin and Carraway, 1973; Hunt and Brown, 1974; Huang et al., 1973; Mastro et al., 1974; Butters and Hughes, 1975; Juliano and Behar-Bannelier, 1975a,b). In using lactoperoxidase for the analysis of surface glycoproteins in cultured hamster cells (line CHO), we have found it advantageous to employ relatively large amounts of enzyme (40 pg/ml) and radioiodine (0.2-0.5 mCi/ml) and to limit the reaction time to 5 minutes at 37°C or 10-15 minutes at room temperature. This seems to give useful levels of labeling (see Fig. 1)without causing a great deal of cell dam-

R. 1. JULIANO

zoo

100

m

FIG.1. Protein staining patterns and autoradiographs of SDS gels of CHO cell plasma membranes. Membranes from CHO cells labeled with 13'1 by the lactoperoxidase method were run on 5.6% slab gels; the gels were stained for protein and prepared for autoradiography. This approach allows a direct comparison of the staining pattern and labeling pattern. Photographic negatives of the dried and stained gels and of the corresponding autoradiographs were scanned on a densitometer. Corresponding bands are indicated by symbols. (a) Radiograph; (b) coomassie blue stain. (See Juliano and Behar-Bannelier, 1975b.) (Reproduced with the permission of the American Chemical Society.)

age, whereas prolonged exposure to protein-free salt solutions results in a rapid loss of cell viability (Juliano and Behar-Bannelier, 1975a; Juliano, unpublished observations). Lactoperoxidase iodination of cells seems inherently a rather inefficient procedure; typically, about 1%of the total label is incorporated into trichloroacetic acid (TCA)precipitable cell-associated radioactivity (it should be noted that, even after the careful washing of labeled cells, a substantial amount of non-TCA-precipitable cell-associated radioactivity remains). Using the conditions described above, we found a level of labeling in the vicinity of 1 x lo5cpm of lz5Iper milligram of membrane protein to be usual (Juliano and Behar-Bannelier, 1975b), although higher levels

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

111

have been reported elsewhere (Hynes, 1973; Butters and Hughes, 1975).

2. GALACTOSEOXIDASE The most commonly used approach to labeling the carbohydrate moieties of the cell surface is the galactose o~idase-[~H]borohydride technique (Steck, 1974; Gahmberg and Hakomori, 1973a). In this system, neuraminidase is used to cleave sialic acid from cell surface glycoproteins; this exposes penultimate galactose residues which are then oxidized to aldehydes by the enzyme galactose oxidase. This is followed by reduction with [3H]borohydride, which regenerates hydroxyl groups and introduces a tritium label into surface galactose residues (see Fig. 2). Earlier studies employing galactose oxidase[3H]borohydrideto label erythrocyte membranes suggested that three

FIG.2. An electron microscope autoradiograph of murine leukemia L-1210 cells in which cell surface galactose and N-acetylgalactosamine moieties were isotopically labeled by NaB3H, reduction after neuramindase and galactose oxidase treatment according to Gahmberg and Hakomori (1973a). Note that the preponderance of silver grains is associated with the cell periphery. x 8500. (Courtesy of R. J. Bernacki and C. W. Porter, Department of Experimental Therapeutics, Roswell Park Memorial Institute, Buffalo, New York.)

112

R. 1. JULIANO

to four glycoproteins existed on the cell surface and that they could be detected by Iactoperoxidase labeling as well as by galactose oxidaseinduced labeling (Carraway, 1975). Recently, however, Gahmberg (1976)reported that at least 20 different glycoproteins can be detected on the surface of the human erythrocyte using galactose oxidase-induced labeling coupled with fluorography. The galactose ~xidase-[~H]borohydridelabeling system has also been applied to studies of the surfaces of normal and transformed fibroblasts (Gahmberg and Hakomori, 197313; Critchley, 1974) and of other cultured cells (Hunt and Brown, 1974; Juliano and Behar-Bannelier, 1975a,b). In an interesting recent study, Gahmberg and Hakomori (1975) probed the surface of hamster NIL cells by examining the effect of lectin binding on galactose oxidase-induced labeling. In using galactose oxidase, as in the case of lactoperoxidase, we have found it advantageous to use fairly large amounts of enzyme and isotope and to minimize the period of labeling so as to avoid loss of cell viability (Juliano and Behar-Bannelier, 1975a). A problem peculiar to techniques which employ [3H]borohydride is that this reagent may react with certain proteins without prior enzymic treatment; thus it is necessary to employ careful controls to ensure that it is indeed galactose residues and not other groups on proteins which are being labeled (Gahmberg, 1976). Two interesting variations on the surface labeling of carbohydrate residues have appeared recently. Itaya et al. (1975) used galactose oxidase to generate cell surface aldehyde moieties and then reacted them with [35S]methionine sulfone hydrazide. This approach, which enables one to introduce a very high-specific-activity 35S label into the cell surface, seems worthy of further development. Datta (1974) employed endogenous sialyltransferase activity to catalyze the incorporation of [14C]sialicacid into cell surfaces using [14C]CMP-sialic acid as a donor. Since there seems to be considerable evidence for the existence of cell surface glycosyltransferases in a variety of cells (Porter and Bernacki, 1975), the use of nucleotide sugars for labeling surface carbohydrate moieties may have quite general utility. A potential problem with this approach might be cleavage of the nucleotide sugars by surface glycosidases, followed by uptake and metabolic incorporation of the free sugar by internal cellular components. 3. TRANSGLUTAMINASES

An interesting new approach to cell surface modification is the use of transglutaminase enzymes which catalyze the formation of y-gluta-

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

113

myl-elysine crosslinks in proteins, or which can b e used to substitute a variety of primary amines onto protein glutamine residues (Dutton and Singer, 1975). Since glutaminases are high-MW entities, the application of these enzymes to intact cells results in cross-linking among surface-exposed proteins and glycoproteins or, in the presence of an excess of a nonpenetrating radioactive or fluorescent amino compound, serves as a surface-labeling system similar to lactoperoxidase. This approach has been most recently applied in studying the intermolecular associations of fibronectin [the large external transformation-sensitive (LETS) protein], a high-MW component of the surface of normal fibroblasts (Keski-Oja et al., 1976). In this case, the blood coagulation factor XIIIa (plasma transglutaminase) was the enzymic reagent employed. 4. IDENTIFICATION OF TRANSMEMBRANE PROTEINS USING SURFACE LABELS Surface labeling with low MW reagents and with lactoperoxidase has been used successfully to demonstrate the transmembrane orientation of certain erythrocyte membrane proteins (reviewed in Carraway, 1975). The general approach has been to label either intact cells or “leaky” membrane vesicles with the surface label reagent, to isolate the component of interest (for example, by cutting the band out of a polyacrylamide gel), and to subject the labeled component to proteolysis and “fingerprint” analyses of its peptides. If the component has a transmembrane orientation, then more peptides should be labeled when the labeling reagent is applied to leaky membranes, where both faces of the membrane are accessible, than when the labeling reagent is applied to intact cells (this is obviously an oversimplification; e.g., see Carraway, 1975). This approach has not been widely applied to cells more complex than erythrocytes and, with one exception, little information is available on the transmembrane orientation of proteins in nucleated animal cells. Hunt and Brown (1975) used lactoperoxidase iodination coupled with proteolytic dissection of “inside-out” membrane vesicles (see Section I1,B) to establish the transmembrane orientation in mouse L cells of a high MW protein which may be similar to fibronectin or the LETS protein.

5. EVALUATION OF SURFACE LABELTECHNIQUES Any surface label study must include an evaluation of the validity of the results. The fact that a particular reagent has been used success-

114

R. L. JULIANO

fully to identify surface macromolecules in one biological system is no guarantee that it will be equally useful in another system. The simplest criterion for deciding whether a particular reagent is truly labeling cell surface components and not labeling internal macromolecules is a comparison of the specific activity of plasma membrane and intracellular proteins after the labeling reaction. In erythrocytes, where it is quite easy to prepare highly purified plasma membranes and where there is an excellent intracellular marker (hemoglobin), the application of this criterion is quite simple. In more complex cell systems where it is more difficult to prepare pure plasma membrane fractions, this aspect of the evaluation is not trivial. Nonetheless, a comparison of the ratio of the activity per milligram of protein in a plasma membrane fraction to that in a cytoplasmic fraction seems to be essential. This ratio should be at least as great as the degree of purification of a known marker enzyme (e.g., Na+,K+-ATPase,or 5’-nucleotidase) in the plasma membrane fraction. One observation which compounds this problem is that some surface label reagents such as lactoperoxidase, for example, seem to prefer to label particulate components rather than soluble components (Juliano and Behar-Bannelier, 1975a). It should be recognized that, if a particular membrane component fails to react with a surface label reagent, this in itself does not constitute evidence that the component is lacking on the outer surface of the cell. In some systems certain outer membrane components are simply unreactive with particular surface labels because of the absence of appropriate reactive moieties or because of shielding phenomena. The best documented example of this is the lack of reactivity of equine erythrocyte glycoprotein with the lactoperoxidase reagent (Carraway et d.,1975). In a similar vein, we observed very substantial differences in the labeling patterns produced by different reagents in cultured hamster cells. Thus, while the major components visualized by lactoperoxidase or by galactose oxidase labeling are above 60,000 apparent MW, the major components visualized by pyridoxal phosphate -[3Hlborohydride labeling are in the 30,000-60,000 MW range (Juliano and Behar-Bannelier, 1975b). It should also be recognized that certain types of reagents may have an unusually high affinity for particular membrane components. Thus, although we originally found SITS and DIDS to be useful surface labels for mammalian cells (Juliano, 1974), we later discovered that in the CHO cell these compounds react primarily with a low-MW entity (<20,000) and scarcely react at all with the surface components visualized by lactoperoxidase and galactose oxidase (Juliano, unpublished observations). A very serious difficulty in the application of surface label tech-

T E C H N I Q U E S FOR THE ANALYSIS O F M E M B R A N E GLYCOPROTEINS

115

niques to nucleated mammalian cells is the loss of cell viability during the labeling reaction. Most mammalian cells are inherently unstable in the protein-free salt solutions used for labeling procedures. As cell death occurs, the interior of the cell becomes available to the putative surface label reagent, and internal labeling occurs (Juliano, 1974). Even if only a few percent of the total cell population dies during the labeling reaction, this may constitute quite a serious problem; the plasma membrane accounts for at most 5% of the total cell mass, hence the internal sites available for labeling in a few dead cells in the population may be as numerous as the surface sites available for labeling in the viable cells which constitute the bulk of the population (see Juliano and Behar-Bannelier, 1975a, for a more complete account). In order to deal with the above-mentioned difficulties, our laboratory has adopted a multifaceted approach to the identification of cell surface components in mammalian cells. We use the following criteria for the positive identification of putative cell surface'proteins and glycoproteins: (1)the component must be reactive with at least two surface label reagents of different chemical specificity; (2) the reactive component must be enriched in a purified plasma membrane fraction; (3)the component must be susceptible to proteolytic digestion of intact, viable cells. From our experience, careful application of these criteria usually results in a correct discrimination between outer surface and internal cellular components. It cannot be emphasized too strongly that careful cell fractionation should go hand in hand with surface label studies, and that the naive application of a surface label followed by solubilization and gel electrophoretic analysis of labeled whole cells does not constitute adequate proof of the surface localization of any given component.

B. Membrane Isolation

The preparation of purified plasma membranes is the usual starting point for the isolation and analysis of membrane glycoproteins. Techniques for subcellular fractionation and membrane isolation have been reviewed extensively elsewhere (DePierre and Karnovsky, 1973; Neville, 1975);in this article, we briefly discuss a few interesting new approaches to membrane purification. Highly purified membrane preparations have been obtained from cultured KB cells by allowing the cells to incorporate latex beads by phagocytosis and then disrupting the cells and isolating the membrane-coated beads (Charalampous et al., 1973).This technique yields

116

R. 1. JULIANO

inside-out vesicles which are apparently sealed; this fact has been exploited nicely by Hunt and Brown (1975)to demonstrate the existence of a transmembrane protein in L cells (see Section 11,A). Another important development has been the evolution of approaches for preparing sealed “transport-competent” membrane vesicles from cultured mammalian cells. Thus, Colombini and Johnstone (1974) prepared vesicles from Ehrlich ascites cells which retained a Na+-dependent amino acid transport capacity, while Hochstadt et al. (1975) prepared vesicles from mouse fibroblasts which engaged in group translocation of ribose. Studies on functionally competent vesicles of this type will undoubtedly be an important new approach in membrane biology. Preparations of inside-out versus “right-side-out” vesicles from erythrocyte ghosts have been of great utility in the red cell field (Steck, 1974). Recently, populations of inside-out and right-side-out vesicles from lymphoid cell membranes were separated based on the lack of affinity of inside-out vesicles for concanavalin (Con A)-Sepharose (Zachowski and Paraf, 1974; Walsh et al., 1976). This technique should be applicable to other cell types and, when used in conjunction with methods to “reseal” membrane vesicles, should be capable of providing interesting material for study. C. Metabolic labeling of Membrane Glycoproteins

Aspects of the metabolism of glycoproteins have been reviewed in detail elsewhere (Warren, 1972; Roseman, 1971), and by Sturgess et al. in this volume. In this section, we simply outline the use of certain precursors in labeling the polypeptide and carbohydrate portions of membrane glycoproteins. In most instances, the compound of choice for labeling cell polypeptides is [35S]methionine. This compound is available with very high specific activity (- 600 Ci/mmole), and the 35Slabel is relatively long-lived and can be conveniently detected by liquid scintillation or autoradiography. Generally, it seems best to incorporate [35Slmethionine under conditions where cellular protein synthesis is proceeding at a normal rate, and this requires the inclusion of “cold” methionine as well as normal amounts of other essential amino acids in the culture medium. In labeling rapidly growing cultured cells, such as CHO, we deem it important to allow the cell to go through at least one doubling time in the presence of isotope, so as to achieve uniform labeling of proteins; CHO cells grown in the presence of [35Slmethionineplus a small amount of “cold” methionine attain a very high specific activity

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

117

of labeling. Other essential amino acids, such as [3H]leucine, also have great utility and can be used in a manner similar to [35S]methionine. A variety of sugars has been used as a precursors for the carbohydrate moieties of glycoproteins, including 3H- and I4C-labeled man-, nose, glucosamine, fucose, and galactose. The metabolic incorporation of exogenous sugars into glycoproteins has been studied recently in several cultured cell systems,by Atkinson and his colleagues (Atkinson, 1975; Muramatsu et al., 1976; Yurchenco and Atkinson, 1975; Ceccarini et al., 1975), as well as by other groups (Kaplan and Moskowitz, 1975; Mattila et al., 1976). [3HlFucose is an excellent label in that it is incorporated almost entirely into glycoproteins, and the label is not distributed to other sugars (Yurchenco and Atkinson, 1975). A liability, however, is that fucose labeling is restricted to certain classes of carbohydrate chains (see Sturgess et al., this volume). Mannose and glucosamine are found almost universally in glycoproteins and are excellent labels for the core region (Ceccarini et al., 1975; Mattila et al., 1976); some conversion of mannose to fucose and of glucosamine to galactosamine and sialic acid is likely to occur (Hughes, 1976, p. 182). In our laboratory we successfully labeled CHO cell glycoproteins with [3Hlglucosamine,using an approach similar to that described for [35S]methioninelabeling, and attained levels of incorporation of 106-107 cpm/mg membrane protein.

111.

FRACTIONATION OF MEMBRANE GLYCOPROTEINS

A first step in the purification and analysis of membrane glycoproteins is to release these components from the membrane matrix in soluble form. There are two general approaches to the solubilization of intrinsic membrane glycoproteins, namely, the use of detergents and the use of denaturing agents such as organic solvents. A. Detergents

The properties of many commonly used detergents and principles for applying these detergents to membrane solubilization have been authoritatively reviewed by Helenius and Simons (1975).This section is in part a synopsis of their article. Detergents in common use can be divided into several broad categories in terms of their interactions with lipids, proteins, and membranes. These categories are (1)long chain ,anionic and cationic sur-

118

R.

1. JULIANO

factants, (2) nonionic surfactants, and (3) bile salts. These various types of detergents possess physical properties which govern their use in membrane solubilization. Thus, in general, long-chain charged surfactants and bile salts have rather high critical micelle concentrations (CMCs) at room temperature, whereas nonionic detergents have much lower CMCs. Bile salts form rather small micelles with MWs on the order of 1000-5000, while long-chain charged surfactants and nonionic surfactants have micellar sizes in the MW range of 10,000 to greater than 100,000. The CMC and micellar size of charged longchain surfactants and bile salts are very sensitive to alterations in pH and ionic strength, while these parameters less strongly affect the physical properties of nonionic surfactants. An interesting new type of nonionic detergent are the alkylglucosides described by Stubbs et al. (1976). All three types of surfactants can interact with, disrupt, and eventually solubilize lipid bilayers, although the mechanism of lipid solubilization may be quite different for bile salts than for the other types of surfactants (Helenius and Simons, 1975). It has been suggested that the interaction of increasing amounts of detergents with lipid bilayers occurs in three stages: stage I-detergent binding and alteration of bilayer properties; stage 11-bilayer-micellar phase transition; stage 111-size decrease of mixed micelles toward that of pure detergent micelles. The three categories of detergents differ markedly in the manner in which they interact with proteins. Long-chain charged surfactants (type I) frequently display'an initial saturable binding to proteins at low detergent concentrations followed by cooperative binding at higher detergent concentrations which involves unfolding and denaturation of the protein (Reynolds and Tanford, 1970; Nozaki et al., 1975). By contrast, nonionic surfactants and bile salts frequently do not induce the cooperative, denaturing binding mode, and thus these agents are more compatible with retention of the biological properties ofproteins (Makinoet al., 1973; Helenius and Simons, 1975). In many cases, protein-protein interactions are not disrupted by type I1 and I11 detergents, and thus many oligomeric membrane enzymes survive solubilization with these agents (for review, see Kagawa, 1972), or at least can be reconstituted upon the readdition of lipid (Kimelberg and Papahadjopoulos, 1972; Rubin and Tzagaloff, 1973). Recently, there have been several detailed studies on the binding of detergents to membrane glycoproteins, and on the hydrodynamic properties of the detergent-protein complexes; this type of analysis has been performed for sodium dodecyl sulfate (SDS) (Grefrath and

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

119

Reynolds, 1974), Triton X-100 (Clarke, 1975), and deoxycholate (Helenius et al., 1976).Of particular interest in these studies is the anomalous binding of detergents to membrane proteins and glycoproteins as compared to their water-soluble counterparts. In terms of the analysis and fractionation of membrane glycoproteins, the great utility of bile salts and nonionic surfactants stems from the fact that they are usually compatible with several important techniques for membrane protein analysis. These surfactants have been used successfully in analytical and separatory procedures, including: (1) lectin affinity chromatography (Allan et al., 1972; Gurd and Mahler, 1974; Schmidt-Ullrich et al., 1975; Nachbar et al., 1976; Nilsson and Waxdal, 1976);(2)immunoprecipitation and immune binding assays (Behar-Bannelier and Juliano, 1976; Zimmerman, 1974; Letarte-Muirhead et al., 1974; Juliano and Li, 1978);(3)antibody affinity chromatography (Letarte-Muirhead et al., 1975; Bridgen et al., 1976); and (4) ligand affinity chromatography (Hudgin et al., 1974; J. Callahan, personal communication). Despite these successful uses of affinity methods in the presence of detergents, the reader should keep in mind that the presence of detergents, even mild ones such as deoxycholate, can alter the biological properties of proteins. For example, Lotan et al. (1977) have reported that several commonly used detergents diminished the binding activities of lectins. Although high concentrations of type I surfactants such as SDS disrupt protein structure, there are several reports of affinity and antibody procedures being used in the presence of low (<0.5%)concentrations of this compound (Zanetta et al., 1975; Kahane et al., 1976). B. Denaturing Solvents

Various denaturing solvent systems have been used to extract the major membrane sialoglycoprotein from human erythrocytes. These include lithium diiodosalicylate (LIS) plus phenol (Marchesi and Andrews, 1971), chloroform-methanol mixtures (Hamaguchi and Cleve, 1972), and aqueous pyridine solutions (Zvilichovsky et al., 1971). These procedures probably result in the destruction.of most membrane enzyme activity and are more appropriate in studies directed toward the chemical analysis rather than the biological functions of membrane glycoproteins. Recently, extractions under denaturing conditions were used to solubilize glycoproteins from L cells (Hunt et al., 1975), milk fat globule membranes (Snow et al., 1977), and Erhlich ascites cells (Rittenhouse et al., 1976). An important point to note is

120

R. 1. JULIANO

that methods such as LIS -phenol extraction and chloroform-methano1 extraction solubilize some of the membrane glycoproteins but denature and precipitate others. Thus, in the erythrocyte, glycophorin is solubilized by the LIS-phenol method, but band 111, the other major membrane glycoprotein, is irreversibly denatured. In a like manner, only one of the two MW classes of L-cell glycoproteins survives LISphenol extraction (Hunt et al., 1975) and, while detergent extracts of Ehrlich cells contain five glycoprotein antigens, LIS extracts contain only three, indicating the selective loss of some glycoprotein components (Rittenhouse et al., 1976). It seems likely that extraction with LIS-phenol, chloroform-methanol, or pyridine preferentially solubilizes highly glycosylated components, perhaps those rich in sialic acid, while other membrane glycoproteins containing lower proportions of carbohydrate are precipitated and lost. A simple approach to the total solubilization and fractionation of membrane proteins under denaturing conditions is the use of fluorinated solvents, such as those routinely employed by peptide chemists. Substances such as hexafluoracetone readily solubilize membranes; the solubilized proteins can then be dialyzed into a more conventional denaturing solution, such as urea or guanidine hydrochloride, and fractionated by conventional means (Juliano, 1972).Direct application of guanidine hydrochloride or other chaotropic agents to membrane solubilization has not been very successful (for review, see Juliano, 1973). C. Fractionation of Membrane Glycoproteins by Column and Polyacrylamide Gel Techniques

Since the starting material for glycoprotein fractionation and isolation is usually a detergent-solubilized membrane preparation, some thought should be given to the choice of a detergent and of techniques for fractionation which are fully compatible. Clearly, fractionation techniques involving differences in molecular charge, such as isoelectric focusing and ion-exchange chromatography, usually require the use of a nonionic detergent. Gel filtration techniques, which fractionate on the basis of size, are probably better served by the use of bile salts which have micellar sizes that are small in relation to the size of membrane proteins, rather than by the use of nonionic detergents which have rather large micelles; this is especially true in cases where the protein of interest could interact with detergent in micellar form as well as with detergent monomers (Clarke, 1975). Another consideration is the stability of oligomeric protein complexes in detergents, in

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

121

relation to the goal of the fractionation scheme. If purification of individual peptides for the purpose of chemical analysis is desired, then use of a type I detergent which denatures protein and disrupts subunit interactions might be indicated. If the solubilization and fractionation of oligomeric complexes are desired, then use of nonionic surfactants may be the method of choice, since there are several reports of the preservation of native oligomeric assemblies from membranes in the presence of nonionic detergents (see, e.g., Yu and Steck, 1975a). Bile salts seem to fall somewhere between these two extremes, since although many proteins remain functional in the presence of these detergents, there are indications that some oligomeric assemblies may be disrupted (Helenius et al., 1976; Allan and Crumpton, 1971). Another important factor in the choice of a detergent and a fractionation scheme is the ease of removal of particular types of detergents. While bile salts are usually readily removed by dialysis, other detergents require more stringent measures. Type I detergents such as SDS can be removed from samples by ion-exchange techniques (Lenard, 197l), while nonionic detergents can be adsorbed onto certain types of bead matrixes (Hollaway, 1973).Frequently, one type of detergent can be readily exchanged for another type which is more appropriate to a particular fractionation step (Helenius and Simons, 1975).

1. GEL FILTRATION CHROMATOGRAPHY Detergent-solubilized membrane glycoproteins often can be fractionated by gel filtration. Thus Snow et al. (1977) used gel filtration in the presence of SDS in the purification of milk fat globule glycoprotein, while Letarte-Muirhead et al. (1974) employed gel filtration in deoxycholate as a step in the purification of Thy 1.1, a glycoprotein antigen, and Allan and Crumpton (1971)used a similar approach to resolve partially lymphocyte membrane proteins. Although, as mentioned above, nonionic detergents are not usually well suited to gel filtration, Furthmayr et al. (1975)used gel filtration in Ammonyx-Lo to separate erythrocyte sialoglycoproteins into two components.

2. ION-EXCHANGE CHROMATOGRAPHY This approach seems to b e of broad utility in separating detergentsolubilized membrane glycoproteins. Thus, Yu and Steck (1975b) used aminoethyl cellulose ion exchangers in the presence of Triton-X for the purification of human erythrocyte band 3. Ion exchange on DEAESephadex in Triton X-100 (Fukuda and Osawa, 1973),and on hydroxy-

122

R. 1. JULIANO

apatite with Tween 20 (Liljas et al., 1976), have been used in the purification of human erythrocyte sialoglycoproteins. Hydroxyapatite chromatography can be carried out even on SDS-solubilized samples, though the elution pattern may reflect the SDS binding capacity of the glycoproteins rather than their intrinsic ionic character (Snow et al., 1977; Moss and Rosenblum, 1972). In a novel approach to fractionation, Shami et al. (1977)exploited the fact that erythrocyte band 3 contains SH residues, while erythrocyte glycophorin does not, and separated these two species on an organomercurial gel in the presence of Triton X-100.

3. GEL ELECTROPHORESIS AND ISOELECTRIC FOCUSING Polyacrylamide gel electrophoresis under denaturing and reducing conditions has become a standard procedure in membrane protein research. Generally either continuous buffer systems in the presence of high (>1%) amounts of SDS (Fairbanks et al., 1971) or discontinuous buffer systems (Neville and Glossman, 1974) in the presence of low (0.1%) SDS are used. In our laboratory, the Fairbanks-type procedure is used routinely for the analysis of membrane glycoproteins (Juliano and Behar-Bannelier, 1975b; Juliano and Ling, 1976), as this procedure seems to avoid problems of glycoprotein aggregation and provides consistent and reliable analyses of high MW glycoproteins. Discontinuous gel systems, however, offer the advantage of higher resolution of the more rapidly migrating membrane species. Recently, systems for two-dimensional analysis of membrane proteins and glycoproteins have been developed, and these promise to enhance greatly the resolving power of polyacrylamide gel techniques. One interesting approach involves isoelectric focusing of membrane proteins in the presence of nonionic detergents and of urea in one dimension, followed by SDS gel electrophoresis in the other dimension (Ames and Nikaido, 1976; Bhakdi et al., 1975); this technique has proven to have high resolution when applied to bacterial and erythrocyte membranes. Another approach involves using electrophoretic separation under denaturing conditions in both dimensions (Anselstetter and Horstmann, 1975; Conrad and Penniston, 1976; Wang and Richards, 1974). D. Lectin Affinity Techniques

The study of cell surface glycoproteins has been considerably facilitated by the use of lectins, the carbohydrate-binding proteins derived

T E C H N I Q U E S F O R THE ANALYSIS O F M E M B R A N E GLYCOPROTEINS

123

most commonly from plant sources. The properties of lectins and their applications in cell biology have been ably discussed in several recent reviews (Lis and Sharon, 1973; Nicolson, 1974; Rapin and Burger, 1974; Brown and Hunt, 1978). This portion of the chapter outlines some of the basic concepts of lectin biochemistry and, in light of these concepts, discusses the applications of lectins to the analysis and fractionation of membrane glycoproteins. Lectins are usually oligomeric proteins with several carbohydratebinding sites; that is, they are multivalent molecules. The interactions of lectins with soluble or membrane-bound glycoproteins can be interdicted by the presence of appropriate sugars known as haptenes which compete for or modify the carbohydrate-binding sites. The haptene sugar specificity displayed by lectins is quite remarkable (Goldstein, 1975), nonetheless it seems clear that the interaction of lectins with glycoproteins involves more than just binding to terminal sugar residues corresponding to specific haptenes. In all probability, lectins bind to complex sequences of sugars in glycoproteins, and thus the details of the sugar linkages and even the nature of the associated polypeptide may be of great importance in lectin-glycoprotein binding phenomena (Kornfeld and Kornfeld, 1970; Young and Leon, 1974; Goldstein, 1975). The affinity of lectins for certain oligosaccharides and/or glycopeptides is often several orders of magnitude greater than their affinity for the most potent free haptene sugar (Young and Leon, 1974; Adair and Kornfeld, 1974; Nagata and Burger, 1974). In a similar vein, it now seems clear that lectins with similar haptene specificities may have vastly different affinities for particular glycoprotein species (Young and Leon, 1974). For example, erythrocyte glycophorin does not bind to Con A but binds to Lens cu2inaris lectin, even though mannosides are appropriate haptenes for both lectins (Findlay, 1974). The biochemical characteristics of the cell surface receptors for lectins are in most cases unknown. A point of interest is whether the binding sites for different lectins reside on different macromolecular species, or whether the same molecule has binding sites for several types of lectins; in all probability, both cases exist in many biological membranes. In the human erythrocyte it seems clear that the glycoproteins which bind Con A, wheat germ agglutinin (WGA), and Ricinus agglutinin represent three distinct populations of macromolecules (Adair and Kornfeld, 1974; Findlay, 1974; Triche et al., 1975). Similarly, it has been proposed that at least some of the receptors for pea lectin and for Con A may reside on different species in murine tumor cells (Allen and Johnson, 1976). However, it is clear that many glycoproteins have the capacity to bind to several lectins; thus erythrocyte gl ycophorin has been reported to bind to WGA, phytohem-

124

R. 1. JULIANO

agglutinin, and other lectins (Fukuda and Osawa, 1973), while the major external glycoprotein of normal fibroblasts (the LETS protein) binds Ricinus lectin as well as Con A (Gahmberg and Hakomori, 1975).The existence on cell surfaces of two or more classes of binding sites with different affinities for lectins has been reported in several instances (Allen and Johnson, 1976; Schmidt-Ullrich et al., 1976; Sandvig et al., 1976; Cuatrecasas, 1973); multiple classes of cellular binding sites may imply the existence of populations of receptors with different affinities for a particular lectin. It has also been reported that, at least in some cases, the binding of lectins to cellular receptors may be treated as a reversible equilibrium phenomenon (Sandvig et al., 1976).As we shall see in Section III,D,2, it may be possible to discriminate between different classes of lectin receptors based on their different equilibrium binding affinities. In addition to their capacity to bind to specific carbohydrate moieties, some lectins have the potential to bind to cell constitutents in a manner which does not involve the carbohydrate recognition sites. Thus Con A can engage in nonspecific interactions which are not reversed by haptene treatment (Goldstein, 1975). Con A coupled to affinity columns was also reported to bind proteins via hydrophobic interactions, although these interactions could be minimized by appropriate immobilization of the lectin onto the affinity support (Davey et al., 1976). 1. LECTINAFFINITY CHROMATOGRAPHY

The most popular approach to the isolation of membrane glycoproteins has been the use of lectin affinity chromatography. Solubilized cells or membrane fractions are passed over columns containing lectins bound to an inert matrix; glycoproteins are bound to the lectin columns, while nonglycosylated proteins pass through; the glycoproteins can then be eluted with an appropriate haptene sugar. This approach has been used to purify partially a variety of glycoproteins from several cellular sources including Ricinus, Con A, wheat germ, and soybean receptors from Ehrlich ascites cells (Nachbar et ul., 1976; Rittenhouse et al., 1976), Con-A receptors from L cells (Hunt and Brown, 1975), Con-A receptors from thymocytes (Letarte-Muirhead et al., 1975; Schmidt-Ullrich et al., 1975), lentil receptors from KB cells (Butters and Hughes, 1975), lentil and WGA receptors from brain (Zanetta et uZ., 1975; Gurd and Mahler, 1974), and a variety of lectin receptors from L-1210 cells (Janson and Burger, 1973). (We use the term “receptor” to denote glycoprotein species which bind particular

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

125

lectins.) Various groups have used lectin affinity columns for the partial purification of glycoproteins from lymphocyte membranes (Nilsson and Waxdal, 1976; Henkart and Fisher, 1975;Allan et al., 1972; Bridgen et al., 1976). The first step in any lectin affinity technique is the preparation of pure, biologically active lectins. Although several lectins are now available commercially, they frequently require additional purification before they are suitable for use. Techniques for the isolation of a variety of lectins have been tabulated in a review by Brown and Hunt (1977),and a large number of isolation procedures has also been described in the volume edited by Ginsberg (1972). To form affinity columns, lectins are usually coupled to an inert supporting matrix. The most commonly used matrix thus far has been cyanogen bromide-activated agarose; however, a wide variety of other supports is possible and their use may in fact be advantageous, as several workers have suggested the possibility of nonspecific hydrophobic interactions between proteins and lectins immobilized on cyanogen bromide agarose (Davey et al., 1976; Nachbar et al., 1976). Lotan et a2. (1977)suggest the use of polyacrylic hydrazide agarose as an alternative matrix; however, various other supports and coupling reactions are available and have been described in detail elsewhere (Wilchek and Jakoby, 1974). There are several important parameters to consider in choosing conditions for affinity chromatography. These include the amount of lectin per unit volume of matrix, the load to be applied to the column in relation to the amount of lectin present, and the conditions for specific haptene-mediated elution of the glycoproteins. While lectin affinity chromatography has been notably successful as a means of separating glycoproteins from nonglycosylated macromolecules, it has not thus far proven to b e of great value in subfractionation of the various glycoproteins likely to be present on cell surfaces. Thus, when affinity columns bearing different lectins were used to fractionate glycoproteins from Ehrlich ascites cells (Nachbar et al., 1976), CHO cells (Gottlieb et al., 1975a), or brain cells (Gurd and Mahler, 1974), the fractions eluted from columns bearing different lectins had many glycoprotein species in common. Although one laboratory (Hunt et al., 1975) has reported the isolation of a purified Con-A receptor glycoprotein from L cells by affinity chromatography, an important part of the purification may stem from the preferential extraction of this component from the membrane rather than from any selectivity in the affinity step. Adair and Kornfeld (1974)pointed out the importance of column load in lectin affinity chromatography; on overloaded columns, they sug-

126

R.

1. JULIANO

gest, high-affinity species bind preferentially to the immobilized lectin, while low-affinity species pass throughthe column. While in most cases the lack of appropriate data precludes an estimation of column loads, it seems likely that many studies on the glycoproteins of nucleated cells have utilized rather scarce membrane material and have probably involved running affinity columns at low membrane protein/lectin ratios. This situation favors the binding of both high- and low-affinity receptors to the column and minimizes the opportunities for useful subfractionation of glycoproteins. A constant problem with affinity chromatography has been the low recovery of material which can be specifically eluted from the columns with haptene (Hunt et d . , 1975; Allan et d., 1972). This phenomenon has been attributed to either hydrophobic interactions with the immobilized lectins (Davey et al., 1976) or to denaturation of the lectins because of the use of inappropriate detergents (Kahane et al., 1976; Lotan et al., 1977).

2. LECTINSTAINING OF GELS Recently, an interesting new approach to the analysis of cellular glycoprotein receptors for lectins has been introduced. In this approach, the glycoproteins are solubilized and run on a polyacrylamide gel under denaturing and reducing conditions; after fixation, the gel is incubated with 1251-labeledlectin. The lectin binds to glycoproteins fixed in the gel, and unbound lectin can be removed by simple washing procedures; the sample is then analyzed by slicing and counting or by radioautography. This approach has been used to study glycoproteins from erythrocytes (Tanner and Anstee, 1975; Robinson et al., 1975), from liver and brain cells (Gurd and Evans, 1976; Gurd, 1977), and from normal and transformed fibroblasts (Burridge, 1976). This technique, while not useful for the purification of lectin receptors, possesses several advantages over affinity chromatography in terms of analytical uses. First, a much larger number of samples can ,be rapidly analyzed, a consideration of some importance in cell biology studies where one might wish to examine the effects of genetic alterations, growth conditions, or viral transformation on the pattern of membrane glycoproteins. Another advantage is that the lectin “staining” technique seems to be somewhat more specific than a f h i t y chromatography in that clear differences in the staining patterns for different lectins are apparent even in complex cell types (Gurd, 1977; Burridge, 1976). One disadvantage of this approach is that it allows one only to examine isolated, denatured glycoprotein species. This

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

127

approach is not of use in studies directed toward the elucidation of oligomeric assemblies in membranes or of possible linkages between membrane glycoproteins and other membrane or cytoskeletal constituents (Nicolson, 1976). This point is emphasized by the studies of Gurd who found that, whereas in intact membranes various lectin receptors had specific topographical relations with each other which resulted in competition between lectins for binding sites, isolated, solubilized glycoproteins failed to display such competitive interactions with lectins (Gurd, 1977).

3. LECTINIMMUNOPRECIPITATION In our laboratory, we recently developed a technique which should be useful for discriminating different subpopulations of membrane glycoproteins and for studying the relationship between surface glycoproteins and other membrane constittitents (Juliano and Li, 1978). This approach, which we have termed lectin immunoprecipitation, begins with the solubilization of lectin receptors from the membrane by means of nonionic detergents or bile salts. The solubilized membrane glycoproteins are then allowed to react with lectins, and the lectin-glycoprotein complexes are precipitated by the addition of an antibody directed against the lectin. The antibody in this system serves merely to ensure the rapid and efficient precipitation of the lectin-receptor complexes. The-precipitates are then solubilized and run on SDS polyacrylamide gels for analysis. This technique, like lectin staining, is not suitable for preparative purposes, but seems well adapted for analytical work. As with the lectin staining described above, it allows one to examine rapidly a large number of samples simultaneously. However, we feel that lectin immunoprecipitation offers some advantages over lectin staining in that the former can be used in connection with chemical cross-linking (Ji, 1976), or other approaches, to study the relationship of glycoproteins to other components in the membrane. In addition, one can use lectin immunoprecipitation to attempt to discriminate high-affinity and low-affinity receptors for any given lectin by varying the ratio of lectin to solubilized glycoprotein in the incubation mixture. We used this technique with some success in the analysis of CHO cell membrane glycoproteins. The CHO cell has three major classes of membrane glycoproteins with approximate MWs of 50,000-65,OOO, 100,000, and 130,000 (these classes are, no doubt, heterogeneous in nature), plus several minor components. When lectin immunoprecipitations of CHO membrane extracts are performed with Con A or WGA

128

R. 1. JULIANO

-RCA

0

01

02

03

01

05

Ob

1

07

08

d9

RELATIVE MOBILITY

FIG.3. Lectin immunoprecipitation of CHO cell membrane glycoproteins. Deoxycholate extracts of membranes from CHO cells metabolically labeled with [3H]glucosamine were treated with Con A, WGA, or R. cmnmunis agglutinin I (RCA I), and the lectin-glycoprotein complexes were precipitated with antisera directed against the lectin. The immunoprecipitates were run in SDS-containing polyacrylamide gels, and the distribution of radioactivity was analyzed by fluorescence-enhanced autoradiography and densitometry. (See Juliano and Li, 1978.)Approximately equal amounts of radioactivity were used for each gel.

at high lectin/membrane glycoprotein ratios, all the major glycoprotein species of the CHO membrane are precipitated. However, at low lectin/glycoprotein ratios, we found that Con A precipitated primarily the 130,000-MWclass of glycoproteins, while wheat germ lectin precipitated primarily the 100,000-MW class, suggesting that these may represent populations with different affinities for the two lectins (see Fig. 3). The lectin immunoprecipitation approach is quite versatile, and it seems likely that more subtle use of this technique will be possible in future endeavors. E. Other Approaches to Membrane Glyeoprotein Fractionation

Some workers have approached the problem of membrane glycoprotein analysis by stripping glycoproteins from intact cells with trypsin, further digesting them with pronase, and analyzing the resultant small glycopeptides by Sephadex chromatography (Buck et al., 1970; Glick et ul., 1973). This approach has been reviewed in detail elsewhere (Glick, 1974) and is not dealt with further here. Immunochemical approaches to the analysis of cell surface glycoproteins seem to offer an impressive degree of specificity and,sensitivity. These approaches, as applied to well-defined antigens of lymphoid cells, have been reviewed in this volume by Letarte. In the present discussion, however, it seems appropriate to mention work

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

129

which has been done with less well-defined antisera prepared against whole membranes or against complex glycoprotein extracts, since even these relatively crude tools can be used profitably. Bjerrum and Bag-Hansen (1976) analyzed the proteins and glycoproteins of the human erythrocyte membrane using crossed immunoelectrophoresis in the presence of a nonionic detergent and employing a serum raised against whole erythrocyte membranes. The technique offers a high degree of resolution and is compatible with histochemical assays for membrane enzymes because of the nondenaturing quality of the detergent used. This approach was also employed by Schmidt-Ullrich et uZ. (1975). In a different vein, Behar-Bannelier and Juliano (1976) used an antiserum against intact CHO cells as a surface probe. They determined that certain membrane proteins immunoprecipitated by native antiserum were not precipitated by antiserum previously adsorbed with intact cells; the same proteins were also reactive with the lactoperoxidase surface label, thus providing two independent tests for the cell surface localization of these particular membrane proteins. Rittenhouse et al. (1976) also used immunochemical approaches in the analysis of Con-A-binding glycoproteins from Ehrlich ascites cells.

IV.

CHEMICAL ANALYSIS OF MEMBRANE GLYCOPROTEINS

After purification of a membrane glycoprotein to the point where it is apparently homogeneous by commonly accepted criteria (SDS gel electrophoresis, for example), it is possible to apply a vast repertoire of techniques derived from protein chemistry and from carbohydrate chemistry to an analysis of the structural details of the macromolecule. A. Preparation of Glycoproteins and Oligosaccharides

A first step in the analysis of the structure of the carbohydrate portion of a glycoprotein is to digest the macromolecule exhaustively with a powerful protease such as pronase so as to obtain oligosaccharide residues with a minimum number of amino acids attached (Spiro, 1973a; Sharon, 1975). The resultant glycopeptides can then be analyzed by gel filtration or ion-exchange chromatography. This type of approach has been applied recently to the analysis of gl ycopeptides and oligosaccharides derived from a liver cell membrane binding glycoprotein (Kawasaki and Ashwell, 1976) and to milk fat globule glycoproteins (Harrison et al., 1975).

130

R. 1. JULIANO

Oligosaccharides, free of amino acids, can be released from O-glycosidic linkages (N-acetylgalactosamine to serine or threonine) by alkaline hydrolysis. When this is done in the presence of borohydride, the N-acetylgalactosamine is converted to N-acetylgalactosaminitol, while the serine and threonine are converted to 2-aminoacrylic and 2-aminocrotonic acid, respectively; these changes may be useful in determining the nature .of the carbohydrate-protein linkage (Sharon, 1975). The oligosaccharides linked N-glycosidically to proteins ( N acetylglucosamine to asparagine) can also be cleaved by alkaline borohydride under more extreme conditions (Sharon, 1975),but an alternative approach is to release core oligosaccharides from such linkages by the use of endoglycosidases (Kohno and Yamashina, 1973; Koide and Muramatsu, 1974; Takasaki and Kobata, 1976). B. Analysis of the Polypeptide Portion of Glycoproteins

Amino acid analysis of membrane glycoproteins by conventional means seems quite routine and has been performed on several species including liver cytochrome b, (Ozols, 1972), erythrocyte glycophorin (Furthmayr et a1., 1975), liver asialoglycoprotein binding protein (Hudgin et aZ., 1974),Con-A receptor from L cells (Hunt et al., 1975), and milk fat globule sialoglycoprotein (Snow et al., 1977). Likewise, N-terminal analysis of membrane glycoproteins seems to present no great problems (Hunt et al., 1975), although several membrane proteins have been reported to have blocked N-termini (see Tanner, this volume). A recently developed technique, which should prove to have great utility for the analysis of membrane protein, involves proteolytic digestion of protein in the presence of SDS, followed by peptide mapping on polyacrylamide gels (Cleveland et al., 1977). Only a very limited number of intrinsic membrane glycoproteins has been subjected to complete or partial sequence analysis. Erythrocyte glycophorin has been fully sequenced by conventional means (manual Edman degradations) by Tomita and Marchesi ( 1975), while cytochrome b, has been partially sequenced (Ozols, 1972).Sequences of portions of influenza virus hemagglutinin and of its precursor have also been determined and have revealed an interesting amino acid palindrome at the site of cleavage of the precursor molecule (Skehel and Waterfield, 1975). Tomita and Marchesi have discussed some of the problems associated with the determination of glycosylated amino acids in the glycophorin sequence. Recently, Bridgen et al. (1976)employed a solid-phase sequencing technique to determine the first 16 residues in human lymphoid cell HL-A and HL-B antigens, using 1

-

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

131

nmole of protein eluted from a polyacrylamide gel. This approach and other microsequencing techniques promise to revolutionize structural determinations for membrane proteins which are difficult to isolate in quantity (Hood, 1977). For a further discussion of microsequencing, see Letarte, this volume. C. Analysis of Individual Sugars

Determination of total hexose and of certain individual sugars, such as fucose and sialic acid in glycoproteins, can be made by colorimetric methods (Dische, 1962; Ashwell, 1966). However, complete analysis of individual sugars requires the release of monosaccharides from glycosidic linkages by means of acid hydrolysis. Since the stability of various monosaccharides in acids, and their reactivity with peptides and amino acids, vary so greatly, the problem of choosing appropriate conditions for hydrolysis is not a trivial one (Sharon, 1975). After appropriate hydrolysis, various methods are available for the determination of individual neutral and amino sugars. The most powerful method is probably gas-liquid chromatography of volatile sugar derivatives (Esselman et al., 1973; Dutton, 1973; Porter, 1975). The most common used volatile forms of sugars are trimethylsylyl derivatives and alditol acetates, although the preparation of methyl glycosides is sometimes advantageous, since this type of derivatization allows the analysis of neutral sugars, hexosamines and sialic acid on a single column (Porter, 1975). Specific neutral sugars can also be determined by enzymic means (Finch et al., 1969), while hexosamines can be resolved with an amino acid analyzer (Spiro, 1973a). D. Determination of the Carbohydrate Structure of Glycoproteins

Analysis of the sequence and linkages of the complex oligosaccharide side chains of glycoproteins is quite a challenging undertaking. The two most powerful approaches to this problem are the use of specific exoglycosidases to cleave particular sugars from the oligosaccharide in a sequential fashion, and the use of periodate oxidation (Smith degradation) to obtain details of structure (Spiro, 1973a; Sharon, 1975). A large number of glyosidases has been purified from plant and animal sources; the preparation of many of these enzymes is detailed in the treatise edited by Ginsberg (1972). Some of these enzymes are now commercially available (tabulated in Sharon, 1975), but many of

132

R. 1.

JULIANO

these products probably require repurification. Before employing glycosidases for the analysis of glycoproteins, it is vital to ascertain their specificity using synthetic substrates. Smith degradation (Goldstein et a1., 1965), involving periodate oxidation, borohydride reduction, and acid hydrolysis, is an extremely useful method for the structural analysis of oligosaccharides. Quantitation of the periodate used, the amount of formaldehyde produced, and the structure of the degradation products gives detailed information on the sequence of sugar residues in the original oligosaccharide. Since analyses using glycosidases or Smith degradation are usually performed on glycopeptides or oligosaccharides rather than on intact membrane glycoproteins, most of the techniques developed for carbohydrate sequencing of soluble glycoproteins can also be applied to membrane glycoproteins. There exist in the literature sophisticated analyses of the carbohydrate structure of soluble proteins, including that of IgE (Baenziger and Kornfeld, 1974),thyroglobulin (Arima and Spiro, 1972), and ovalbumin (Lee and Scocca, 1972). The carbohydrate sequences for several membrane glycoproteins have now also been partially or fully established. The structure of the N-glycosidically and O-glycosidically linked carbohydrates of human erythrocyte glycophorin were established b y Thomas and Winzler (1969, 1971) and by Kornfeld and Kornfeld (1970). Kawaski and Ashwell (1976) used a combination of glycosidase treatment and Smith degradation to establish the structure of two glycopeptides derived from hepatic binding protein. Codington et al. (1975)employed Smith degradation, coupled with the use of specific lectins to determine partially the carbohydrate structure of a major surface glycoprotein from murine mammary carcinoma cells. An interesting preliminary approach to the carbohydrate structure of Semliki Forest virus glycoproteins was that of Mattila et al. (1976),who used the incorporation of different radiolabeled sugars to reveal two classes of glycopeptides in the virus glycoproteins. Recently there has been a great deal of interest in the carbohydrate structure of viral glycoproteins (Atkinson et al., 1976); this topic is reviewed in more detail by Compans in this volume.

E. MWr of Membrane Glycoproteins

A major problem of membrane glycoprotein analysis has been the inability to determine correct MWs for these macromolecules. Polyacrylamide gel electrophoresis in the presence of SDS has been a rapid and convenient tool for establishing the polypeptide MWs of soluble and nonglycosylated proteins; however, the binding of SDS

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

133

by glycoproteins and by membrane proteins is anomalous (Grefrath and Reynolds, 1974), thus invalidating the SDS gel technique for these classes of macromolecules. A helpful empirical modification of the SDS gel technique for glycoproteins makes use of the observation that the apparent MW of a glycoprotein converges to a fixed value in gels of increasing acrylamide content (Segrest and Jackson, 1973). However, this technique has no firm theoretical foundation and has been validated for only a limited number of glycoproteins. More recently, the problem of establishing true MWs for detergent-solubilized membrane glycoproteins has been approached through the use of analytical ultracentrifugation, and some reliable techniques having solid biophysical foundations have been developed (Clarke, 1975; Reynolds and Tanford, 1976). The problem of the physical characterization of detergent-solubilized membrane proteins has been reviewed in detail by Tanford and Reynolds (1976). V.

GENETIC ANALYSIS OF MEMBRANE GLYCOPROTEINS

An interesting new approach is the use of somatic cell genetics to explore the structure and biosynthesis of membrane glycoproteins. Several workers used lectins as toxic agents in the selection of lectinresistant clones of cultured mammalian cells. This has been done primarily in CHO, BHK, and mouse lymphoma cells for the lectins phytohemagglutinin (Stanley et al., 1975a), Ricinus agglutinin (Gottlieb et al., 1975a; Hyman et al., 1974; Meager et al., 1975), and Con A (R. Baker, personal communication). Several lectin-resistant clones display a markedly reduced lectinbinding capacity (Stanley et aZ., 1975a; Gottlieb et aZ., 1975) and altered cell surface glycoproteins (Juliano and Stanley, 1975; Gottlieb et al., 1975). In some cases, however, reduced sensitivity to and binding capacity for one lectin are accompanied by increased sensitivity to other lectins (Stanley et al., 1975b; Briles et al., 1977). Several lectinresistant clones have been reported to lack specific glycosyltransferase activities whose absence can account for the observed phenotypes (Stanley et al., 1975c; Gottlieb et al., 1975; Meager et al., 1975). Mutants of this type may b e extremely useful in elucidating the details of glycoprotein biosynthesis. Various workers have also reported lectin-resistant clones which fail to display marked differences from parental cells in terms of lectin binding or in patterns of surface glycoproteins (Briles et al., 1977; Stanley et al., 1975b). The lesion in the latter clones may involve modification of the uptake of lectins or modification of the regulation of

134

R. 1. JULIANO

glycoprotein biosynthesis. Besides providing a new tool for the study of membrane glycoprotein biosynthesis, the various classes of lectinresistant clones of mammalian cells should prove extremely valuable in studying such diverse phenomena as viral replication (Schlesinger et al., 1976), cell adhesion (Edwards et al., 1976; Juliano, 1978), and the action of toxins. In a similar vein, several workers have isolated mammalian cell variants defective in the expression of glycoprotein membrane antigens (Hyman and Stallings, 19’16; Rajan et al., 1976). Another interesting example of the relationship between membrane biochemistry and cell genetics is the case of sublines of mouse tumors with differences in cell surface carbohydrates which parallel differences in malignancy (Sanford et al., 1973; Shin et al., 1975; Codington et al., 1975). ACKNOWLEDGMENTS I especially thank Mrs. Doris Wills for invaluable assistance in the preparation of this chapter. In addition, I am grateful to the following for generously providing reprints or preprints of articles in press: Jay Brown, Ralph Bernacki, Stuart Komfeld, Garth Nicolson, Gilbert Ashwell, S. I. Hakomori, Paul Atkinson, and James Curd. This work was supported by the Medical Research Council of Canada and the National Cancer Institute of Canada. REFERENCES Adair, W., and Kornfeld, S. (1974).Isolation of the receptors for wheat germ agglutinin and the Ricinus communis lectins from human erythrocytes using afEnity chromatography. ]. Biol. Chem. 249,4696-4704. Allan, D., and Crumpton, M. J. (1971).Solubilization of pig lymphocyte plasma membrane and fractionation of some of the components. Biochem. 1. 123,967-975. Allan, D., Auger, J., and Crumpton, M. J. (1972).Glycoprotein receptors for Con A isolated from pig lymphocyte plasma membrane by affinity chromatography in sodium deoxycholate. Nature (London),New Biol. 236,23-25. Allen, H. J., and Johnson, E. A. Z. (1976).Studies on 6C3HED murine ascites tumor cell receptors for mannosyl binding lectins. Biochim. Biophys. Acta 436,557-566. Ames, C . E.-L., and Nikaido, K. (1976).Two-dimensional gel electrophoresis of membrane proteins. Biochemistry 15,716-723. Anselstetter, V., and Horstmann, H.-J. (1975). Two-dimensional polyacrylamide-gel electrophoresis of the proteins and glycoproteins of the human erythrocyte. Membr. Eur. J. Biochem. 56,259-269. Arima, T., and Spiro, R. G. (1972).Studies on the carbohydrate units of thyroglobulin. Structure of the mannose-N-acetylglucosarnineunit (unit A) of the human and calf proteins. I. Biol. Chem. 247, 1836-1848. Ashwell, G. (1966). New colormetric methods of sugar analysis. In “Complex Carbohydrates” (E. Freufeld and V. Ginsberg, eds.), Methods in Enzymology, Vol. VII, pp. 85-95. Academic Press, New York. Atkinson, P. H. (1975).Synthesis and assembly of HeLa, cell plasma membrane glycoproteins and proteins. 1.B i d . Chem. 250,2123-2134.

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

135

Atkinson, P. H., Moyer, S. A., and Summers, D. F. (1976).Assembly of vesicular stomatitis virus glycoprotein and matrix protein into HeLa cell plasma membranes. J . Mol. Biol. 102,613-631. Baenziger, J., and Kornfeld, S. (1974).Structure of the carbohydrate units of IgE immunoglobulin. I. Over-all composition, glycopeptide isolation and structure of the high mannose oligosaccharide unit. J . Biol. Chem. 249, 1889-1896. Behar-Bannelier, M., and Juliano, R. L. (1976).Surface polypeptides of the Chinese hamster ovary cell: An immunochemical study. Can.J . Biochem. 54,185-191. Bhakdi, S . , Knufermann, H., and Wallach, D. F. H. (1975).Two-dimensional separation of erythrocyte membrane proteins. Biochim. Biophys. Acta 397,550-557. Bjermm, 0. J., and BBg-Hansen, T. C. (1976).The immunochemical approach to the characterization of membrane proteins: Human erythrocyte membrane proteins analysed as a model system. Biochim. Biophys. Acta 455,66-89. Bretscher, M.S. (1971).A major protein which spans the human erythrocyte membrane. J. Mol. Biol. 59,351-357. Bridgen, J., Snary, D., Crumpton, M. J., Barnstable, C., Goodfellow, P., and Bodner, W. F. (1976).Isolation and N-terminal amino acid sequence of membrane-bound human HLA-A and HLA-B antigens. Nature (London)261,200-205. Briles, E.B., Li, E., and Kornfeld, S. (1977).Isolation ofwheat germ agglutinin-resistant clones of Chinese hamster ovary cells deficient in membrane sialic acid and galact0se.J. Biol. Chem. 252, 1107-1116. Brown, J. C., and Hunt, R. C. (1978).Lectins. Int. Reu. Cytol. In press. Buck, C. A., Glick, M. S., and Warren, L. (1970).A comparative study of glycoproteins from surface control and Rous sarcoma virus transformed hamster cells. Biochemistry 9,4567-4576. Burridge, K. (1976).Changes in cellular glycoproteins after transformation: Identification of specific glycoproteins and antigens in sodium dodecyl sulfate ge!s. Proc. Natl. Acad. Sci. U.S.A. 73,4457-4461. Butters, T. D., and Hughes, R. C. (1975).Surface labelling of human tumor KB cells: Iodination and fractionation of membrane glycoproteins. Biochem. J . 150, 59-

69. Cabantchik, Z.I., and Rothstein, A. (1974).Membrane proteins related to anion permeability of human red blood cells. I. Localization of disulfonic stilbene binding sites in proteins involved in permeation. J . Membr. Biol. 15,207-226. Cabantchik, Z.I., Knauf, P. A., Ostwald, T., Markus, H., Davidson, L., Breuer, W., and Rothstein, A. (1976).The interaction of an anion photoreactive probe with the anion transport system of the human red blood cell. Biochim. Biophys. Acta 455, 526-

537. Carraway, K. L. (1975).Covalent labeling of membranes. Biochim. Biophys. Acta 415,

479-410. Carraway, K.L., Colton, D. G., Shin, B. C., and Triplett, R. B. (1975).Species variability in the modification of erythrocyte surface proteins by enzymatic probes. Biochim. Biophys. Acta 382, 181-192. Ceccarini, C., Muramatsu, T., Tsawg, J., and Atkinson, P. (1975).Growth dependent alterations in oligomannosyl cores of glycopeptides. Proc. Natl. Acad. Sci. U.S. 72,

3139-3143. Charalampous, F. C., Gonatas, N. K., and Melbourne, A. D. (1973).Isolation and properties of the plasma membrane of KB ce1ls.J. Cell Biol. 59,421-435. Clarke, S. (1975).The size and detergent binding of membrane proteins.]. Biol. Chem.

250,5459-5469. Cleveland, D. W., Fischer, S. G., Kirschman, M. W., and Laemmli, V. K. (1977). Peptide

136

R. 1. JULIANO

mapping by limited proteolysis in sodium dodecyl sulfate and analysis by gel electrophoresis. J. Biol. Chem. 252, 1102-1106. Codington, J. F., Linsley, K. B., and Jeanloz, €3. W. (1975).Immunochemical and chemical investigations of the structure of glycoprotein fragments obtained from epiglycanin, a glycoprotein at the surface of the TA3-Ha cancer cell. Carbohydr. Res. 40, 171-182.

Colombini, M., and Johnstone, R. M. (1974). Na+-dependent amino acid transport in plasma membrane vesicles from Ehrlich ascites ce1ls.J. Membr. Biol. 15,261-276. Conrad, M. H., and Penniston, J. T. (1976). Resolution of erythrocyte membrane proteins by two-dimensional electrophoresis. J. Biol. Chem. 251,253-255. Cook, G. M. W., and Stoddart, R. W. (1973). “Surface Carbohydrates of the Eucaryotic Cell.” Academic Press, New York. Critchley, D. R. (1974). Cell surface proteins of NIL1 hamster fibroblasts labeled by a galactose oxidase, tritiated borohydride method. Cell 3, 121-125. Cuatrecasas, P. (1973). Interaction of wheat g e m agglutinin and concanavalin A with isolated fat cells. Biochemistry 12, 1312-1323. Datta, P. (1974). Labeling of the external surface of hamster and mouse fibroblasts with [‘4C]sialic acid. Biochemistry 13,3987-3991. Davey, M. W., Sulkowski, E., and Carter, W. A. (1976). Binding of human fibroblast interferon to concanavalin A-agarose. Involvement of carbohydrate recognition and hydrophobic interaction. Biochemistry 15,704-713. DePieme, J. W., and Kamovsky, M. L. (1973). Plasma membranes of mammalian ce1ls.J. Cell Biol. 56,275-303. Dische, Z . (1962).Color reactions of Carbohydrates. Methods Carbohydr. Chem. 1,477514.

Dutton, A., and Singer, S. J. (1975). Crosslinking and labeling of membrane proteins by transglutaminase-catalyzed reactions. Proc. Natl. Acad. Sci. U S A . 72,2568-2571. Dutton, G. G. S. (1973). Application of gas-liquid chromatography to carbohydrates: Part 1. In “Advances in Carbohydrate Chemistry and Biochemistry” (R. S. Tipson and D. Horton, eds.), Vol. 28, pp. 12-101. Academic Press, New York. Edwards; J. G., Dysart, J. M., and Hughes, R. C. (1976). Cellular adhesiveness reduced in ricin-resistant hamster fibroblasts. Nature (London)264,66-68. Esselman, W. J., Laine, R., and Sweeley, C. C. (1973). Isolation and characterization of glycosphingolipids. In “Complex Carbohydrates,” Part B (V. Ginsburg, ed.), Methods in Enzymology, Vol. 28, pp. 140-156. Academic Press, New York. Fairbanks, G., Steck, T. L., and Wallach, D. F. H. (1971). Electrophoretic analysis of major polypeptides of the human erythrocyte membrane. Biochemistry 10,26062617.

Finch, P. R., Yue, R., Schachter, H., and Moscarello, M. A. (1969). Enzymic methods for the microassay of D-Mannose, D-Glucose, D-Galactose, and L-Fucose from acid hydrolyzates of glycoproteins. Anal. Biochem. 31,296-305. Findlay, J. B. C. (1974). The receptor of concanavalin A and Lens culinaris phytohemagglutinin in the membrane of the human erythrocyte.]. Biol. Chem. 249,43984403.

Fukuda, M., and Osawa, T. (1973). Isolation and characterization of a glycoprotein from human group 0 erythrocyte membrane. J . Biol. Chem. 248,5100-5105. Furthmayr, H., Tomita, M., and Marchesi, V. T. (1975). Fractionation of the major sialoglycopeptides of the human red blood cell membrane. Biochem. Biophys. Res. Commun. 65,113-121. Gahmberg, C . G. (1976).External labeling of human erythrocyte glycoproteins.1. Biol. Chem. 251,510-515.

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

137

Gahmberg, C. G., and Hakomori, S. I. (1973a). External labeling of cell surface galactose and galactosamine in glycolipid and glycoprotein of human erythrocytes. J. Biol. Chem. 248,4311-4317. Gahmberg, C. G., and Hakomori, S. I. (1973b). Altered growth behavior of malignant cells associated with changes in externally labeled glycoprotein and glycolipid. Proc. Natl. Acad. Sci. U S A . 70,3329-3333. Gahmberg, C. G., and Hakomori, S. I. (1975). Surface carbohydrates of hamster fibroblasts. 11. Interaction of hamster NIL cell surfaces with Ricinus communis lectin and concanavalin A as revealed by surface galactosyllabe1.J. Biol. Chem. 250,24472451. Ginsberg, V. (ed.) (1972). “Complex Carbohydrates,” Part B. Methods in Enzymology, Vol. 28. Academic Press, New York. Click, M. C. (1974). Isolation and characterization of surface membrane glycoproteins. In “Methods in Membrane Biology” (E. D. Kom, ed.), Vol. 2, pp. 157-204. Plenum, New York. Click, M. C., Kimhi, Y., and Littauer, U. S. (1973). Glycopeptides from surface membranes of neuroblastoma cells. Proc. Natl. Acad. Sci. U.S.A. 70, 1682-1687. Goldstein, I. J. (1975).In “Concanavalin A” (T. K. Chowdhury and A. K. Weiss, eds.), pp. 35-54. Plenum, New York. Goldstein, I. J., Hay, G. W., Lewis, B. A., and Smith, F. (1965). Controlled degradation of polysaccharides by periodate oxidation, reduction and hydrolysis. Methods in Carbohydrate Chemistry, (R. L. Whistler, ed.), Vol. 5, pp. 361-370. Gottlieb, C., Baenziger, J., and Komfeld, S. (1975a). Deficient uridine diphosphate-Nacetylg1ucosamine:glycoprotein N-acetylglucosaminyltransferase activity in a clone of Chinese hamster ovary cells with altered surface glycoproteins. J. Biol. Chem. 250,3303-3339. Gottlieb, C., Skinner, St. A. M., and Kornfeld, S. (1975b). Isolation of a clone of Chinese hamster ovary cells deficient in plant lectin-binding sites. Proc. Natl. Acad. Sci. U.S.A. 71, 1078-1082. Cottschalk, A. (1972). “The Glycoproteins: Their Composition, Structure and Function.” Elsevier, Amsterdam. Grefrath, S. P., and Reynolds, J. A. (1974). The molecular weight of the major glycoprotein from the human erythrocyte membrane. Proc. Natl. Acad. Sci. U S A . 71,39133916. Curd, J. W. (1977).Synaptic plasma membrane glycoproteins: Molecular identification of lectin receptors. Biochemistry 16,369-374. Curd, J. W., and Evans, H. W. (1976). Identification of liver plasma membrane glycoproteins which bind to 12sI-labelledconcanavalin A following electrophoresis in sodium dodecyl sulfate. Can. ]. Biochem. %, 477-480. Curd, J. W., and Mahler, H. R. (1974). Fractionation of synaptic plasma membrane glycoproteins by lectin affinity chromatography. Biochemistry 13,5193-5198. Hamaguchi, H., and Cleve, H. (1972). Solubilization of human erythrocyte membrane glycoproteins and separation of the M N glycoprotein from a glycoprotein with I, S and A activity. Biochim. Biophys. Acta 278,271-280. Harrison, R., Higginbotham, J. D., and Newman, R. (1975). Sialoglycopeptides from bovine milk fat globule membrane. Biochim. Biophys. Acta 389,449-463. Helenius, A., and Simons, K. (1975). Solubilization of membranes by detergents. Biochim. Biophys. Acta 415,29-79. Helenius, A,, Fries, E., Garoff, H., and Simons, K. (1976). Solubilization of the Semliki Forest virus membrane with sodium deoxycholate. Biochim. Biophys. Acta 436, 319-334.

138

R. 1. JULIANO

Henkart, P., and Fisher, R. I. (1975).Characterization of the lymphocyte surface receptors for Con A and PHA.J. Immunol. 114,710-771. Hochstadt, J., Quinlan, D., Rodger, R., Li, C. C., and Pound, D. (1975).Use of isolated membrane vesicles in transport studies. In “Methods in Membrane Biology” (E.D. Kom, ed.), Vol. 5, pp. 117-162 Plenum, New York. Hogg, N. M. (1974). A comparison of membrane proteins of normal and transformed cells by lactoperoxidase labeling. Proc. Nutl. Acud. Sci. U.S.A. 71,489-492. Hollaway, P. W. (1973).A simple procedure for removal of Triton X-100 from protein samples. Anal. Biochem 53,304-308. Hood, L., Cecha, M., Ewald, S., Hunkapiller, M., and McMillen, M. (1977).Structure and evolution of gene products coded by the major histocompatibility complex of mammals. J . Supramolecular Struct. Suppl. 1, p. 1. Huang, C.-C., Tsai, C.-M., and Canellakis, E. S. (1973).lodination of cell membranes. 11. Characterization of HeLa cell membrane surface proteins. Biochim. Biophys. Acta 332,59-68. Hudgin, R. L., Pricer, W. E., Jr., and Ashwell, G. (1974).The isolation and properties of a rabbit liver binding protein specific for asialogylcoproteins. J. Biol. Chem. 249, 5536-5543. Hughes, R. C. (1976).“Membrane Glycoproteins.” Butterworth, London. Hunt, R. C., and Brown, J. C. (1974).Surface glycoproteins of mouse L cells. Biochemistry 13,22-28. Hunt, R. C., and Brown, J. C. (1975). Identification of a high molecular weight transmembrane protein in mouse L ce1ls.J. Mol. Biol. 97,413-422. Hunt, R. C., Bullis, C. M., and Brown, J. C. (1975).Isolation of a concanavalin A receptor from mouse L cells. Biochemistry 14, 109-115. Hyman, R., and Stallings, V. (1976).Characterization of a TL- variant of a homozygous TL+ mouse lymphoma. Zmmunogenetics 3,75-84. Hyman, R., LaCorbiere, M., Stavarek, S., and Nicolson, G. (1974). Derivation of lymphosome variants with reduced sensitivity to plant lectins.]. Natl. Cancer Inst. 52, 963-969. Hynes, R. 0. (1973).Alteration of cell surface proteins by viral transformation and by proteolysis. Proc. Natl. Acad. Sci. U.S.A. 70,3170-3174. Itaya, K., Gahmberg, C. G., and Hakomori, S . 4 . (1975).Cell surface labeling of erythrocyte glycoproteins by galactose oxidase and a Mn++-catalyzed coupling reaction with methionine sulfone hydrazide. Biochem. Biophys. Res. Commun. 64, 10281034. Janson, V. I., and Burger, M. M. (1973).Isolation and characterization of agglutinin receptor sites. 11. Isolation and partial purification of a surface membrane receptor for wheat germ agglutinin. Biochim. Biophys. Acta 291, 127. Ji, T. H. (1976).Crosslinking of lectins and receptors in membranes with heterobifunctional crosslinking reagents. In “Membranes and Neoplasia” (V. Marchesi, ed.), pp. 171-178. Alan Liss, New York. Juliano, R. L. (1972).The solubilization and fractionation of human erythrocyte membrane proteins. Biochim. Biophys. Acta 266,301-306. Juliano, R. L. (1973). The proteins of the erythrocyte membrane. Biochim. Biophys. Acts 300,341-378. Juliano, R. L. (1974).Isothiocyanostilbene disulfonates are useful labels for the surface proteins of cultured mammalian cells. Erp. Cell Res. 86, 181-184. Juliano, R. L. (1978). Adhesion and detachment characteristics of Chinese hamster cell membrane mutants. J. Cell Biol. 76,43-49.

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

139

Juliano, R. L., and Behar-Bannelip, M. (19754.An evaluation of techniques for labelling the surface proteins of cultured mammalian cells. Biochim. Biophys. Acta 375,

249-267. Juliano, R. L., and Behar-Bannelier, M. (1975b).Surface polypeptides of the cultured Chinese hamster ovary cell. Biochemistry 14,3816-3825. Juliano, R. L., and Li, G. (1978).Lectin binding glycoproteins of CHO cells: partial fractionation of the receptors for concanavalin A and wheat germ agglutinin using a lectin immunoprecipitation technique. Biochemistry (in press). Juliano, R. L., and Ling, V. (1976). A surface glycoprotein modulating drug permeability in Chinese hamster ovary cell mutants. Biochim. Biophys. Acta 455, 152-162. Juliano, R. L., and Stanley, P. (1975).Altered cell surface glycoproteins in phytohemagglutinin-resistant mutants of Chinese hamster ovary cells. Biochim. Biophys. Acta 389,401-406. Kagawa, Y. (1972).Reconstitution of oxidative phosphorylation. Biochim. Biophys. Acta

265,241-296. Kahane, I., Furthmayr, H., and Marchesi, V. T. (1976).Isolation of membrane glycoproteins by affinity chromatography in the presence of detergents. Biochim. Biophys. Acta 426,464-476. Kahn, C . R. (1976).Membrane receptors for hormones and neurotransmitters. J . Cell Biol. 70, 261-286. Kaplan, J., and Moskowitz, M. (1975).Studies on the turnover of plasma membranes in cultured mammalian cells. 11. Demonstration of heterogeneous rates of turnover for plasma membrane proteins and glycoproteins. Biochim. Biophys. Acta 389, 306-

313. Kawasaki, T.,and Ashwell, G. (1976).Carbohydrate structure of glycopeptides isolated from an hepatic membrane-binding protein specific for asialog1ycoproteins.J. Biol. Chem. 251,5292-5299. Keski-Oja, J., Mosher, D. F., and Vaheri, A. (1976).Cross-linking of a major fibroblast surface-associated glycoprotein (fibronectin) catalyzed by blood coagulation factor XIII. Cell 9, 29-35. Kimelberg, H.K., and Papahadjopoulos, D. (1972). Phospholipid requirements for (Na2+ + K+)-ATPase activity: Head group specificity and fatty acid fluidity. Biochim. Biophys. Acta 282,277-292. Kohno, M., and Yamashina, I. (1973).4-~-Aspartylglycosylamineamidohydrolase (glycosy1 asparaginase) from hog kidney. In “Complex Carbohydrates,” Part B (V. Ginsburg, ed.), Methods in Enzymology, Vol. 28, pp. 786-788. Academic Press, New York. Koide, N., and Muramatsu, T. (1974).P-N-Acetylglucosaminidaseacting on carbohydrate moieties of glycoproteins. J. Biol. Chem. 249,4897-4904. Kornfeld, R.,and Kornfeld, S. (1970).The structure of a phytohemagglutinin receptor site from human erythrocytes. J . Biol. Chem. 245,2536-2545. Lee, Y. C., and Scocca, J. R. (1972).A common structural unit in asparagine-oligosaccharides of several glycoproteins from different sources. J. Biol. Chem. 247,5753-

5758. Lenard, J. (1971).Rapid and effective removal of sodium dodecyl sulfate from proteins. Biochem. Biophys. Res. Commun. 45,662-668. Letarte-Muirhead, M., Acton, R. T., and Williams, A. F. (1974).Preliminary characterization of Thy-1.1 and Ag-B antigens from rat tissues solubilized in detergents. Biochem. J. 143,51-61. Letarte-Muirhead, M., Barclay, A. N., and Williams, A. F. (1975).Purification of the

140

R. 1. JULIANO

Thy-1 molecule, a major cell-surface glycoprotein of rat thymocytes. Biochem. J . 151,685-697. Liljas, L., Lundahl, P., and Hjerten, S. (1976).The major sialoglycoprotein of the human erythrocyte membrane: Release with a non-ionic detergent and purification. Biochim. Biophys. Actu 426,526-534. Lis, H., and Sharon, N. (1973).The biochemistry of plant lectins (phytohemagglutinins). Annu. Rev. Biochem. 42,541-475. Lotan, R., Beattie. G., Hubbell, W., and Nicolson, G. L. (1977).Activities of lectins and their immobilized derivatives in detergent solutions. Implications on the use of lectin affinity chromatography for the purification of membrane glycoproteins. Biochemistry 16, 1787-1794. Makino, S., Reynolds, J. A,, and Tanford, C. (1973).The binding of deoxycholate and Triton X-100to proteins. J. Biol. Chem. 428,4926-4932. Marchalonis, J . J., Cone, R. E., and Santer, V. (1971).Enzymatic iodination: A probe for accessible surface proteins of normal and neoplastic lymphocytes. Biochem. J. 124, 921-927. Marchesi, V. T., and Andrews, E. P. (1971).Glycoproteins: Isolation from cell membranes with lithium diiodosalicylate. Science 174, 1247-1248. Marshall, R. D. (1972).Glycoproteins. Annu. Reu. Biochem. 41,673-702. Mastro, A. M., Beer, C. T., and Mueller, G. C. (1974).Iodination of plasma membrane proteins of BHK cells in different growth states. Biochim. Biophys. Actu 352,3851. Mattila, K., Luukkonen, A., and Renkonen, 0. (1976).Protein-bound oligosaccharide of Semliki Forest virus. Biochim. Biophys. Actu 419,435-444. Meager, A., Chanukit, A. U., Nairn, R., and Hughes, R. C. (1975).Ricin resistance in baby hamster kidney cells. Nature (London)257, 137-139. Morrison, M., Bayse, G. S., and Webster, R. G. (1971).Use of lactoperoxidase catalyzed iodination in immunochemical studies. lrnunnochemistry 8,289-297. Moss, B., and Rosenblum, E. N. (1972).Hydroxylapatite chromatography of protein sodium dodecyl sulfate complexes: A new method for separation of polypeptide subunits. J . Biol. Chem. 247,5194-5198. Muramatsu, T., Koide, N., Ceccarini, C., and Atkinson, P. H. (1976).Characterization of mannose-labeled glycopeptides from human diploid cells and their growth-dependent alterations. J . Biol. Chem. 251,4673-4679. Nachbar, M. S., Oppenheim, J. D., and Aull, F. (1976).Interactions of lectins with plasma membrane glycoproteins of the Ehrlich ascites carcinoma cell. Biochim. Biophys. Actu 419,512-529. Nagata, Y.,and Burger, M. M. (1974).Wheat germ agglutinin: Molecular characteristics and specificity for sugar binding.]. Biol. Chem. 249,3116-3122. Neville, D.M. (1975).Isolation of cell surface membrane fractions from mammalian cells and organs. In “Methods in Membrane Biology” (E. D. Kom, ed.), Vol. 3, pp. 1-49.Plenum, New York. Neville, D. M., and Glossman, H. (1974).Molecular weight determination ofmembrane protein and glycoprotein subunits by discontinuous gel electrophoresis in dodecyl sulfate. In “Biomembranes,” Part B (S. Fleischer and L. Packer, eds.), Methods in Enzymology, Vol. 32,pp. 92-1’02.Academic Press, New York. Nicolson, G. L. (1974).The interactions of lectins with animal cell surfaces. Int. Reu. Cytol. 39,89-189. Nicolson, G . (1976).Transmembrane control of receptors on normal and tumor cells. I. Cytoplasmic influences over cell surface components. Biochim. Biophys. Actu 457, 57-108.

TECHNIQUES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

141

Nilsson, S. F., and Waxdal, M. J. (1976). Isolation and identification of the major concanavalin A binding glycoproteins from murine lymphocytes. Biochemistry 15,26982704. Nozaki, Y., Reynolds, J. A., and Tanford, C. (1975). The interaction of a cationic detergent with bovine serum albumin and other proteins.]. Biol. Chem. 249,4452-4459. Ozols, J. (1972).Cytochrorne b, from a normal human liver. J . Biol. Chem. 247,22422245. Phillips, D. R. (1972).Effect of trypsin on the exposed polypeptides and glycoproteins in the human platelet membrane. Biochemistry 11,4582-4588. Porter, C. W., and Bemacki, R. J. (1975). Ultrastructural evidence for ectoglycosyltransferase systems. Nature (London) 256,648-650. Porter, W. H. (1975).Application of nitrous acid deamination of hexosamines to the simultaneous GLC determination of neutral and amino sugars in glycoproteins. Anal. Biochem. 63,27-43. Rajan, T. V., Nathenson, S. G., and Schafi, M. D. (1976). Regulatory variants for the expression of H-2 antigens: Isolation and characterizati0n.J. Natl. Cancer Znst. 56, 1221-1228. Rapin, A. M., and Burger, M. M. (1974). Tumor cell surfaces: general alteration; caused by agglutinins (E. Klein and S. Weinhouse, eds.),Adu. Cancer Res. 20, 1-91. Reynolds, J. A., and Tanford, C. (1970).The gross conformation ofprotein-sodium dodecyl sulfate complexes. J . Biol. Chem. 245, 5161-5165. Reynolds, J. A., and Tanford, C. (1976). Determination of molecular weight of the protein moiety in protein-detergent complexes without direct knowledge of detergent binding. Proc. Natl. Acad. Sci. U.S.A. 73,4467-4470. Rifkin, D. B., Compans, R. W., and Reich, E. (1972).A specific labeling procedure for proteins on the outer surface of membranes.]. Biol. Chern. 247,6432-6437. Rittenhouse, H. G., Benian, G., Rittenhouse, J. W., Hansen, E. R., and Boyd, L. E. (1976). Concanavalin A receptors in Ehrlich cells. In “Membranes and Neoplasia” (V. Marchesi, ed.), pp. 203-213. Alan Liss, New York. Robinson, P. J., Bull, F. G., Anderton, B. H., and Roitt, I. M. (1975).Direct autoradiographic visualization in SDS-gels of lectin-binding components of the human erythrocyte membrane. FEES Lett. 58,330-333. Roseman, S. (1971). The synthesis of complex carbohydrate by multilglycosyltransferase systems and their potential function in intracellular adhesion. Chem. Phys. Lipids 5,270-297. Rothstein, A., Cabantchik, Z. I., and Knauf, P. (1976). Mechanism of anion transport in red blood cells: Role of membrane proteins. Fed. Proc., Fed. Am. SOC.Exp. Biol. 35, 3-10. Rubin, M. S., and Tzagoloff, A. (1973). Assembly of the mitochondria1 membrane system. IX. Purification, characterization and subunit structure of yeast and beef cytochrome oxidase. J. Biol. Chem. 248,4269-4274. Sandvig, I. C., Olsnes, S., and Phil, A. (1976). Kinetics and binding of toxic lectins and ricin to surface receptors of human cells. J. Biol. Chem. 251,3977-3989. Sanford, B. H., Codington, J. F., Jeanhoz, R. W., and Palmer, P. D. (1973). Transplantability and antigenicity of two sublines of the TA3 tumor. J. Immunol. 110, 12331237. Schlesinger, S., Gottlieb, C., Feil, P., Gelb, N., and Kernf, S. (1976). Growth of enveloped RNA viruses in a line of Chinese hamster ovary cells with deficient N-acetylglucosaminyltransferase activity. J . Virol. 17,239-246. Schmidt-Ullrich, R., Wallach, D. F. H., and Hendricks, J. (1975). Concanavalin-A-reactive protein of rabbit thymocyte plasma membranes: Analysis by crossed immune

142

R. 1. JULIANO

electrophoresis and sodium dodecyl sulfate/polyacrylamide gel electrophoresis. Biochim. Biophys. Acta 382,295-310. Schmidt-Ullrich, R., Wallach, D. F., and Hendricks, J. (1976).Interaction of concanavalin A in the rabbit thymocyte plasma membrane. Biochim. Biophys. Acta 443,487600. Segrest, J. P., and Jackson, R. L. (1973).Molecular weight determinations of glycoproteins by polyacrylamide gel electrophoresis in sodium dodecyl sulfate. In “Complex Carbohydrates,” Part B (V. Ginsburg, ed.), Methods in Enzymology, Vol. 28, pp. 54-63.Academic Press, New York. Shami, Y., Ship, S., and Rothstein, A. (1977).Rapid quantitative separation of the major glycoproteins (PAS 1,2and 3)from other human red cell membrane proteins in a non-denaturing medium by affinity chromatography. Anal. Biochem. 80,438-445. Sharon, N. (1975).“Complex Carbohydrates: Their Chemistry, Biosynthesis and Function.” Addison Wesley, Reading, Massachusetts. Shin, B. C., and Carraway, K. (1973).Cell surface constituents of carcoma 180 ascites tumor cells. Biochim. Biophys. Acta 330,254-268. Shin, B. C., Ebner, K. E., Hudson, B. G., and Carraway, K. L. (1975).Membrane glycoprotein differences between normal lactating mammary tissue and the R3230 AC mammary tumor. Cancer Res. 35, 1135-1140. Skehel, J. J., and Waterfield, M. D. (1975).Studies on the primary structure of the influenza virus hemagglutinin. Proc. Natl. Acad. Sci. U.S.A. 72,93-97. Snow, L. D.,Colton, D. G., and Carraway, K. L. (1977).Purification and properties of the major sialoglycoprotein of the milk fat globule membrane. Arch. Biochem. Biophys. 179,690-697. Spiro, R. G.(1973a).Study of carbohydrates of glycoproteins. I n “Complex Carbohydrates,” Part B (V. Ginsberg, ed.), Methods in Enzymology, Vol. 28, pp. 3-42. Academic Press, New York. Spiro, R. G. (1973b).’Glycoproteins. Adu. Protein Chem. 27,349-407. Stanley, P., Caillibot, V., and Siminovitch, L. (1975a).Stable alterations at the cell membrane of Chinese hamster ovary cells resistant to the cytotoxicity ofphytohemagglutinin. Som. Cell Gen. 1,3-26. Stanley, P., Caillibot, V., and Siminovitch, L. (1975b).Selection and characterization of eight phenotypically distinct lines of lectin resistant Chinese hamster ovary cells. Cell 6, 121-128. Stanley, P., Narasimhan, S., Siminovitch, L., and Schachter, H. (1975~). Chinese hamster ovary cells selected for resistance to the cytotoxicity of phytohemagglutinin are deficient in a UDP-N-acetyglucosamine-glycoproteinN-acetylglucosaminyltransferase activity. Proc. Natl. Acad. Sci. U S A . 72,3323-3327. Staros, J., and Richards, F. M. (1974).Phytochemical labelling of the surface proteins of human erythrocytes. Biochemistry 13,2720-2722. Steck, T. L.(1974).Preparation of impermeable inside out and right side out vesicles from erythrocyte membranes. I n “Methods in Membrane Biology (E. D. Kom, ed.), Vol. 2,pp. 245-282.Plenum, New York. Stubbs, G . W., Smith, H. G., Jr., and Litman, B. J. (1976).Alkyl glucosides as effective solubilizing agents for bovine rhodopsin: A comparison with several commonly used detergents. Biochim. Biophys. Acta 426,46-56. Takasaki, S., and Kobata, A. (1976).Purification and characterization of an endo-p-galactosidase produced by Diplococcus pneumoniae. J . Biol. Chem. 251,3603-3609. Tanford, C., and Reynolds, J. A. (1976).Characterization of membrane proteins in detergent solutions. Biochim. Biophys. Acta 457, 133-170.

TECHNlQU.ES FOR THE ANALYSIS OF MEMBRANE GLYCOPROTEINS

143

Tanner, M. J. A,, and Anstee, D. J. (1975).A method for the direct demonstration of the lectin-binding components of the human erythrocyte membrane. Biochem. J . 153, 265-270. Tanner, M. J. A,, Boxer, D. H., Cumming, J., and Verrier-Jones, J. (1974).A set of proteins common to the circulation human platelet and lymphocyte. Biochem. J . 141, 907-916. Teng, N. N. H., and Chen, L. B. (1975).The role of surface proteins in cell proliferation as studied with thrombin and other proteases. Proc. Natl. Acad. Sci. U S A . 72,413417. Teng, N. N. H., and Chen, L. B. (1976).Thrombin-sensitive surface protein of cultured chick embryo cells. Nature (London)259,578-580. Thomas, D. B., and Winzler, R. (1969).Structural studies on human erythrocyte glycoproteins: Alkali labile oligosaccharides. 1. Biol. Chem. 244,5943-5946. Thomas, D. B., and Winzler, R. (1971). Structure of glycoproteins of human erythrocytes: Alkali stable oligosaccharides. Biochem. J . 124,55-59. Tomita, M., and Marchesi, V. (1975).Amino acid sequence and oligosaccharide attachment sites of human erythrocyte glycophorin. Proc. Natl. Acad. Sci. U S A . 72,2964 -2968. Triche, T. J., Tillack, T. W., and Kornfeld, S. (1975).Localization of the binding sites for the Ricinus communis, Agaricus bisporus and wheat germ lectins on human erythrocyte membranes. Biochim. Biophys. Acta 394,540-549. Trosper, T., and Levy, D. (1974).Characterization of the surface protein components in adipocyte plasma membranes. Biochemistry 13,428-4290. Vitetta, E. S., and Uhr, J. W. (1975).Immunoglobulins and alloantigens on the surface of lymphoid cells. Biochim. Biophys. Acta 415, 253-271. Vitetta, E. S., Baur, S., and Uhr, J. (1971). Cell surface immunoglobulin. 11. Isolation and characterization of immunology from mouse splenic lymphocytes.]. E x p . Med. 134,242-264. Walsh, F. S., Barber, B. H., and Crumpton, M. J. (1976).Preparation of inside-out vesicles of pig lymphocyte plasma membrane. Biochemistry 15,3557-3563. Wang, K., and Richards, F. M. (1974).An approach to nearest neighbor analysis of membrane proteins. ]. Biol. Chem. 249,8005-8015. Warren, L. (1972). The biosynthesis of metabolism of amino sugars and amino sugar containing heterosaccharides. In “Glycoproteins: Their composition, Structure and Function” (A. Gottschalk, ed.), pp. 1097-1 126. Elsevier, Amsterdam. Wilchek, M., and Jakoby, W. B. (1974). The literature on affinity chromatography. In “Affinity Techniques: Enzyme Purification,” Part B (W. B. Jakoby and M. Wilchek, eds.), Methods in Enzymology, Vol. 34, pp. 3-10. Academic Press, New York. Wray, V. P., and Perdue, J. F. (1974).Protein and glycoprotein subunit composition of plasma membrane from chick embryo fibroblasts. J . Biol. Chem. 249, 11891197. Yamada, K. M., and Weston, J. A. (1974). Isolation of a major cell surface glycoprotein from fibroblasts. Proc. Natl. Acad. Sci. U.S.A. 71,3492-3496. Young, N. M., and Leon, M. A. (1974).The affinity of concanavalin A andLens culinaris hemagglutinin for glycopeptides. Biochim. Biophys. Acta 365,418-425. Yu, J., and Steck, T. L. (1975a).Associations of band 3, the predominant polymer of the human erythrocyte membrane. ]. Biol. Chem. 250,9176-9184. Yu, J., and Steck, T. L. (1975b).Isolation and characterization of band 3, the predominant polypeptide of the human erythrocyte membrane.]. Biol. Chern. 250,91709175.

144

R. 1. JULIANO

Yurchenco, P. D., and Atkinson, P. H. (1975). Fucosyl glycoprotein of precursor pools in HeLa cells. Biochemistry 14,3107-3110. Zachowski, A,, and Pard', A. (1974).Use of concanavalin A polymer to isolate right-sideout vesicles of purified plasma membrane from eukaryotic cells. Biochem. Biophys. Res. Commun. 57,787. Zanetta, J. P.,Morgan, I. G., and Combos, G . (1975).Synaptosomal plasma membrane glycoproteins: Fractionation by d n i t y chromatography on concanavalin A. Brian Res. 83,337-348. Zimmerman, B. (1974).Lymphocyte plasma membrane. I. Characterization of an antigenic plasma membrane constituent on rat thymus ce1ls.J. Irnmunol. 112,249-259. Zvilichovsky, B., Gallop, P. M., and Blumenfeld, 0. (1971). Isolation of a glycoproteinglycolipid fraction from human erythrocyte membranes. Biochem. Biophys. Res. Commun. 44,1234-1243.

.

CURRENT TOPICS IN MEMBRANES AND TRANSPORT VOLUME 11

Glycoprotein Membrane Enzymes JOHN R . RZORDAN Research Institute The Hospital for Sick Children and Department of Clinical Biochemistry Uniuersity of Toronto Toronto. Canada

AND GORDON G . FORSTNER Research Znstitute The Hospital for Sick Children and Department of Physiology University of Toronto Toronto. Canada

I . Introduction . . . . . . . . . . . . . . . . . . . I1. Specific Enzymes (Table I) . . . . . . . . . . . . . . .

A . Glycosyltransferases . . . . . . . . . . . . . . . B. Nucleotide Pyrophosphatase . . . . . . . . . . . . . C . 5'-Nucleotidase . . . . . . . . . . . . . . . . . D . (Na+ + K+)M$+-ATPase . . . . . . . . . . . . . . E . Cytochrome b, Reductase . . . . . . . . . . . . . . F . Acetylcholinesterase . . . . . . . . . . . . . . . G . Alkaline Phosphatase . . . . . . . . . . . . . . . H . Brush Border Hydrolases . . . . . . . . . . . . . . I . Lysosomal Acid Hydrolases . . . . . . . . . . . . . 111. Membrane Association . . . . . . . . . . . . . . . . A . Nature of' Association . . . . . . . . . . . . . . . B. Sidedness and Topology . . . . . . . . . . . . . . IV . Structure . . . . . . . . . . . . . . . . . . . . A . Amino Acid Composition . . . . . . . . . . . . . . B . Monosaccharide Composition . . . . . . . . . . . . C. Functional Role of Carbohydrate . . . . . . . . . . . V . Functional Interrelationships . . . . . . . . . . . . . . A . Modification of Recognition Phenomena . . . . . . . . . B. Membrane Transport . . . . . . . . . . . . . . . VI Biosynthetic and Developmental Aspects . . . . . . . . . . A Biosynthesis . . . . . . . . . . . . . . . . . . B Developmental Changes . . . . . . . . . . . . . . VII Are All Ectoenzymes Glycoproteins? . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . .

. .

. .

Copyright @ 1978 hy Academic Press. Inc .

All rights of reproduction in any form reserved. ISBN 0-12-153311-5

146 147 147 157 158 161 164 164 166 170 183 188 188 191 194 194 195 195 201 201 203 206 207 208 210 210 145

146

J. R. RIORDAN AND G. G. FORSTNER

I.

INTRODUCTION

The universal carbohydrate or “sugar coating” possessed by the plazma membranes of mammalian cells (Rambourg et al., 1966) has been of great interest to biologists since Bennett’s (1963)original description of the glycocalyx in 1963. Glycoproteins on cell surfaces were ascribed various functions relating to cell adhesion, recognition phenomena, and scavenger roles, each of which has extreme potential significance. Perhaps not surprisingly, since many of the techniques employed to isolate and study glycoproteins are destructive to biological function, the concept that some and possibly many surface glycoproteins are functioning enzymes has developed rather slowly. When glycoprotein enzymes were first reviewed in 1972 (Jutisz and de la Llosa), only 2 of the 14 listed were membrane-bound (cholinesterase and alkaline phosphatase). In contrast, this chapter, although sampling and describing only the best characterized membrane-bound glycoenzymes, lists an additional 15 which were unrecognized as glycoproteins at the start of the decade. Many of the additions are bound to the brush border membranes of intestinal and renal proximal tubular cells, long recognized for their very thick periodic acid-Schiff (PAS)-positive surface coats (Leblond, 1950).In the intestine, enzyme glycoproteins account for a major portion of the coat as labeled by radioactive glycoprotein precursors (Forstner, 1971). Indeed, in chese membranes, all the enzymes extracted and purified to date have turned out to be substantially glycosylated when examined for carbohydrate, and it is gradually becoming apparent from their relatively low purification factors that a sizable percentage of the membrane protein, perhaps a third, consists of glycoenzymes. The case is not so clear elsewhere, although lysosomal enzymes seem invariably to be glycoproteins, and it is premature to conclude that the majority of membrane glycoproteins represent functioning enzymes. Nevertheless, it seems obvious that glycoprotein enzymes are important parts of cell membranes, contributing to both their structure and function. In this chapter membrane enzymes known to be glycoproteins are described individually in a sequence concentrating initially on enzymes of general distribution, followed by brush border hydrolases, and last by acid hydrolases of the lysosome. Such a treatment will, we hope, form a sound base for subsequent generalizations with respect to membrane localization, structural peculiarities, the role of carbohydrate in enzyme action, functional features, and developmental changes. Throughout, structural features are related to current con-

GLYCOPROTEIN MEMBRANE ENZYMES

147

cepts of membrane organization and assembly (Rothman and Lenard, 1977; Rothman and Lodish, 1977; Katz et al., 1977; Wirth et al., 1977; Sturgess et al., this volume), so as to be consistent with the overall theme of this volume. Despite the risk of contributing to confusion, it is necessary at the outset to briefly refer to the current terminology used to describe the sidedness of membrane enzymes and membrane proteins in general. Plasma membrane enzymes which have their substrate sites oriented to the external surface of the cell (extracytoplasmically) were termed ectoenzymes several years ago (dePierre and Karnovsky, 1973). In contradistinction, it seems reasonable to classify those membrane enzymes with a cytoplasmic orientation of active sites as endo-enzymes. Recently, however, Rothman and Lenard (1977) have chosen to classify membrane proteins in general as either ecto-proteins or endo-proteins. The former group are those membrane proteins which “have substantial hydrophilic mass projecting beyond the extracytoplasmic surface of the bilayer”. In view of the apparently constant asymmetry of membrane glycoproteins, they would all fall within this category. Hence, a membrane endo-enzyme (substrate site cytoplasmically directed) such as the (Na+ K+)Mg2+ATPaseis a transmembrane glycoprotein and on this basis would be an ecto-protein according to Rothman and Lenard (1977). Their classification defines endo-proteins as those which “do not project beyond the extra-cytoplasmic surface of the bilayer, and may have most of their mass associated with the cytoplasmic side of the membrane”. No membrane glycoproteins, enzymic or otherwise, could qualify as endo-proteins. Thus, while all membrane glycoprotein enzymes appear to be ecto-proteins, it remains useful to categorize them as either ecto- or endo- according to the sidedness of their active sites. Clearly, the latter must be transmembrane proteins, whereas, the former may or may not be. Of the enzymes described in detail below, two are endoenzymes (Na+ + K+)Mg+ATPaseand cytochrome b, reductase); the remainder are ectoenzymes.

+

II. SPECIFIC ENZYMES (TABLE I) A. GIycosyItransferases

These enzymes, which are responsible for the addition of sugars during the biosynthesis of oligosaccharide chains, are localized predominantly in the membranes of the endoplasmic reticulum and

GLYCOPROTEIN MEMBRANE ENZYMES

151

Golgi apparatus (Schachter et al,, 1970; Schachter and Roden, 1973). Some activity is present in mitochondria (Louisot and Morelis, 1973). The possibility of localization in the plasma membrane is controversial (Keenan and Morrk, 1975) and is discussed in detail. Soluble glycosyltransferases present in body fluids such as serum and milk have yielded to purification more easily than membrane-bound forms and have been characterized. In cases where they have been purified and analyzed?the enzymes have been found to be glycoproteins (Lehman et al., 1975; Podolsky and Weiser, 1975; Fraser and Mookerjea, 1976; Smith and Brew, 1977). This discussion concentrates on two aspects of these enzymes: the characterization, membrane association, and glycoprotein nature of those which have been purified, and the controversial evidence regarding the degree of their localization at the external surface of the plasma membrane. Although these enzymes generally have not been as well characterized as some of the others described in this chapter, they serve to illustrate some of the problems encountered in the study of membrane glycoprotein enzymes. 1. PROPERTIES

The soluble galactosyltransferase from milk was purified by gel filtration and affinity chromatography on a-lactalbumin linked to agarose (Andrews, 1970; Trayer and Hill, 1971; Mawal et al., 1971; Geren et al., 1976). This enzyme is a glycoprotein with a MW of 58,000 (Magee et al., 1973). As a result of proteolytic degradation a lower-MW (42,000) form appears without loss of carbohydrate. The enzyme contains approximately 2.2%mannose, 2.8%galactose? 2.4% N-acetylglucosamine (GlcNAc), and 3.0% sialic acid (Lehman et al., 1975). More recently a similar enzyme has been solubilized and purified from Golgi membranes of lactating sheep mammary glands (Smith and Brew, 1977).These authors suggest that proteolytic cleavage may yield the secreted soluble enzyme. The galactosyltransferase from rabbit erythrocyte ghosts was purified to apparent homogeneity and partially characterized (Podolsky et al., 1974; Podolsky and Weiser, 1975). An enrichment of about 500fold was achieved by preparative polyacrylamide gel electrophoresis of a high-speed (105,OOOg;1 h) supernatant from extensively sonicated ghost membranes. The enzyme eluted from the gels contained carbohydrate and had a MW of 54,000 as judged by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Double-diffusion assays against concanavalin A (Con A) showed visible precipitation; Con A (133 pg/ml) caused 50%inhibition of the membrane-bound en-

152

J. R. RIORDAN AND G. G. FORSTNER

zyme. A 10-fold purification was obtained by affinity chromatography of sonicated membranes on Con A-Sepharose. A second affinity step using UDP-Sepharose produced a single glycoprotein enriched 620fold in specific activity which stained on periodic acid-Schiff treatment. The amino acid analysis showed a large amount of asparagine (13%).Of a total carbohydrate content of about 15% approximately 7% was mannose, 7% galactose, 4% GlcNAc, and 2% sialic acid. In addition to the evidence for direct interaction with Con A given above, it was found that the purified enzyme could bind to type-0 human erythrocytes and conferred Con-A-mediated agglutinability on them. Pretreatinent of the enzyme with a-mannosidase rendered it ineffective in this respect. Fraser and Mookerjea (1976) purified galactosyltransferase from serum and liver microsomes of rats. The serum enzyme was enriched 6000- to 7000-fold using a-lactalbumin-Sepharose, yielding a homogeneous glycoprotein with a MW of 43,000, in agreement with data for other purified soluble galactosyltransferases (Powell and Brew, 1974). Although these workers (Fraser and Mookerjea, 1976) state that purification of the microsomal membrane enzyme can also be attained using a-lactalbumin-Sepharose, they describe only a 6-fold purification by gel filtration of Sephadex G-200. This was obtained from the portion of the membrane enzyme “solubilized” b y 200 mM NaCl. On the basis of the much larger size of this enzyme as compared to the serum enzyme, as indicated by Sephadex G-200 chromatography, it was suggested that the membrane enzyme possesses a hydrophobic “tail,” as has been shown to be the case for several other membrane enzymes (see Section 111,A). More recently these workers have obtained a 680-fold purification of the enzyme from microsomal membranes solubilized with 1% Triton X-100 (Fraser and Mookerjea, 1977). In addition to the purifications of membrane glycosyltransferases referred to above, an N-acetylgalactosaminyltransferase from blood group-A plasma has been purified in an elegant manner, utilizing its ability to bind to unmodified Sepharose (Whitehead et al., 1974).Since galactosyltransferase is the only glycosyltransferase which has been purified from membranes, the glycoprotein nature of others has not been proven. In fact, the lack of purified enzymes until recently has been one of the major factors contributing to the controversy regarding the surface plasma membrane localization of glycosyltransferases. The application of labeled monospecific antibodies to pure enzyme would contribute to resolution of the subcellular localization problem. As mentioned above, there is good evidence that, at least in secretory cells

GLYCOPROTEIN MEMBRANE ENZYMES

153

such as liver, most of the glycosyltransferases are attached to the membranes of the endoplasmic reticulum and Golgi apparatus (Schachter et al., 1970; Schachter and Roden, 1973). 2. SURFACELOCATION With Roseman’s (1970) postulate that ectoglycosyltransferases could serve as mediators of intercellular adhesion, attention was focused on their localization at the cell surface. Roseman and his colleagues provided some evidence for the existence of galactosyltransferases at the surfaces of embryonic neural retinal cells (Roth et al., 1971a,b; McGuire, 1972). Since that time a large number of investigations with a wide variety of tissue and cell types has been carried out with the aim of either supporting or refuting the surface localization theory and of assessing possible functional roles. These, in addition to intercellular adhesion, have included repair of surface complex carbohydrates (Bernacki, 1974) and recognition of glycoproteins (Jamieson et al., 1971) and cells (McLean and Bosmann, 1975). We restrict ourselves here primarily to a consideration of evidence for or against the presence of these enzymes at the cell surface. First, an enzyme can be termed “ecto” on the basis of observed activity on intact cells only if the plasma membrane is impermeable to substrates. In the case of glycosyltransferases this condition certainly holds for macromolecular acceptors such as partially completed glycoproteins and glycopeptides, but does not hold for monosaccharide acceptors. Cell membranes are generally quite impermeable to nucleotides (Glynn, 1968) and therefore also to nucleotide sugar donors. However, nucleotide sugars can be hydrolyzed to yield sugar phosphates and free sugars. The sugars can then enter the cell and be incorporated into oligosaccharides intracellularly. It is then difficult to distinguish incorporation into intracellular molecules by this route from direct transfer to endogenous surface acceptors. Hydrolysis of nucleotide sugars has been observed at the surface of several different cell types in which surface transferases have been studied. Included in this category are lymphocytes (Verbert et al., 1976), fibroblasts (Patt and Grimes, 1976), and neural retinal cells (Roth et al., 1971b). One means of determining whether or not this hydrolysis and uptake of free sugar are responsible for incorporation into endogenous acceptors is to check for inhibition by high concentrations of added hydrolysis products. With ascites cells (Porter and Bernacki, 1975), neural retinal cells (McGuire, 1972), lymphocytes (Painter and White, 1976; Verbert et al., 1976), and fibroblasts (Patt and Grimes, 1974, 1975),

154

J. R. RIORDAN AND G. G. FORSTNER

high concentrations of the free sugar being transferred did not influence the incorporation, indicating that it was directly transferred from nucleotide sugar to acceptor. Otherwise the excess added sugar would have inhibited incorporation by competing for entry into the cell with the radioactively labeled sugar released by nucleotide sugar hydrolysis. Confirmatory evidence was provided by the failure of inhibitors of sugar transport such as phlorizin to affect incorporation (Verbert et al., 1976). In at least one instance the possibility of phagocytosis of an exogenous acceptor was considered but, by using an acceptor linked to agarose, uptake by this mechanism was found not to occur in lymphocytes (Cacan et al., 1976). The properties of the plasma membrane in regard to permeability to substrates of course are entirely different in damaged or nonviable cells. It is difficult to imagine an experimental situation in which at least some such cells are not present in the total population. Some investigators have made estimates of this proportion. Patt and Grimes (1976) stated that as many as 20% of the BALB/c fibroblasts in suspensions used for mannosyltransferase assays were nonviable. A similar situation existed with neural retinal cells (Roth et al., 1971b), whereas about 10% of rat lymphocytes used to assay galactosyltransferase were nonviable (Cacan et al., 1976). Deppert and Walter (1977) have claimed that no glycosyltransferase activity can be detected on cells grown in monolayer culture, the activity observed when such cells are suspended resulting from alterations during suspension. In addition to altered permeability of damaged cells, the enzymes themselves might "leak" from the cells. In fact, enzymes might even be released in some fashion from the surfaces of intact viable cells. However, in experiments where this was checked, no activity was found in the media in which the cells were incubated (Lamont et al., 1974; Bernacki, 1974; Roth et al., 1971b). In cases where the transfer of sugars to endogenous surface acceptors is apparently being assayed, it should be possible to create more acceptor sites by treating the cells with appropriate glycosidases. For example, neuraminidase treatment of Ehrlich ascites cells increased the incorporation of sialic acid into endogenous acceptor by a factor of 6 (Irwin and Anastassiades, 1975). Porter and Bernacki (1975) observed an identical degree of enhancement in another line of ascites cells. To obtain evidence that transfer to an acceptor has occurred on or external to the plasma membrane it is necessary to isolate it from these locations. However, this has rarely been done. Cacan et al. (1976)

GLYCOPROTEIN MEMBRANE ENZYMES

155

demonstrated the transfer of galactose from UDP galactose to ovomucoid-Sepharose which did not enter the lymphocytes. In an analogous manner, modification of activity by nonpenetrating inhibitors has been used as evidence that certain enzymes are surface-located (Edelson and Cohn, 1976.). Glycosyltransferases are sensitive to plant lectins (see Section IV,C,l), but it usually has not been determined that modification has occurred under conditions where the lectin is not endocytosed. Another approach to demonstrating that sugars have been attached to the cell surface takes advantage of the formation of new lectin reactor sites. As an example, galactosylation of the rat lymphocyte surface increased the number of soybean agglutinin-binding sites about twofold (Verbert et al., 1976). The quantity of a transferase at the surface of a cell population compared to the total activity in the cells or in subcellular fractions may help to assess the validity of a certain level of activity being attributed to the cell surface. In fact, in some instances (Patt and Grimes, 1974) where this type of quantitation was apparently utilized in support of surface localization, the demonstration of nearly equal levels of activity in intact and disrupted cells implies that essentially all the activity is located at the surface. Although this possibility has not been disproved for nonsecretory cells, it seems highly unlikely, particularly in view of recent evidence that membrane glycoproteins seem to use the same biosynthetic mechanism and route as secretory gl ycoproteins (Katz et al., 1977; Wirth et al., 1977). In the case of the biosynthesis of a vesicular stomatitis virus-coded glycoprotein of the host cell plasma membrane, glycosylation occurs during transit through intracellular membranes and is completed well before the protein appears in the surface membrane (Knipe et al., 1977a,b). Perhaps the most convincing evidence in favor of cell surface glycosyltransferase and acceptor activities comes from the ultrastructural work of Porter and Bernacki (1975).These workers showed that, when murine leukemic ascites cells were incubated with CMP-L3HIN-acetylneuraminic acid, (CMP-L3H1NANA)and analyzed by electron microscope autoradiography, most of the 3H was present over the plasma membrane. About 90% of the tritium could be removed by neuraminidase treatment, and treatment prior to incubation increased labeling sixfold. Of the total radioactivity, 86% was recovered as sialic acid and 14% as CMP-NANA. A 100-fold excess of sialic acid, mannose, or mannosamine was without influence on labeling. There was no indication of intracellular labeling before labeling of the plasma membrane. In addition to experiments with whole cells, glycosyltransferases

156

J. R. RIORDAN AND G. G . FORSTNER

have of course been assayed in isolated plasma membranes. While such measurements reveal the presence of activity (Pricer and Ashwell, 1971; Riordan et aZ., 1974; Munro et al., 1975; Kartner et al., 1977), the levels are much lower than in microsomes or Golgi bodies and are not greatly enriched over the levels in cell or tissue homogenates. These findings do not exclude the presence of low levels of truly plasma membrane-associated activities. It would be useful to access this possibility using electron microscope autoradiography with isolated plasma membranes. Even though they would be inevitably contaminated to some extent with intracellular membranes, it may be possible to identify morphologically the type of membranes with which silver grains are associated. Currently, there is little reason to believe that a large proportion of the glycosyltransferase activity known to be present in the endoplasmic reticulum and Golgi apparatus of secretory cells is shifted to the surface plasma membrane in other cell types. Arguments that relatively small amounts of plasma membrane enzymes contribute to specific cell surface functions have been difficult to prove or refute. There is a need for quantitative experiments in which well-characterized glycosyltransferase activity in different fractions of a cell claimed to have surface transferases is determined. In the case of rat liver, plasma membranes contained very low levels of fucosyl-, sialyl-, galactosyland N-acetylglucosaminyltransferases relative to the amounts present in the Golgi apparatus (Munro et d.,1975). Furthermore, it is imperative that purified enzymes be obtained from the various membranous subcellular fractions and compared. A knowledge of the degree of specificity of antibodies to these purified enzymes will be of assistance in quantitative localization experiments. Discussions of the functional roles of surfaces transferases should become more intelligible with these tasks accomplished. The role of carbohydrate in membrane glycosyltransferases shown to be glycoproteins (Podolsky and Weiser, 1975; Fraser and Mookerjea, 1976) is not known. Limited attempts to gain information on this topic have been made using glycosidases and lectins. For example, removal of the sialic acid from purified milk galactosyltransferase with neuraminidase was shown to have no effect on activity (Lehman et al., 1975). Some of the mannose residues present in the galactosyltransferase of rabbit erythrocyte membranes are accessible to direct interaction with Con A (Podolsky and Weiser, 1975). As a result, activity was inhibited about 50% at Con-A concentrations above 25 pg/ml. (GlcNAc and N-acetylgalactosamine (GalNAc) transferase activities were unaffected.) In contrast, the galactosyltransferase of liver Golgi

GLYCOPROTEIN MEMBRANE ENZYMES

157

membranes is activated as much as 100% b y the binding of high concentrations of Con A (Young et al., 1976a,b). A mitogenic dose of Con A (5 pg/ml) caused a large increase in the galactosyltransferase of thymus lymphocytes after 72 hours (Lamont et al., 1974). Sialyltransferase, apparently at the surface of mouse thymocytes, was greatly enhanced by 5-10 pg/ml Con A within 1 hour; this activation was not sensitive to cycloheximide (Painter and White, 1976). Transfer of galactose, mannose, GlcNAc, and sialic acid in cultured fibroblasts was apparently insensitive to Con A, wheat germ agglutinin (WGA), and phytohemagglutinin (PHA) at concentrations of 2.5 mg/ml (Patt and Grimes, 1974). Thus the information available indicates that glycosyltransferases can be inhibited, stimulated, or unaffected by plant lectins. Hence the findings to date using glycosidases and lectins as probes have not been particularly revealing with respect to clarification of the contribution of carbohydrate to enzymic function. It seems reasonable to assume that the carbohydrate of a glycosyltransferase may serve to orient it so that both the oligosaccharide moieties and substrate sites are correctly directed toward the cisternal surface of the endoplasmic reticulum and Golgi apparatus and possibly toward the external surface of the plasma membrane. B. Nucleotide Pyrophosphatase

The pyrophosphate bond of various nucleotides is rapidly split by the hemoglobin-free perfused rat liver (Domschke et al., 1971; Liersch et al., 1971). The enzyme which is responsible for this activity has been solubilized by digestion with crude pancreatic lipase (Decker and Bischoff, 1972), trypsin (Bischoff et al., 1975), and N-dodecyl sarcosinate (Evans et al., 1973; Elovson, 1977). Evans (1974) demonstrated that hepatic nucleotide pyrophosphatase could be labeled with 1311 by incubating hepatocytes with lactoperoxidase, and that it was therefore located on the external surface of the plasma membrane. In a careful subfractionation study, Bischoff et al. (1975) showed that the endoplasmic reticulum contained a similar enzyme which on parallel isolation and purification with the plasma membrane enzyme had essentially the same catalytic and physical properties. Nucleotide pyrophosphatase hydrolyzes p-nitrophenylthymidine 5’-monophosphate plus a variety of purine and pyrimidine nucleotides, yielding 5’-nucleoside monophosphates, but it does not cleave nucleic acids, CAMP, or phosphate monoesters (Evans et al., 1973; Bischoff et ul., 1975). The enzyme requires Mg2+or Ca2+and is inhibited by EDTA. Maximal activity occurs at an alkaline pH (9.6-9.7)

158

J. R. RIORDAN AND E. G. FORSTNER

(Evans et al., 1973; Bischoff et al., 1975). Enzymes with similar properties have been described in a variety of tissues (Razzel, 1963). The detergent-solubilized enzyme from mouse liver is a sialoglycoprotein containing 22 moles of sialic acid and 51 moles of glucosamine per mole (Evans et al., 1973). Glucose, mannose, galactose, and fucose, but not GalNAc, were demonstrated qualitatively by gas-liquid chromatography. Neutral carbohydrate determined with anthrone amounted to 5%by weight. In rat liver both the plasma membrane and endoplasmic reticulum enzymes stained strongly with periodatefuchsin. There seems little doubt therefore that nucleotide pyrophosphatase is a glycoprotein. The MWs of the trypsin-solubilized and detergent-solubilized enzymes have been estimated by SDS-PAGE and are quite similar: 137,000 (rat liver, Bischoff et al., 1975) and 128,000 (mouse liver, Evans et al., 1973). Comparative studies have not been performed on enzymes isolated by both techniques from the same tissue. Subunits have not been described. There is very little or no cysteine (Evans et al., 1973), which probably accounts for the enzyme’s resistance to thiols. In common with many glycoprotein ectoenzymes the polarity of the constituent amino acid (49%)is higher than that of many integral membrane proteins (Table 11).The effect of modification of carbohydrate structure on the enzyme has not been tested.

C. 5’-Nucleotidare 5’-Nucleotidase has been extensively studied as an ectoenzyme (DePierre and Karnowsky, 1973,1974;Fleit et al., 1975; Newby et al., 1975; Woo and Manery, 1975; Carraway et al., 1976; Stefanovic et al., 1976). As with nucleotide pyrophosphatase, cytochemical evidence has shown the enzyme also to be present in membranes of the endoplasmic reticulum (Widnell, 1972) and the Golgi apparatus (Farquhar et al., 1974), although in lesser amounts. A soluble form of this enzyme exists in some tissues (Ipata, 1968; Itoh et d., 1968; Naito et d., 1974), and it is excreted by some microorganisms (Brownlee and Heath, 1975). The membrane-associated enzyme cannot be dissociated from the membrane by the disruption of ionic interactions, but it is amenable to detergent solubilization (Songet al., 1962).The latter property has been utilized in all purifications of the enzyme thus far. The enzyme was first purified from isolated membranes b y Widnell and Unkeless (1968),who achieved enrichments in specific activity of about 300- and 75-fold relative to that of rat liver microsomes and plasma membranes, respectively. PAGE in the presence of phenol,

160

J. R . RlORDAN AND G. G. FORSTNER

acetic acid, and urea indicated the presence of one major protein band. Phospholipids, primarily sphingomyelin, remained bound to the enzyme. The second purification procedure provided the first evidence that the enzyme was a glycoprotein. With the use of a different selection of conventional protein purification procedures, an enzyme of similar purity but lower specific activity was obtained from detergent-solubilized mouse liver plasma membranes (Evans and Gurd, 1973). The protein band visualized after PAGE also stained for carbohydrate with the periodate-Schiff method. Although purification of the enzyme from liver plasma membranes has now been reported by at least two other laboratories (Nakamura, 1976; Slavik et al., 1977), the carbohydrate composition has not yet been described, possibly because yields have not exceeded 100-200 pg even in fairly largescale preparations. The enzyme has a MW of roughly 150,000 (Evans and Gurd, 1973; Nakamura, 1976)and may be a dimer of two subunits approximately one-half that size (Evans and Gurd, 1973; Slavik et al., 1977). Other evidence for the glycoprotein nature of the enzyme has come from studies of its sensitivity to lectins (Riordan and Slavik, 1974a,b; Zachowski and Paraf, 1974; Carraway et al., 1975; Riordan and Slavik, 1975; Stefanovic et al., 1975; Slavik et al., 1977). While it is not possible to determine that modification of membrane-bound enzyme activity results from the direct binding of lectin to the enzyme (for a more detailed discussion see Section IV,C), the sensitivity of the purified enzyme to inhibition by lectins indicates this to be the case (Slavik et al., 1977). The purified enzyme remains susceptible to inhibition by Con A, WGA, and Ricinus communis agglutinin. Dimeric succinyl Con A is incapable of inhibition. Antisera to whole-liver plasma membranes (Gurd and Evans, 1974) or purified 5’-nucleotidase (Riemer and Widnell, 1975) inhibits activity in a manner probably analogous to that described for lectins (Riordan et al., 1977). No evidence has appeared on the probable glycoprotein nature of the enzyme purified from fat cell plasma membranes (Newby et d., 1975). Haptene-reversible binding of the enzyme from deoxycholate-solubilized lymphocyte plasma membranes to Lens culinaris lectin (attached to agarose) provides some indication of carbohydrate in the 5’nucleotidase in these membranes (Hayman and Crumpton, 1972). Indirect evidence for the presence of carbohydrate, as well as its lack of contribution to activity, was also obtained using an N-acetylglucosaminyltransferase-deficient clone of Chinese hamster ovary cells (Gottlieb et al., 1975). Plasma membrane 5’-nucleotidase activity was the same in parent and deficient cell lines. Despite this, the detergent-sol-

GLYCOPROTEIN MEMBRANE ENZYMES

161

ubilized enzyme from the parent cells bound specifically to R . communis agglutinin-agarose affinity columns, whereas the enzyme from the deficient cell line did not. Treatment of rat liver plasma membranes with neuraminidase or pgalactosidase individually or sequentially does not influence the activity of the enzyme (Riordan, unpublished observations). Similarly, activity at the surface of cultured astroblasts was unaffected by neuraminidase (Trams et al., 1976).

D. (Na+

+ K+)Mg+-ATPare

This enzyme, by virtue of its structure and orientation across the lipid bilayer of the plasma membrane of cells, couples the vectorial transport of Na+ and K+ to the hydrolysis of ATP (for reviews, see Skou, 1973; Dahl and Hokin, 1974; Glynn and Karlish, 1975; Albers, 1976). Skou (1957) provided the first evidence for the existence of such an enzyme. During the course of studies on the kinetic and allosteric properties of particulate forms of the enzyme from a variety of tissues, experience with the use of detergent (Skou, 1962) and salt (Nakao et al., 1965; Boegman et al., 1970) extraction procedures was gained. During the 1970s use has been made of these findings in purification of the enzyme from tissues in which it is concentrated. Even from these sources yields have been low, hindering analytical work. Studies of what have been termed the partial reactions of the enzyme have contributed greatly to the widely accepted notion that the enzyme is synonymous with the mechanism of active transport of Na+ and K+ [for a partial survey, see Ann. N . Y . Acad. Sci. 242, 80-317 (1974)l. Reactions which have been analyzed individually include binding of ATP and Na+- and Mg2+-dependentphosphorylation on the internal side of the membrane, K+-stimulated dephosphorylation, and binding of cardiac glycosides which occurs near the K+ site on the outside of the membrane. Toule and Copenhaver (1970) obtained the first soluble enzyme of high specific activity from rabbit kidney. The following year Kyte (1971) achieved the first satisfactorily homogeneous preparation of Na+,K+,Mg2+-ATPasefrom canine renal medulla. The enzyme consisted of two subunits, the smaller of which was a glycoprotein. The larger subunit had a MW of between 84,000 and 139,000 depending on the method of measurement, had glycine as the N-terminal amino acid, and consisted of 47% nonpolar amino acids (Kyte, 1972). The small chain was between 35,000 and 57,000 daltons in mass and had

162

1. R. RIORDAN AND G. G. FORSTNER

an N-terminal alanine and 44%apolar amino acids. The carbohydrate content of the enzyme from dogfish rectal gland was 10-15% (Hokin, 1974) with about 4% amino sugars, 3% neutral sugars, and 2% sialic acid (Kyte, 1972) in the renal medulla enzyme. Enzymes of somewhat higher specific activity but exhibiting similar subunit structure have been obtained from pig brain (Nakao et al., 1974),Squalus rectal gland (Hokin et al., 1973; Hokin, 1974),and rabbit kidney (Jdrgensen, 1974). In all cases the purified enzyme contained phospholipid and existed in vesicular form. Recent evidence from negative staining and freeze-fracture electron microscopy shows the surface of the vesicles to exhibit a uniform population of particles 95-100 8, in diameter (van Winkle et al., 1976).The MW of the native enzyme has been estimated at approximately 300,000 (Kyte, 1972). If this is so, it seems reasonable that it may exist as a tetramer consisting of two large and two small subunits. The fact that incorporation of the pure enzyme into phospholipid vesicles (Goldin and Tong, 1974; Hilden et al., 1974; Shamoo, 1974) restores active transport has given further credence to the belief that the enzyme is the transport system. None of the partial reactions outlined above and no other aspects of enzyme activity or ion transport have been attributed to the glycoprotein subunit. The large nonglycosylated subunit possesses both the internal phosphorylation site (Dahl and Hokin, 1974) and the externally oriented cardiac glycoside binding site (Ruoho and Kyte, 1974). However, there are at least four different types of evidence that the glycoprotein is a necessary and integral part of the enzyme. The first was provided by cochromatography with the large chain on gel filtration (Kyte, 1971) or Con A-Sepharose affinity columns (Marshall, 1976) in the presence of detergent. Second, the two chains were cross-linked with dimethyl suberimidate (Kyte, 1972), indicating their very close proximity in the membrane. Third, even though the large chain is the site which is phosphorylated, an intact enzyme is required for phosphorylation (Fahn et al., 1966; Fahn, 1968; Collins and Albers, 1972). Fourth, antibodies raised against the glycoprotein subunit inhibited both soluble (Jean et al., 1975) and membrane-bound (Rheeand Hokin, 1975) forms of the enzyme. As well as implicating the glycoprotein in the functioning of the enzyme, antibodies have provided other clues to the enzymic mechanism (Kyte, 1974). Antisera raised against whole enzyme from brain inhibit Na+-dependent phosphorylation but not K+-activated p-nitrophenylphosphatase (Askari and Rao, 1972). Antibodies to the holoenzyme from canine renal medulla (Smith and Wagner, 1975) or pig kidney (Jargensen et al., 1973) inhibit ion transport if they have access

GLYCOPROTEIN MEMBRANE ENZYMES

163

to the inner side of the membrane. Immunoglobulins from other antisera to dog kidney enzyme (Kyte, 1972) bind to the enzyme without inhibiting it. This finding makes it energetically unlikely that large movements of the enzyme through the bilayer are involved in the transport process (Singer, 1974b). Efforts to ascertain the role of the glycoprotein began only recently. For example, Churchill and Hokin (1976) suggest that slightly reduced susceptibility of the glycoprotein chain to proteolysis by trypsin or chymotrypsin in the presence of Na+ and/or K+ means that the glycoprotein may participate in binding the ions. No implications were made regarding ion translocation, presumably because of lack of evidence that this subunit traverses the membrane (although it may) and the likelihood that at least the carbohydrate moiety is fixed in an externally oriented direction (Rothman and Lenard, 1977). Experiments with lectins thus far have been contradictory and hence have contributed little to our understanding of the function of the glycoprotein. High concentrations (3 mg/ml) of Con A precipitated 40% of a Triton X-100-solubilized enzyme from dogfish salt gland without affecting activity (Marshall, 1976). Swann et al. (1975)showed marked inhibition of the eel electric organ enzyme at high concentrations of the same lectin, but only a very small proportion was reversed by a haptene inhibitor of binding of Con A. Ouabain binding, however, remained unaffected. Low Con-A concentrations (<25 pg/ml) apparently caused up to 50% inhibition of the liver plasma membrane enzyme (Luly and Emmelot, 1975). Earlier work aimed at understanding the mitogenic effect of lectins on lymphocytes showed activation of the enzyme in microsomes at 3 pg/ml Con A (Pommier et al., 1975) but inhibition at 33 pg/ml (Novogrodsky, 1972).No systematic efforts to utilize enzymic or chemical modifications of carbohydrate structure to assess its potential function in the enzyme have been reported yet. At present, it is only possible to utilize the process of elimination regarding the possible involvement of the glycoprotein subunit in either the ionophoric or regulatory properties of the enzyme complex. No knowledge of the molecular mechanisms or sites of ion binding or of the mechanism of ion translocation is available. Similarly, nothing is known about the reception of regulatory stimuli such as hormones to which the enzyme is known to be sensitive (Gavryck et al., 1975; Luly and Emmelot, 1975; Albers, 1976). In this regard there is evidence that certain cell surface receptors for polypeptide hormones such as insulin are probably glycoproteins (Cuatrecasas, 1972). Other possible roles include localization and orientation within the membrane, but little definite evidence is available.

164

J. R. RIORDAN AND G. G. FORSTNER

E. Cytochrome b5 Reductase

This enzyme, which transfers electrons from NADH to cytochrome

b5in the catalysis of fatty acid desaturation (Mathews and Czerwinski, 1976), is discussed by Sturgess et al. in this volume. With respect to its glycoprotein nature the enzyme from ox liver microsomes has been shown to stain for carbohydrate (Panfili et d . , 1971).The catalytic site is oriented toward the outer surface of the microsomal membrane (Strittmatter et d., 1972). Microsomal glycoproteins in general have their carbohydrate directed inward (Nicolson et al., 1972). Therefore this enzyme is probably a transmembrane protein. Rothman and Lenard (1977),in classifying the enzyme as an extrinsic protein, apparently did not consider the evidence for the sidedness of the substrate site and carbohydrate to be conclusive. In this regard further analysis of the carbohydrate portion of the enzyme is urgently required.

F.

Acetylcholinesterase

This enzyme and the acetylcholine receptor (de Robertis and Schacht, 1974) are membrane glycoproteins believed to be intimately involved in the control of ion currents across excitable membranes (Potter, 1970; Neumann and Nachmansohn, 1975).At cholinergic synapses acetylcholine is hydrolyzed by the enzyme and thereby prevented from exerting its action on the receptor (Ferry and Marshall, 1973).The role of this enzyme, which catalyzes the,hydrolysis of choline esters, in the function of nerve and muscle tissue has been studied for most of this century. Similar enzymes with different substrate specificities are also present in erythrocyte membranes (Berman, 1973) and in plasma (Main et aZ., 1972).Enzymes from several different sources exist as mixtures of multiple molecular forms which have been detected primarily as having different sedimentation coefficients (Rosenberry, 1976) and isoelectric points (Gurd, 1976). The enzyme isolated from the electric organ of Electrophorus eZectricus (Dudai et al., 1973; Rosenberry et d., 1974) may be present as 8, 11, 14, and 18s forms. These apparently have MWs of approximately 300,000,430,000,780,000and 1,100,000 and consist of one tetramer only, the tail plus two tetramers, and the tail plus three tetramers, respectively. No differences in specific enzyme activities or catalytic properties among these four forms are apparent. Furthermore, SDS-PAGE shows that the 14 and 18s species have the same polypeptide composition as the 11s (Rosenberry et al., 1974). The

GLYCOPROTEIN MEMBRANE ENZYMES

165

subunit monomer has a MW of approximately 75,000 (Rosenberry, 1975). It apparently consists of two glycosylated polypeptides having MWs of approximately 50,000 and 22,000 (Rosenberry et al., 1974). The various forms of the purified enzyme have been visualized in the electronmicroscope (Massoulie et al., 1971; Rieger et al., 1973). The distribution of these forms depends upon the methods of isolation and storage of the enzyme. The action of endogenous protease while standing for long periods, or added trypsin, results in a homogeneous 11s enzyme (Taylor et al., 1974), whereas high-ionic-strength extracts of fresh electric organ tissue yield a stable mixture of the 8, 14, and 18s forms which are not naturally interconvertible (Massoulie and Rieger, 1969). In addition to this multiplicity of forms the 14 and 18s species aggregate into very large structures (70-100s) at low ionic strength (Massoulie and Rieger, 1969). There is some indication that the intermolecular reactions involved in the interconversions of these different oligomeric structures may be related to the association of the enzyme with the membrane (Dudai and Silman, 1973). It has been suggested that some conditions which either favor or prevent conversion of the 14 and 18s species to the 70 to 100s aggregates may similarly affect the association of the enzyme with the membrane (Rosenberry, 1976). For example, high ionic strength, which is used to extract the enzyme from the membrane, also prevents the 14 and 18s species from aggregating into the 70 to 100s forms. It is postulated that in vivo aggregation (possibly to form the 70 to 100s complex) may prevent the lateral translational diffusion membrane proteins normally undergo, and as a result restrict the enzyme to the synaptic region where it is concentrated (Grafinus et al., 1971; Dudai and Silman, 1974). It has been proposed that the specific arrangement of the enzyme within the membrane with respect to that of the acetylcholine receptor may constitute the “basic excitation unit” (Neumann et al., 1973). If this is true, it is an excellent example of “membrane coupling” of function involving glycoproteins. Although proteases have been widely used in purification, the ability of the enzyme to be extracted with salt has led to its classification as an extrinsic as opposed to an integral membrane protein (Singer, 1974a). Nevertheless, salt extraction, particularly from some mammalian tissues, is facilitated by nonionic detergents (Hall, 1973);Dudai and Silman, 1974).A recent report claims a specific association of the enzyme with cardiolipin of the bovine erythrocyte membrane (Beauregard and Roufogalis, 1977). The multiple forms of the enzyme can also be resolved by isoelectric focusing (Brodbeck et al., 1973; Gurd, 1976). Use of this means of

166

J. R. RIORDAN AND G. G. FORSTNER

separation has been made in the few attempts to assess possible differences in the carbohydrate content of the different species. The carbohydrate content of cholinesterase from Torpedo is indicated in Table 111.In one case (Rosenberry et al., 1974) an attempt was made to quantitate the relative amounts of carbohydrate in the different polypeptide bands seen on SDS-PAGE by calculating the ratios of areas of PAS-stained to coomassie blue-stained peaks. On this fairly crude basis the contents were generally similar. Powell et al. (1973) had suggested that the carbohydrate content of the elongated tail-associated forms of the enzymes was different from that of the globular forms. While this remains a possibility, more recent work by Wiedmar et al. (1974), Gurd (1976),and Taylor et al. (1974)has failed to show any differences in terms of reactivity with plant lectins. These investigators, as well as several others, are in agreement that several different lectins are without influence on the activity of the enzyme, although binding occurs readily (Neunier et al., 1974; Bon and Rieger, 1975). Gurd (1976) examined lectin interactions with mammalian brain enzyme quite thoroughly and found that the isoelectric points of all species were shifted upward by Con-A or L. culinaris lectin, whereas WGA and R . communis lectins had little effect. The degree of binding by affinity columns of these lectins indicated differential retention with respect to lectin but not enzyme species. Brodbeck et al. (1973) found that the removal of certain sialic residues from the enzyme of plaice muscle altered the apparent substrate specificity. Neuraminidase treatment caused each of the species to shift to higher isoelectric points with no change in catalytic or inhibitory properties. However, extensive further treatments caused yet a further upward shift, and the enzyme species focusing near pH 10 lost their ability to hydrolyze aliphatic acid choline esters such as benzoylcholine and butyrylcholine while activity with acetylcholine as substrate remained undiminished. These investigators interpreted the observations in terms of “closing up” of the substrate site so that the larger, more hydrophobic substrates were no longer accommodated. The suggestion that the negatively charged sialic acid residues may contribute to the specificity of the substrate site is appealing, however, there is as yet no information on the likelihood of the uncharged sugars having similar effects. G. Alkaline Phorphatare

Alkaline phosphatases are relatively nonspecific orthophosphate monoester phosphohydrolases widely distributed in mammalian

168

J. R. RIORDAN AND G. G. FORSTNER

cells. Each is a zinc metalloenzyme, containing 4 gm/mole, with a variable Mg2+ requirement (Cathala et al., 1975a,b; Ohkubo et al., 1974) for full activity. Pyrophosphatase (Fernley and Walker, 1957; Moss et al., 1967; Brenna et al., 1975) and ATPase (Fernley and Walker, 1957; Ohkubo et al., 1974) activities have been described in some preparations. Organ-specific isoenzymes can be differentiated b y specific inhibitors, heat, and urea, and their detection has been of use in the diagnosis of malignancies and diseases of the liver and bone (Fishman, 1974). Excellent comprehensive reviews are available (Fernley, 1971; Fishman, 1974; Chen, 1976). Alkaline phosphatases are usually concentrated in plasma membranes. However, an intracellular membrane localization has been reported in the intestinal epithelium (Sandborn and Makita, 1969), placenta (Hulstaert et al., 1973), kidney proximal tubule (Moelbert et al., 1960),and eosinophilic leukocyte (Makita and Sandborn, 1970). In the liver the enzyme is present predominantly in the plasma membranes of the bile canaliculi (Fishman and Lin, 1973) and does not extend appreciably to the basal and lateral plasma membranes beyond the tight junctions. In HeLa cells (Bosmann et al., 1968; Singer and Fishman, 1974)and ovarian cancer cells (Sasaki and Fishman, 1973)the enzyme is restricted to the plasma membrane. Interestingly, when cancer cells synthesizing the Regan isoenzyme were studied, intracellular activity was dominant, suggesting that localization may vary with the stage of cell differentiation (Sasaki and Fishman, 1973). Intestinal, renal, placental, and hepatic isoenzymes have been purified extensively, almost invariably following extraction with n-butanol. In contrast to Escherichia coli alkaline phosphatase, which is not a glycoprotein (Kelley et al., 1973), mammalian alkaline phosphatases contain substantial carbohydrate (Tables I and 111). Purified intestinal alkaline phosphatase was found to be an asialoglycoprotein (Fosset et al., 1974), as expected from earlier observations which showed that electrophoretic migration of the human isoenzyme was not affected by neuraminidase (Moss et al., 1966). Although the majority of intestinal alkaline phosphatase contains no sialic acid, a neuraminidase-sensitive subfraction was found in chick duodenum (Chang and Moog, 1972), and Komoda and Sakagishi (1976a,b) found a trace of sialic acid in a human fetal intestine alkaline phosphatase. The dominant placental isoenzyme is, in contrast, a sialoglycoprotein (Ghosh et al., 1967; Badger and Sussmann, 1976). Inhibition of hemagglutination by H-1 and PR8 viruses, which depends upon binding to sialic acid groups, has been demonstrated (Ghosh and Usategui-

GLYCOPROTEIN MEMBRANE ENZYMES

169

Gomez, 1969; Ghosh et al., 1971).Ghosh et al. (1974)found that galactose, mannose, fucose, and glucose were present in molar ratios 0f4:3:0.6:0.8. Kidney alkaline phosphatase (bovine) contained a minimum of 15% carbohydrate with 11.8% hexosamine and 3.6% neutral sugars (Cathala et al., 1975a). Two pure enzymes, F, and F2, differing only in PI values, hence in electrophoretic migration, were identified. After treatment with neuraminidase, both forms behaved identically on electrophoresis, suggesting that the differences were due to their sialic acid content. The hepatic isoenzyme also appears to be a sialoglycoprotein in that its electrophoretic migration is neuraminidase-sensitive (Badger and Sussman, 1976). Badger and Sussman (1976) found 62 glucosamine residues per 1000 residues of amino acid in an hepatic preparation purified 10,000-fold, compared to 18 residues in placental alkaline phosphatase. Both human liver and intestinal alkaline phosphatases are bound by Con A, and partial purifications have been achieved with the aid of Con A-Sepharose 4B chromatography (Komoda and Sakagishi, 1976a,b). Komoda and Sakagishi (1976a,b) explored the effect of lectins on activity in some detail (see Section IV,C). Mammalian alkaline phosphatases are dimers consisting of identical subunits (Cathala et al., 1975a; Badger and Sussman, 1976). Reported MWs for the dimer have ranged from 116,000 for the human placental enzyme (Gottlieb and Sussman, 1968) to 172,000 for the bovine kidney enzyme (Cathala et d.,1975a). Monomers have been readily obtained in guanidine hydrochloride or in SDS with reducing agents (Gottlieb and Sussman, 1968; Ohkubo et al., 1974; Cathala et al., 1975a; Badger and Sussman, 1976; Malik and Buttenvorth, 1976). The human placental dimer was stable between pH 4.7 and 10.3, but dissociated at pH 2.3 and 10.5 to monomeric forms or aggregate (Sussman and Gottlieb, 1969). Intestinal (Fosset et al., 1974), kidney (Cathala et al., 1975a),andE. coli (Lazdunski and Lazdunski, 1966) alkaline phosphatases also appear to form monomers at extremes of pH. Monomeric forms have thus far proved to be inactive (Sussman and Gottlieb, 1969; Cathala et al., 1975a). Labeling with 32P after quenching the phosphorylated derivative formed with saturating concentrations of [32PlAMP,and stop-flow analysis of the hydrolysis of 2P-dinitrophenyl phosphate in the alkaline pH range, indicate that only one of two possible active sites is phosphorylated at any one instant. A mechanism has been suggested whereby each subunit contains an active site which interacts coopera-

170

J. R. RIORDAN AND G. G. FORSTNER

tively with its neighbor during hydrolysis of the phosphate ester (Chappelet-Tordo et ul., 1974; Cathala et ul., 1975b). Multiple forms of intestinal alkaline phosphatase have been reported by many investigators (Srikantaiah and Radhakrishnan, 1970; Saini and Done, 1972; Chang and Moog, 1972; Nayudu and Hercus, 1974; Malik and Butterworth, 1976). One form appeared to be a sialoglycoprotein in the chicken (Chang and Moog, 1972),but neuraminidase was without effect on three forms found by Srikantaiah and Radhakrishnan (1970) and Malik and Butterworth (1976).The heterogeneity may be due to a combination of aggregation (Chang and Moog, 1972), accompanying lipid (Nayudu and Hercus, 1974), limited proteolysis occurring during preparation (Malik and Butteworth, 1976), or carbohydrate microheterogeneity (Malik and Butterworth, 1976). H. Brush Border Hydrolases

The brush border plasma membranes of the intestine and kidney contain alkaline phosphatase and quite possibly many of the other glycoprotein enzymes already discussed. The following section concentrates, however, on a group of enzymes almost exclusively identified with these membranes because of their high activity in this region. In a few cases, such as hepatic aminopeptidase, where highly purified enzymes from other organs have been studied, their description is included with that of their brush border conterpart. 1. THE SUCRASE-ISOMALTOSE COMPLEX

Sucrase (sucrose glucohydrolase) and isomaltase (oligo-1,6-glucosidase) activities exist in a two-enzyme complex (Kolinska and Semenza, 1967)attached to the outer surface of the intestinal brush border membrane (Johnson, 1967). Two active centers are present (Kolinska and Semenza, 1967), and the separation of individual sucrase and isomaltase units has been achieved (Cogoli et al., 1973; Conklin et aZ., 1975; Braun et al., 1975). More properly the complex might be considered to consist of two intimately linked maltases (a-1,4-glucosidases), since each unit probably has maltase activity (Kolinska and Semenza, 1967), with separate specificity for either sucrose or isomaltose. The two units are tightly coupled. Both are absent in the early stages of 'intestinal development and subsequently appear and increase simultaneously during maturation (Rubino et al., 1964). Their topographical distributions along the length of the intestine are

GLYCOPROTEIN MEMBRANE ENZ-YMES

171

also identical (Antonowicz et al., 1974; Rubino et aZ., 1964; Asp et al., 1975). The two units seem to be controlled by a common gene, since hereditary sucrase deficiency is always associated with a corresponding isomaltase deficiency (Auricchio et al., 1965a; Eggermont and Hers, 1969). However, Eggexmont and Hers (1969)reported evidence of a residual isomaltase unit in the sucrase-isomaltase-deficient intestine. There is a possibility therefore, as suggested by Conklin et al. (1975), that isomaltase may be a product of an independent gene but dependent upon association with sucrase for its full expression. The complex is thought to be part of a particle having a diameter of 40-60 A which can be visualized on the exterior surface of the microvillous plasma membrane by negative staining (Oda and Seki, 1966; Johnson, 1967; Maestracci, 1976) and which is readily removed by exposure to papain (Johnson, 1967; Maestracci, 1976). These particles are not apparent after freeze-etching, which demonstrates instead a closely packed unbranched filamentous coat projecting from the external leaflet of the plasma membrane (Swift and Mukherjee, 1976). The shape of the complex, if indeed it projects from the surface of the plasma membrane, is therefore in doubt. Gitzelmann et al. (1970) argued that rabbit sucrase-isomaltase is an integral membrane protein on the basis of experiments with a ferritin-labeled antibody to the complex. Unfortunately the membranes were trypsinized and crosslinked with Formalin before exposure to the labeled antibody, and therefore the proximity of label and membrane on which their conclusions are based may have been influenced by the collapse and entanglement of surface proteins. Sucrase is not present in active form in immature cells of the intestinal crypt (Nordstrom et al., 1967; Das and Gray, 1969; Herbst and Koldovsky, 1972) or on the basolateral membranes of the mature enterocyte (Fujita et al., 1972). Sucrase-isomaltase is similar to another intestinal brush border enzyme, lactasephlorizin hydrolase (Diedrich, 1973), in that it is not present on renal brush borders (Berger and Sacktor, 1970). The papain-solubilized complex has been isolated in a state of high purity from rabbit (Kolinska and Semenza, 1967; Cogoli et al., 1972; Mosimann et al., 1973) and human (Conklin et al., 1975) intestine. The rabbit enzyme had a MW of 221,000 and an axial ratio of 4 (Mosimann et al., 1973),suggesting that its shape was relatively elongated. An axial ratio of 8.0 was found for the human enzyme (Conklin et al., 1975). The carbohydrate content was approximately 14%, consisting of glucosamine, galactosamine, mannose, galactose, glucose, and fucose (Cogoli et al., 1972) (Table I and Table 111). No sialic acid was found, as is also characteristic of maltase (Kelly and Alpers, 1973a)and

172

J. R. RIORDAN AND G. G. FORSTNER

alkaline phosphatase (Fosset et al., 1974).Nevertheless, the isoelectric point was low (ph 4.0), presumably because of the high content of acidic amino acid residues (Cogoli et al., 1972).The rabbit complex has also been isolated in a detergent-solubilized form using Triton (Sigrist et al., 1975).The amino acid and carbohydrate composition was almost identical to that of the papain-solubilized enzyme. Treatment of the papain-solubilized enzyme with 5 M guanidine hydrochloride and 0.1 M mercaptoethanol at 70°C dissociated the complex into two inactive subunits with a MW of 112,000 (Mosimann et al., 1973). Prolonged dialysis against a Verona1 buffer (pH 9.0) followed by a 30-minute exposure to pH 9.6 at 37°C resulted in the separation of an isomaltase unit with approximately twice the specific activity of the original complex, plus a second unit with very low sucrase activity (Cogoli et al., 1973).The isomaltase subunit had a MW of 113,000 by sedimentation velocity analysis and was not split further by urea, SDS, thiol reagents or alkylation (Mossimann et d.,1973). A more satisfactory separation of the catalytically active subunits was achieved by treating the complex with citraconic anhydride in order to acylate free amino groups (Braun et al., 1975). A sucrase unit with 50% activity was obtained and, following deacylation under mildly acidic conditions, activity was increased to approximately 75% of the original level. Both subunits are glycoproteins (Quaroni et al., 1975). Interestingly the isomaltase subunit obtained through citraconylation contained more sugar than that obtained by alkali treatment (Braun et al., 1975), suggesting that some of the oligosaccharide chains may be O-glycosidically linked to the protein core. Conklin et al. (1975) purified sucrase-isomaltase from human autopsy material, achieving solubilization by autolysis of the intact tissue for 1 hour at 37°C. Under these conditions small amounts of free sucrase and isomaltase are formed with MWs of 130,000 and 120,000, respectively. Partial dissociation to subunits was achieved by dialysis of the holoenzyme in 6 M urea and 0.01 M P-mercaptoethanol for 24 hours. The detergent-solubilized form of the rabbit enzyme has been studied superficially (Sigrist et al., 1975). It has a pronounced tendency to self-aggregate, but after digestion with papain yields a complex which is similar to the papain-solubilized form, plus low MW fragments. Recent work suggests that an apolar fragment is released from the N-terminus of the isomaltase unit, indicating that it is this unit which is customarily anchored to the plasma membrane (Semenza, personal communication).

GLYCOPROTEIN MEMBRANE ENZYMES

173

2. MALTASE-GLUCOAMYLASE

The enzyme is a broad-specificity, neutral a l - 4 glucosidase very similar in substrate specificity to the well-characterized acid a l - 4 glucosidase of liver (Jeffrey et al., 1970) and muscle (Palmar, 1971) lysosomes. In contrast to pancreatic amylase, intestinal maltase-glucoamylase liberates glucose from starch and glycogen (Alpers and Solin, 1970). In humans the enzyme has the highest affinity for linear oligosaccharides containing nine al-4-linked glucose residues (Kelly and Alpers, 1973a). In contrast, in rabbits, the highest affinities are reserved for maltose and starch (Sivakami and Radhakrishnan, 1976). A small, but constant a l - 6 glucosidase activity is also present (Sivakami and Radhakrishnan, 1976). 'The enzyme is exposed to the outer surface of the brush border membranes, and its active portion is readily removed by papain without affecting the composition and structure of the lipid bilayer (Johnson, 1967; Louvard e t al., 1975a). A similar enzyme is present on the external surface of brush borders of the renal tubule (Stevenson, 1973), and neutral a1,4 glucosidase activity has been described in liver (Lejeune et al., 1963) and muscle membranes (Angelini and Engel, 1973). Maltase-glucoamylase is the largest component of the intestinal brush border membrane in the rat (Galand an4 Forstner, 1974), hamster (Critchley et al., 1975), and rabbit (Sivakanii and Radhakrishnan, 1976). The enzyme accounts for approximately 27% of the total maltase activity of the human intestine (Kelly and Alpers, 1973a) and perhaps 80% of the maltase activity of the rat intestine (Galand and Forstner, 1974). It is the most heat-stable of the intestinal disaccharidases (Auricchio et al., 1965b), resisting inactivation for 40 minutes at 50°C (Galand and Forstner, 1974). In humans (Kelly and Alpers, 1973a)and monkeys (Seetharam et al., 1970) both maltase and glucoamylase are identically heat-inactivated. Human maltase-glucoamylase appears to have a single catalytic site for both starch and maltase (Kelly and Alpers, 1973a). In rabbits, kinetic studies with mixed substrates and a variety of inhibitors suggest that there may be two active sites (Sivakami and Radhakrishnan, 1976). MW estimates have varied depending on species and techniques. In the rat, the MW of the papain-solubilized enzyme appears to be in the range 400,000-500,000 (Galand and Forstner, 1974; Maestracci et al., 1975; Flanagan and Forstner, 1978)and in humans 200,000-300,000 (Kelly and Alpers, 1973a). Human maltase-glucoamylase contained 32-38% carbohydrate by weight, and on the basis of its amino acid

1 74

J. R. RIORDAN AND G. G. FORSTNER

and carbohydrate composition had a calculated partial specific volume of 0.684, more reminiscent of mucins than most enzymes (Kelly and Alpers, 1973a).A very high fucose content of 20% was found, with 0.6% mannose, 5%galactose, 5%glucosamine, and 2% galactosamine. There was no sialic acid, a feature which this enzyme shares with several intestinal brush border hydrolases including alkaline phosphatase (Fosset et al., 1974) and sucrase (Cogoli et al., 1972). The oligosaccharides attached to the enzyme have blood group activity. A highly pure rat maltase-glucoamylase contained 16.7% carbohydrate (Kelly and Alpers, 1973b). The spectrum of individual monosaccharides was much more typical of other membrane glycoproteins (Table I). We have also purified the rat enzyme (Flanagan and Forstner, 1978) and found a similar carbohydrate content of 20%, consisting of fucose, mannose, galactose, glucosamine, and galactosamine in a molar ratio of 1:3:1:4:1. In both preparations analyzed by Kelly and Alpers (1973b) small amounts of xylose were found, possibly representing contamination with proteoglycans or bacterial polysaccharides. Xylose was not present in our preparation (Flanagan and Forstner, 1978). Rat maltase dissociates readily when heated beyond 87°C in the presence of SDS (Flanagan and Forstner, 1976) and when subjected to an acid milieu below pH 5.0 (Flanagan and Forstner, 1977). Heat and pH inactivation profiles coincide exactly with dissociation. Unfortunately dissociation complexes aggregate and are difficult to study. In SDS or urea polyacrylamide gels, five inactive species varying in MW between approximately 143,000 and 468,000 can be separated. This pattern is achieved by both approaches to dissociation and appears to represent a series of very stable aggregates of one or two monomeric species. During gradual heat dissociation active intermediate complexes of MW lower than that of the original enzyme are formed. Of major interest is the possibility that maltase and glucoamylase activities are separable, since acid a-glucosidase of liver can be dissociated by 3 it4 urea into glucoamylase and maltase subunits (Belenki and Rosenfeld, 1972). This problem has not been resolved to date. A detergent-solubilized form of the enzyme was partially purified by Maroux and Louvard (1976). On cleavage by trypsin an apolar segment with an approximate MW of 10,000 was obtained. The active portion of the enzyme may be linked to such an apolar anchor in the lipid bilayer. Confirmation awaits evidence that the detergent-solubilized enzyme can be obtained with the same degree of purity as the proteolytically solubilized form since entrapment of neighboring bilayer components within detergent micelles is not easily excluded.

GLYCOPROTEIN MEMBRANE ENZYMES

1 75

3. THE LACTASE-PHLORIZIN HYDROLASE COMPLEX The ability to digest the milk sugar lactose is an almost obligatory requirement for all suckling mammals. The enzyme responsible for this activity is an intestinal brush border P-galactosidase with a pH optimum of 5.5-6.0 (Asp and Dahlqvist, 1968; Gray and Santiago, 1969) which attains its highest activity at birth and tends in most species to diminish sharply at weaning (Koldofsky, 1969).It is not present in the kidney brush border (Berger and Sacktor, 1970; Diedrich, 1973). An important feature of intestinal lactase, analogous to the coupling of sucrase-isomaltase and maltase-glucoamylase activities, is its close association with a second enzymic activity specific for phlorizin hydrolase, a P-glucoside (Malathi and Crane, 1969; Schlegel-Haueter et al., 1972; Colombo et aZ., 1973). The presence of two active centers has been established by competitive substrate experiments (LorenzMeyer et al., 1972; Colombo et al., 1973; Ramaswamy and Radhakrishnan, 1975b), response to inhibitors (Malathi and Crane, 1969; Ramaswamy and Radhakrishnan, 1975b),and heat inactivation (Schlegel-Haueter et al., 1972). Highly purified preparations from the rat (Colombo et ul., 1973; Leese and Semenza, 1973; Birkenmeier and Alpers, 1974), rabbit (Ramaswamy and Radhakrishnan, 1975a), and monkey (Ramaswamy and Radhakrishnan, 1975b) all possessed phlorizin hydrolase activity, and in the rat the ratio of lactase to phlorizin hydrolase remained constant over a 70-fold range of purification (Colombo et al., 1973). Both lactase and phlorizin hydrolase are more active at birth than in the adult (Colombo et al., 1973),and both activities are absent in human lactase deficiency (Lorenz-Meyer et al., 1972). Thus there is good evidence that the two enzymes are genetically and developmentally linked. However, Birkenmeier and Alpers (1974) found that the two activities varied somewhat independently in the neonatal period in the rat, while Maestracci et al. (1975) reported migration at slightly different rates on polyacrylamide gels containing brush border membrane enzymes solubilized by SDS. It is still not certain therefore that fortuitous copurification, perhaps of two enzymes with a common subunit such as is seen with hexosaminidases A and B (see Section 11,1,3), can be excluded. Leese and Semenza (1973) have suggested that glycosylceramides may constitute the natural substrate for phlorizin hydrolase. The coincidence of phlorizin hydrolase and glycosylceramidase activity may be species-dependent, since in the monkey the lactase site appears to hydroiyze cerebroside more rapidly than the phlorizin hydrolase site

176

J. R. RIORDAN AND G. G. FORSTNER

(Ramaswamy and Radhakrishnan, 1975b). Like lactose, phlorizin is hydrolyzed optimally at neutral pH (Malathi and Crane, 1969; Ramaswamy and Radhakrishnan, 1975a). Rat phlorizin hydrolase appears to be very similar to the glycosyl and galactosylceramide-cleavingenzyme isolated from rat intestine by Brady et al. (1965).A glycosylceramidase with similar properties, including optimal activity at pH 6.0, has been isolated and purified from human placenta (Pentchev et al., 1973). Although lactose (Brady et al., 1965) and cellobiose (Pentchev et al., 1973; Brady et al., 1975) inhibited glycosylceramidase activity, the presence of associated lactase activity was not examined with either preparation. The lactase-phlorizin hydrolase complex is efficiently released from the external surface of the brush border membrane by papain (Maestracci, 1976). In fact, with the exception of Brady et al. (1965), who solubilized their glycosylceramide-cleavingenzyme with sodium cholate, all investigators have used proteolytic enzymes to obtain soluble preparations for purification. Rat lactase-phlorizin hydrolase contained 17% carbohydrate (Birkenmeier and Alpers, 1974), and no sialic acid was found. Ramaswamy and Radhakrishnan (197513) reported that monkey lactasephlorizin hydrolase interacted strongly with Con A. The rat complex was rich in acidic amino acids and contained only 35% hydrophobic residues (Birkenmeier and Alpers, 1974). In human intestine Gray and Santiago (1969) found a MW of 280,000 by sucrose density gradient centrifugation. Birkenmeier and Alpers (1974) found a dominant species by PAGE electrophoresis after solubilization with SDS and 1% mercaptoethanol with a MW of 132,000. The complex may therefore contain two subunits of similar size. Brady et al. (1965) and Pentchev et al. (1973) extracted glycosylceramidase activity with sodium taurocholate. Monomers of 60,000 MW were found with the purified placental enzyme in SDS polyacrylamide gels containing mercaptoethanol. 4. ENTEROKINASE Enterokinase is a serine endopeptidase present in the intestinal brush border membrane (Nordstrom and Dahlqvist, 1971; Louvard et al., 1973; Schmitz et al., 1974) and perhaps to a lesser degree in other subcellular membrane fractions (Schmitz et al., 1974). Enterokinase initiates pancreatic proteolysis within the intestinal lumen by cleaving Lys-Ile residues 6 and 7 in the N-terminus of trypsinogen, releas-

GLYCOPROTEIN MEMBRANE ENZYMES

1 77

ing active trypsin and an inactive hexapeptide (Maroux et al., 1971). Specificity is provided by the (Asp),-Lys sequence of residues 2 to 6 for which enterokinase has a higher affinity than trypsin or other intestinal endopeptidases (Maroux et al., 1971). Catalysis is enhanced by bile salts in concentrations in excess of the critical micellar concentrations (Hadorn et al., 1971; Nordstrom, 1972). Enterokinase deficiency occurs as an inherited recessive disorder in humans and is associated with a dramatic failure of protein digestion (Haworth et al., 1975). Activation of trypsinogen occurs at the luminal surface of the intestinal membrane, and therefore enterokinase is clearly an ectoenzyme. It is released with extreme facility by papain (Nordstrom, 1972; Louvard et al., 1973), more rapidly than lactase, maltase, or sucrase (Louvard et al., 1973).In contrast to other brush border enzymes trypsin, chymotrypsin, and bile salts are also effective solubilizing agents (Nordstrom, 1972; Hadorn et al., 1971). Intestinal perfusion with the enteric hormone CCK-PZ releases enterokinase (Gotze et al., 1972). The mechanism is unknown but is associated with the simultaneous release of alkaline phosphatase (Warnes et al., 1969; Gotze et al., 1972). The enzyme has been purified 1000-fold from porcine intestine following extraction with 2.5% deoxycholate (Baratti et al., 1973). It contained 37% carbohydrate, consisting of 20% neutral sugar, 15% amino sugar, and 2% sialic acid. A MW of 197,000 was determined by sedimentation equilibrium analysis. After reduction two subunits of 134,000 and 62,000 MW were found on SDS-PAGE. Both units appeared to be equally rich in carbohydrate. Neither unit was dissociated further by treatment with 6 M guanidine hydrochloride, and only the smaller subunit was labeled b y [32Pldii~~pr~pylpho~ph~flu~ridate ([32PlDFP).It seems probable therefore that the enzyme consists of two discrete units, catalytic activity being restricted to the smaller. Barnes and Elmslie (1974) showed that the trypsinogen-specific site which recognizes the (Asp), sequence can be destroyed by heating without affecting the catalytic site. The specificity site has not been localized to either subunit.

5. OLIGOAMINOPEPTIDASE Intestinal (Maroux et al., 1973; Kim and Brophy, 1976) and renal (Thomas and Kinne, 1972) brush border membranes contain neutral aminopeptidases which catalyze the hydrolysis of N-terminal L-amino

178

J. R. RIORDAN AND G. G. FORSTNER

acids (Kim et al., 1976) from peptide substrates ranging in length from two to eight amino acids. The greatest activity is reserved for tripeptides or longer (George and Kenny, 1973; Peters, 1973; Kim et al., 1976), so that the enzymes are true oligopeptidases. There is no C-terminal or endopeptidase activity (Peters, 1973; Kim et al., 1976). Oligoaminopeptidases are metalloenzymes containing 2 gm-atoms of zinc per mole (Wacker et al., 1971; Maroux et al., 1973). Aminoacyl P-napthylamides and p-nitroanilides are hydrolyzed rapidly and constitute convenient but rather nonspecific substrates. Intestinal oligoaminopeptidase is an ectoenzyme anchored to the plasma membrane by a nonpolar segment (Maroux and Louvard, 1976; Louvard et al., 1976; see Section V,B,4,a). The enzyme is rapidly released from brush border membranes by papain in both the intestine (Maroux et al., 1973) and the kidney (George and Kenny, 1973).Louvard et al. (1975b) showed that the free and bound forms of the intestinal enzyme were inhibited to the same degree by a specific aminopeptidase antibody. It is very likely therefore that the active site is fully exposed at the external surface of the bilayer. Several oligoaminopeptidases have been purified. Hog intestinal brush border oligoaminopeptidase was solubilized from membrane fragments by trypsin and purified by ammonium sulfate fractionation and anion-exchange chromatography (Maroux et al., 1973). The enzyme was a glycoprotein, containing 23% carbohydrate and, in contrast to intestinal alkaline phosphatase (Fosset et al., 1974) sucraseisomaltase (Cogoli et al., 1972), maltase-glucoamylase (Kelly and Alpers, 1973a), sialic acid was present. The MW as determined by equilibrium centrifugation using the meniscus depletion method was approximately 248,000. After treatment with SDS and mercaptoethanol for 18 hours at room temperature four bands were seen on PAGE with approximate MWs of 129,000,98,000,91,000, and 45,500. Kim et al. (1976) isolated the enzyme from the rat intestine following solubilization with Triton X-100. Two forms, both glycoproteins, were found which appeared to differ only in carbohydrate content. The faster F form contained 19% carbohydrate, and the S or slow form, 23%. Sialic acid residues may have accounted for the difference in electrophoretic mobility. The F enzyme contained 9 moleslmole, while the S form contained 6 moles/mole. Glucosamine, galactosamine, glucose, mannose, galactose, fucose, and sialic acid were identified in an approximate molar ratio of 1:0.1: 1:0.5:0.5 :3 :0.1 in the F enzyme. The S enzyme contained higher quantities of mannose, galactose, and galactosamine. The F and S enzymes were immunologically identical and could not be distinguished as to pH optimum, heat stability, sub-

GLYCOPROTEIN MEMBRANE ENZYMES

179

strate specificity, or metal ion requirements, suggesting that the minor differences in carbohydrate composition had little to do with function (Kim et aZ.,1976). The MW of the F and S enzymes was 280,000, and in SDS with mercaptoethanol two subunits of 140,000 MW appeared. PHA-P gave a precipitin line with the two enzymes on immunodiffusion plates, whereas Con A, WGA, and ricin failed to do S O (Kim et al., 1976). PHA-P inhibited activity to a maximum of 30% for peptidase F and 40% for peptidase S, but inhibition varied somewhat with different substrates. Peptidase S was also precipitated more completely by the agglutinin. These differences may be the result of larger amounts of terminal GlcNAc residues in peptidase S but are interesting because they suggest that free movement of carbohydrate side chains may be necessary for optimum catalytic activity. The renal brush border enzyme has been designated aminopeptidase M (George and Kenny, 1973). It is probably identical to the enzyme purified by Wacker et d . (1971) from pig kidney particulate material after solubilization with trypsin. The enzyme contained 20% carbohydrate, and the MW was 280,000. After prolonged exposure to SDS plus mercaptoethanol, three bands were separated on SDSPAGE with approximate MWs of 60,000, 100,000, and 140,000. The identity of the intestinal and renal enzymes was established beyond reasonable doubt by Vannier et d.(1976),who isolated and compared the oligoaminopeptidases from porcine intestinal and renal brush borders under identical conditions both in the trypsin- and detergent-solubilized forms. The physical and structural properties of the two enzymes were found to be remarkably similar (Table IV). With the use of a quantitative immunological technique, six crossreacting determinants were identified in the two enzymes, indicating a very high degree of homology. Polar peptides of approximately 9000 MW and similar amino acid composition (polarity indexes 35 and 37) could be cleaved from the detergent solubilized forms of both enzymes, indicating that the portions of the enzyme buried in the membrane are also likely to be quite similar. Starnes and Behal (1974) purified a high MW aminopeptidase from human liver following solubilization by prolonged autolysis. It was a zinc metalloenzyme activated by Co2+(Garner and Behal, 1974). Unfortunately, the substrate specificity for oligopeptides was not assessed. The liver aminopeptidase contained 17.5%carbohydrate, consisting of 4.3% glucosamine, 9.0% hexose, and 4.1% sialic acid. The MW was 235,000 by sedimentation equilibrium analysis, and the monomer in denaturing solvents had a MW of 118,000.

180

J. R. RIORDAN AND G. G. FORSTNER

TABLE IV COMPARISON OF SOME MOLECULAR PROPERTIES OF RENAL AND INTESTINAL AMINOPEPTIDASES~ Aminopeptidase Property

Renal

Intestinal

Molecular weight (Yphantis), 6 = 0.693 for both proteins Sugars, trypsin form (%) Neutral sugar Amino sugar Sialic acid Number of subunits, trypsin form MW of subunits

233,000

248,000

14 7 1.5 3 130,000 96,000 49,000

15 8 0.3 3 130,000 96,000 49,000

(Gly),-Ser Gly-Ser-Val

(Ala),-Ser Ala-Ser-Val

N-Terminal residues Detergent form Trypsin form From Vannier et al. (1976).

6. ~-GLUTAMYLTFUNSPEPTIDASE y-Glutamyltranspeptidase catalyzes the transfer of a y-glutamyl moiety from multiple peptide donors to a wide range of amino acid and peptide acceptors. The enzyme is present on plasma membranes of many epithelial tissues which seem to be particularly active in transport and absorption, such as the brush border membranes of the intestine and kidney, and plasma membranes from the choroid plexus, ciliary body, bile ducts, and intrapancreatic ductules of the pancreas (Meister et al., 1976).The enzyme is linked functionally to the y-glutamyl cycle (Meister, 1973; Orlowski and Meister, 1970),the degradation of extracellular glutathione (Elce and Broxmeyer, 1976), and the biosynthesis of mercapturic acids (Elce, 1970).Meister and colleagues have argued that it might also translocate amino acids across biological membranes (Meister, 1973; Orlowski and Meister, 1970).y-Glutamyltranspeptidase and phosphate-independent, maleate-stimulated glutaminase appear to be identical enzymes in the rat (Tate and Meister, 1975; Curthoys and Kuhlenschmidt, 1975).In the presence of maleate (Tate and Meister, 1975), or below pH 6.0 (Curthoys and Kuhlenschmidt, 1975),y-glutamyltranspeptidase functions principally as a glutaminase, perhaps contributing to renal ammonia formation during

GLYCOPROTEIN MEMBRANE ENZYMES

181

acidosis. Renal y-glutamyltranspeptidase displays no phosphate-independent glutaminase activity in humans (Miller et al., 1976). More detailed information is presenfed in a recent review (Meister et al., 1976). y-Glutamyltranspeptidase is a membrane ectoenzyme. It is readily released from intestine (Auricchio et al., 1972) and renal brush borders (Curthoys and Kuhlenschmidt, 1975) by papain. Incubation of isolated kidney cells with papain results in selective release of the enzyme without the corresponding appearance of cytoplasmic markers (Kuhlenschmidt and Curthoys, 1975). Following the initial description of the enzyme by Hanes et al. (1950) detergent-solubilized preparations from kidney were purified approximately 100-fold by Szewczuk and Baranowski (1963) and Orlowski and Meister (1965). In retrospect neither of these preparations was pure. In 1968 Katunuma et al. (1968) isolated a phosphate-independent glutaminase released from an ammonium sulfate precipitate of rat kidney homogenate by bromelain, which was subsequently purified 817-fold. Tate and Meister (1975) showed that treatment of the detergent-solubilized enzyme with bromelain produced a smaller species with an altered composition and an 8-fold higher specific activity. The light or bromelain-treated form from rat kidney has a MW in the range 68,000-70,000 (Tate and Meister, 1975). Leibach and Binkley (1968)prepared the enzyme from hog kidney by ficin digestion and found a MW of 80,000. The human kidney enzyme, extracted with 1% sodium deoxycholate and purified 740-fo1d7had an apparent MW of 90,000 on Sephadex-G2OO chromatography and SDS-PAGE (Miller et al., 1976). Hog, bovine, and human kidney y-glutamyltranspeptidases (Szewczuk and Baranowski, 1963; Orlowski and Meister, 1965; Szewczuk and Connell, 1964; Miller et al., 1976) have been characterized as glycoproteins. Con A binds both heavy (detergentsolubilized) and light forms of the rat enzyme (Tate and Meister, 1975). Chromatography on Con A-Sepharose has been used by three groups of investigators to purify the enzymes (Tate and Meister, 1975; Hughey and Curthoys, 1976; Miller et al., 1976).The light bromelaintreated kidney enzyme contained 18.5% carbohydrate (10% hexose, 7% aminohexose, and 1.5% sialic acid), whereas the carbohydrate content of the detergent form was slightly higher, particularly that of hexosamine. The percentage of polar amino acids in the detergent- and bromelain-solubilized forms did not differ significantly (Tate and Meister, 1975). Hughey and Curthoys (1976) recently succeeded in purifying a Triton X-100-solubilized heavy form 300-fold in the rat to a state with approximately two-thirds of the specific activity of a corresponding pa-

182

J. R. RIORDAN AND G. G. FORSTNER

pain-solubilized light enzyme (Hughey and Curthoys, 1976). The Triton X-100-solubilized transpeptidase was soluble only in the presence of detergents and aggregated just below the critical micellar concentration of Triton X-100. By determining the amount of Triton X-100 bound by the enzyme in solution, and assuming a hybrid partial specific volume calculated from that of the Triton X-100 micelle and the amino acid and carbohydrate composition of the papain-solubilized enzyme, a MW of 169,000 was found for the Triton-enzyme complex. On the basis of the Triton-binding data, the glycoprotein portion of this MW was calculated to be 87,000. In contrast, the papain-solubilized enzyme had a MW of 69,000, suggesting that it lacked a peptide of approximately 18,000 MW which was present in the detergent-solubilized enzyme. Although this evidence suggests that the detergentsolubilized enzyme may possess a sizable nonpolar peptide responsible for Triton binding and possibly for holding the enzyme in the brush border membrane, its significance is clouded by the extrapolations required to determine the MW of the detergent-enzyme complex. In fact, when the complex was subjected to SDS-PAGE in the presence of reducing agents, two peptides with MWs of 27,000 and 54,000 were formed which were almost identical to peptides of 27,000 and 51,000 MW obtained with the papain-solubilized enzyme. Although the SDS-PAGE results suggest that papain could conceivably cleave a peptide of 3000 daltons from the larger subunit of the detergent-solubilized enzyme, the conclusion that such a peptide exists seems somewhat premature. It is perhaps significant that Miller et al. (1976) were unable to show that papain altered the mobility on SDSPAGE of their highly purified human kidney enzyme. The light subunit both in rat (Inoue et al., 1977; Tate and Meister, 1977) and human (Tate and Meister, 1977) kidney contains the y-glutamyl binding site, whereas the “large” subunit stains much more strongly with the PAS reagent (Hughey and Curthoys, 1976) and presumably contains the majority of the carbohydrate. If the carbohydrate content can be equated with maximal interaction with an aqueous environment, these observations suggest that the light or “catalytic” subunit may be more closely linked to the membrane and presumably more favorably situated for a translocating role. 7. NEUTRAL METALLOENDOPEPTIDASEO F RABBIT KIDNEY BRUSH BORDER This enzyme was initially identified in the brush border of the renal tubule by George and Kenney (1973)and purified by Kerr and Kenney

GLYCOPROTEIN MEMBRANE ENZYMES

183

(1974a), using its affinity for the B chain of insulin as a convenient assay. Like intestinal enterokinase it is a neutral endopeptidase, but unlike enterokinase it is not affected by DIFP (Kerr and Kenney, 1974b) and prefers bonds involving the a-amino group of hydrophobic residues provided they are not C-terminal. The enzyme was solubilized by treatment with toluene and trypsin (Kerr and Kenney, 1974a) and chromatographed on Sephadex G-200, DEAE-cellulose, and hydroxyapatite columns until homogeneous as judged by PAGE and ultracentrifugation. The enzyme gave a strongly positive reaction for carbohydrate with the PAS reagent. Hexose was determined in three preparations with the phenol-sulfuric acid method. A mean of 165 pg hexose per milligram of protein was found. The enzyme appears to consist of a single peptide chain of 93,000 MW by sedimentation equilibrium, which was not dissociated further by thiol reagents, SDS, or guanidine hydrochloride. One zinc atom per mole was found by atomic absorption spectrophotometry, and this metal was the most active in restoring activity after chelation. 1. Lysosomal Acid Hydrolases

A great many acid hydrolases appear to be glycoproteins, as judged by their altered electrophoretic mobility after treatment with neuraminidase (Goldstone et al., 1971, Goldstone and Koenig, 1970; Stevens et al., 1973; Norden and O’Brien, 1973; Beutler et al., 1973; Stevens et aZ., 1976) and their absorption to immobilized Con A (Beutler et al., 1975b, Bishayee and Bachhawat, 1974; Yang and Srivastava, 1975; Balasubramanian and Bachhawat, 1975, 1976; Norden et al., 1974). Several have also been purified to a state of homogeneity, permitting analyses of their carbohydrate content. The following is a brief review of a few of the best characterized enzymes. More detailed information on general aspects can b e found in excellent treatises (Dingle and Fell, 1973; Hers and Van Hoof, 1973). 1. P-GLUCURONIDASE

Lysosomal P-glucuronidase from liver is a glycoprotein. In the rat, a preparation purified 8400-fold contained 0.72% hexosamine (Stahl and Touster, 1971)according to determinations made after hydrolysis under conditions which were not optimized. In bovine liver, a preparation purified 4400-fold (Plapp and Cole, 1966) contained 3-6% neutral hexose plus some glucosamine.

184

J. R. RIORDAN AND G. G. FORSTNER

Extraction does not appear to be difficult. Plapp and Cole (1966) encouraged autolysis by incubating the crude homogenate at 37°C for 11 days, but Stahl and Touster (1971) achieved 90% extraction of lysosomes with a dilute 5 mh4 tris-phosphate buffer at 4°C for 24 hours. The rat enzyme has a pH optimum of 4.5 with 10+ M phenolphthalein glucuronide as substrate and a PI of 5.8. A MW of 265,000280,000 was calculated from sucrose density gradient centrifugation, gel filtration, and PAGE techniques (Stahl and Touster, 1971) and agrees well with the estimate of 280,000 obtained by equilibrium ultracentrifugation (Plapp and Cole, 1966).After incubation at 37°C for 90 minutes in 1% SDS and 1% mercaptoethanol a single band of MW 75,000 was found on SDA-PAGE, suggesting that the enzyme is normally a tetramer. In an interesting study Plapp and Cole (1967)demonstrated impressive charge heterogeneity of bovine liver glucuronidase on DEAE-cellulose and showed that retention correlated with a gradual increase in hexose content. 2. GM,-GANGLIOSIDE-P-GALACTOSIDASE GM,-ganglioside-P-galactosidase removes the terminal galactose residue from GM,-ganglioside and lactosylceramide but is relatively inactive against galactosylceramide (Callahan and Gerrie, 1975; Tanaka and Suzuki, 1977).Absence of this activity accounts for a fatal lysosoma1 storage disease of children associated with tissue accumulation of GM,-ganglioside and several polysaccharides (Okada and O’Brien, 1968; O’Brien et al., 1965; Suzuki, 1968; Callahan and Wolfe, 1970). Distler and Jourdian (1973) showed that the enzyme from bovine testes removed galactose from desialyzed orosomucoid and from the linkage region of a nasal proteoglycan “core” preparation as well as keratin sulfate. Optimum activity occurs at acid pH (Tanaka and Suzuki, 1977). A second acid galactosidase, which is relatively inactive with GM,-ganglioside but hydrolyzes galactosyl- and lactosylceramides (Tanaka and Suzuki, 1977), has been described but has not yet been characterized as a glycoprotein. Norden and O’Brien (1973) found two forms (A and B) of GMl-ganglioside-P-galactosidase in human liver on agarose gel filtration and showed that the electrophoretic migration of both was reduced by prior treatment with neuraminidase. Subsequently they demonstrated that both forms were bound in excess of 95% to Con A-agarose columns and specifically eluted with a-methyl-D-mannoside (Norden and O’Brien, 1974). Coelution of GM,-ganglioside-P-galactosidase and 4-methylumbelliferyl-~-galactosidase activities was observed. In

GLYCOPROTEIN MEMBRANE ENZYMES

185

contrast to the acidic activity, neutral hepatic p-galactosidase was not bound by Con A-agarose (Norden and O’Brien, 1974).In experiments with immobilized WGA only 60% of the enzymic activity was specifically bound, suggesting that the acid P-galactosidases may exhibit oligosaccharide heterogeneity. The A form was purified 17,000-fold from human liver (Norden et al., 1974).It had an apparent MW on SDS gel electrophoresis of 65,00075,000. Like the acid P-galactosidase isolated from intestine by Asp and Dahlqvist (1968), it was inhibited by p-chloromercuribenzoate and hydrolyzed lactose as well as synthetic P-galactosides. P-Fucosides and a-L-arabinosides were also hydrolyzed, but not galactosyl- or lactosylceramides. Antisera to the A enzyme quantitatively precipitated the B form but had no effect on the galactosylceramide-pgalactosidase from liver. Failure to hydrolyze lactosylceramide is probably explained on the basis of inappropriate assay conditions, since optimum conditions vary for each substrate (Tanaka and Suzuki, 1977). 3. p-D-N-ACETYLHEXOSAMINIDASES

The hexosaminidases have been studied in great detail as a consequence of their genetic deletion in Tay Sachs disease, Sandhoff‘s disease and juvenile GM,-gangliosidosis. In these diseases GM,-ganglioside accumulates in tissues by virtue of a failure to cleave the terminal GalNAc residue (Tallman et al., 1972).Two major forms of hexosaminidase, A and B, are found in human tissues (Robinson and Stirling, 1968), while a third, hexosaminidase S, is prominent in the tissues of patients with Sandhoff’s disease (Ikonne et al., 1975; Beutler and Kuhl, 1975). In a series of articles (Srivastavaand Beutler, 1974; Beutler and Kuhl, 1975; Beutler et al., 1975a, 1976; Lee and Yoshida, 1976 and Srivastava et al. (1974a,b) convincing evidence has been accumulated to show that hexosaminidases B and S are homopolymers of subunits P and a , respectively, while hexosaminidase A contains both a and p subunits. Hexosaminidases A and B both adhere to Con A-Sepharose and are eluted with a-methyl-D-mannoside (Beutler and Kuhl, 1975) but, in contrast to an earlier interpretation (Robinson and Stirling, 1968),contain no sialic acid (Lee and Yoshida, 1976) and are not affected by neuraminidase (Srivastava et d . , 197413). Hexosaminidase A contains approximately 5% carbohydrate, while hexosaminidase B contains approximately 9%. The difference appears to be accounted for by the fact that the p subunit is much more heavily glycosylated than the a

186

J. R. RIORDAN AND G. G. FORSTNER

subunit (Lee and Yoshida, 1976). Hexosaminidase S (the a-unit homopolymer) obtained from the liver of patients with Sandhoff‘s disease was found to bind poorly to Con A, as might be predicted from the apparent paucity of mannosyl residues (Lee and Yoshida, 1976). Purified hexosaminidase A had a MW of 110,000 b y equilibrium sedimentation and appeared to be a tetramer. The enzyme was easily released from lysosomes by sonication. On purification the activity of hexosaminidases A and B toward GM,-ganglioside was lost, although activity toward synthetic substrates such as 4-methyhmbelliferyl-~-DN-acetylglucosaminidase was well preserved (Srivastava et al., 1974b). 4. ARYLSULFATASES Most tissues contain two lysosomal arylsulfatases, A and B. The natural substrates for arylsulfatase A are cerebroside sulfate and sulfated glycosaminoglycans. The natural substrate for arylsulfatase B is possibly the GalNAc 4-O-sulfate found in chondroitin 4-sulfate and dennatan sulfate (O’Brien et al., 1974).Enzyme deficiencies are found in hereditary neurological disorders characterized by the accumulation of metachromatic sulfate-rich material (Austen, 1973) or mucopolysaccharides (O’Brien et al., 1974). Both enzymes are glycoproteins (Nichol and Roy, 1965, Graham and Roy, 1973; Balasubramanian and Bachhawat, 1975,1976).The suggestion has been made that arylsulfatase A is converted to arylsulfatase B b y digestion with neuraminidase (Goldstone et al., 1971), but this has been refuted (Graham and Roy, 1973), and in fact the two enzymes have quite different properties (Table V). The enzymes are easily extracted in aqueous solutions (Nichol and Roy, 1964; Balasubramanian and Bachhawat, 1976) and therefore appear to be loosely attached to lysosomal membranes. Zinc acetate chiefly precipitates arylsulfatase B, and this property has been used as a convenient initial step in the isolation of A and B enzymes from brain (Balasubramanian and Bachhawat, 1975, 1976). a. Arylsulfutase A . The enzyme from ox liver contained 9.5% carbohydrate (Graham and Roy, 1973). Both its catalytic and sedimentation properties were unaffected by removal of 98% of the sialic acid residues. The pH optimum was 5.6. The enzyme was inactivated b y exposure to temperatures in excess of 60°C for 5 minutes. Mehl and Jatzkewitz (1964) showed that only 0.12% as much sulfate was released from sulfatide as from nitrocatechol sulfate under standard assay conditions, but a heat-stable, nondialyzable cofactor increased cerebroside sulfatase activity 12-fold. The identity of the cofactor is unknown. Balasubramanian and Bachhawat (1975) purified the enzyme from

187

GLYCOPROTEIN MEMBRANE ENZYMES

TABLE V COMPARISON OF SOME PROPERTIES OF ARYLSULFATASES A AND B Property

Arylsulfatase A

Arylsulfatase B

Hydrolysis of cerebroside sulfate

Yes

No

K m (d) p-Nitrocatechol sulfate

Negative

0.38" 1.9 11" 4Sa-60,00Od Positive

No inhibition Weak activation

Inhibition Sixfold activation

0.6O

O.4gb p-Nitrophenyl sulfate MW Charge at pH 7.5' Effect of sodium chloride" With p-nitrocatechol sulfate With p-nitrophenyl sulfate a

45"

107b- 122,00OC

Balasubramanian and Bachhawat (1976). Graham and Roy (1973). Balasubramanian and Bachhawat (1975). Bleszynski and Roy (1973).

sheep brain, taking advantage of specific binding to Con A-Sepharose. The enzyme was not bound to R . communis agglutinin, suggesting that it lacked accessible P-galactose residues (Nicolson, 1973). Analyses suggested a neutral sugar content of%%, greatly in excess of that found in ox liver (Graham and Roy, 1973), and considerable amounts of glucose were found. In agreement with previous findings (Nichol and Roy, 1964; Woodward et al., 1973) the enzyme was a monomer at pH 7.5 but polymerized at pH 5.0. Human arylsulfatase A from a variety of sources exhibits multiple bands on isoelectric focusing, with isoelectric points varying from pH 4.4 to 4.9 (Stevens et al., 1976). In liver neuraminidase eliminated three of six bands, suggesting that sialic acid accounts for some but not all of the heterogeneity. b. Arylsulfatase B . The enzyme was purified 1344-fold from sheep brain but still contained additional protein (Balasubramanian and Bachhawat, 1976). The neutral sugar content was 11.7%with 0.4%sialic acid. An optimum pH of 5.6 identical with that of arylsulfatase A was found. Sulfate, sulfite, and pyrophosphate were inhibitory, while 6-fold activation was found in 0.2 it4 sodium chloride. In ox brain, Bleszynski and Roy (1973) separated seven possible isoenzymes by ion-exchange chromatography, but similar heterogeneity was not found in the sheep brain enzyme (Balasubramanian and Bachhawat,

1976).

188

J. R. RIORDAN AND G. G. FORSTNER

111.

MEMBRANE ASSOCIATION

A. Nature of Association

The glycoprotein enzymes described in Table I range from peripheral to integral membrane components on the basis of the ciassification of Singer (1974b). This gradation in extent of membrane association is indicated schematically in Fig. 1. Acetylcholinesterase (class VII, Fig. I), which can be dissociated from membranes of electric organ by high ionic strength (Massoulie and Rieger, 1969), is apparently associated primarily via electrostatic interactions and is therefore extrinsic or peripheral. The fact that proteases can also release a fully active enzyme (Taylor et al., 1974) suggests that cleavage can occur between a portion of the enzyme containing the active site and the portion involved electrostatically with the membrane. The

I

I1

Ill

IV

v

VI

VII

FIG.1. Schematic collective representation of glycoprotein enzymes in a hypothetical lipid bilayer membrane with -their carbohydrate oriented extracytoplasmically. I,

(Na+ + K+)M@+-ATPase;11, cytochrome b, reductase; 111, glycosyltransferases or alkaline phosphatase; IV, 5’-nucleotidase; V, transport system for sugars, nucleosides, or amino acids; VI, brush border hydrolases or nucleotide pyrophosphatase; VII, acetylcholinesterase. w , Substrate sites; P, substrate site on a glycosyltransferase for transfer of sugar to a glycoprotein acceptor; L, site for transfer to a glycolipid acceptor. It is not implied that the same enzyme possesses both sites. The K+ and cardiac glycoside-binding sites on the extracytoplasmic side of the large subunit of (Na+ K+)M@+-ATPaseare indicated in structure I. The unstippled portions of structures 111, IV, and VI indicate that the degree to which these enzymes penetrate the cytoplasmic leaflet of the lipid bilayer is unknown.

+

GLYCOPROTEIN MEMBRANE ENZYMES

189

only indication of any hydrophobic interaction of the enzyme with the membrane comes from experiments showing that theefficiency of salt extraction was increased by the presence of a neutral detergent (Dudai and Silman, 1974). However, the detergent need not participate directly in disruption of enzyme-membrane interactions but may instead render them more sensitive to salt by diminishing other intermolecular associations which normally contribute to the overall arrangement of membrane components. In addition to acetylcholine esterase, the lysosomal enzymes pglucuronidase and arylsulfatase are loosely attached to membranes. Although most other enzymes depend on apolar interactions with membrane, some, including primarily brush border hydrolases (class VI, Fig. l),do so via only a small portion of their protein. Small apolar peptides have been released by proteases from detergent-solubilized forms of aminopeptidase and maltase (Maroux and Louvard, 1976)and conceivably represent segments capable of insertion into the plasma membrane. Both y-glutamyltranspeptidase (Hughey and Curthoys, 1976) and sucrase-isomaltase (Sigrist et al., 1975) appear capable of conversion from large detergent-soluble forms by treatment with proteases, perhaps undergoing elimination of apolar peptides in the process. As discussed in the sections dealing with the respective enzymes, the conclusions to be derived from proteolytic experiments are completely dependent upon the purity of the detergent-solubilized preparations, since the segment cleaved is usually not larger than 5% of the total molecule and might easily be mimicked by 5% contamination of the enzymes by other membrane proteins. Deductions based on MW comparisons are also suspect and difficult to define because of problems with aggregation and micelle formation. The strongest supporting evidence for an apolar footpiece has been provided for aminopeptidase by Desnuelle, Louvard, and colleagues. These workers showed that aminopeptidase could be labeled from the inner surface of the plasma membrane using a photogenerated NAP-Fab probe (Louvard et al., 1976), and that the labeled fragment was subsequently released from the detergent-solubilized enzyme by trypsin (see Section V,B,4,a). Thus a segment of the enzyme must pass completely through the membrane, as is the case with glycophorin (Tomita and Marchesi, 1975). The same group also showed that several antigenic sites on the detergent-solubilized enzyme which are available to highly purified monospecific antibody are apparently buried within the membrane, since they are not reactive when the enzyme is in situ (Vannier et al., 1976; Louvard et al., 197513).

190

J. R. RIORDAN AND G. G. FORSTNER

The hepatic nucleotide pyrophosphatase (class VI, Fig. 1)is also associated with membrane in a manner which permits solubilization by either proteases (Bischoff et al., 1975) or detergent (Evans et al., 1973). Attempts at salt extraction have not been reported but presumably are not effective. The nucleoside-monophosphate hydrolase, 5’nucleotidase (class IV; Fig. l ) , has been purified only from detergentsolubilized membranes (Widnell and Unkeless, 1968; Evans and Gurd, 1973; Nakamura, 1976; Slavik et al., 1977). It is not dissociated from membrane by ionic alterations, and no reports on removal by proteases have appeared, although both papain and trypsin apparently activate the membrane-bound enzyme (Nakamura, 1976). The purified detergent-soluble enzyme has bound lipid, but activity is fairly insensitive to perturbations of bulk membrane lipid (Emmelot and Bos, 1966). The glycosyltransferases (class 111, Fig. 1) are firmly embedded in lipid of microsomal and Golgi membranes. Although Fraser and Mookerjea (1976) released about 25% of the galactosyltransferase from rat liver microsomes with 0.2 it4 sodium chloride, this enzyme and other gl ycosyltransferases are generally integral proteins (Danwalder et al., 1972). Activities are markedly enhanced by the addition of detergent (Schachter et al., 1970), but attempts at separation from other components of the solubilized membranes often results in inactivation. The temperature dependency of sialyl- and galactosyltransferases of rat liver Golgi membranes is suggestive of a strong influence of membrane lipid (Mitranic et al., 1976). An intimate involvement of the transferases responsible for addition of the more proximal sugars of oligosaccharide chains with membrane lipids may be required to permit association with dolichol lipid substrates (Lucas et al., 1975; Chen et al., 1975; Pless and Lennarz, 1975). In one case where reasonably well-documented purification of a membrane-bound glycosyltransferase was achieved (Podolsky and Weiser, 1975), extensive sonication of rabbit erythrocyte membranes was used to obtain a “solubilized” enzyme as judged by failure to sediment at 105,000 g for 1hour. However, centrifugation at much higher forces and electron microscope examination would be required to determine whether or not the enzyme remains associated with very small membrane vesicles generated on sonication. The use of proteases to attempt removal of glycosyltransferases from membrane has not been reported, probably because the current belief is that they are firmly embedded in the lipid bilayer. Nevertheless, there is a need to test the effects of proteases on the glycosyltransferases of the endoplasmic reticulum and Golgi apparatus.

GLYCOPROTEIN MEMBRANE ENZYMES

191

Alkaline phosphatase seems to be a similarly integral membrane protein, since it is not removed by proteases and is solubilized with relative difficulty by detergents. Among enzymes definitely associated with membrane by way of penetration of at least a portion of the protein in the bilayer, cytochrome b, reductase (class 11, Fig. 1) is perhaps the best characterized example of an enzyme with a hydrophobic “membrane-binding appendage” (Spatz and Strittmatter, 1973). As described by Sturgess et al. (this volume), comparison of detergent- and protease-dissociated forms of the enzyme provide reasonably strong evidence that such a hydrophobic peptide does append this enzyme to the membrane in a manner which provides localization but no restriction of catalysis in (class I, Fig. 1)is a glycothe aqueous phase. (Na+ K+)Mg“?+-ATPase protein enzyme totally integrated in membrane lipid and dependent on it for activity. Of course, it is a transmembrane enzyme, hence penetrates the full thickness of the bilayer. Many years of work were required before the correct conditions of ionic strength and choice and concentration of detergent required for purification of the solubilized enzyme were achieved (see Section 11,D). The specific lipid requirements of the enzyme have been investigated in great detail (Albers, 1976). In enzymes from several different sources a specific need for phosphatidylserine has been established.

+

0. Sidedness and Topology

It is clear from Section I1 that all the plasma membrane enzymes dealt with have their substrate sites extracytoplasmically oriented except Na+,K+,Mg2+-ATPase.Cytochrome b, reductase apparently acts at the cytoplasmic surface of the endoplasmic reticulum. Therefore in these two cases the carbohydrate and active sites are located on opposite sides of the membrane. In all the other enzymes both active sites and carbohydrate are directed extracytoplasmically (see Section 11; Fig. 1). The large number of glycoprotein ectoenzymes on the surface of the intestinal and renal brush borders makes them excellent organelles for studying enzyme-membrane relationships. All the enzymes are intrinsic membrane proteins in the sense that they are not easily extracted with salt solutions. Their relationships have been probed kinetically with proteases and with detergents. The active sites of intestinal sucrase, maltase, alkaline phosphatase (Miller and Crane, 1961),and lactase (Malathi and Crane, 1968) lie ex-

192

J. R. RIORDAN AND G. G. FORSTNER

ternal to the Naf-dependent carrier for glucose and galactose, which presumably spans the lipid bilayer. Many of the intestinal glycoprotein enzymes can be removed from the plasma membrane by proteases without disrupting the underlying membrane bilayer (Johnson, 1967; Oda and Seki, 1966; Louvard et aZ., 1975a; Maestracci, 1976). Highly specific bonds or sites of attachment must be involved, since papain and elastase release enzymes very efficiently, while trypsin and chymotrypsin are ineffective (Maestracci, 1976). The most rapidly and completely released enzymes are maltase, leucylnaphthylamidase [predominantly oligoaminopeptidase], sucrase, glutamyltransferase, and lactase in descending order ranging from 75% effectiveness to 32% effectiveness after exposure to papain for 15minutes (Maestracci, 1976).These results do not necessarily imply variable degrees of surface exposure, since susceptibility could depend equally on the substrate specificity of the protease. Indeed, with proteases other than papain, quite different susceptibilities have been seen. For example, 36% of the leucylnaphthylamidase activity was released by trypsin (Maestracci, 1976), whereas the remainder of the papain-sensitive enzymes were not affected. In addition, elastase, although it was as efficient as papain in releasing most of the susceptible enzymes, was not effective in releasing glutamyltransferase (Maestracci, 1976). The experiments with papain, nevertheless, define a group of enzymes which must be linked to the underlying plasma membrane by a bond which is at least accessible to a relatively bulky protease. Current evidence we have reviewed for maltase, sucrase, and oligoaminopeptidase indicates that the portion of the enzyme remaining in the membrane is extremely small in relation to the released active head groups, perhaps less than 5% of the total bulk. One must assume therefore that the comparatively huge head group projects relatively freely from the bilayer so as to expose the linkage region. Negatively stained brush border membrane preparations reveal surface knobs 45 %, in diameter projecting outward approximately twice their diameter from the membrane. These knobs disappear after treatment with papain (Maestracci, 1976) and presumably represent released enzymes (Johnson, 1967; Maestracci, 1976). Unfortunately, knoblike structures were not seen by freeze-etching techniques, although a palisadelike fibrillar coat was demonstrated (Swift and Mukherjee, 1976).At the moment the relationship between these two structures is conjectural. Physical studies suggest that sucrase-isomaltase is an elongated molecule Mosimann et al., 1973; Conklin et aZ., 1975), and perhaps its natural state is that of a fibrillar

GLYCOPROTEIN MEMBRANE ENZYMES

193

protein projecting from the membrane. It would be interesting to determine whether or not linear projections seen by freeze-etching techniques disappear on treating the brush border membrane with papain. In contrast to the readily released enzymes, intestinal trehalase and alkaline phosphatase are very resistant to the action of papain, elastase, trypsin, and chymotrypsin (Maestracci, 1976). Both enzymes have been isolated with the aid of detergents and purified to homogeneity (Fosset et al., 1974; Sasajima et al., 1975). Intestinal alkaline phosphatase is a glycoprotein containing 12% carbohydrate. Trehalase is less certainly characterized, but Seetharam et al. (1976) have suggested that it too is a glycoprotein containing 8% carbohydrate. If so, both enzymes probably possess exposed polar segments at the membrane surface, which either are not susceptible to proteases or lack the active center. Lack of susceptibility to proteolytic digestion is not explainable on the basis of a high-carbohydrate content. Both maltase and oligoaminopeptidase are very sensitive to proteolytic solubilization but contain a high percentage of carbohydrate (Table 111). Interestingly, the resistance of these enzymes to proteolytic release is paralleled by comparatively poor solubilization by detergents (Louvard et d . , 1975a; Sasajima et d., 1975; Boedeker et d, 1976). Sasajima et al. (1975) showed that 1% Triton X-100 released 80% of the intestinal sucrase immediately, but only 40% of the trehalase and no alkaline phosphatase. Louvard et al. (1975a) found that 1%Triton X100 released 80-100% of the aminopeptidase and maltase activities overnight at 4"C, but only 30-40% of the alkaline phosphatase. In a similar study Boedeker et al. (1976) found that alkaline phosphatase was much more slowly solubilized from microvillus membranes by Triton X-100and deoxycholate than other disaccharidases. The evidence suggests therefore that the intestinal surface hydrolases fall into two groups which can be distinguished consistently by their response to proteases and detergents. Members of the first group represent the majority of the well-characterized enzymes (class VI, Fig. 1) and are easily removed by detergents and proteolytic agents; the second group consists at the moment of trehalase and alkaline phosphatase (class 111, Fig. l),and its members appear to be more deeply embedded within the plasma membrane. Similar studies suggest that the glycoprotein ectoenzymes on the renal brush border are arranged in a similar manner. Thomas and Kinne (1972) found that aminopeptidase was readily solubilized from rat kidney brush border fractions with either papain or Triton X-100, while alkaline phosphatase was released by neither. Neither enzyme

194

J. R. RIORDAN AND G. G. FORSTNER

was released by lithium chloride (1.5 gm/100 ml). Booth and Kenny (1976) found that aminopeptidase M, dipeptidylpeptidase IVYand neutral endopeptidase were not extractable by salt but were solubilized by papain. In agreement with previous results alkaline phosphatase was not affected by papain. Little information is available regarding the distribution of any of the enzymes in the plane parallel to that of the membrane. Essentially, aside from absolute sidedness, little knowledge is available on the intramembranous arrangement of the enzymes. Chemical crosslinking experiments of the type described by Wang and Richards (1974) should permit the definition of some intermolecular relationships. It would be of interest to know how closely to each other the acetylcholinesterase and acetylcholine receptors are situated in the postsynaptic membrane. Similarly, relationships among different glycosyltransferases in the microsomal and Golgi membranes would provide insight into the mechanism of sequential addition of sugars to growing oligosaccharide chains.

IV.

STRUCTURE

A. Amino Acid Composition

Amino acid analyses have been performed on many isolated glycoprotein enzymes. In general it seems fair to conclude that the technique has not revealed characteristic peculiarities which might distinguish one enzyme from another. Capaldi and Vanderkooi (1972) suggested that the polarity index expressed as the sum of the residue mole percentages of aspartic acid, asparagine, glutamic acid, lysine, serine, arginine, threonine, and histidine, might be a useful parameter for the characterization of membrane proteins. Only 2% of 205 soluble proteins had polarity indexes below 40%, whereas 47% of 19membrane proteins fell within this category. Table I1 lists the polarity indexes of various membrane glycoprotein enzymes. In spite of the fact that most must be judged integral proteins on the basis of their resistance to salt extraction, or evidence of tight membrane association, all have high indexes more characteristic of soluble proteins. Interestingly, detergent extraction, which might be expected to preserve apolar portions of the enzymes, does not seem to be associated with diminished polarity. As discussed previously, it is likely that apolar segments are relatively small in proportion to the size of each enzyme.

195

GLYCOPROTEIN MEMBRANE ENZYMES

6. Monosaccharide Composition

In none of the enzymes considered has the sequence of monosaccharides or the linkages between them been determined. However, the sugar compositions of many of the enzymes are known (Table 111). For nucleotide pyrophosphatase, (Na+ K+)Mg+-ATPase, acetylcholinesterase, alkaline phosphatase, y-glutamyltranspeptidase, enterokinase, oligoaminopeptidase, p-glucuronidase, and hexosaminidase, only sialic acid and the total amounts of neutral sugars and amino sugars have been quantitated. There is no suggestion of similar ratios among these three quantities in these enzymes. With alkaline phosphatase, for example, the amino sugar/neutral sugar ratio is 1 : 1for the enzyme from calf intestine and 3 : 1 for that from bovine kidney. The significance, if any, of such tissue and species differences is not known. For enzymes in which each of the sugar residues has been measured the compositions are not markedly different from those of “serum-type’’ glycoproteins (Neuberger et al., 1972) or nonenzymic membrane glycoproteins (Kawasaki and Ashwell, 1976). One notable difference, however, is the presence of GalNAc in most of the intestinal brush border hydrolases. It is interesting that this sugar, which may participate in the 0-glycosidic linkage to serine or threonine residues in “mucin-type” glycoproteins, is consistently present in membrane glycoprotein enzymes isolated from a tissue which synthesizes large amounts of mucins. However, the residue has also been detected in nucleotide pyrophosphatase of liver plasma membranes. Significant amounts are also present in glycophorin, the major glycoprotein of the human erythrocyte plasma membrane (Furthmayer and Marchesi, 1977), but not in other nonenzymic membrane glycoproteins analyzed to date. The apparent presence of glucose in sucrase-isomaltase, oligoaminopeptidase, and arylsulfatase A is significant in view of the fact that glucose had been thought to be absent from animal glycoproteins. However, recently there have been suggestions that small amounts of this sugar are present in many of the lipid-linked intermediates in glycoprotein biosynthesis (Spiro et al., 1976).

+

C. Functional Role of Carbohydrate

There is clearly neither any evidence to suggest nor any a priori reason to assume that the carbohydrate moiety of glycoprotein enzymes is directly involved in catalytic function. However, there are reasons to suspect that the oligosaccharide chains may participate in one or

1 96

J. R. RIORDAN AND G. G. FORSTNER

more of the following functions: (1) modification or regulation of enzyme activity, (2) maintenance of three-dimensional structure, (3) maintenance of a specific orientation within the membrane consistent with function, and (4)protection from proteolysis. The first of these has been tested directly by assessing the influence on activity of binding of specific plant lectins and treatment with glycosidases. The findings of such studies veiy likely relate to points (2) and (3)as well. The degree to which many glycoprotein membrane enzymes retain activity after their removal from the membrane by proteases (see Section 111) attests to the resistance of at least their active site areas to proteolysis. However, the relative sensitivity before and after removal of carbohydrate has in most instances not been tested, so that its role in protection is unclear. In the case of one glycopeptide from the major erythrocyte membrane glycoprotein, further cleavage by trypsin or chymotrypsin occurs after removal of sialic acid with neuraminidase (Furthmayer and Marchesi, 1977). Hence a portion of an oligosaccharide can apparently block protease-sensitive bonds. Nevertheless, it is also possible that the active site regions simply do not contain accessible protease-sensitive bonds.

1. AS PROBED WITH LECTINS Lectins have been widely used in studies of the chemical and physical state of cell surface plasma membranes and their molecular constituents (Lis and Sharon, 1973; Rapin and Burger, 1974; Berlin et al., 1974; Nicholson, 1976; Lis and Sharon, 1977). In general, use has been made of the abilities of lectins to serve as qualitative and quantitative markers of the oligosaccharides which serve as their receptors, to alter the lateral arrangement of these receptors in the plane of the membrane, as well as the physical state of membrane lipid, and to aggregate cells, subcellular membranes, and isolated glycoproteins. Among the various alterations of membrane function which result, changes in several enzymes have been observed. The details of the responses of the enzymes described in Section I1 are not reiterated here, but instead an attempt is made to indicate the types of information about the enzymes which can be gained. When a lectin binds to the oligosaccharide of a membrane enzyme, activity may be inhibited, stimulated, or unaffected. If no change occurs, this may indicate that the enzyme does not have the carbohydrate structure required for binding or that binding has occurred but does not influence substrate binding and catalysis. If either inhibition or stimulation occurs, it may result either directly from the binding of lectin to the carbohydrate of the enzyme itself or indirectly from a

GLYCOPROTEIN MEMBRANE EN2YMES

197

change in the enzyme’s membranous environment on binding to other complex carbohydrates of the membrane. Hence a change in the activity of a membrane-associated enzyme on the binding of lectin does not necessarily provide evidence that the enzyme is a glycoprotein. However, if the response to lectin is retained when the enzyme is purified from the membrane, it seems reasonable to conclude that the enzyme is a glycoprotein. Conversely, if the response is lost after purification, it is likely that modification has resulted from an indirect membranous change. The mechanisms of either direct or indirect modification of activity are not known, but in the case of a direct interaction it is not difficult to imagine binding of the large lectin molecule, with or without subsequent intermolecular cross-linking, resulting in inhibition. Stimulation of activity due to direct binding would presumably be the consequence of a conformational change favorable to substrate binding or catalysis. This can occur even with a soluble glycoprotein enzyme; Con A binding to Mucor miehei acid protease causes a 30%increase in proteolytic activity with an accompanying conformational change (Rickert and McBride-Warren, 1976). Indirect modifications of activity could also result from steric restrictions due to lectin binding to a closely adjoining glycoprotein or gl- tolipid. Alternatively, activities of membrane enzymes are sensitive t l t alterations in the fluidity of their local lipid environment (Kimc \erg!, 1977), and lectin binding has been shown to elicit fluidity changes in some instances (Barnett et al., 1974; Toyshima and Osawa, 1975). The response to lectins of only a few membrane enzymes has been systematically investigated. This has been done with the 5’-nucleotidase of plasma membranes from liver (Riordan and Slavik, 1974a,b, 1975; Slavik et al., 1977)and mammary gland (C. A. C. Carraway et al., 1975; K. L. Carraway et al., 1976; Carraway and Carraway, 1976). The enzyme of liver membranes responded to the binding of increasing amounts of some lectins such as Con A in a biphasic manner, exhibiting enhanced activity at low lectin concentrations which is overcome by marked inhibition at higher concentrations. The stimulatory phase is not observed with detergent-solubilized or purified enzyme (Riordan and Slavik, 1974a; Slavik et al., 1977), hence must be the indirect result of a change in the membrane. The inhibitory phase is retained and in fact occurs at lower lectin concentrations with solubilized enzyme from both liver (Riordan and Slavik, 1974b) and mammary (Carraway et d., 1976) membranes. Furthermore, the purified enzyme from liver is extremely sensitive to inhibition (Slavik et al., 1977),providing additional evidence that the enzyme is a glycoprotein. The reaction of Mg2+-ATPase,another ectoenzyme of liver and

198

J. R. RIORDAN AND G. G. FORSTNER

Lectin Concentratlon

FrG.2. Response of 5’-nucleotidase (dashed line) and Mg*+-ATPase(solid line) of liver plasma membranes to the binding of increasing concentrations of Con A. WGA and R . communis agglutin have similar effects. ACT, “area of cooperative transition” where the activities of these two enzymes change in a cooperative manner. [‘Z51]Con-A binding to the membranes is positively cooperative over this range (Riordan et al., 1977).

mammary membranes, with Con A and other lectins has aIso been described (Carraway et al., 1975; Riordan and Slavik, 1977). Low lectin concentrations have little influence, but at higher concentration the response is apparently the reciprocal of that of 5’-nucleotidase (Fig. 2); i.e., activity is strongly stimulated. Hence, while 5’-nucleotidase is inhibited by direct binding of lectin to its carbohydrate moiety, Mg2+ATPase is simultaneously stimulated. Since the latter has not been purified, it is not known whether or not it is a glycoprotein, hence it was not considered in Section 11. It is, however, a plasma membrane ectoenzyme (DePierre and Karnowsky, 1974; Trams and Lauter, 1974),hence likely to be a glycoprotein (see Sections VI and VII). The properties of lectin-induced activation of the liver membrane enzyme have been investigated (Riordan et al., 1977) and found to be quite different from those of the inhibitory phase of 5‘-nucleotidase modification. Stimulation of Mg2+-ATPase occurred only at temperatures above approximately 30°C and was abolished by removal of phospholipid either with detergent or phospholipase. There was apparently a specific requirement for phosphatidylcholine (Riordan et al., 1977). Reconstitution was possible with this lipid or a total membrane lipid extract (Riordan, J. R., unpublished observations). Because of difficulty in obtaining a pure enzyme, it is not known whether the increase in activity is due to direct lectin binding to it or to the result of a membrane change. In either case there is a strict requirement for phospholipid which could either maintain the enzyme in a conformation amenable to activation on direct binding or participate in the mediation of

GLYCOPROTEIN MEMBRANE ENZYMES

199

an environmental change on the binding of lectin to receptors other than the enzyme. However, inhibition of 5’-nucleotidase is totally independent of membrane lipid. Although mechanisms are not yet totally understood at the molecular level, the responses of plasma membrane 5’-nucleotidase and Mg*+-ATPaseillustrate at least two ways in which membrane enzymes can be modified on the binding of lectins to surface carbohydrate. Komoda and Sakagishi (1976a,b) demonstrated that different organspecific isoenzymes of alkaline phosphatase respond differently on the binding of Con A. The enzyme from human intestine is activated nearly twofold by Con A (0.02 pM), while the human hepatic enzyme is slightly inhibited (about 10%). Interestingly, the intestinal enzyme contains two less mannose residues than the hepatic enzyme. The enzyme from liver has 12-22 moles/mole sialic acid, while the intestinal enzyme has virtually none. Removal of this sialic acid with neuraminidase caused a stimulation of activity of magnitude similar to that caused in the intestinal enzyme by Con A. Con A had no further effect on the hepatic enzyme following the neuraminidase-induced activation. All these changes apparently resulted from a direct interaction of lectin with enzyme, which aIso brought about an increased resistance to inactivation by heat or trypsin treatment. As indicated in Section 11, several membrane glycoprotein enzymes have not yet been purified in sufficient quantities to permit analysis of carbohydrate composition and structure. Therefore the demonstration of specific binding by a plant lectin of known specificity can provide the first clues to sugar composition and structure. A negative response to a lectin provides little information, since it is possible that an enzyme contains carbohydrate which is complementary to the binding site of the lectin but that accessibility is limited. However, if an enzyme is retained by an immobilized lectin affinity chromatography column, bound by a radioactively labeled lectin, or modified in its activity, then some information on carbohydrate content is obtained. Hence 5’-nucleotidase on the basis of its reaction with Con A very likely has oligosaccharide chains containing mannose residues in aglycosidic linkage and unsubstituted at C-2, C-4, and C-6 (Ogataet al., 1975; Kornfeld and Ferris, 1975).Sensitivity of the purified enzyme to WGA and R . communis agglutinin also indicates the presence of GlcNAc (and possibly sialic acid) and galactose or GalNAc. Since soybean agglutinin with specificity for GalNAc had little influence on the enzyme, the effect of R. communis agglutinin probably resulted from binding to galactose residues. Oligoaminopeptidase from rat intestine was precipitated by PHA-P (Kim et al., 1976), suggesting the presence

200

J. R. RIORDAN AND G. G. FORSTNER

of an oligosaccharide sequence of the type D-GalP1+4GlcNAc * Man (Lis and Sharon, 1973). Con A, WGA, and R . communis agglutinin did not precipitate the enzyme, indicating that mannose, GluNAc and galactosylgalactose residues are probably not accessible to the lectins. Lysosomal hydrolases including hexosaminidases (Beutler and Kuhl, 1975), and arylsulfatases (Balasubramanian and Bachhawat, 1975) bind to Con A-Sepharose and are eluted by a mannose analog, indicating that they possess this sugar in a position which is accessible to the lectin. The fact that some lectin specificities are very broad and in some cases not yet totally defined limits their use in structure determination. It can certainly not be expected that useful information will be obtained unless several lectins with different specificities are utilized under circumstances where binding is optimal. Reversal of any effect by low MW haptenes is also absolutely necessary in order to exclude changes due to nonspecific lectin associations not related to binding to carbohydrate.

2. AS PROBED WITH GLYCOSIDASES As with lectins, results of experiments in which membrane-bound enzymes are subjected to enzymic or chemical modification are likely to be ambiguous and may still be so even when the enzymes have been purified. Nevertheless, at least one glycosidase, neuraminidase, presumably because of its availability, has been applied to many of the enzymes discussed in this chapter (Section 11).In most instances potential changes in activity and/or electrophoretic mobility have been monitored. In the case of acetylcholinesterase, a possible role of sialic acid in determining substrate specificity was suggested (Section 11,F).Removal of the sialic acid residues most resistant to the action of neuraminidase altered the substrate specificity of the enzyme, such that hydrolysis of acetylcholine was unchanged but more hydrophobic substrates were no longer hydrolyzed at all (Brodbeck et al., 1973). Alkaline phosphatase from tissues in which it contains sialic acid is influenced by removal of the terminal sugar. For example, in bovine kidney, where two forms of the enzyme having different isoelectric points exist, neuraminidase treatment results in conversion to a single form (Cathala et al., 1975a). Neuraminidase also changes the electrophoretic mobility of the hepatic enzyme (Badger and Sussman, 1976). More interestingly this results in removal of the substrate inhibition which the enzyme normally exhibits (Komoda and Sakagishi,

GLYCOPROTEIN MEMBRANE ENZYMES

20 1

1976a). Hence in the case of this enzyme and acetylcholinesterase, sialic acid residues seem to be involved in regulating the active site. The activity of the other plasma membrane monophosphate ester hydrolase, 5’-nucleotidase, is apparently unaffected by neuraminidase treatment (Trams et al., 1976; Riordan, unpublished observations). Neuraminidase has been widely used in studies of lysosomal hydrolases. The electrophoretic mobility of acid P-galactosidase is reduced on exposure to neuraminidase (Norden and O’Brien, 1973). A similar response of aryl sulfatase (Goldstone et al., 1971), hexosaminidase (Tallman et al., 1972), and a-galactosidase was observed and initially used to suggest that removal of sialic acid caused conversion from the A to the B forms of these enzymes. However, the product of the digestion of the A forms with neuraminidase turned out not to be identical to the naturally occurring B forms (Graham and Roy, 1973). This points strongly to the need for caution in interpreting the changes observed when utilizing glycosidases as probes of glycoprotein enzyme structure and function. In addition to the need for complete characterization of the product, which these studies emphasized, it is important to confirm the nominal purity and specificity of the glycosidase being used. It is of course also necessary that the monosaccharide expected to be released by the particular glycosidase being utilized is measured quantitatively. Highly purified glycosidases other than neuraminidase are not generally available commercially, and therefore few studies in which other sugars have been removed have been done. An exception is the demonstration b y Wacker (1974) that the microsomal aminopeptidase of pig kidney retained its substrate specificity and catalytic activity after removal of almost all the sialic acid, 60% of the neutral sugars, and 45% of the DGlcNAc by exhaustive treatment with neuraminidase, a-mannosidase, and P-N-acetylglucosaminidase. Although the judicious application of chemical modifications of carbohydrate structure such as those described by Noonan and by Juliano in this volume should provide valuable information on the role of carbohydrate in membrane glycoprotein enzymes, no such experiments have been reported. V.

FUNCTIONAL INTERRELATIONSHIPS

A. Modification of Recognition Phenomena

Some membrane enzymes are involved indirectly in recognition phenomena by modifying ligands which interact with cell surface re-

202

J. R. RIORDAN AND G. G. FORSTNER

ceptors. A good example of this type of function is exhibited by acetylcholinesterase which hydrolyzes the neurotransmitter, acetylcholine. The enzyme therefore has the capacity to terminate and possibly to modulate the signal for increased ionic conductance. The fact that the receptor and enzyme are localized at the surface of the postsynaptic junction qualifies them as “a coupled membrane unit” and raises the interesting but vexing question of how the two are compartmentalized in a combined fashion with respect to their common substrate. The evidence for and against the presence of glycosyltransferases at the external surface of plasma membranes and their possible role in recognition was considered in detail in Section I1,A. Although a major portion of the cell’s complement of these enzymes does not reside at this surface location, the question of whether or not the relatively small amounts localized there could contribute to transient contacts (Roseman, 1970) between cells remains open. Certainly these enzymes, by virtue of their substrate sites, have the potential to bind carbohydrate which may be at the surface of a separate cell. In at least one instance, carbohydrate of the glycoprotein enzyme galactosyltransferase served as a lectin receptor enabling Con A to agglutinate the erythrocytes possessing the enzyme at its surface (Podolsky et al., 1974). Alkaline phosphatase is generally considered a nonspecific “scavenger-type’’ enzyme with no precise function other than hydrolysis of a wide variety of monophosphate esters at the cell’s external surface. This broad degradative role is consistent with the observation that its synthesis is induced simultaneously with that of several lysosomal hydrolases (Hosli et al., 1976). The apparent concerted synthesis of glycoprotein enzymes of these two different subcellular localizations is interesting from the point of view of the biosynthetic mechanisms discussed in Section V1,A. There is some very recent suggestive evidence that this enzyme might also play a regulatory role in the uptake of secreted lysosomal enzymes by fibroblasts (Kaplan et aZ., 1977). Although experimental evidence is only preliminary and indirect, it is proposed that a hexose phosphate moiety of P-glucuronidase is recognized by a cell surface component as a signal for uptake. The addition of exogenous alkaline phosphatase prevents uptake, apparently by hydrolysis of the hexose phosphate. It seems possible that endogenous cell surface alkaline phosphatase may normally regulate uptake of the enzyme by cleaving the phosphate from the hexose on P-glucuronidase when its concentration exceeds the K, for the reaction. However, much further experimentation is required to determine whether or not this hypothesis has any relation to fact.

203

GLYCOPROTEIN MEMBRANE ENZYMES

B. Membrane Transport

Of the enzymes with which we have dealt some are involved directly in transport across membranes and some prepare their substrates for transport by converting them to products which can be handled by other transport systems. (Na+ K+)M$+-ATPase is the best characterized true transporting enzyme. Of the brush border hydrolases, sucrase, maltase, and possibly lactase are directly involved in transport, and aminopeptidase may be capable of directly carrying out transport of amino acids (see Section V,B,4,a).In addition, brush border hydrolases are involved indirectly in transport, since they degrade their substrates to products which can be translocated by membrane carriers for sugars and amino acids. Analogously, 5’-nucleotidase converts nucleotides to nucleosides to which membranes are permeable. The transfer of electrons in which cytochrome b, reductase is involved may traverse the microsomal membrane. The way in which these individual enzymes contribute to transport is considered in the following Sections in somewhat more detail.

+

1. Na+

ANDK+

IONS

The active transport of these ions against an electrochemical gradient is accomplished by the (Na+ K+)M$+-ATPase. As discussed in Section II,D, the enzyme consists of two subunits both of which very likely span the plasma membrane. The mechanisms of neither the ion translocation nor the conversion of the energy of hydrolysis of ATP to a force which drives this translocation are completely understood. Hence it does not seem unreasonable to postulate that Na+ and K+ may utilize an aqueous cleft between the two subunits as a transmembrane route, although no evidence bearing directly on this possibility has yet been described. The proposal has been made that the glycosylated subunit may be involved in binding of the cations (Churchill and Hokin, 1976), since most of the other functions of the enzyme complex have been ascribed to the larger unglycosylated protein.

+

2. NUCLEOSIDES Plasma membranes are permeable to nucleosides, and the properties of their translocation have been extensively investigated (Plageman and Richey, 1974), whereas nucleotides are unable to traverse the membrane intact. Hence, when a nucleoside monophosphate is encountered at the external surface of a cell, it is dephosphorylated by

204

J. R. RIORDAN AND G. G. FORSTNER

ecto-5’-nucleotidase, and the nucleoside product may then be taken up. In this sense 5’-nucleotidase may be considered as contributing to the transport of nucleosides. In fact, in cells such as leukocytes which do not synthesize purine nucleotides de nooo, this pathway of recovery of extracellular nucleosides for rephosphorylation intracellularly is an important source of cell nucleotides. This has been clearly demonstrated using lymphocytes from patients with chronic lymphocytic leukemia which are deficient in 5’-nucleotidase (Fleit et aZ., 1975). The uptake of adenosine from AMP by these cells was greatly diminished compared to that by cells having the enzyme. Surface nucleotide pyrophosphatase also is able to contribute higher nucleotides to this pathway, producing nucleoside monophosphates as substrates for 5’-nucleotidase. 3. SUGARS:DISACCHARIDASE-RELATED TRANSPORT

About 5-10% of the sugars provided to the intestine as disaccharides enter the intestinal cell by a mechanism independent of the classic Na+-dependent carrier from monosaccharides (Malathi et al., 1973; Ramaswamy et al., 1974). Maltase, lactase, isomaltase, sucrase, and trehalase contribute monosaccharides to this alternative pathway. When black lipid membranes are formed from emulsions containing the isolated sucrase-isomaltase complex, their permeability to glucose and fructose derived from sucrose is greatly enhanced (Storelli et al., 1972) Enhanced permeability does not require sodium and is not seen with monosaccharides added to one side of the membrane in free form (Storelli et aZ., 1972). Enhanced transport of sugars derived from sucrose has also been demonstrated in liposomes reconstituted with sucrase-isomaltase (Semenza, 1976). Transport of glucose derived rom maltose is perhaps a function of the maltase activity of the sucrase-isomaltase complex, but the system does not appear to be confined to a single disaccharidase, since there is good evidence that trehalase at least functions as an independent translocater. Trehalose does not inhibit the transport of glucose derived from sucrose, nor do other disaccharides inhibit the transport of glucose derived from trehalose (Ramaswamy et al., 1974). Glucose cleaved from lactose is also transported by a disaccharidase-dependent system, but it is not yet clear whether or not galactose takes part. If it does, one would have to conclude that at least three of the intestinal disaccharidases contain sodium-independent translocating systems. Interestingly, both native (Ramaswamy et al., 1974) and reconstituted (Semenza, 1976) transporting systems are not greatly inhibited by tris, whereas sucrase, iso-

GLYCOPROTEIN MEMBRANE ENZYMES

205

maltase, and other disaccharidases are strongly inhibited (Dahlqvist, 1961). Semenza has postulated that transport-sensitive enzyme molecules may exist as topologically distinct species partially buried in the lipid bilayer or perhaps spanning it so that they are sterically unavailable to the tris ions. This is somewhat at variance with our current concepts which assign the carbohydrate-containing and active site-associated hydrophilic portions of the disaccharides to an exposed surface location, but is consistent with the fact that the papain-solubilized enzyme, presumably lacking its apolar membrane-bound segment, acts perfectly well as a translocator in reconstitution experiments (Storelli et al., 1972). A transport glycoprotein ought to span the membrane in some manner (Singer, 1974a; Ho and Guidotti, 1975). A detergent-solubilized maltase preparation has been shown to contain a protease-sensitive apolar peptide (Maroux and Louvard, 1976), but it is still uncertain whether or not this enzyme consists of a hydrophilic active portion plus a membrane-associating hydrophobic tail. The glycoprotein headpiece released by papain could, by itself, contain apolar peptide sequences which might serve to bury portions of the enzyme within the lipid bilayer under the proper circumstances. Presumably burial would entail some conformational change in the whole molecule, perhaps reducing the role of the buried segment as a hydrolase.

4. AMINOACIDS AND PEPTIDES

a. Aminopeptidase-Related Transport. Small peptides are transported in intact form by mammalian intestine (Burston et al., 1972), but there is currently no evidence that surface peptidases are involved. It has been postulated, however, that the brush border oligoaminopeptidase of the intestine takes part in the transport of peptidederived amino acids (Ugolev, 1972; Maroux et al., 1973; Louvard et al., 1976). Maroux and Louvard (1976) isolated a peptide with a MW of approximately 10,000 from the detergent-purified oligoaminopeptidase after digestion with trypsin. The amino acids of the peptide had a polarity index of 35 (Maroux and Louvard, 1976), which is quite low as might be expected of an apolar segment, and N-terminal analyses suggested that the segment belonged to only one subunit of the enzyme. In a very elegant experiment employing a Fab fragment covalently linked to 4-fluoro-3-nitrophenylazide,a portion of the oligoaminopeptidase was labeled from within brush border vesicles. The labeled fragment was released by trypsin from the detergent-solubilized aminopeptidase, suggesting that it contained the apolar segment and,

206

J. R. RIORDAN AND G. G. FORSTNER

since it was labeled on the interior surface of the membrane, the segment must span it. The oligoaminopeptidase therefore appears to satisfy one of the minimum requirements for a transport glycoprotein in that a portion of the molecule passes entirely through the membrane. b. y-Glutamyltranspeptidase-Related Transport. Meister and colleagues have argued in several publications (Meister, 1973, 1974; Meister et a1., 1976) that y-glutamyltranspeptidase might function in amino acid transport via the y-glutamyl cycle. The postulate involves the binding of amino acid to the enzyme on the external surface of the membrane, while a separate y-glutamyl site interacts with intracellular glutathione at the internal surface. A y-glutamyl amino acid product is then formed with concurrent release of the amino acid from its primary binding site and the product is translocated to the interior of the cell where the amino acid is released by the action of a cytoplasmic glutamylcyclotransferase, The y-glutamyl cycle enzymes, 5-0x0prolinase, glutamylcysteine synthetase, and glutathione synthetase, subsequently regenerate glutathione. The mechanism supposes a transmembrane extension of the enzyme, and this has received partial support from the work of Hughey and Curthoys (1976)which suggests that the detergent-solubilized enzyme possesses a small peptide tail. There is no evidence that the y-glutamyl-binding site resides in this tail, however; in fact the bromelain-solubilized enzyme is an excellent acceptor for glutathione (Tate and Meister, 1975). Patients who cannot synthesize glutamylcysteine and cannot generate glutathione adequately, however, exhibit aminoaciduria (Konrad et aZ., 1972). Furthermore, maleate which promotes the glutaminase function of the enzyme, weakening its affinity for amino acid acceptors (Tate and Meister, 1975), causes aminoaciduria when administered to animals (Rosenberg and Segal, 1964). Although alternative explanations no doubt exist for these phenomena, they provide at least some circumstantial evidence that y-glutamyltranspeptidase has a transport role.

VI.

BIOSYNTHETIC AND DEVELOPMENTAL ASPECTS

This topic is dealt with in detail for membrane glycoproteins in general by Sturgess et aZ. in this volume. However, there are some cases where studies of the biosynthesis of membrane glycoprotein enzymes have been informative because their catalytic activity provides an additional means of assessing their presence at various stages of development. The appearance of activity which can be assayed either cytochemically in whole cells or tissues or biochemically in isolated

GLYCOPROTEIN MEMBRANE ENZYMES

207

subcellular fractions can thus be compared with the incorporation of radioactively labeled precursors of both protein and carbohydrate. A. Biosynthesis

At least three of the enzymes we have considered provide some clues to biosynthetic pathways which membrane glycoprotein enzymes may follow. First, 5’-nucleotidase has been shown cytochemically to be localized in the rough endoplasmic reticulum (Widnell, 1972) and Golgi apparatus (Farquhar et al., 1974), as well at the plasma membrane where it is concentrated. These observations could be interpreted as supportive of the so-called membrane flow model for biogenesis of plasma membrane constituents (Palade, 1959). However, there is no quantitative evidence that the enzyme associated with either of the intracellular organelles is a precursor of the plasma membrane enzyme. In fact, the sidedness of the enzyme in the various locations is not consistent with a flow of enzyme from the endoplasmic reticulum through the Golgi and into the plasma membrane by way of a fusion and inversion of Golgi vesicles into the cell surface membrane. Activity appeared to be situated at the cytoplasmic surface of rough microsomal vesicles and the heaviest Golgi vesicle subfractions. In two lighter Golgi vesicle fractions activity was at the extracytoplasmic surface, as it is on the plasma membrane. No other experimental evidence is available bearing on the relationships (if any) between these various enzyme pools. However, significant amounts of the enzyme are not detected in nonparticulate fractions, so there is no support for the existence of soluble precursors of any of the membrane-bound forms. Very recently, Elovson (1977)claimed in the case of another plasma membrane glycoprotein enzyme of liver, nucleotide pyrophosphatase, that there is a precursor-product relationship between rough microsoma1 and plasma membrane localized forms. However, kinetics of precursor incorporation into Golgi-associated enzyme were similar to those of the plasma membrane enzyme; hence there was no indication that the Golgi enzyme was a precursor of that on the plasma membrane. Rothman and Lenard (1977) suggest that cytochrome b, reductase, which has its active site oriented to the external surface of smooth microsomes, is not an integral transmembrane protein and is probably not synthesized on membrane-bound polysomes. They propose that the enzyme may be synthesized at the cytoplasmic surface of the en-

208

1. R. RIORDAN AND G. G. FORSTNER

doplasmic reticulum, utilizing loosely associated cytoplasmic ribosomes. However, they seem to have neglected the fact that the enzyme is a glycoprotein (Panfili et d.,1971), the carbohydrate of which is likely directed to the extracytoplasmic side of the endoplasmic reticulum, apparently like that of all the glycoproteins of this organelle (Nicolson and Singer, 1974). If, as seems to be the case, the enzyme has its substrate site directed toward the cytoplasm and its carbohydrate directed extracytoplasmically, then it must be a transmembrane protein (Fig. 1).Therefore its synthesis probably proceeds by penetration of its N-terminus through the rough endoplasmic reticulum membrane so that it can be glycosylated internally while its substrate site is formed near the C-terminus on the ribosomal side of the membrane. B. Developmental Changes

Like other surface components the glycoprotein enzyme population might be expected to change with cell and tissue differentiation. In most species the synthesis of many of the brush border hydrolases including alkaline phosphatase, sucrase, maltase, trehalase, and oligoaminopeptidase (probably) in the intestine is repressed in utero and postnatally for a variable period up to weaning. Even in the adult intestine these enzymes are synthesized only by the mature epithelial cells and not by cells resident in intestinal crypts (Nordstrom et al., 1967). Both thyroxine and cortisone stimulate the synthesis of intestinal brush border enzymes in the suckling animal (Yeh and Moog, 1975). Etzlar and Moog (1966) and Dubs et al. (1975) have presented immunological evidence which suggests that synthesis of an inactive precursor may precede the appearance of the ectoenzymes. Change with development is not confined to the intestine. In the kidney glutamyltranspeptidase exhibits little activity in the fetus, but in the liver, lung, and brain, fetal activity is much higher than in the adult (Albert et al., 1970a,b; Tate and Meister, 1975). The profound changes in the glycoenzyme population might be expected to have a significant effect on the characteristics of the surface carbohydrate coat. This is quantitatively evident in the intestine from the work of Yeh and Moog (1975), which demonstrates a close correlation between the development of brush border enzymes and the appearance of prominent PAS staining in the intestinal glycocalyx. Potentially, of course, the introduction of a particular glycoenzyme into a membrane surface offers an opportunity for the insertion of new surface properties which could modulate intercellular adhesion, cell growth, and organogenesis. Sela et al. (1972) found, for example, that two transformed hamster cell

GLYCOPROTEIN MEMBRANE ENZYMES

209

lines were no longer able to cleave nucleotide pyrophosphate, indicating that pyrophosphatase had been deleted. Podolsky and Weiser (1975) introduced galactosyltransferase carrying receptors for Con A onto the surface of erythrocytes and imparted Con-A agglutinability to the cells. Kelly and Alpers (1973b) identified a very active blood group-A activity in purified intestinal maltase which presumably accounts for a significant portion of the alloantigenicity of the intestinal surface. Thus glycoenzymes have the capacity to change surface antigenicity and binding characteristics. Other evidence exists, however, which suggests that a particular carbohydrate attribute may vary in a specific enzyme during stages of development. Kottgen et al. (1976) found that fetal glutamyltransferase failed to bind to Con A-Sepharose, whereas in adult liver and intestine the enzyme was bound strongly. Neuraminidase converted the fetal enzyme to the adult ConA-reactive type. Similarly alkaline phosphatase isolated from fetal intestine bound Con A poorly and appeared to contain sialic acid, while the enzyme in adult intestine bound Con A and, as noted previously, lacked sialic acid (Komoda and Sakagishi, 197613). Specific attributes may therefore depend more on glycosyltranferase activity in a particular tissue than on the enzymes that are being synthesized. A systematic investigation of lectin binding in specific fetal and adult enzymes isolated from a single membrane such as the intestinal brush border where several enzymes are available in pure form, would do much to indicate which of these two alternatives is generally operative. During early development y-glutamyltransferase is found in a cytoplasmic location in the renal tubule, whereas in the adult the enzyme is predominantly bound to the microvillus plasma membrane (Meister et al., 1976). Recently it has become apparent that a similar situation exists for many of the disaccharidases in the intestine (Galand and Forstner, 1974). The best characterized example is intestinal neutral maltase-glucoamylase where approximately 50% of the enzyme in suckling rat intestine is soluble rather than membrane-bound. The soluble maltase-glucoamylase is identical to the brush border enzyme antigenically (Flanagan and Forstner, 1978), in size, in substrate affinity, in susceptibility to tris inhibition, and in heat sensitivity (Galand and Forstner, 1974). At weaning, or in response to hydrocortisone, the ratio of soluble to membrane-bound activity falls sharply. The suggestion has been made that the soluble form is a precursor of the membrane-bound form which either lacks a binding site or is synthesized in excess of the capacity of the immature plasma membrane to accept it. Since in the first instance a binding site might require a specific carbohydrate sequence, analysis of lectin-binding affinities and carbohydrate composition would be of great interest.

210

J. R. RIORDAN AND G. G. FORSTNER

It is noteworthy that maltase-glucoamylase, oligoaminopeptidase (Maroux and Louvard, 1976), and sucrase-isomaltase (Semenza, personal communication) are apparently anchored to the plasma membrane by apolar N-terminal segments. Since it is the N terminus which should be translocated first to the extracytoplasmic side of the rough endoplasmic reticulum according to the recently developed signal hypothesis (Blobel and Dobberstein, 1975a,b),it is possible that the soluble enzymes might result from cleavage and loss of the apolar segment after translocation. The analogy to the action of signalase, which converts translocated proteins to soluble secretory products, is obvious. Equally obvious is the implication that the anchor and signal sequences may be identical. If this were the case, membrane insertion would depend not only on recognition but on the preservation of an intact signal sequence through resistance to the action of various signalases. Loss of soluble maltase activity with maturation might then be expected to coincide with the loss of a relatively specific signalase or the development of a resistant signal sequence.

VII.

ARE ALL ECTOENZYMES GLYCOPROTEINS?

It is at present possible to state that of those ectoenzymes which have been purified and analyzed, there are none which are free of carbohydrate. Since the list of s w h enzymes is becoming fairly extensive (section 11),it is tempting to conclude that in fact all ectoenzymes are glycoproteins. It is possible that the precise mechanism of insertion may preclude nonglycosylated proteins serving as ectoenzymes. Membrane endoenzymes may also be glycoproteins as best illustrated by the (Na+ + K+)Mg2+ATPaseand probably also by cytochrome b5 reductase. Glycoprotein endoenzymes are transmembrane proteins and may also rely on their carbohydrate for maintenance of this configuration. Hence, in the case of both ecto- and endoenzymes, a major role of carbohydrate appears to be in localization and orientation of the molecules in a manner compatible with their function. Oligosaccharide chains are also probably involved in more subtle regulatory modifications. Strong evidence for the latter is still forthcoming. REFERENCES Alhers, R. W. (1976).The sodium plus potassium transport ATPase.In “The Enzymes of Biological Membranes. Vol. 3: Membrane Transport” (A. Martonosi, ed.),pp. 283301. Plenum, New York.

GLYCOPROTEIN MEMBRANE ENZYMES

21 1

Albert, Z., Rzucildo, Z., and Ztarzyk, H. (1970a). Biochemical and histochemical investigation of the y-glytamyltranspeptidase in embryomal and adult organs of man. Acta Histochem. 37,74-79. Albert, Z., Rzucildo, Z., and Starzyk, H. (1970b). Comparative biochemical and histochemical studies on the activity of y-glutamyltranspeptidase in the organs of fetuses, newborns and adult rats. Actrr Histochem. 37,34-39. Alpers, D., and Solin, M. (1970). The characterization of rat intestinal amylase. Gustroenterology 58,833-842. Andrea, G., Marzi, A., and Auricchio, S. (1976).a-Glutamyl-P-naphthylamide hydrolase of rabbit small intestine: Localization on the brush border and separation from other brush border peptidases. Biochim. Bioph!ys. Acta 419,42-50. Andrews, P. (1970). Purification of lactose synthetase A protein from human milk and demonstration of its interaction with a-lactalbumin. FEBS Lett. 9,297-300. Angelini, C . , and Engel, A. (1973). Subcellular distribution of acid and neutral a-glucosidases in normal, acid maltase deficient and myophosphorylase deficient human skeletal muscle. Arch. Biochem. Biophys. 156,350-355. Antonowicz, I., Chang, S., and Grand, R. (1974). Development and distribution of lysosoma1 enzymes and disaccharidases in human fetal intestine. Gastroenterology 67, 51-58. Askari, A., and Rao, S. N. (1972). Na+,K+-ATPasecomplex: Effects of anticomplex antibody on the partial reactions catalyzed by the complex. Biochem. Biophys. Res. Commun. 49, 1323-1328. Asp, N., and Dahlqvist, A. (1968).Rat small intestinal P-galactosidases: Kinetic studies with three separated fractions. Biochem. J . 110, 143-150. Asp, N., Gudmand-Hoyer, E., Andersen, B., Berg, N., and Dahlqvist, A. (1975). Distribution of disaccharidases, alkaline phosphatase, and some intracellular enzymes along the human small intestine. Sccind. /. Gastroenterol. 10,647-651. Auricchio, S., Rubino, A., Prader, A., Ray, I., Jos,J,, Frezal, J., and Davidson, M. (1965). Intestinal glycosidase activities in congenital malabsorption of disaccharides.]. Pediat. 66,555-564. Auricchio, S., Semenza, G., and Rubino, A. (1965b). Multiplicity of human intestinal disaccharidases. 11. Characterization of individual maltases. Biochim. Biophys. Acta 96,498-507. Auricchio, S., Pierro, M., Andria, G., and Deritis, G. (1972). Enzymic activities of the brush border membrane of rat intestine hydrolyzing P-naphthylamides of amino acids, leucinamide and depeptides. Biochim. Biophys. Acta 274,420-425. Austen, J. (1973). Metachromatic leukodystrophy (sulfatide lipidosis). I n “Lysosomes and Storage Diseases” (H. Hers and F. VenHoof, eds.), pp. 412-435. Academic Press, New York. Badger, K. S., and Sussman, H. H. (1976). Structural evidence that human liver and placental alkaline phosphatase isoenzyme are coded by different genes. Proc. Natl. Acad. Sci. U.S.A. 73, 2201-2205. Balasubramanian, K., and Bachhawat, B. (1975). Purification, properties and glycoprotein nature of arylsulfatase A from sheep brain. Biochim. Biophys. Acta 403, 113-121. Balasubramanian, K., and Bachhawat, B. (1976). Partial purification, properties and glycoprotein nature of arylsulphatase B from sheep brain./. Neurochem. 27,485-492. Baratti, J., Maroux, S., Louvard, D., and Desnuelle, P. (1973). On porcine enterokinase: Further purification and some molecular properties. Biochim. Biophys. Acta 315, 147-161.

21 2

J. R. RIORDAN A N D 0 . G. FORSTNER

Barnes, R. J., and Elmslie, R. G. (1974). The active site of porcine enteropeptidase selective inactivation of the peptidase activity. Biochim. Biophys. Acta 350,495-498. Barnett, R. E., Scott, R. E., Furcht, L. T., and Kersey, J. H. (1974). Evidence that mitogenic lectins induce changes in lymphocyte membrane fluidity. Nature (London) 249,465-466. Belenki, D., and Rosenfeld, E. (1972). Effect of urea on enzymatic activity and electrophoretic mobility of acid y-amylase (a-glucosidase).Biochem. Biophys. Res. Commun. 46,443-448. Bennett, H. (1963). Morphological aspects of extracellular polysaccharides. J . Histochem. Cytochem. 11,14-23. Berger, S., and Sacktor, B. (1970). Isolation and biochemical characterization of brush borders from rabbit kidney. J . Cell Biol. 47,637-645. Berlin, R. D., Oliver, J. M., Ukena, T. E., and Yin, H. H. (1974). Control of cell surface topography. Nature (London)247,45-46. Berman, J. D. (1973). Structural properties of acetylcholinesterase from eel electric tissue and bovine erythrocyte membranes. Biochemistry 12, 1710-1715. Bemacki, R. J. (1974). Plasma membrane ectoglycosyltransferase activity of L1210 murine leukemic cells. J . Cell Physiol. 83,457-466. Beutler, E., and Kuhl, W. (1975). Subunit structure of human hexosaminidase verified: Interconvertibility of hexosaminidase isoenzymes. Nature (London)258,262-264. Beutler, E., Guinto, E., and Kuhl, W. (1973). Variability of a-galactosidase A and B in different tissues of man. Am. J. Hum. Genet. 25,42-46. Beutler, E., Kuhl, W., and Comings, D. (1975a). Hexosaminidase isoenzyme in type 0 GM,-gangliosidosis (and Hoff-Jatzkewitz disease).Am.J . Hum. Genet. 27,628-638. Beutler, E., Guinto, E., and Kuhl, W. (1975b). Placental acid hydrolase purification on concanavalin A-Sepharose. Lab. Methods 85,672-677. Beutler, E., Yoshida, A., Kuhl, W., and Lee, J. (1976). The subunits of hexosaminidase A. Biochem. J . 159,541-543. Birkenmeier, E., and Alpers, D. H. (1974). Enzymatic properties of rat lactase-phlorizin hydrolase. Biochim. Biophys. Actu 350, 100-112. Bischoff, E., Tran-Thi, T., and Decker, K. (1975). Nucleotide phyrophosphatase of rat liver. Eur. J . Biochem. 51,353-356. Bishayee, S., and Bachhawat, B. (1974). Con A-Sepharose affinity chromatography: A study on the glycoprotein nature of brain acid hydrolases. Neurobiology 4,48-56. Bleszynski, W., and Roy, A. (1973). Some properties of the sulphatase B of ox brain. Biochim. Biophys. Acta 317, 164-171. Blobel, G., and Dobberstein, B. (1975a).Transfer of proteins across membranes. I. Presence of proteolytically processed and unprocessed nascent immunoglobulin light chains on membrane-bound 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. ]. Cell Biol. 67,852-862. Boedeker, E., Higgins, Y., Donaldson, R., and Small, D. (1976). Selective solubilization of enzyme proteins from microvillus membrane (MVM) using nonionic and anionic detergents. Custroenterology 70,865. (Abstr.) Boegman, R. J., Manery, J. F., and Pinteric, L. (1970). The separation and partial purification of membrane bound (Na+ + K+) dependent M e + ATPase from frog skeletal muscle. Biochim. Biophys. Acta 203,506-530.

GLYCOPROTEIN MEMBRANE ENZYMES

21 3

Bon, S., and Rieger, F. (1975).Interactions between lectins and electric eel acetylcholinesterase. F E E S Lett. 53,282-284. Booth, A., and Kenny, A. (1976). Proteins of the kidney microvillus membrane. Biochem. J . 159,395-407. Bosmann, H. B., Hogopien, A,, and Eylar, E . (1968). Cellular membranes: The isolation and characterization of the plasma and smooth membranes of HeLa cells. Arch. Biochem. Biophys. 128,51-69. Bouregard, G. and Roufogalis, B. D. (1977). The role of tightly bound phospholipid in the activity of erythrocyte acetylcholinesterase. Biochem. Biophys. Res. Commun. 77,211-219. Brady, R., Gal, A., Kanfer, J., and Bradley, R. (1965). The metabolism of glucocerebrosides. 11. Purification and properties of a glycosyl- and galactosylceramide-cleaving enzyme from rat intestinal tissue.J. Biol. Chem. 240,3766-3770. Braun, H., Cogoli, A., and Semenza, G. (1975). Dissociation of small intestinal sucraseisomaltase complex into enzymatically active subunits. Eur. J . Biochem. 52, 475-480. Brenna, O., Perella, M., Pace, M., and Pietta, P. (1975).Affinity chromatography purification of alkaline phosphatase from calf intestine. Biochern. J . 151,291-296. Brodbeck, U., Gentinetta, R., and Lundin, S. J. (1973). Multiple forms of a cholinesterase from body muscle of plaice (Pleuronectes platessa) and possible role of sialic acid in cholinesterase reaction specificity. Acta Chem. Scund. 27, 561-572. Brownlee, S. T., and Heath, E. C. (1975). An extracellular 5'-nucleotidase with both monoesterase and diesterase activity from Micrococcus sodonensis. Archu. Biochem. Biophys. 166, 1-7. Burston, D., Addison, J., and Matthews, D. (1972). Uptake of dipeptides containing basic and acidic amino acids by rat small intestine in oitro. Clin. Sci. 43, 823837. Cacan, R., Verbert, A., and Montreuil, J. (1976). New evidence fop cell surface galactosyltransferase. FEES Lett. 63, 102-106. Callahan: J. W., and Genie, J. (1975). Purification of GM,-ganglioside and ceramidelactoside P-galactosidase from rabbit brain. Biochim. Biophys. Acta 391, 141153. Callahan, J . W., and Wolfe, L. S. (1970).Isolation and characterization of keratin sulfates from the liver of a patient with GM,-gangliosidosis type I. Biochim. Biophys. Acta 215,527-543. Capaldi, R., and Vanderkooi, G . (1972). The low polarity of many membrane proteins. Proc. Natl. Acad. Sci. U.S.A. 69,930-932. Carraway, C. A. C., Jett, G., and Carraway, K. L. (1975). Cooperative effects in the perturbation of membrane enzymes by concanavalin A. Biochem. Biophys. Res. Commun. 67, 1301-1306. Carraway. K. L. and Carraway. C. A. (1976). Concanavalin A perturbation of membrane enzymes of mammary g1and.J. Supramol. Struct. pp. 121-126. Carraway, K. L., Fogle, D. D., Chestnut, R. W., Huggins, J. W., and Carraway, C. A. C. (1976).Ecto-enzymes of mammary gland and its tumours: Lectin inhibition of 5'nucleotidase of the 13762 rat mammary ascites carcinoma. J . Biol. Chem. 251, 6173-6178. Cathala, G . , Bmnel, C., Chappelet-Tordo, D., and Lazdunski, M. (1975a). Bovine kidney alkaline phosphatase: Purification, subunit struchire and metalloenzyme properties.J. Biol. Chem. 250,6040-6045.

214

J. R. RIORDAN A N D G. G. FORSTNER

Cathala, G., Brunel, C., Chappelet-Tordo, D., and Lazdunski, M. (1975b).Bovine kidney alkaline phosphatase: Catalytic properties, subunit interactions in the catalytic process and mechanism of MgZ+ stimulation. J. Biol: Chem. 250,6046-6053. Chang, C., and Moog, F. (1972).Alkaline phosphatases of the chicken duodenum: Isolation and partial characterization of the multiple forms of duodenal phosphatase in pre and post hatching stages. Biochim. Biophys. Acta 250, 154-165. Chappelet-Tordo, D., Fo-Set, M., Iwatsuho, M., Cache, C., and Lazdunski, M. (1974). Intestinal alkaline phosphatase: Catalytic properties and half of sites reactivity. Biochemistry 13, 1788-1795. Chen, S. T. (1976).Alkaline phosphatase. In “Front: Gastrointest. Res. vol2” (L. Vander Reis, ed.), pp. 109-149, Karger, Basel. Chen, W. W., Lennarz, W. J., Tarention, A. L., and Maley, F. (1975).A lipid-linked oligosaccharide intermediate in glycoprotein synthesis in oviduct: Structural studies on the oligosaccharide chain. J . Biol. Chern. 250,7006-7013. Churchill, L., and Hokin, L. E. (1976).The susceptibility of the glycoprotein from the purified (Na+,K+)-activatedadenosine triphosphatase to tryptic and chymotryptic degradation with and without Na+ and K+. Biochim. Biophys. Acta 434,258-264. Ciliv, G. and Ozand, P. T. (1972).Human erythrocyte acetylcholmesterase purification, properties and kinetic behaviour. Biochtm. Biophys. Acta u)4, 136-156. Cogoli, A., Mosimann, H., Vock, C., Von Balthazar, A., and Semenza, G. (1972).A simplified procedure for the isolation of sucrase-isomaltase complex from rabbit intestine. Eur. J. Biochem. 30,7-14. Cogoli, A., Eberle, A., Sigrist, H., Joss, C., Robinson, E., Mosimann, H., and Semenza, G. (1973).Subunits of the small intestinal sucrase-isomaltase complex and separation of its enzymatically active isomaltase moiety. Eur. J . Biochem. 33,4048. Collins, R. C., and Albers, R. W. (1972). The phosphoryl acceptor protein of Na-KATPase from various tissues. J . Neurochem. 19, 1209-1213. Colombo, V., Lorenz-Meyer, H., and Semenza, G. (1973).Small intestinal phlorizin hydroiase: The P-glycosidase complex. Biochim. Biophys. Acta 327,412-424. Conklin, K., Yamshiro, K., and Gray, G. (1975). Human intestinal sucrase-isomaltase identification of free sucrase and isomaltase and cleavage of the hybird into active distinct subunits. J . Biol. Chem. 250,5735-5741. Critchley, D., Howell, K., and Eichholz, A. (1975). Solubilization of brush borders of hamster small intestine and fractionation of some of the components. Biochim. Biophys. Acta 394,361-376. Cuatrecasas, P. (1972).The nature of insulin-receptor interactions. In “Insulin Action” (I. B. Fritz, ed.), pp. 137-170. Academic Press, New York. Curthoys, N. P., and Kuhlenschmidt, T. (1975). Phosphate independent glutaminase from rat kidney: Partial purification and identity with y-glutamyltranspeptidase. J . Biol. Chem. 250,2099-2105. Dahl, J. L., and Hokin, L. E. (1974).The sodium-potassium adenosinetriphosphatase. Annu. Rev. Biochem. 43,327-356. Dalqvist, A. (1961).Determination of maltase and isomaltase activities with a glucoseoxidase reagent. Biochem. J. 80,547-551. Danwalder, M., Whaley, W. G., and Kaphart, J. E. (1972). Functional aspects of the Golgi apparatus. Subcell. Biochem. 1,225-275. Das, B., and Gray, G. (1969).Protein synthesis in small intestine. Localization and correlation with DNA synthesis and sucrase activity. Biochim. Biophys. Acta 195, 255-257. Decker, K.,and Bischoff, E. (1972).Purification and properties of nucleotide pyrophosphatase from rat liver plasma membranes. FEBS Lett. 21,95-98. DePierre, J. W., and Kamovsky, M. L. (1973). Plasma membranes of the mammalian

GLYCOPROTEIN MEMBRANE ENZYMES

215

cell: A review of methods for their characterization and isolation. I . Cell Biol. 56, 275-303. DePierre, J. W., and Karnovsky, M. L. (1974).Ectoenzymes of the guinea pig polymorphontlclear leukocyte. I. Evidence for an ecto-adenosine monophosphatase, -adenosine triphosphatase, and p-nitrophenyl phosphatase. J . Biol. Chem. 249, 71 11-7120. Deppert, W., and Walter, C . (1977).Cell surface glycosyltransferase: Do they exist?J. Supramol. Struct. Suppl. 1,37. d e Robertis, E., and Schacht, J., eds. (1974). “Neurochemistry of Cholinergic Receptors.” Raven, New York. Diedrich, D. F. (1973). On the absence of phorizin hydrolase in renal brush border membranes. Arch. Biochem. Biophys. 153, 155-161. Dingle, J., and Fell, H., eds. (1973).“Lysosomes in Biology and Pathology”. North-Holland Publ., Amsterdam. Distler, J., and Jourdian, G. (1973). The purification and properties of P-galactosidase from bovine testis. J. Biol. Chem. 248,6772-6780. Domschke, N., Liersch, M., and Decker, K. (1971).Lack of permeation of coenzyme A from blood into liver cells. Hoppe-Seyler’s2. Physiol. Chem. 352,85-88. Dubs, R., Bitzelmann, R., Steinmann, B.,and Lindenmann, J. (1975).Catalytically inactive sucrase antigen of rabbit small intestine: The enzyme precursor. Helu. Paediat. Acta 30,89-102. Dudai, Y., and Silman, I. (1973).The effect of Ca2+ on interaction of acetylcholinesterase with subcellular fractions of electric organ tissue from the electric eel. FEBS Lett. 30,49-52. Dudai, Y., and Silman, I. (1974).The effects of solubilization procedures on the release and molecular state of acetylcholinesterase from electric organ tissue from the electric eel. J . Neurochem. 23, 1177-1187. Dudai, Y., Herzberg, M., and Silman, I. (1973).Molecular structures of acetylcholinesterase from electric organ tissue of the electric eel. Proc. Natl. Acad. Sci. U S A . 70, 2473-2476. Edelson, P. J., and Cohn, Z. A. (1976).5’-Nucleotidase activity of mouse peritoneal macrophages. J. E x p . Med. 144,1596-1608. Eggermont, E., and Hers, H. (1969).The sedimentation properties of the intestinal aglucosidases of normal human subjects and of patients with sucrose intolerance. Eur. J . Biochem. 9,488-496. Elce, J. S . (1970). Metabolism of a glutathione conjugate of 2-hydroxyoestradiol by rat liver and kidney preparations in vitro. Biochem. J. 116,913-917. Elce, J. S., and Broxmeyer, B. (1976).y-Glutamyltransferase of rat kidney: Simultaneous glutathione. Bioassay of the hydrolysis and transfer reactions with (gl~tamate-’~C) chem. J. 153,223-232. Elovson, J. (1977).Biogenesis of liver plasma membrane glycoproteins. J . Supramol. Stmct. Suppl. 1, 19. Emmelot, P., and Bos, C. J. (1966). Studies on plasma membranes. 111. Mg”-ATPase (Na+ + K+-Mg2+)ATPase and 5‘-nucleotidase activity on plasma membranes isolated from rat liver. Biochim. Biophys. Acta 120, 369-382. Etzlar, M., and Moog, F. (1966).Inactive alkaline phosphatase in the nursling duodenum: Immunological evidence. Science 154, 1037-1038. Evans, W. H. (1974).Nucleotide pyrophosphatase, a glycoprotein located in the hepatocyte surface. Nature (London)250,391-394. Evans, W. H., and Curd, J. W. (1973). Properties of a 5’-nucleotidase purified from mouse liver plasma membrane. Biochem. J . 133, 189-199. Evans, W., Hood, D., and Curd, J. W. (1973).Purification and properties of a mouse liver

216

J. R. RIORDAN AND G. G. FORSTNER

plasma membrane glycoprotein hydrolysing nucleotide pyrophospbate and phosphodiester bonds. Biochem. J. 135,819-826. Fahn, S. (1968). Sodium-potassium-activated adenosine triphosphatase of electric organ. J . Biol. Chem. 243, 1993-2002. Fahn, S., Koval, G. J., and Albers, R. W. (1966). Sodium-potassium-activated adenosine triphosphatase of Electrophorus electric organ. I. An associated sodium-activated transphosphorylation. J . Biol. Chem. 241, 1882-1889. Farquhar, M. G., Bergeron, J. J. M., and Palade, G. E. (1974). Cytochemistry of Golgi fractions prepared from rat liver. J . Cell Biol. 60,8-25. Fernley, H. (1971). Mammalian alkaline phosphatases. I n “The Enzymes” (P. Boyer, ed.), Vol. 4, pp. 417-447. Academic Press, New York. Fernley, H., and Walker, P. (1957). Studies on alkaline phosphatase: Inhibition by phosphatase derivatives and substrate specificity. Biochem. J . 104, 101-108. Ferry, C. B., and Marshall, A. R. (1973). Nature of physiologically important cholinesterase at mammalian end plates. Eur. J. Pharmacol. 23, 111-114. Fishman, W. H. (1974). Perspectives on alkaline phosphatase isoenzymes. Am. J . Med. 56,617-650. Fishman, W. H., and Lin, C. W. (1973). “Membrane Phosphohydrolases, Metabolic Conjugation and Metabolic Hydrolysis,” p. 387. Academic Press, New York. Flanagan, P. R. and Forstner, G. G. (1976). Dissociation of rat intestinal maltase-glucoamylase. Fed. Proc. Fed. Am. SOC. E x p . Biol. 35,284. Flanagan, P. R., and Forstner, G. G. (1977). The enzymic activity of transient sodium dodecyl sulphate-protein complexes of rat intestinal maltase-glucoamylase formed during dissociation. Biochem. SOC. Trans. 5, 1542-1543. Flanagan, P. R., and Forstner, G. G. (1978). Rat intestinal maltase-glucoamylase. Purification of the enzyme and its anomalous dissociation by heat or low pH. Biochem. J . (in press). Fleit, H., Conklyn, M., Stebbins, R. D., and Silber, R. (1975). Function of 5‘-nucleotidase in the uptake of adenosine from AMP by human 1ymphocytes.J. Biol. Chem. 250,8889-8892. Forstner, G. G. (1971).Release of intestinal surface-membrane glycoproteins associated with enzyme activity by brief digestion with papain. Biochem. J . 121,781-789. Fosset, M., Chappelet-Tordo, D., and Lazdunski, M. (1974). Intestinal alkaline phosphatase: Physical properties and quaternary structure. Biochemistry 13, 17831788. Fraser, I. H., and Mookerjea, S. (1976). Studies on the purification and properties of UDP-galactose-glycoprotein galactosyltransferase from rat liver and serum. Biochem. J . 156,347-355. Fraser, I. H. and Mookerjea, S. (1977).Purification of membrane-bound galactosyltransferase from rat liver microsomal fractions. Biochem. J . 164,541-547. Fujita, M., Ohta, H., Kawai, K., Matsui, H., and Nakao, M. (1972). Differential isolation of microvillous and basolateral plasma membranes from intestinal mucosa. Biochim. Biophys. Acta 274,336-347. Furthmayer, H., and Marchesi, V. T., (1977). Sialoglycoproteins of human erythrocyte membranes.J. Supramol. Struct., Suppl. 1, 1. Galand, G., and Forstner, G. (1974). Soluble neutral and acid maltases in the suckling rat intestine. Biochem. J . 144,281-292. Gamer, C. W., and Behal, F. J. (1974). Human liver aminopeptidase: Role of metal ions in mechanism of action. Biochemistq 13,3227-3232. Gavryck, W. A., Moore, R. D., and Thompson, R. C. (1975). Effect of insulin upon membrane-bound (Na+ K+)-ATPase extracted from skeletal muscle. J . Physiol. (London) 2 5 2 , 4 3 4 8 .

+

GLYCOPROTEIN MEMBRANE ENZYMES

217

G‘eorge, S. G., and Kenny, A. J. (1973). Studies on the enzymology of a purified preparation of brush border from rabbit kidney. Biochem. J . 134,43-57. Geren, C. R., Magee, S. C., and Ebner, K. E. (1976). Hydrophobic chromatography of galactosyltransferase. Arch. Biochem. Biophys. 172, 149-155. Ghosh, N., and Usategui-Gomez, M. (1969).Viral haemogglutination inhibitory potency of human placental alkaline phosphatase. Biochim. Biophys. Acta 177,565-571. Ghosh, N., Goldman, S., and Fishman, W. (1967). Human placental alkaline phosphatase, a sialoprotein. Enzymologia 33, 113-124. Ghosh, N., Ruckenstein, A., Labowsky, C., and Cox, P. (1971). Human placental alkaline phosphatase, an inhibitor of haemagglutination by PR8 influenza A virus. J . Clin. Invest. 50,35a. Ghosh, N., Cox, R., and Winzler, R. (1974). Characterization of monosaccharides and human placental alkaline phosphatase by gas liquid chromatography. Biochim. Biophys. Acta 343,638-640. Gitzelmann, R., Bachi, T., Binz, H., Lindenmann, J., and Semenza, G. (1970). Localization of rabbit intestinal sucrase with ferritin antibody complexes. Biochim. Biophys. Acta 196,20-28. Glynn, I. M. (1968). Membrane ATPase and cation transport. Br. Med. Bull. 24, 165169. Glynn, I. M., and Karlish, S. J. D. (1975). The sodium pump. Annu. Rev. Physiol. 37, 13-55. Goldin, S. M., and Tong, S. W. (1974). Reconstitution of active transport catalysed by the purified sodium and potassium ion-stimulated adenosine triphosphatase from canine renal medulla. J . Biol. Chem. 249, 5907-5915. Goldstone, A., and Koenig, H. (1970). Lysosomal hydrolases are glycoproteins. Life Sci. 9,1341-1345. Goldstone, A., Konecny, P., and Koenig, H. (1971).Lysosomal hydrolases: Conversion of acidic to basic forms by neuraminidase. F E E S Lett. 13,68-72. Gottlieb, A., and Sussman, H. (1968). Human placental alkaline phosphatase: Molecular weight and subunit structure. Biochim. Biophys. Acta 160, 167-171. Gottlieb, C., Baenziger, J., and Kornfeld, S. (1975). Deficient uridine diphosphate-Nacetylg1ucosamine:glycoprotein N-acetylglucosaminyltransferase activity in a clone of Chinese hamster ovary cells with altered surface glycoproteins. J . Biol. Chem. 250,3303-3309. Gotze, H., Adelson, J., Hadorn, B., Porhnann, R., and Troesch, V. (1972). Hormone elicited enzyme release by the small intestinal wall. Gut 13,471-476. Grafinus, M. A., Bond, H. E., and Millar, D. B. (1971).Acetylcholinesterase interaction with a lipoprotein matrix. Eur. J . Biochem. 22,382-390. Graham, E., and Roy, A. (1973). The sulphatase of ox liver. XVII. Sulphatase as a glycoprotein. Biochim. Biophys. Acta 329,88-92. Gray, G., and Santiago, N. (1969). Intestinal P-galactosidases. I. Separation and characterization of three enzymes in normal human intestine.]. Clin. Invest. 48,716-728. Curd, J. W. (1976). Fractionation of rat brain acetylcholinesterase by lectin affinity chromatography. J . Neurochem. 27,1257-1259. Curd, J. W., and Evans, W. H. (1974).Distribution of liver plasma membrane 5’-nucleotidase as indicated by its reaction with anti-plasma membrane serum. Arch. Biochem. Biophys. 164,305-311. Hadorn, B., Steiner, N., Sumuda, C., and Peters, T. J. (1971). Intestinal enterokinase: Mechanisms of secretion into the lumen of the intestine. Lancet 1, 165-166. Hall, Z. W. (1973); Multiple forms of acetylcholinesterase and their distribution in end-plate and non-endplate regions of rat diaphragm muscle. /. Neurobiol. 4, 343462.

21 8

J. R. RIORDAN AND G. 0. FORSTNER

Hanes, C., Hird, F., and Isherwood, F. (1950).Synthesis of peptides in enzymatic reactions involving glutathione. Nature (London) 166,288-292. Haworth, J. C., Hadom, B., Gourley, B., Presad, A., and Troesch, V. (1975).Intestinal enterokinase deficiency, occurrence in two sibs and age dependency of clinical expression. Arch. Dis. Child. 50,277-282. Hayman, M., and Crumpton, M. J. (1972).Isolation of glycoproteins from pig lymphocyte plasma membrane using Lens culinaris phytohemagglutinins. Biochem. Biophys. Res. Commun. 47,923-930. Herbst, J., and Koldovsky, 0. (1972).Cell migration and cortisone induction of sucrase activity in jejunum and ileum. Biochem. J . 126,471-476. Hers, H., and Van Hoof, F., eds. (1973).“Lysosomes and Storage Diseases.” Academic Press, New York. Hilden, S., Rhee, H., and Hokin, L. E. (1974).Sodium transport by phospholipid vesicles containing purified sodium and potassium ion-activated adenosine triphosphatase. J . Biol. Chem. 249,7432-7440. Ho, M., and Guidotti, G. (1975).Membrane protein from human erythrocytes involved in anion exchange. J . Biol. Chem. 250,675-683. Hosli, P., Erickson, R. P., and Vogt, E. (1976).Prospects for prenatal diagnosis of cystic fibrosis: Induction of biochemical abnormalities in fibroblasts from patients with cystic fibrosis by a urinary glycoprotein. Biochem. Biophys. Res. Commun. 73,209216. Hokin, L. E. (1974).Purification and properties of the Na+ + K+ activated ATPase and reconstitution of Na+ transport. Ann. N.Y. Acad. Sci. 242, 12-23. Hokin, L. E., Dahl, J. L., Deupree, J. D., Dixon, J. F., Hackney, J. F., and Perdue, J. F. (1973). Studies on the characterization of the sodium-potassium transport adenosine triphosphatase. J. Biol. Chem. 248,2593-2605. Hughey, R. P., and Curthoys, N. P. (1976).Comparison of the size and physical properties of y-glutamyltranspeptidase purified from rat kidney following solubilization with papain or with Triton X-100.1. Biol. Chem. 24,7863-7870. Hulstaert, C. E., Torringa, J. L., Koudstaal, J., Hardonk, M. J., and Molenaar, I. (1973). The characteristic distribution of alkaline phosphatase in the full term human placenta. Gynecol. Invest. 4,23-30. Iknonne, J., Ratzzi, M., and Desnick, R. (1975).Characterization of Hex S, the major residual beta-hexosaminidase activity in type 0 GM, gangliosidosis (Sandhoff-Jatzkewitz disease). Am. J . Hum. Genet. 27,639-650. Inoue, M., Horiuchi, S., and Morino, Y. (1977).Affinity labeling of rat-kidney y-glutamyl transpeptidase. Eur. J. Biochem. 73,335-342. Ipata, P. L. (1968). Sheep brain 5‘-nucleotidase: Some enzymatic properties and allosteric inhibition by nucleoside triphosphates. Biochemistry 7 , 507-515. Irwin, D., and Anastassiades, T. P. (1975).Cell surface sialoglycopeptide metabolism and surface glycosyl transferase activity in the Ehrlich ascites cell. Can.J. Biochem. 53,895-902. Itoh, R., Mitsui, A,, and Tsushima, K. (1968).Properties of 5’-nucleotidase from hepatic tissue of higher animals. J. Biochem. (Tokyo) 63, 165-169. Jamieson, G. A., Urban, C. L., and Barber, A. J. (1971). Enzymatic basis for platelet: Collagen adhesion as the primary step in haemostasis. Nature (London),New Biol. 234,5-7. Jean, D. H., Albers, R. W., and Koval, G. J. (1975). Sodium-potassium-activated adenosine triphosphatase of Electrophorus electric organ. J . Biol: Chem. 250, 1035-1040. Jeffrey, P. L., Brown, D. H., and Illingsworth-Brown, B. (1970). Studies of lysosomal

GLYCOPROTEIN MEMBRANE ENZYMES

219

a-glucosidase. I. Purification and properties of the rat liver enzyme. Biochemistry 9, 1403-1415. Jhrgensen, P. L. (1974). Purification and characterization of (Na+ + K+) ATPase. Biochim. Biophys. Actu 356,36-52. Jbrgensen, P. L., Hansen, O., Glynn, I. M., and Cavieres, J. D. (1973). Antibodies to pig kidney (Na+ + K+)-ATPase inhibit the Na+ pump in human red cells provided they have access to the inner surface of the membrane. Biochim. Biophys. Actu 291, 795-800. Johnson, C. (1967). Disaccharidase localization in hamster intestinal brush borders. Science 155,1670-1672. Jutisz, M., and de la Llosa, P. (1972). Glycoproteins as hormones and as enzymes. In “Glycoproteins, Their Composition, Structure and Function” (A. Bottschalk, ed.), pp. 1019-1062, Elsevier, Amsterdam. Kaplan, A., Achord, D. T., and Sly, W. S. (1977). Phosphohexosyl components of a lysosoma1 enzyme are recognized by pinocytosis receptors on human fibroblasts. Proc. Natl. Acad. Sci. U.S.A. 74,2026-2030. Kartner, N., Alon, N., Swift, M., Buchwald, M., and Riordan, J. R. (1977). Isolation of plasma membranes from human skin fibroblasts. J. Membr. Biol. 36, 191-211. Katunuma, N., Katsunuma, T., Tomino, I., and Matsuda, Y. (1968). Regulation of glutaminase activity and differentiation of the isoenzyme during development. Adu. Enzyme Regul. 6,227-242. Katz, K., Wirth, D., and Lodish, H. (1977). Biosynthesis and maturation of virus membrane glycoproteins. J . Suprumol. Struct., Suppl. 1, 13. Kawasaki, T., and Ashwell, G. (1976). Carbohydrate structure of glycopeptides isolated from an hepatic membrane-binding protein specific for asialoglycoproteins. J. Biol. Chem. 251,5292-5299. Keenan, T. W., and MorrB, D. J. (1975).Glycosyltransferases: Do they exist on the surface membrane of mammalian cells? FEBS Lett. 5 5 , 8 4 3 . Kelley, P., Neumann, P., Shriefer, K., Cancedda, F., Schlesinger, M., and Bradshaw, R. (1973). Amino acid sequence of E . coli alkaline phosphatase and variations between two isoenzymes. Biochemistry 12,3499-3503. Kelly, J., and Alpers, D. (1973a).Properties of human intestinal glucoamylase. Biochim. Biophys. Actu 315, 113-120. Kelly, J., and Alpers, D. (197313).Blood group antigeneity of purified human intestinal disaccharidases. J . B i d . Chem. 248,8216-8221. Kerr, M. A., and Kenney, A. J. (1974a).The purification and specificity of a neutral endopeptidase from rabbit kidney brush border. Biochem. J. 137,477-488. Kerr, M. A., and Kenney, A. J. (1974b).The molecular weight and properties of a neutral metallo-endopeptidase from rabbit kidney brush border. Biochem. J. 137,489-495. Kim, Y. S., and Brophy, E. J. (1976). Rat intestinal brush border membrane peptidases. I. Solubilization, purification, and physicochemical properties of two different forms of the enzyme. J . Biol. Chem. 251,3199-3205. Kim, Y. S., Brophy, E. J., and Nicholson, J. A. (1976). Rat intestinal brush border membrane peptidases. 11. Enzymatic properties, immunochemistry and interactions with lectins of two different forms of the enzyme.J. B i d . Chem. 251,3206-3212. Kimelberg, H. K. (1978). The influence of membrane fluidity on the activity of membrane-bound enzymes. In “Cell Surface Reviews. Vol. 3: Dynamic Aspects of Cell Surface Organization” (G. Poste and G. L. Nicolson, eds.), In Press. North-Holland Publ., Amsterdam. Knipe, D. M., Baltimore, D., and Lodish, H. F. (1977a). Localization of two cellular forms of the vesicular stomatitis viral glycoprotein. J . Virol. 21, 1121-1127.

220

J. R. RIORDAN AND G. G. FORSTNER

Knipe, D. M., Lodish, H. F., and Baltimore, D. (1977b).Separate pathways of maturation of the major structural proteins of vesicular stomatitis virus.]. Virol. 21, 11281139. Koldovsky, 0. (1969).Digestion of disaccharides. In “Development of the Functions of the Small Intestine in Mammals and Man,” pp. 44-59. Karger, Basel. Kolinska, J., and Semenza, G. (1967).Studies on intestinal sucrase and on intestinal sugar transport. V. Isolation and properties of sucrase-isomaltase from rabbit small intestine. Biochim. Biophys. Acta 146, 181-195. Komoda, T., and Sakagishi, Y. (1976a).Partial purification and some properties of human liver alkaline ghosphatase. Biochim. Biophys. Acta 438, 138-152. Komoda, T., and Sakagishi, Y. (1976b).Partial purification of human intestinal alkaline phosphatase with affinity chromatography: Some properties and interaction of concanavalin A with alkaline phosphatase. Biochim. Biophys. Acta 445,645-660. Konrad, P., Richards, F., Valentine, W., and Paglia, D. (1972).y-Glutamyl-cysteine synthetase deficiency: A cause of hereditary hemolytic anemia. N. Engl. J . Med. 286,557-561. Kornfeld, R., and Ferris, C. (1975).Interaction of immunglobulin glycopeptides with concanavalin A. J . Biol. Chem. 250,2614-2619. Kottgen, E., Reutter, W., and Gerok, W. (1976).Two different y-glutamyltransferases during development of liver and small intestine: A fetal (sialo) and an adult (asialo) glycoprotein. Biochem. Biophys. Res. Commun. 72,61-66. Kuhlenschmidt, T., and Curthoys, N. P. (1975).Subcellular localization of rat kidney phosphate independent glutaminase. Arch. Biochem. Biophys. 167,519-524. Kyte, J . (1971).Enzymatic action of sialidase of Vibrio cholerae on brain gangliosides above and below the critical micelle concentration.]. Biol. Chem. 246,7642-7648. Kyte, J. (1972).Properties of the two polypeptides of sodium- and potassium- dependent adenosine triphosphatase. 1.Biol. Chem. 247,7642-7649. Kyte, J . (1974).The reactions of sodium and potassium ion-activated adenine triphosphatase with specific antibodies. J . Biol. Chem. 249,3652-3600. Lamont, J. T., Perrotto, J. L., Weiser, M. W., and Isselbacher, K. J. (1974).Cell surface galactosyltransferase and Iectin agglutination of thymus and spleen lymphocytes. Proc. Natl. Acad. Sci. U.S.A. 71,3726-3730. Lazdunski, C., and Lazdunski, M. (1966).etude cinetique du mecanisme d’action catalytique de la phosphatase alcaline d’Escherichia coli. Biochim. Biophys. Acta 113, 551-565. Leblond, C . P. (1950).Distribution of periodic acid-reactive carbohydrates in adult rat. Am. J . Anat. 86, 1-49. Lee, J., and Yoshida, A. (1976).Purification and characterization of human hexosaminidases A and B. Biochem. J. 159,535-539. Leese, H. J., and Semenza, G. (1973).On the identity between the small intestinal enzymes phlorizin hydrolase and glycosylceramidase. 1. Biol. Chern. 248, 81708173. Lehman, E. D., Hudson, B. G., and Ebner, K. E. (1975).Studies on carbohydrate s t r u c ture of bovine milk galactosyltransferase. FEBS Lett. 54,65-68. Leibach, F. H.,and Binkley, F. (1968).y-Glutamyltransferase of swine kidney. Arch. Biochem. Biophys. 127,292-301. Lejeune, N., Thines-Sempaux, D., and Hers, H. (1963).Tissue fraction studies. 16.Intracellular distribution of properties of a-glucosidases in rat liver. Biochem. ]. 86, 16-21. Liersch, M., Groteluschen, H., and Decker, K. (1971).Zur Frage der Permeation von NAD in die Leberzelle. Hoppe-Seyler’s 2. Phystol. Chem. 352,267-274.

GLYCOPROTEIN MEMBRANE ENZYMES

221

Lis, H., and Sharon, N. (1973).The biochemistry of plant lectins (phytohemagglutinins). Annu. Rev. Biochem. 42,541-574. Lis, H. and Sharon, N. (1977). Lectins: Their chemistry and applications to immunology. In “The Antigens” (M. Sela, ed.) vol. 4. Academic Press, New York. Lorenz-Meyer, H., Blum, A., Haemmerli, H., and Semenza, G. (1972).A second enzyme defect in acquired lactase deficiency: Lack of small intestinal phlorizin hydrolase. Eur. J . Clin. Inuest. 2,326-331. Louisot, P., and Morelis, R. (1973). Biosynthese des glycoproteines dans les mitochondries. I. J h d e d e I’activite mannosyl-transferase mitochondriale. Biochimie 55, 671-677.

Louvard, D., Maroux, S., Baratti, J., and Desnuelle, P. (1973). On the distribution of enterokinase in porcine intestine and on its subcellular localization. Biochim. Biophys. Acta 309,127-137. Louvard, D., Maroux, S., Vennier, C. H., and Desnuelle, P. (1975a).Topological studies on the hydrolases bound to the intestinal brush border membrane. Biochim. Biophys. Acta 375,236-248. Louvard, D., Maroux, S., and Desnuelle, P. (1975b). Topological studies on the hydrolases bound to the intestinal brush border membrane. 11. Interactions of free bound aminopeptidase with a specific antibody. Biochim. Biophys. Acta 389,389-400. Louvard, D., Semeriva, M., and Maraow, S. (1976). The brush border aminopeptidase, a transmembrane protein as probed by macromolecular photolabel1ing.J. Mol. Biol. 106,1023-1035.

Lucas, J. J., Waechter, C. J., and Lennarz, W. J. (1975). The participation of lipid-linked oligosaccharide in synthesis of membrane glycoproteins. J . Biol. Chem. 250,19922002.

Luly, P., and Emmelot, P. (1975). Studies on plasma membranes. XXIII. Hormone-like action of concanavalin A on liver plasma membranes: Inhibition of (Na+ + K+) ATPase. Chem.-Biol.Interact. 11,377-385. McGuire, E. (1972). A possible role for carbohydrates in cell-cell adhesion. I n “Membrane Research” (C. F. Fox, ed.), pp. 347-368. Academic Press, New York. McLean, R. J., and Bosmann, H. B. (1975). Cell-cell interactions: Enhancement of glycosyltransferase ectoenzyme systems during Chlamydomonas gamete contact. Proc. Natl. Acad. Sci. U.S.A. 72,310-313. Maestracci, D. (1976). Enzymic solubilization of the human intestinal brush border membrane enzymes. Biochim. Biophys. Acta 433,469-481. Maestracci, D., Preiser, H., Hedges, T., Schmitz, J., and Crane, R. K.(1975).Enzymes of the human intestinal brush border membrane: Identification after gel electrophoretic separation. Biochim. Biophys. Acta 382, 147-156. Magee, S. C., Mawal, R., and Ebner, K. E. (1973). Proteolytic conversion of the molecular forms of bovine milk galactosyl transferase. J . B i d . Chem. 248, 7565-7569. Main, A. R., Tarkan, E., Aull, J. L., and Soucie, W. G. (1972). Purification of horse serum cholinesterase by preparative polyacrylamide gel electrophoresis. J. Biol. Chem. 247,566-571.

Makita, T., and Sandborn, E. B. (1970).The ultrastructural localization of adenosine triphosphatase and alkaline phosphatase activity in eosinophilic leukocytes. Histochemie 24,99-105. Malathi, P., and Crane, R. K.(1968). Spatial relationship between intestinal disaccharidases and the active transport system for sugars. Biochim. Biophys. Acta 163,275277.

Malathi, P., and Crane, R. K. (1969). Phlorizin hydrolase: A 8-glucosidase of hamster intestinal brush border membrane. Biochim. Biophys. Acta 173,245-256.

222

J. R. RIORDAN AND G. G. FORSTNER

Malathi, P., Ramaswamy, K., Caspary, W.,and Crane, R. (1973).Studies on the transport of glucose from disaccharides by hamster small intestine in oitro. I. Evidence for a disaccharidase-related transport system. Blochim. Biophys. Acta 307,613-666. Malik, N., and Butterworth, P. J. (1976).Molecular properties of rat intestinal alkaline phosphatase. Biochim. Biophys. Acta 446,105-114. Maroux, S., and Louvard, D. (1976).On the hydrophobic part of aminopeptidases and maltases which bind the enzyme to the intestinal brush border membrane. Biochim. Biophys. Acta 419, 189-195. Maroux, S., Baratti, J., and Desnuelle, P. (1971).Purification and specificity of porcine enterokinase. J . Biol. Chem. 246,5031-5039. Maroux, S., Louvard, D., and Baratti, J. (1973).The amino peptidase from hog intestinal brush border. Biochim. Biophys. Acta 321,282-295. Marshall, M. 0. (1976).Studies on the glycoprotein component of (Na+ + K+)-ATPase from dogfish salt gland. Binding to concanavalin A and removal of sialic acid by neuraminidase. Biochim. Biophys. Acta 455,837-848. Massoulie, J.,and Rieger, F. (1969).Acetylcholinesterase in electric organs of fish (Torpedo and gymnotid) and membrane systems. Eur. J . Biochem. 11,441-455. Massoulie, J., Rieger, F., and Bon. S. (1971).EspBces acetylcholinesterasiques globulaires et allong6es des organes electriques de Poisson. Eur. J . Biochem. 21, 542-551. Mathews, F. S., and Czerwinski, E. W. (1976).Cytochrome b, and cytochrome b, reductase from a chemical and x-ray diffraction viewpoint. In “The Enzymes of Biological Membranes. Vol. 4: Electron Transport Systems and Receptors” (A. Martonosi, ed.), pp. 143-194. Plenum, New York. Mawal, R., Morrison, J. F., and Ebner, K. E. (1971). Studies on galactosyltransferase: Detection of enzyme-reactant complexes by means of affinity chromatography. J . Biol. Chem. 246,7106-7109. Mehl, E., and Jatzkewitz, H. (1W).Eine Cerebrosidsulfatase aus Schweineniere. Hoppe-Seyler’s Z . Physiol. Chem. 339,260-276. Meister, A. (1973).On the enzymology of acid transport. Science 180,33-39. Meister, A. (1974).Glutathione, metabolism and function via the y-glutamyl cycle. Life Sci. 15,177-190. Meister, A., Tate, S., and Ross, L. (1976). Membrane-bound y-glutamyltranspeptidase in enzymes of biological membranes. In “The Enzymes of Biological Membranes. Vol. 3: Membrane Transport” (A. Martinosi, ed.), pp. 315-347. Plenum, New York. Miller, D., and Crane, R. (1961).The digestive function of the epithelium of small intestine. I. An intracellular locus of dissacharide and sugar phosphate ester hydrolysis. Biochim. Biophys. Acta 52,281-293. Miller, S., Awasthi, Y., and Srivastava, S . (1976).Studies of human kidney y-glutamyl transpeptidase: Purification and structural, kinetic and immunological properties.]. Biol. Chem. 251,2271-2278. Mitranic, M., Sturgess, J. M., and Moscarello, M. A. (1976).The effect of temperature on the galactmyl and sialyl transferases and on the ultrastructure of Golgi membranes. Biochim. Biophys. Acta 443, 190-197. Moelbert, E., Duspiva, F., and Deimling, 0. (1960).Histochemical localization of phosphatases in tubular epithelial cells of the mouse kidney in the electron microscopic picture. Histochemie 2,5-22. Mosimann, H., Semenza, G., and Sund, H. (1973).Hydrodynamic properties of the sucrase-isomaltase complex from rabbit small intestine. Eur. J . Biochem. 36,489-494.

GLYCOPROTEIN MEMBRANE ENZYMES

223

Moss, D., Eaton, R., Smith: J., and Whitby, L. (1966). Alteration in the electrophoretic mobility of alkaline phosphatases after treatment with neuraminidase. Biochem. J. 98,320-330. Moss, D., Eaton, R., Smith, J., and Whitby, L. (1967). Association of inorganic pyrophos.. . phatase activity with human alkaline-phosphatase preparations. Biochem. J. 102, 53-57. Munro, J. R., Narasimhan, S., Wetmore, S., Riordan, J . R., and Schachter, H. (1975). Intracellular localization of GDP-L-fucose:glycoprotein and CMP-sialic acid:apolipoprotein glycosyltransferases in rat and pork livers. Arch. Biochem. Biophys. 169, 269-277. Naito, Y., Itoh, R., and Tsushima, K. (1974).5‘-Nucleotidase of chicken liver: A comparison of soluble 5’-nucleotidase activities in chicken and rat liver. Int. J. Biochem. 5, 807-810. Nakamura, S. (1976). Effect of sodium deoxycholate on 5’-nucleotidase. Biochim. Biophys. Acfa 426,339-347. Nakao, M., Nakao, T., Hara, Y., Nagai, F., Yagasaki, S., Koi, M., Nakagawa, A., and Kawai, K. (1974). Purification and properties of (Na+ + KC)-ATPasefrom pig brain. Ann. N.Y. Acad. Sci. 242,24-35. Nakao, T., Toshima, K., Nagano, K., and Nakao, M. (1965). Highly specific sodium-potassium-activated adenosine triphosphatase from various tissues of rabbit. Biochem. Biophys. Res. Commun. 19,755-758. Nayudu P., and Hercus, F. (1974). Molecular heterogeneity of most duodenal alkaline phosphatase. Biochem. J. 141,93-101. Neuberger, A., Gottschalk, A., Marshall, R. D., and Spiro, R. G. (1972). Carbohydratepeptide linkages in glycoproteins and methods for their elucidation. In “Glycoproteins: Their Composition, Structure and Function,” Part A (A. Gottschalk, ed.), pp. 450-486. Elsevier, Amsterdam. Neumann, E., and Nachmansohn, D. (1975). Nerve excitability-Towards an integrating concept. In “Biomembranes” (L. A. Manson, ed.), pp. 99-166. Plenum, New York. Neumann, E., Nachmansohn, D., and Katchalsky, A. (1973). An attempt at an integral interpretation of nerve excitability. Proc. Natl. Acad. Sci. U S A . 70,727-731. Neunier, J. C., Sealock, R., Olsen, R., and Changeux, J. P. (1974). Purification and properties of the cholinergic receptor protein from electrophorus electricus electric tissue. Eur. 1.Biochem. 45,371-394. Newby, A. C., Luzio, J. P., and Hales, C. N. (1975). The properties and extra-cellular location of 5’-nucleotidase of the rat fat-cell plasma membrane. Biochem. J . 146, 625-633. Nichol, L., and Roy, A. (1964). The sulphatase of ox liver VIII. The sedimentation of purified sulphatase A. J. Biochem. (Tokyo) 55,643-651. Nichol, L., and Roy, A. (1965). The sulfate of ox liver. IX. The polymerization of sulfatase A. Biochem. 4,386-396. Nicolson, G., Lacorbiere, M., and Delmonte, P. (1972). Outer membrane terminal saccharides of bovine liver nuclei and mitochondria. E r p . Cell. Res. 71,468-473. Nicolson, G. (1973). Neuraminidase “unmasking” and failure of trypsin to “unmask” Bgalactose-like sites on erythrocyte, lymphoma, and normal virus-transformed fibroblast cell membranes. J. Nat. Cancer Znst. 50, 1443-1451. Nicolson, G. L., and Singer, S. J. (1974). The distribution and asymmetry of mammalian cell surface saccharides utilizing femtin-conjugated plant agglutinins as specific saccharide stains. J. Cell. B i d . 60, 236-248.

224

J. R. RIORDAN AND G. G. FORSTNER

Norden, A. G., and O’Brien, J. S. (1973). Ganglioside GM, B galactosidase: Studies in human liver and brain. Arch. Biochem. Biophys. 159,383-392. Norden, A. G., and OBrien, J. S. (1974). Binding ofhuman liver P-galactosidase to plant lectins immobilized in agarose. Biochem. Biophys. Res. Commun. 56, 193-198. Norden, A. G., Tennant, L., and O’Brien, J. S. (1974).GM, ganglioside B galactosidase A. Purification and studies of the enzyme from human liver. /. B i d . Chem. 249, 7969-7976. Nordstrom, C. (1972). Enzymic release of enteropeptidase from isolated rat duodenal brush borders. Biochim. Biophys. Acta 268,711-718. Nordstrom, C., and Dahlqvist, A. (1971). Rat enterokinase: The effect of ions and the localization in the intestine. Biochim. Biophys. Acta 242,209-225. Nordstrom, C., Dahlqvist, A,, and Josefson, L. (1967). Quantitative determination of enzymes in different parts of the villi and crypts of rat small intestine. /. Histochem. Cytochem. 15,713-721. Novogrodsky, A. (1972). Concanavalin A stimulation of rat lymphocyte ATPase. Biochim. Biophys. Acta 266,343-349. O’Brien, J. F., Cautz, M., and Spranger, J. (1974). Maroteaux-Lamy disease (mucopolysaccharidosis VI), subtype A: Deficiency of a N-acetylgalactosamine-4-sulfatase. Biochem. Biophys. Res. Commun. 60,1170-1177. O’Brien, J. S., Stem, M., Landing, B., O’Brien, J. K.,and Donnell, G. (1965). General gangliosidosis: Another unborn error of ganglioside metabolism. Am. J. Dis. Child. 109,338-346. Oda, T., and Seki, S. (1966). Molecular basis of structure and function of the plasma membrane of the microvilli of intestinal epithelial cells. Int. Congr. Electron Micros., 6th, Kyoto pp. 387-388. Ogata, S., Muramatsu, T., and Kobata, A. (1975). Fractionation of glycopeptides by affinity column chromatography on concanavalin A-Sepharose. J . Biochem. (Tokyo) 78, 687-696. Ohkubo, A., Langerman, N., and Kaplan, M. M. (1974). Rat liver alkaline phosphatase: Purification and properties. J . Biol. Chem. 249,7174-7180. Okada, S., and OBrien, J. S.(1968). Generalized gangliosidosis: Beta-galactosidase deficiency. Science 160,1002-1004. Orlowski, M., and Meister, A. (1965). Isolation of y-glutamyltranspeptidasefrom hog kidney. J . Biol. Chem. 240,338-347. Orlowski, M., and Meister, A. (1970). The y-glutamyl cycle: A possible transport system for amino acids. Proc. Natl. Acad. Sci. U.S.A. 67, 1248-1255. Painter, R. G., and White, A. (1976). Effect of concanavalin A on expression of cell surface sialyltransferase activity of mouse thymocytes. Proc. Natl. Acad. Sci. U.S.A.73, 837-841. Palade, G . E. (1959).Functional changes in the structure of cell components. In “Subcellular Particles” (T. Hayashi, ed.), pp. 64-83. Ronald Press, New York. Palmer, T. (1971). The substrate specificity of acid a-glucosidase from rabbit muscle. Biochem. J. 124,701-711. Panfili, E., Sottocasa, G. L., and d e Bernard, B. (1971). A high molecular weight form of NADH-cytochrome b, reductase from ox liver microsomes. Biochim. Biophys. Actu 253,323-331. Patt, L. M., and Grimes, W. J. (1974). Cell surface glycolipid and glycoprotein glycosylbansferases of normal and transformed cells. J. B i d . Chem. 249,4157-4165. Patt, L. M., and Grimes, W. J. (1975). Ectoglycosyltransferase activity in suspensions and monolayers of cultured fibroblasts. Biochem. Biophys. Res. Commun. 67, 483-490.

GLYCOPROTEIN MEMBRANE ENZYMES

225

Patt, L. M., and Grimes, W. J. (1976). Formation of mannosyl-lipids by a n ectomannosyl transferase in suspensions of BALB/c fibroblasts. Biochim. Biophys. Acta 444, 97-107. Pentchev, P., Brady, R., Hibbert, S., Gal, A., and Shapiro, D. (1973). Isolation and characterization of glucocerebrosidase from human placental tissue.]. Biol. Chem. 248, 5256-5261. Peters, T. J. (1973).The hydrolysis of glycine oligopeptides by guinea pig intestinal mucosa and by isolated brush borders. Clin. Sci. Mol. Med. 45,805-816. Plageman, P. G . W., and Richey, D. P. (1974). Transport of nucleosides, nucleic acid bases, choline and glucose by animal cells in culture. Biochim. Biophys. Acta 344, 263-305. Plapp, B., and Cole, R. (1966). Purification and characterization of bovine liver @-glucuronidase. Arch. Biochem. Biophys. 116, 193-206. Plapp, B., and Cole, R. (1967). Demonstration and partial characterization of multiple forms of bovine liver @-glucuronidase.Biochemistry 6,3676-3681. Pless, D. D., and Lennarz, W. J. (1975). A lipid-linked oligosaccharide intermediate in glycoprotein synthesis: Characterization of [Man-’*C] glycoproteins labeled from [Man-I4C]oligosaccharide-lipid and GDP-[I4C]Man.]. B i d . Chem. 250,70147019. Podolsky, D. K., and Weiser, M. W. (1975). Role of cell membrane galactosyltransferase in concanavalin A agglutination of erythrocytes. Biochem. J . 146,213-221. Podolsky, D. K., Weiser, M. W., LaMont, J. T., and Isselbacher, K. J. (1974).Galactosyltransferase and concanavalin A agglutination of cells. Proc. Natl. Acad. Sci. U.S.A. 71,904-908. Pommier, G., Ripert, G., Azoulay, E., and Depieds, R. (1975). Effects of concanavalin A on membrane-bound enzymes from mouse lymphocytes. Biochim. Biophys. Acta 389,483494. Porter, C. W., and Bernacki, R. J. (1975). Ultrastruchiral evidence for ectoglycosyltransferase systems. Nature (London)256,648-650. Potter, L. T. (1970).Synthesis, storage and release of [‘4C]acetylcholine in isolated rat diaphragm muscles. J . PhysioZ. (London)206, 145-166. Powell, J. T., and Brew, K. (1974). The preparation and characterization of two forms of bovine gala.ctosy1 transferase. Eur. J . Biochem. 48,217-228. Powell, J. T., Bon, S., Rieger, F., and Massoulie, J. (1973). Electrophorous acetylcholinesterase-A glycoprotein: Molecular weight of its subunits.. FEBS Lett. 36, 17-22. Pricer, W. E., Jr., and Ashwell, G . (1971). The binding of desialylated glycoproteins to plasma membranes of rat liver. J. B i d . Chem. 246,48254833. Quaroni, A., Gershon-Quaroni, E., and Semenza, G. (1975).Tryptic digestion of native small intestinal sucrase-isomaltase complex: Isolation of sucrase subunit. Eur. J . Biochem. 52,481-486. Ramaswamy, K., Malathi, P., Caspary, W., and Crane, R. (1974). Studies on the transport of glucose from disaccharides by hamster small intestine in uitro. 11. Characteristics of the disaccharidase-related transport system. Biochim. Biophys. Acta 345,39-48. Ramaswamy, S . , and Radhakrishnan, A. (1975a). Lactase and phlorizin hydrolase from rabbit small intestine. Ind. J . Biochem. Biophys. 12, 199-203. Ramaswamy, S., and Radhakrishnan, A. (197511) Lactase-phlorizin hydrolase complex from monkey small intestine: Purification, properties and evidence for two catalytic sites. Biochim. Biophys. Acta 403,446455. Rambourg, A., Neutra, M., and Leblond, G. (1966). Presence of a “cell coat” rich i n carbohydrate at the surface of cells in the rat. Anat. Rec. 154,41-71.

226

J. R. RIORDAN AND 0 . 0 . FORSTNER

Rapin, A. M. C., and Burger, M. M. (1974). Tumour cell surfaces: General alterations detected by agglutinins. Cancer Res. 20, 1-91. Razzel, W. E. (1963).Phosphodiesterases. I n “Preparation and Assay of Enzymes” (S. P. Colowick and N. 0. Kaplan, eds.), Methods in Enzymology, Vol. 6, pp. 236-258. Academic Press, New York. Rhee, H. M., and Hokin, L. E. (1975). Inhibition of the purified sodium-potassium activated adenosinetriphosphatase from the rectal gland of Squalus acathias by antibody against the glycoprotein subunit. Biochem. Biophys. Res. Commun. 63,11391145. Rickert, W. S., and McBride-Warren, P. A. (1976). Acid proteases from species of Mucor. 111. Interaction with concanavalin A and concanavalin A Sepharose. Can. J . Biochem. 54,120-129. Rieger, F., Bon, S., Massoulie, J., and Cartaud, J. (1973). Observation par microscopie electronique des formes allongees et globulairs d e l’acetylcholinesterase d e gymnote (Electrophorus electricus).Eur. ]. Biochem. 34,539-547. Riemer, B. L., and Widnell, C. C. (1975). The demonstration of a specific 5’-nucleotidase activity in rat tissues. Arch. Biochem. Biophys. 171,343-347. Riordan, J. R., and Slavik, M. (1974a). Interaction of concanavalin A with liver plasma membranes. Proc. Con. Fed. Biol. SOC.17, 113. Riordan, J. R., and Slavik, M. (1974b). Interactions of lectins with membrane glycoproteins: Effects of concanavalin A on 5’-nucleotidase. Biochim. Biophys. Acta 373, 356-360. Riordan, J. R., and Slavik, M. (1975).Modification by lectins of the ectoenzyme 5’-nucleotidase before and after its purification from plasma membranes. Proc. Can. Fed. Biol. SOC. 18, 18. Riordan, J. R., and Slavik, M. (1977). Mechanism of activation of plasma membrane Mg2+-ATPase by plant lectins. J . Supramol. Struct. Suppl. 1, 9. Riordan, J. R., Mitchell, L., and Slavik, M. (1974). The binding of asialoglycoprotein to isolated Golgi apparatus. Biochem. Biophys. Res. Commun. 59, 1373-1379. Riordan, J. R., Slavik, M., and Kartner, N. (1977). Nature of the lectin-induced activation of plasma membrane Mg2+ ATPase. J. Biol. Chem. 252,5449-5455. Robinson, D., and Stirling, J. (1968).N-Acetyl-P-glucosaminidases in human spleen. Biochem. J . 107,321-327. Roseman, S. (1970). The synthesis of complex carbohydrates by multiglycosyltransferase systems and their potential function in intercellular adhesion. Chem. Phys. Lipids 5,270-297. Rosenberg, L., and Segal, S. (1964).Maleic acid-induced inhibition of amino acid transport in rat kidney. Biochem. ]. 92,245-352. Rosenbemy, T. L. (1976).Acetylcholinesterase. In “The Enzymes of Biological Membranes” (A. Martonosi, ed.), pp. 331-363. Plenum, New York. Rosenberry, T. L. (1975). Acetylcholinesterase. Adv. Enzymol. Relat. Areas Mol. Biol. 43,103-218. Rosenberry, T. L., Chen, Y. T., and Bock, E. (1974). Structure of 11s acetylcholinesterase: Subunit composition. Biochemistry 13,3068-3079. Roth, S . , McGuire, E. J., and Roseman, S. (1971a). An assay for intercellular adhesive specificity. J. Cell Biol. 51, 525-535. Roth, S., McGuire, E. J., and Roseman, S. (1971b). Evidence for cell-surface glycosyltransferase: Their potential role in cellular recognition. J . Cell Biol. 51,536-547. 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.

GLYCOPROTEIN MEMBRANE ENZYMES

227

Rothman, J. E. and Lodish, H. F. (1977). Synchronized transmembrane insertion and glycosylation of a nascent membrane protein. Nature 269,775-780. Rubino, A., Zimbalatti, F., and Auricchio, S. (1964).Intestinal disaccharidase activities in adult and suckling rats. Biochim. Biophys. Acta 92,305-311. Ruoho, A., and Kyte, J. (1974). Photo affinity labelling of the ouabain-binding site on (Na+ + K+) adenosinetriphosphatase. Proc. Natl. Acad. Sci. U S A . 71,2352-2356. Saini, P. K., and Done, J. (1972). The diversity of alkaline phosphatase from rat intestine: Isolation and purification of the enzyme(s). Biochim. Biophys. Acta 258, 147-153. Sandhorn, E. B., and Makita, T. (1969). The effect of dimethyl sulfoxide on aldehyde fixation and the localization of alkaline phosphatase activity in tissues.]. Cell Biol. 43, 121a-122a. (Abstr.) Sasajima, K., Kawachi, T., Sato, S., and Sugimara, T. (1975).Purification and properties of a,a-trehalase from the mucosa of rat small intestine. Biochim. Biophys. Acta 403, 139-146. Sasaki, M., and Fishman, W. H. (1973). Ultrastructural studies on Regan and non-Regan isoenzymes of alkaline phosphatase in human ovarian cancer cells. Cancer Res. 33, 3008-3018. Schachter, H., and Roden, L. (1973). The biosynthesis of animal glycoproteins. In “Metabolic Conjugation and Metabolic Hydrolysis” (W. H. Fishman, ed.), Vol. 3, pp. 1149. Academic Press, New York. Schachter, H., Jabbal, I., Hudgin, R., Pinderic, L., McGuire, E., and Roseman, S. (1970). Intracellular localization of liver sugar nucleotide glycoprotein glycosyltransferases in a Golgi-rich fraction.]. Biol. Chem. 245, 1090-1100. Schlegel-Haueter, S., Hore, P., Kerry, K. R., and Semenza, G. (1972).T h e preparation of lactase and glucoamylase of rat small intestine. Biochim. Biophys. Acta 258, 506-519. Schmitz, J., Preiser, H., Mestracci, D., Crane, R. K., Troesch, V., and Hadorn, B. (1974). Subcellular localization of enterokinase in human small intestine. Biochim. Biophys. Acta 343,435-439. Seetharam, B., Swaminathan, N., and Radhakrishnan, N. (1970). Studies on mammalian glucoamylases with special reference to monkey intestinal glucoamylase. Biochem. J . 117, 939-946. Seetharam, B., Brimme, N., Goodwin, C., and Alpers, D. (1976). Differential sensitivity of intestinal brush border enzymes to pancreatic and lysosomal proteases. Life Sci. 18,89-96. Sela, B., Lis, H., and Sachs, L. (1972). Enzymatic hydrolysis of uridine diphosphate-Nby noracetyl-D-galactosamine and uridine diphosphate-N-acetyl-D-glucosamine mal cells and blocks in this hydrolysis in transformed cells and their revertants.]. Biol. Chem. 247,7585-7590. Semenza, G . (1976).Small intestinal disaccharidases: Their properties and role as sugar translocators across natural and artificial membranes. In “The Enzymes of Biological Membranes, Vol. 3: Membrane Transport” (A. Martonosi, ed.), pp. 349-376. Plenum, New York. Shamoo, A. E. (1974). Isolation of a sodium-dependent ionophore from (Na+ + K+)ATPase preparations. Ann. N.Y. Acad. Sci. 242,389-405. Sigrist, H., Ronner, P., and Semenza, G. (1975).A hydrophobic form ofthe small intestinal sucrase-isomaltase complex. Biochim. Biophys. Acta 406,433-446. Singer, R., and Fishman, W. (1974).Characterization of two HELA sublines: TCRC-1 produced Regan isoenzyme and TCRC-2 non-Regan isoenzyme. J . Cell Biol. 60, 777-780.

228

J. R. RIORDAN AND G. G. FORSTNER

Singer, S. J. (1974a). The molecular organization of membranes. Annu. Reu. Biochem. 43,805-833. Singer, S. J. (1974b). The molecular organization of biological membranes. In “Structure and Function of Biological Membranes” (J. J. Rothfield, ed.), pp. 145-222. Academic Press, New York. Singer, S. J. (1974~).Molecular biology of cellular membranes with applications to immunology. Adu. Immunol. 19, 1-66. Sivakami, S., and Radhakrishnan, A. (1976). Kinetic studies on glucoamylase of rabbit small intestine. Biochem. J. 153,321-327. Skou, J. C. (1957).The influence of some cations on an adenosine triphosphatase from peripheral nerves. Biochim. Biophys. Actu 23,394-401. Skou, J. C. (1962).Preparation from mammalian brain and kidney ofthe enzyme system involved in active transport of Na+ and Kt. Biochim. Biophys. Acta 58,314-325. Skou, J. C. (1973). The relationship of the (Na+ + K+)-activated enzyme system to transport of sodium and potassium across the cell membrane. Bioenergetics 4, 1-30. Slavik, M., Kartner, N., and Riordan, J. R. (1977). Lectin-induced inhibition of plasma membrane 5’-nucleotidase: Sensitivity of purified enzyme. Biochem. Biophys. Res. Commun. 75,342-349. Smith, C. A. and Brew, K. (1977). Isolation and characteristics of galactosyltransferase from Golgi membranes of lactating sheep mammary glands. J . Biol. Chem. 252, 7294-7299. Smith, T. W., and Wagner, H., Jr. (1975). Effects of (Na+ + K+)-ATPase-specific antibodies on enzymatic activity and monovalent cation transport. J . Membr. Biol. 25, 341-360. Song, C. S., Tandler, B., and Bodansky, 0.(1967). Solubilization of 5’-nucleotidase from human liver and reassociation of its activity into membranous vesicles. Biochem. Med. 1,100-109. Spatz, L., and Strittmatter, P. (1973). A form of reduced nicotinamide adenine dinucleotide-cytochrome b, reductase containing both the catalytic site and an additional hydrophobic membrane-binding segment. J . Biol. Chem. 248,793-799. Spiro, R. G . , Spiro, M. J., and Bhoyroo, V. D. (1976). Lipid saccharide intermediates in glycoprotein biosynthesis. 11. Studies of the structure of oligosaccharide-lipid from thyroid. J. Biol. Chem. 251,6409-6419. Srikantaiah, M., and Radhakrishnan, A. (1970). Studies on the metabolism of vitamin B6 in the small intestine. Part 11. Heterogeneity of monkey intestinal alkaline phosphatase. Ind. J. Biochem. 7, 145-150. Srivastava, S., and Beutler, E. (1974). Studies on human P-D-N-acetylhexosaminidases. 111. Biochemical genetics of Tay-Sachs and Sandhoffs diseases.J . Biol. Chem. 249, 2054-2057. Srivastava, S . , Awasthi, Y., Yoshida, A., and Beutler, E. (1974a). Studies on human P-DN-acetylhexosaminidases. I. Purification and properties. J . Biol. Chem. 249, 2043-2048. Srivastava, S., Yoshida, A,, Awasthi, Y., and Beutler, E. (1974b). Studies on human P-DN-acetylhexosaminidases. 11. Kinetic and structural properties. J. Biol. Chem. 249, 2049-2053. Stahl, P., and Touster, A. (1971). P-Glucosidase of rat liver lysosomes: Purification, properties, subunits. J . Biol. Chem. 246,5398-5406. Starnes, W. L., and Behal, F. H. (1974). Human liver aminopeptidase: The amino acid and carbohydrate content, and some physical properties of a sialic acid containing glycoprotein. Biochemistq 13,3221-3233.

GLYCOPROTEIN MEMBRANE ENZYMES

229

Stefanovic, V., Mandel, P., and Rosenberg, A. (1975).Activation of acetyl- and butyrylcholinesterase by enzymatic removal of sialic acid from intact neuroblastoma and astroblastonia cells in culture. Biochemistry 14, 5257-5260. Stefanovic, V., Mandel, P., and Rosenberg, A. (1976).Ecto-5’-nucleotidase of intact cultured C6 rat glioma cells. J. Biol. Chem. 251,3900-3905. Stevens, R., Hartman, M., Fluharty, A., and Kihara, H. (1973).A second form of arylsulfatase A in human urine. Biochim. Biophys. Acta 302,338-344. Stevens, R., Fluharty, A., Killgrove, A., and Kihara, H. (1976). Mucoheterogeneity of arylsulfatase A from human tissues. Biochim. Biophys. Acta 445,661-671. Stevenson, F. K. (1973).Studies on the cell surface maltase of the rabbit renal cortex. Biochim. Biophys. Acta 311,409-416. Storelli, C., Vogeli, H., and Semenza, G . (1972). Reconstitution of a sucrase-mediated sugar transport system in lipid membranes. FEBS Lett. 24,287-292. Strittmatter, P. (1961).The binding of pyridine nucleotides to cytochrome b, aporeductase. J . Biol. Chem. 236,2336-2341. Strittmatter, P., Rogers, M. J., and Spatz, L. (1972).The binding of cytochrome b, to liver microsomes. J . Biol. Chem. 247,7188-7194. Sussman, H. H., and Guttlieb, A. J. (1969).Human placental alkaline phosphatase. 11. Molecular and subunit properties of the enzyme. Biochim. Biophys. Acta 194,170179. Suzuki, K. (1968). Cerebral GM,-gangliosidosis: Chemical pathology of visceral organs. Science 159, 1471-1472. Swann, A. C., Daniel, A., Albers, R. W., and Koval, G. J. (1975).Interactions of lectins with (Na+ + K+)-ATPase of eel electric organ. Biochin. Biophys. Acta 401, 299306. Swift, J. G., and Mukherjee, T. M. (1976).Demonstration of t h e fuzzy surface coat of rat intestinal microvilli by freeze-etching.J . Cell Biol. 69,491-494. Szewczuk, A., and Baranowski, T. (1963). Purification and properties of y-glutamyl transpeptidase from beef kidney. Biochem. 2. 338,317-329. Szewczuk, A., and Connell, G. (1964).The effect of neuraminidase on the properties of gamma-glutamyl transpeptidase. Biochim. Biophys. Acta 83,218-223. Tallman, J., Johnson, W., and Brady, R. (1972).The metabolism of Tay-Sachs ganglioside: Catabolic studies with lysosomal enzymes from normal and Tay-Sachs brain tissue.]. Clin. Inuest. 51,2339-2345. Tanaka, H., and Suzuki, K. (1977).Substrate specificities of the two genetically distinct human brain P-galactosidases. Bruin Res. 122,325-335. Tate, S., and Meister, A. (1975).Identity of maleate-stimulated glutaminase with glutaniyltranspeptidase in rat kidney. J . Biol. Chem. 250,4619-4627. Tate, S., and Meister, A. (1977).Affinity labeling of y-glutamyltranspeptidase and location of the y-glutamyl binding site on the light subunit. Proc. Natl. Acad. Sci. U.S.A. 74,931-935. Taylor, P., Jones, J. W., and Jacobs, N. M. (1974).Acetylcholinesterase from Torpedo: Characterization of an enzyme species isolated b y lytic procedures. MoZ. PharmaC O ~ 10,78-92. . Thomas, L., and Kinne, R. (1972). Studies on the arrangement of aminopeptidase and alkaline phosphatase in the microvilli of isolated brush border of rat kidney. Biochim. Biophys. Acta 255, 114-125. Tomita, M., and Marchesi, V. T. (1975). Amino-acid sequence and oligosaccharide attachment sites of human erythrocyte glycophorin. Proc. Natl. Acad. Sci. U S A . 72, 2964-2968. Toule, D. W., and Copenhaver, J. H., Jr. (1970). Partial purification of a soluble

230

J. R. RIORDAN AND G. G. FORSTNER

+

(NaC K+)-dependent ATPase from rabbit kidney. Biochim. Biophys. Acta 203, 124-132. Toyoshima, S., and Osawa, T. (1975). Lectins from Wisteriajuoribunda seeds and their effect on membrane fluidity of human peripheral 1ymphocytes.J. Biol. Chem. 250, 1655- 1660. Trams, E.G., and Lauter, C. J. (1974). On the sidedness of plasma membrane enzymes. Biochim. Biophys. Acta 345,180-197. Trams, E. G., Lauter, C. J., and Banfield, W. G. (1976).On the activation of plasma membrane ecto-enzymes by treatment with neuraminidase. J . Neurochem. 27, 10351042. Trayer, I. P., and Hill, R. L. (1971). The purification and properties of the protein of lactose synthetase. J . Biol. Chem. 246,6666-6675. Ugolev, A. M. (1972). “Peptide Transport in Bacteria and Mammalian Gut,” pp. 123137. Elsevier, Amsterdam. Vannier, C., Louvard, D., Maroux, S., and Desneulle, P. (1976).Structure and topological homdogy between porcine intestinal and renal brush border aminopeptidases. Biochim. Biophys. Acta 455, 185-199. Van Winkle, W. B., Lane, L. K., and Schwartz, A. (1976). The subunit fine structure of isolated, purified Na+,K+-adenosine triphosphatase. Exp. Cell Res. 100,291-296. Verbert, A., Cacan, R., and Montrevil, J. (1976). Ectogalactosyl transferase: Presence of enzyme and acceptors on the rat lymphocyte cell surface. Eur. 1.Biochem. 70,4953. Wacker, H. (1974). The role of carbohydrate moieties in the activity and properties of aminopeptidase from pig kidney. Biochim. Biophys. Acta 334,417-422. Wacker, H., Lehky, P., Fischer, E., and Stein, E. (1971). Physical and chemical characterization of pig kidney particulate aminopeptidase. Helu. Chim.Acta 54,473-485. Wang, K., and Richards, F. M. (1974). An approach to nearest neighbor analysis of membrane proteins: Application to the human erythrocyte membrane of a method employing cleavable cross-linkages. J. Biol. Chem. 249,8005-8018. Warnes, T., Hine, P., and Kay, G. (1969). Alkaline phosphatase in duodenal juice following secretion and pancreozymin. Gut 10, 1049. (Abstr.) Whitehead, J. S., Bella, A., Jr., and Kim, Y. S. (1974). An N-acetylgalactosaminyltransferase from human blood group A plasma. I. Purification and agarose binding properties. ]. Biol. Chem. 249,3442-3447. Widnell, C. C. (1972). Cytochemical localization of 5’-nucleotidase in subcellular fractions isolated from rat liver. I. The origin of 5’-nucleotidase activity in microsomes. 1.Cell Biol. 52,542-558. Widnell, C. C., and Unkeless, J. C. (1968). Partial purification of a lipoprotein with 5‘nucleotidase activity from membranes of rat liver cells. Proc. Natl. Acad. Sci. U.S.A. 6ly105O-1057. Wiedmar, T., Gentinetta, R., and Brodbeck, U. (1974). Binding of acetylcholinesterases to concanavalin A. FEBS Lett. 47,260-262. Wirth, D. F., Katz, F., Lodish, H. F., and Robbins, P. B. (1977). How a single Sindbis virus in RNA directs the synthesis of one soluble protein and two integral membrane glycoproteins.]. Supramol. Struct. Suppl. 1, 21. Woo, Y.T., and Manery, J. F. (1975). 5’-Nucleotidase: An ecto-enzyme of frog skeletal muscle. Biochim. Biophys. Acta 397, 144-152. Woodward, M., Dodgson, K., Hook, G., and Rose, F. (1973). Problems associated with the assay of arylsulphatases A and B of rat tissues. Biochem. J . 134, 183-190. Yang, C., and Srivastava, P. (1975). Purification of bull sperm hyaluronidase by concanavalin A affinity chromatography. Biochim. Biophys. Acta 391,382-387.

GLYCOPROTEIN MEMBRANE ENZYMES

23 1

Yeh, K., and Moog, F. (1975). Development of the small intestine in the hypophysectomized rat. Deu. Biol. 47,173-184. Young, M., Riordan, J. R., and Moscarello, M. A. (1976a). Sensitivity of Golgi galactosyl transferase to concanavalin A. Proc. Int. Congr. Biochem., 10th p. 229. (Abstr.) Young, M. E.M., Moscarello, M. A., and Riordan, J. R. (1976b). Concanavalin A binding to membranes of the Colgi apparatus and resultant modification of galactosyltransferase activity. j . Biol. Chem. 251,5860-5865. Zachowski, A. and Pard, A. (1974). Use of concanavlin. A polymer to isolate right side-out vesicles of purified plasma membrane from eukaryotic cells. Biochem. Biophys. Res. Commun. 57,787-792.

This Page Intentionally Left Blank

CURRENT TOPICS IN MEMBRANES AND TRANSPORT, VOLUME

11

Membrane Glycoproteins of Enveloped Viruses RICHARD W. COMPANS AND MAURICE C . KEMP Department of Microbiology University of Alabama in Birmingham The Medical Center Birmingham, Alabama

. . . . . . . . . . A. Glycoproteins . . . . . . . . . . . B. Lipid Bilayer . . . . . . . . . . . C. Carbohydrate-Free Proteins . . . . . . D. Carbohydrates . . . . . . . . . . Arrangement of Viral Envelope Components . . Structure and Function of Viral Glycoproteins . . A. Fine Structure of Envelope Glycoproteins . B. Functions of Viral Proteins . . . . . .

I. Membranes of Lipid-Containing Viruses 11. Components of Viral Membranes

111. IV.

V.

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

. . . . . . . . . . , , . . . . . . . . . . . . . . . . C. Function of Carbohydrates in Viral Glycoproteins . , . D. Effects of Viral Proteins on Lipid Bilayer Structure . . . Assembly of Viral Membranes . . . . . . . . . . . A. Cellular Sites of Maturation . . . . . . . . . .

. . . . . . , , . . . . . . . . . . . . . . . . . . . . . . . . . B. Synthesis and Intracellular Localization of Envelope Proteins. . . C. Sequence of Events in Viral Assembly . . . . , , . . . . D. Macromolecular Interactions in Membrane Assembly . . . . . References . . . . . . . . . . . . . . . . . . . . 1.

233 236 236 238 239 240 240 242 242 250 255 258 260 260 26 1 263 266 268

MEMBRANES OF LIPID-CONTAINING VIRUSES

Lipid-containing animal viruses, including members of at least 10 different major groups, possess components in their envelopes which are similar to those of other biological membranes: lipids, proteins, and carbohydrates linked to specific subsets of lipids or proteins. Investigation of the fine structure, arrangement, and assembly of these viral membrane components is a field that has recently attracted increasing attention, both because of intrinsic interest in these viruses, many of which are of great public health importance, and because viral envelopes have several unique advantages in membrane re233 Copyright @ 1978 by Academic Press, Inc. All rights 01 reproduction in any tarm reserved. ISBN n - i z - 1 m i 1 - 5

234

R. W. COMPANS AND M. C. KEMP

search. The most important advantage is structural simplicity, in that lipid-containing viruses possess a small number of virus-coded polypeptides and glycoproteins as membrane components. Sufficient quantities of many enveloped viruses can be purified for detailed biochemical analysis, including primary structure studies of viral membrane proteins like those now underway. Purification procedures are rapid and simple, and a fairly homogeneous population of particles can be obtained free of contaminating cellular membranes. For the simpler lipid-containing viruses, the possibility therefore exists for complete structural determination of a biological membrane, and such information should contribute greatly to an understanding of the molecular details of membrane structure and interactions among membrane components in general. Other advantages of enveloped viruses in studies of membrane structure and biogenesis include the ease of biosynthetic labeling of viruses grown in cell culture with specific radioactive precursors and the availability of mutants in defined gene products, some of which are proving to be useful in the analysis of viral membrane assembly. Since envelope components are integral parts of cellular membranes during the assembly of virus particles (Compans et al., 1966; Compans and Choppin, 1971),studies of the synthesis and mechanism of insertion of these components into membranes should provide useful information pertaining to the biogenesis of cellular membranes. Many viral systems offer the additional advantage that host cell biosynthesis is inhibited during infection, so that it is possible to analyze the synthesis of membrane components in a cell in which only a small number of virus-coded membrane components are being produced and to follow the intracellular migration of specific membrane glycoproteins by radiolabeling procedures. Finally, the ability to prepare viruses with specific modifications in either lipid or protein composition arises from the fact that viral envelope lipids are derived from the plasma membrane of the host cell, whereas viral proteins are entirely virus-coded. Thus by growing different virus types in the same cell it is possible to prepare a membrane in which the protein composition can b e varied while the lipid composition remains constant. Similarly, by growing the same virus in a series of different cells, viral membranes are obtained which possess the same set of proteins but vary in lipid composition. These approaches have been useful in studies of lipid-protein interactions in viral membranes (Landsberger et al., 1973). Several recent reviews are available concerning both the experimental and theoretical aspects of viral envelope structure and assem-

236

R. W. COMPANS A N D M. C. KEMP

bly (Lenard and Compans, 1974; Blough and Tiffany, 1975; Wagner, 1975; Choppin and Compans, 1975; Compans and Choppin, 1975; Klenk, 1974).In this chapter, we emphasize recent information on the glycoprotein components of enveloped viruses and endeavor to point out specific findings on viral envelopes which we believe to be of broad significance. In Table I are listed the 10 established major groups, or families, of lipid-containing viruses of vertebrates, and examples of the best studied members of certain groups. For several families, only limited information is available on viral envelope structure, whereas other groups have been the object of intensive study. We concentrate our discussion on the glycoproteins of the most intensively studied virus groups, which are the first four groups of RNA viruses listed in Table I: myxoviruses, paramyxoviruses, rhabdoviruses, and togaviruses. We also discuss selected findings for other virus groups where these have revealed unusual or important aspects of viral membrane structure or assembly.

II. COMPONENTS

OF VIRAL MEMBRANES

Although enveloped viruses of different major groups vary in size and shape, as well as in the MWs of their structural polypeptides, there are general similarities in the types of polypeptide components present in virions. The types of structural components found in viral membranes are summarized briefly in the following discussion, and further detailed information is available in the recent reviews cited above. A. Glycoproteins

All the enveloped viruses studied to date possess one or more glycoprotein species. The presence of carbohydrate covalently linked to proteins has usually been demonstrated by the incorporation of a radioactive precursor such as glucosamine or fucose into viral polypeptides, which is resolved by sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis. The specificity of labeling by these precursors is shown by the lack of their incorporation into carbohydrate-free polypeptides of the virion (Straws et al., 1970). The number of distinct glycoprotein species present in the virion varies for various virus groups. As shown in Table 11, there is also marked variation in the glycoprotein content of virions when expressed as a percent of

237

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

TABLE I1 PROTEIN COMPOSITION OF

LIPID-CONTAININGVIRUSES

Percent of total protein Carbohydrate-free proteins

Virus type

Number of glycoproteins

Glycoproteins

M Protein

Others

Toga Bunya

3 2

76 64

24 36

Oncorna, B type Coronaa

3

58.4

41.6

4

58

-

42

Paramyxo

2

39.5

14.0

46.5

Rhabdo

1

34.2

28.8

37

Myxo

2

33.4

36

30.6

1-2 13 1-2

27.5 19.7 7

Arenab HerpesC Oncorna, C typed

72.5 80.3 93

Source of data Garoff et al. (1974) Obijeski et al. (1976) Yagi and Compans (1977) Hierholzer et al. (1972) Lamb and Mahy (1975) Bishop and Roy (1972) Compans and Choppin (1975) Vezza et al. (1977) Heine et al. (1974) Famulari et al. (1976)

Classification of glycoproteins was stated to be only tentative. A single glycoprotein was detected in the arenaviruses Tacaribe and Tamiami (Gard et al., 1977), whereas the data given are for Pichinde virions which contain two glycoproteins. The values represent our calculations from the autoradiograph scan data presented in the reference. The values given represent our calculations from the data presented in the reference. a

total viral protein. As discussed in Section V, these data suggest that glycoproteins of the various virus groups may play a greater or lesser role in assembly and maintenance of the viral envelope structure. However, glycoproteins are essential components for viral infectivity even where they constitute only a minor fraction of the mass of the virion, since they are necessary for the initial events in the viral replication cycle. The fine structure, biological functions, and orgapization

238

R. W. COMPANS AND M. C. KEMP

of glycoproteins in the viral envelope are discussed in detail in Section IV. The carbohydrates of viral glycoproteins are specified in large part by host cell transferases, whereas the amino acid sequences are coded b y the viral genome. Glycoproteins of the same virus may exhibit host cell-dependent differences in electrophoretic mobility, which are due to differences in the carbohydrate components (Haslam et al., 1970; Compans et al., 1970; Schulze, 1970). Thus viral glycoproteins may be useful probes to detect differences in glycosylation among various cell types. An important exception to the general finding of host cell-specified carbohydrates is found in myxoviruses and paramyxoviruses. Glycoproteins of these viruses lack sialic acid (Klenk et al., 1970a,b), which is believed to be a result of the virus-specified neuraminidase incorporated as a structural component of these virions. 8. lipid Bilayer

All enveloped viruses contain lipid as a major structural component, and available evidence indicates that all the lipid is contained in a single bilayer structure which forms the matrix of the limiting membrane of the virion. Investigations using biophysical methods including xray diffraction (Harrison et al., 1971) and electron spin resonance (Landsberger et al., 1971, 1973; Landsberger and Compans, 1976) indicate that viral lipids are arranged in a bilayer. With electron microscopy, a well-defined unit membrane is observed at the location of the bilayer. The virus derives its lipids from the host cell membrane where virus maturation occurs, which is the plasma membrane in most instances. Therefore a given virus can exhibit marked variation in lipid composition when grown in different cell types, which reflects the composition of the host cell plasma membrane in each instance (Klenk and Choppin, 1969, 1970; Quigley et al., 1971). Apart from minor differences in carbohydrates of glycoproteins, virion proteins are indistinguishable when the virus is propagated in a variety of cells; therefore there appears to be little or no determining influence of viral proteins on the composition of the lipid bilayer. Since the viral envelope is continuous with the host cell membrane during morphogenesis, and membrane lipids are characterized by a high rate of lateral diffusion in the plane of the membrane, the similarity in lipid composition between viral envelopes and plasma membranes is not surprising. The distribution of various lipid classes on the internal and external

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

239

sides of the bilayer has recently been investigated in influenza virions using phospholipase digestion and phospholipid exchange proteins (Tsai and Lenard, 1975; Rothman et al., 1976). The results indicate that lipids are distributed asymmetrically, with the majority of the phosphatidylinositol and about half of the phosphatidylcholine in the external half of the bilayer and most of the phosphatidylethanolamine and phosphatidylserine in the inner half. These distributions are thought to reflect a similar distribution in the host cell plasma membrane. Since less than half of the total phospholipids were accessible b y either approach, it was suggested that glycolipids constitute a significant part of the outer monolayer of the viral membrane. The presence of glycolipids in various enveloped viruses has been demonstrated by chemical analyses (Klenk and Choppin, 1970), and the types of glycolipids present also reflect those of the host cell. Myxoand paramyxoviruses lack neuraminic acid residues in their glycolipids, which again is a likely result of their neuraminidase activity (Klenk and Choppin, 1970; Klenk et al., 1970b). The reactivity of viral glycolipids toward specific lectins demonstrates that their carbohydrates are exposed on the external surface of the bilayer (Klenk et al., 1972a). C. Carbohydrate-Free Proteins

All polypeptides located internal to the viral lipid bilayer are devoid of carbohydrate, and they may be subdivided into several types according to their structural location and functions. All enveloped viruses possess at least one polypeptide species closely associated with the nucleic acid to form a nucleocapsid or core structure, which is termed the nucleocapsid protein. In addition, viruses with helical nucleocapsids possess another major internal protein, termed the membrane protein, that appears to be associated with the internal surface of the lipid bilayer. Most nucleocapsid proteins are not directly involved in interaction with the lipid bilayer, although the nucleocapsids of togaviruses are roughly spherical structures that appear to be closely apposed to the inner surface of the bilayer. The internal membrane (M) proteins of myxo-, paramyxo-, and rhabdoviruses occupy a similar location in the viral envelope. It appears that these proteins are primarily responsible for conferring considerable rigidity on the viral envelope, as compared to that of host cell membranes of similar lipid composition (Stoeffel and Bister, 1975; Landsberger and Compans, 1976; Lenard et al., 1976). More complex viruses possess additional internal polypeptides as

240

R. W. COMPANS AND M. C. KEMP

major components, but their precise locations in the virion have not been established. In addition, minor internal protein components are frequently observed in virions which possess enzymic functions, such as transcriptase activity. Such minor polypeptides are not thought to play a direct role in the structure or assembly of the viral envelope. D. Carbohydrates

It is well established that carbohydrate chains are covalently linked to glycoproteins and glycolipids of the viral envelope, and both of these components were described above. However, it is uncertain whether or not other complex carbohydrates serve a function in viral envelope structure. The presence of sulfated mucopolysaccharides in highly purified virus preparations of several major groups was recently observed by radiolabeling virions with sulfate (Compans and Pinter, 1975; Pinter and Compans, 1975). These sulfated components are derived from the host cell, since they can be labeled selectively by the growth of cells in 35S0,-containing medium prior to virus infection. In the case of influenza virus, it is possible that these components represent a host cell antigen described to be associated with purified virus particles (Knight, 1944). Labeled sulfated mucopolysaccharides can be removed from virus preparations by digestion with hyaluronidase or trypsin without any effect on viral infectivity, suggesting that these components are not essential (Pinter and Compans, 1975). It is likely that they associate with the external surface of the virion during the process of maturation.

111.

ARRANGEMENT OF VIRAL ENVELOPE COMPONENTS

Enveloped viruses share many common features in the organization of their structural components, as indicated by several approaches which include electron microscopy, surface-labeling and proteolytic digestion experiments, and isolation of subviral components. Much of the evidence pertaining to the arrangement of membrane components has been discussed in other reviews cited above and is only briefly summarized in this section. The organization of the viral membrane corresponds in many respects to the fluid mosaic membrane model (Singer and Nicolson, 1972). Viral glycoproteins appear to be integral membrane proteins which are exposed on the external surface of a lipid bilayer. The external location of the glycoproteins is indicated by their sensitivity to

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

24 1

proteolytic digestion and reactivity toward surface-labeling reagents (see review in Lenard and Compans, 1974). Partial penetration of the glycoproteins into the bilayer is suggested by the fact that segments of the glycoproteins of several virus groups are found to be resistant to protease (Gahmberg et al., 1972; Mudd, 1974, Schloemer and Wagner, 1975b; Lenard et al., 1976). In contrast to the glycoproteins, the carbohydrate-free polypeptides of enveloped viruses are located internally to the lipid bilayer, as indicated by their resistance to protease treatment and lack of reactivity toward surface-labeling reagents. These internal proteins are also not reactive toward specific antibodies without disruption of the viral envelope. The best examples of nonglycosylated proteins which appear to be intimately associated with the internal surface of the envelope are the M proteins of myxo-, paramyxo-, and rhabdoviruses. However, whether these polypeptides associate with the bilayer as integral membrane proteins, or as peripheral membrane proteins such as spectrin in erythrocyte membranes, has not been clearly established. In Fig. 1 the schematic cross section of an influenza virion shows the general arrangement of its structural components. The extent of penetration of the glycoproteins into the bilayer is not indicated but,

FIG.1. Schematic cross section of an influenza virion, depicting the arrangement of the major structural polypeptides. The rodlike HA spikes and the more complex NA spikes are composed of HA and NA polypeptides, respectively. These glycoproteins are exposed at the external surface of the lipid bilayer, and segments may penetrate the bilayer (not shown). The internal membrane-associated M protein appears to form a closely packed layer beneath the bilayer. T h e helical nucleocapsids are found as multiple discrete segments, each containing an RNA molecule coated with a major nucleocapsid protein, (NP). The nucleocapsids probably interact directly with the M protein during virion assembly. (From Compans and Choppin, 1975.)

242

R. W. COMPANS AND M. C. KEMP

as discussed in Section IV, it is likely that a hydrophobic segment penetrates into the bilayer and may even traverse it. A closely packed layer of M protein is depicted on the internal surface of the bilayer in accordance with available information, and the helical nucleocapsids are depicted as twisted hairpinlike structures (Compans et al., 1972). It is likely that the nucleocapsids interact with the M proteins during assembly and in so doing participate in determining virion size and shape. Little is known about these interactions, and they are not indicated in Fig. 1.

IV.

STRUCTURE AND FUNCTION OF VIRAL GLYCOPROTEINS

A. Fine Structure of Envelope Glycoproteins

In this section we summarize the available information on the detailed structure of the glycoproteins of four virus groups. The infonnation obtained includes the size and shape of viral glycoproteins, the number of polypeptide chains in the complete glycoprotein structure, and compositional data on the polypeptide and oligosaccharide portions of the molecules. Detailed structural information on glycoproteins has been obtained for only a few virus types and is obviously incomplete even for the best studied viral systems.

1. INFLUENZA VIRUS GLYCOPROTEINS Influenza viruses possess two glycoproteins with distinct biochemical and morphological properties: hemaggluti-nin (HA) and neuraminidase (NA). The HA of influenza virus iii one of the best studied viral glycoproteins. Electron microscope studies of the isolated spike structure have shown that it is a rod-shaped triangular prism approximately 14 nm long and 5 nm wide (Laver and Valentine, 1969).The HA spike projects radially from the virion envelope, and it is probably a trimer having a MW of about 225,000 daltons (Schulze, 1975; Wiley et al., 1977). HA glycoproteins can exist in one of two alternative forms: a single polypeptide with a MW of 75,000 daltons designated HA, or two polypeptides with MWs of approximately 50,000 and 25,000 daltons designated HA1 and HA2, respectively. HA, and HA, are proteolytic cleavage products of HA, which remain cross-linked by disulfide bonds (Lazarowitz et al., 1971; Laver, 1971).The extent of cleavage of HA to HA, and HA2 may vary from none to complete cleavage of all HA polypeptides (Lazarowitz et al., 1973).The extent of cleavage de-

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

243

pends upon several factors including the host cell, the virus strain, and the presence or absence of proteases in the culture medium. The cleavage of HA may not be a requirement for infectivity of all virus strains, and spike morphology is not altered detectably by cleavage of HA to HA, and HA,, but recent studies have shown that cleavage enhances the infectivity of some influenza virus strains (Klenk et al., 1975; Lazarowitz and Choppin, 1975). The tryptic peptide maps of HA, and HA, are distinct, and the only notable difference in the amino acid composition of the two polypeptides is the proline content (Laver, 1971). The proline content of HA, is at least six times higher than that of HA,. HA, is tightly folded on itself, probably reflecting the high proline content. A portion of the HA spike is released by protease treatment as a soluble protein which can be crystallized (Brand and Skehel, 1972). Functional HA spikes have been isolated from several influenza strains by treating virions with ionic or nonionic detergents. After removal of the detergent, the spikes aggregate, forming rosettelike structures, suggesting that the base of the HA spike is hydrophobic in nature (Laver and Valentine, 1969).Available evidence indicates that the HA spike is amphipathic and that a portion of it is buried within the lipid bilayer of the viral envelope. This is indicated b y the finding that protease-treated virions retained HA, or portions thereof while having no identifiable spikes (Compans et nl., 1970; Lenard et al.,

1976). The HA, glycoprotein therefore contains the hydrophobic end of the spike. This glycoprotein aggregates even in the presence of guanidine hydrochloride (Laver, 1971). The composition of a 50-residue peptide associated with the lipid bilayer, which is removed from the carboxyl terminus of HA2 by bromelain, has been estimated, and 21 of the amino acids are hydrophobic in nature, whereas 14 may be charged (Skehel and Waterfield, 1975). The carboxyl terminus of this peptide is thought to be buried within the lipid bilayer. The aminoterminal end of HA, seems to extend radially from the virion surface, and a sequence of 10 residues from the amino-terminal end of HA, is highly conserved in both type-A and type-B influenza virions (Skehel and Waterfield, 1975). The hydrophobicity of HA, suggests that it is similar to other membrane-associated glycoprotein segments (Segrest et al., 1972; Nakashima et al., 1975), but freeze-fracture studies of the influenza virion indicate that extensive portions of HA and NA do not penetrate through the lipid bilayer (Bachi et al., 1969). Circular dichroism studies of HA spikes isolated intact and those cleaved from the virion surface may be useful in further defining the secondary structure of the hydrophobic regions.

244

R. W. COMPANS A N D M. C. KEMP

Until recently, little information was available concerning the carbohydrate portion of the HA glycoprotein. About 80% of the carbohydrate attached to viral proteins was shown to be linked to HA; the glycosy1 moieties were shown to be comprised of glucosamine, mannose, galactose and fucose at a molar ratio of 6 :4 : 1: 1, respectively, and the total carbohydrate of HA, determined by chemical means, was estimated to be approximately 12,000 daltons (Laver, 1971). It was also found that HA, of influenza A virions contains fucose, whereas HA, of influenza B (GL1760) does not (Choppin et al., 1975). Recently the glycopeptides obtained by pronase digestion of influenza A viruses have been characterized (Schwarz et al., 1977; Nakamura and Compans, 1978b). Both groups showed that HA possesses type I (oligosaccharide side-chain comprised of glucosamine, mannose, galactose, fucose and sialic acid) and type I1 (oligosaccharides comprised of glucosamine and mannose) glycopeptides, but the distribution of each type on HA, and HA, was shown to depend upon the virus strain. Schwarz et al. (1977) showed that HA, of two avian influenza viruses possessed only type I glycopeptides, whereas both type I and I1 glycopeptides were shown to be present on HA,. However, Nakamura and Compans (197813)using the WSN strain of influenza showed that HA, possessed both type I and I1 glycopeptides whereas HA, contained only type I glycopeptides. Since it is likely that HA, occupies a similar structural location in WSN and avian hemagglutinin proteins, these results suggest that specific amino acid sequences determine whether a type I or type I1 oligosaccharide is added at a particular site on the glycoprotein. The host cell type may also determine the type of oligosaccharide chains added to a given viral glycoprotein. Thus, although HA, of WSN strain influenza virions grown in MDBK cells contained only type I glycopeptides, both type I and I1 glycopeptides were found in HA, when grown in CEF cells (Nakamura and Compans, 1978~). Schwarz et al. (1977) showed that the type I HA glycopeptides of fowl plague and N influenza strains grown in chicken embryo fibroblasts (CEF) had a molecular weight of approximately 2,600 daltons, whereas type I1 glycopeptides from the same source has a molecular weight of approximately 2,000 daltons. The molecular weights of type I and I1 glycopeptides of the WSN strain of influenza virus were also found to have approximately this size (Nakamura and Compans, 1978b). HA glycopeptides of virus grown in MDBK cells were slightly larger in size than those of virus grown in CEF cells. Based upon the estimated carbohydrate content (12,000 daltons) of the HA glycoprotein obtained by Laver (1971)and Schwarz and Klenk (1974) and the

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

245

size estimates of the type I and I1 glycopeptides of influenza virus grown in MDBK cells, it was estimated that HA, contains a single type I glycopeptide whereas HA, possesses two type I and one or two type I1 oligosaccharide side-chains for the WSN strain (Nakamura and Compans, 1978b). The NA spike, as shown in Fig. 1, also projects radially from the lipid bilayer of the influenza virion but is morphologically distinguishable from the HA spike (Laver and Valentine, 1969). The NA spike is elongated and has a square knoblike structure at one end. The oblong head is approximately 5 x 8.5 nm, and it is attached to a fiber approximately 10 nm long. The NA spike has a MW of approximately 240,000 daltons and is comprised of four NA polypeptides each having a MW of approximately 58,000 daltons; two NA polypeptides appear to be linked by disulfide bonds to form dimers, which are thought in turn to aggregate by noncovalent bonds to form the tetrameric spike (Bucher and Kilbourne, 1972; Lazdins et al., 1972). The amino acid composition of the NA has been determined, and it has a cysteine content significantly higher than that of the other viral polypeptides (Laver and Baker, 1972). Influenza NA is in some ways morphologically similar to enzymes involved in sugar metabolism in the gut, as discussed by Forstner and Riordan in this volume. After trypsin treatment a tetramer of four coplaner subunits of 4 X 4 x 4 nm was isolated (Wrigley et al., 1973). This structure appears to correspond to the knob, and it is enzymically active. The knob structure can no longer aggregate with itself or with HA molecules, indicating that the hydrophobic regions of the molecule have been removed by proteolysis. The MW of the portion of the NA monomer which remains associated with the envelope after trypsin treatment was estimated to be 7000 daltons by Lazdins et al. (1972) and 12,000 daltons by Wrigley et al. (1973).Presumably, this portion functions to attach the NA molecule to the virion and plays no role in the enzymic properties. Furthermore, the portion of the spike which remains associated with the viral envelope is more highly glycosylated than the knob-shaped portion of the NA spike with enzymic activity (Lazdins et al., 1972). NA is present in influenza virions in smaller amounts than HA, in a ratio of about three to four HA polypeptides for each NA polypeptide. Little information is available on the carbohydrate components of the NA. Recently, both the HA and NA of influenza virions were shown to be sulfated glycoproteins (Compans and Pinter, 1975).The sulfate appears to be covalently linked to the oligosaccharide chains of viral gly-

246

R. W. COMPANS AND M. C. KEMP

coproteins (Nakamura and Compans, 1977, 1978a). Glycoproteins of enveloped viruses of all the other major groups studied also are sulfated, whereas carbohydrate-free polypeptides are not (Pinter and Compans, 1975; Kaplan and Ben-Porat, 1976).

2. RHABDOVIRUSG PROTEIN Rhabdoviruses are covered with closely spaced glycoprotein spikes approximately 10 nm in length as shown by electron microscope studies. Rhabdoviruses possess only a single glycoprotein species termed the G protein, and for vesicular stomatitis virus (VSV) it has a MW of approximately 69,000 daltons (Wagner, 1975).Carhvright et d . (1972) have postulated that only a single glycoprotein molecule constitutes the spike structure. The G proteins of rhabdoviruses are amphipathic, like the glycoproteins of influenza virions. After protease treatment of VSV a fragment of the G protein was demonstrated to be associated with the intact virion (Mudd, 1974). More recently, Schloemer and Wagner (1975a) isolated a small nonglycosylated portion of the G protein from the envelope of protease-treated VSV virions. The fragment was found to have a MW of 5200 daltons, approximately equivalent to 50 amino acids. This is similar to the estimated size of the portion of the HA2 protein of the influenza virion thought to be buried within the lipid bilayer. Amino acid analysis of the G-protein fragment showed that it contained a preponderance of hydrophobic amino acids. As is the case for the membrane-associated portions of influenza HA, the hydrophobic fragment of the G protein is long enough to penedate the lipid bilayer. Conclusive evidence for such penetration has not been obtained, but cross-linking experiments with glutaraldehyde suggest that interactions may occur between G proteins and internal M proteins (Brown et d . , 1974). The carbohydrates linked to the G protein of VSV have been studied in detail. The glycoprotein is 9-10% carbohydrate by weight and contains mannose, galactose, N-acetylglucosamine, and neuraminic acid as the major sugar components, with lesser amounts of N-acetylgalactosamine and fucose (McSharry and Wagner, 1971; Burge and Huang, 1970. Etchison and Holland, 1974a). The size and composition of the carbohydrate moieties per VSV glycoprotein are variable, depending upon the cell type in which the virions are grown (Burge and Huang, 1970; Etchison and Holland, 1974a). Likewise, the sequence of the carbohydrates within the glycosyl side-chains may exhibit cell depen-

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

247

dence (Moyer and Summers, 1974). The monosaccharide composition of the oligosaccharide side-chains is similar to that of influenza virus (Etchison and Holland, 1974b), with the exception that sialic acid is present on the termini. Klenk et aZ. (1970b) demonstrated the presence of sialic acid on the envelope of VS virions by treating the virus with colloidal iron hydroxide which stains sialic acid residues. The stain was shown to bind to the envelope of VSV, whereas it did not bind to influenza virions or paramyxoviruses as shown by electron microscopy. More recent studies have shown that the glycosyl moieties are not terminated by sialic acid when VSV is grown in mosquito cells, because these cells lack sialyl transferase (Schloemer and Wagner,

1975b). Preliminary studies indicated that from 3-5 cyanogen bromide peptides of the VSV G-protein may be glycosylated (Wagner, 1975). However, Etchison and Holland (1974a,b) have calculated that there are only 2 glycopeptides of 3000-3400 daltons in each G protein molecule. More recent studies by Etchison et al. (1977)have indicated that the VSV G-protein possesses two identical glycopeptides. The number average molecular weight of the glycopeptides was estimated to be 3150 by gel filtration analysis and 3450 based on composition of the amino acid and sugar residues. Additional information concerning the structure of the glycopeptide was obtained by sequential chemical and enzymatic degradation. These results indicate that the glycopeptides are acidic type I glycopeptides with two or three mannose branches terminating in sialic acid. Moyer and co-workers (1976) have obtained evidence that the glycosyl residues are covalently linked to an asparagine residue of the G protein, and the sequence of the monosaccharides that constitute the oligosaccharide side-chains has been determined (Hunt and Summers, 1976b).

3. TOGAVIRUS GLYCOPROTEINS Togavinises are small, spherical enveloped viruses 50-60 nm in diameter with a core structure that appears to be icosahedral. Although there are many members of the togavirus group, the best studied are Sindbis virus and Semliki Forest virus (SFV). Sindbis virus possesses two glycoproteins designated El and E,; both have a MW of approximately 50,000 daltons, and they are not linked by disulfide bonds because they can be separated under nonreducing conditions and without alkylation (Schlesingeret al., 1972). SFV is similar to Sindbis virus in that it also has 2 glycoproteins of similar molecular weight designated El and E2, but a third glycoprotein designated E, has also been

248

R. W. COMPANS A N D M. C. KEMP

detected. The molecular weights of El, E,, and E, were estimated to be 49,000,52,000 and 10,000 daltons, respectively (Garoffet al., 1974). E, and E, are synthesized as a common precursor protein (NVP68) that is cleaved to yield E, and E,. El, EP,and E, of SFV are present in equimolar ratios as are E, and E2 of Sindbis virus. The arrangement and relationship of the glycoproteins within the spike structure of the Sindbis and SFV virion are yet to be resolved. Garoff and Simons (1974) and Garoff (1974) have used the cross-linking agent dimethyl suberimidate to study the interrelationships of the glycoproteins of SFV. El and E, were most readily cross-linked, but steric hindrance may have prevented the cross-linking of El, E2, and E,. Recent studies by Jones et al. (1977)have suggested that El and E, of Sindbis virus, like El, EZ,and E, of SFV, probably constitute the glycoprotein spike, since the cleavage of PEP,a precursor of E,, does not occur in temperature sensitive mutants of complementation groups including that thought to represent El. Furthermore, cleavage is inhibited by antibodies directed against either El or E,. These data suggest that PE, and El may exist as a complex in the membrane of the infected cell, and presumably also that El and E, remain as a complex in the viral envelope. Protease treatment of SFV cleaves the glycoproteins, El and E,, and residual segments can be isolated from the envelope of spikeless particles (Utermann and Simons, 1974). The amino acid composition of each of these peptides demonstrates that they are enriched in hydrophobic amino acids. The residual peptides have a MW of approximately 5000 daltons; hence they are comprised of about 50 amino acids. Thus, like the glycoproteins of influenza virus and VSV, togavirus glycoproteins are amphipathic in nature and appear to possess a peptide of similar size embedded within the viral envelope. When SFV was treated with high concentrations of dimethyl suberimidate, both the tail fragments of El and Ez were cross-linked with the nucleocapsid protein, which supports the conclusion that the hydrophobic segments, of El and E, may penetrate through the lipid bilayer and interact with the nucleocapsid (Garoff and Simons, 1974). Carbohydrate analysis of SFV showed that El contains about 18 moles of monosaccharide, E2 about 28 moles and E, about 22 moles (Garoff et al., 1974). E2 seemed to be particularly rich in mannose. It has recently been reported that the SFV glycoproteins El, E,, and E, are differentially glycosylated (Mattila et al., 1976). El and E, were shown to contain, on the average, one type A glycosyl side chain. E2 was shown to contain one type A glycosyl side chain and possibly one or two B-type glycosyl side chains. A-type oligosaccharides are complex structures containing fucose, galactose, mannose, and N-acetyl-

M E M B R A N E GLYCOPROTEINS OF ENVELOPED VIRUSES

249

glucosamine, whereas B-type oligosaccharides contain only mannose and N-acetylglucosamine (Johnson and Clamp, 1971). The apparent MWs of the A-type units of E, and E, were 3400 and 4000 daltons, respectively. The B-type side chains of E, had an apparent MW of 2000 daltons, and the A-type units an approximate MW of 3100 daltons. Similar studies have been performed with Sindbis virus, and Keegstra et al. (1975)showed that one type-A and one type-B residue were attached to each of the two different glycoproteins. The extent of completion of the glycosyl side chains was shown to be dependent in part upon the cell in which the virus was grown (Burge and Huang, 1970; Keegstra et ul., 1975). BHK-21 cells were able to complete type-A side chains (i.e., to add galactose and terminal sialic acid), whereas chicken embryo fibroblasts were quite inefficient in adding sialic acid (Keegstra et al., 1975). Both SFV and Sindbis virus possess hemagglutinating activity. Recent studies by Dalrymple et al. (1976) localized this activity to the El glycoprotein of Sindbis virus, whereas the E2 glycoprotein possesses the antigenic determinants which react with neutralizing antibody. 4. PARAMYXOVIRUS GLYCOPROTEINS

The virions of paramyxoviruses are covered with glycoprotein surface projections approximately 10 nm in length. Two functionally distinct types of glycoproteins have been isolated from several members of the paramyxovirus group (Scheid et al., 1972; Scheid and Choppin, 1973; 1974). In contrast to the situation in influenza viruses, which possess hemagglutinating and neuraminidase activities on distinct glycoprotein molecules, both of these activities are associated with the larger of the two glycoprotein species in paramyxoviruses, which is designated the HN glycoprotein. The smaller glycoprotein component in paramyxoviruses is associated with cell fusion and hemolysis activities and is designated the F glycoprotein. Purified glycoproteins of SV5 form rosettelike clusters in the absence of detergents, suggesting that the spikes have hydrophobic bases (Scheid et al., 1972). The aggregates formed are morphologically distinguishable, the HN protein forming berrylike aggregates while the F protein forms distinct rosettelike clusters consisting of radiating spikes 10-13 nm in length with distinct terminal knobs. In the three best studied paramyxoviruses, SV5, Newcastle disease virus (NDV), and Sendai virus, the H N glycoprotein has a MW range of 65,000-74,000 daltons, and the HN spike solubilized by detergent treatment sediments at 8.9s (Scheid et al., 1972). Although the exact size and fine structure of the HN spike remain to be determined, it has

250

R. W. COMPANS AND M. C. KEMP

been suggested that each morphological spike contains at least two glycoprotein monomers (Choppin and Compans, 1975). In some strains of NDV, an 82,000-dalton precursor of the HN glycoprotein designated HNo is incorporated into virions (Nagai and Klenk, 1977). These particles have reduced hemagglutinating and neuraminidase activities which are activated upon proteolytic cleavage, which produces the 74,000 dalton HN molecule. In SVS, NDV, and Sendai virions, the F glycoprotein has a MW of -56,000 daltons, and the F spike that is solubilized by treatment of SV5 virions with Triton X-100 has a sedimentation coefficient of approximately 6.7s (Scheid et al., 1972). In Sendai virions (Homma and Ohuchi, 1973; Scheid and Choppin, 1974) and certain strains of NDV (Nagai et al., 1976) grown in some cell types, the F glycoprotein is not found, and a larger precursor molecule designated Fois observed. Recent studies have indicated that proteolytic cleavage of the Foglycoprotein yields two cleavage products, which have bekn designated F1and F, (Shimizu et al., 1974; Nagai et al., 1976; Scheid and Choppin, 1977).The F2 cleavage product is more highly glycosylated than F1,and appears to be located on the distal end of the spike; it contains a blocked N-terminal as is also found in the uncleaved F, glycoprotein (Scheid and Choppin, 1977). Thus the cleavage of F, generates a free N-terminal on the F, segment of the glycoprotein, and this appears to expose a new hydrophobic region of the molecule which may be important for virus-induced cell fusion. The F, glycoprotein is inactive in cell fusion and hemolysis and can be converted into the active F glycoprotein by proteolytic cleavage, with concomitant activation of cell fusion and hemolysis. It is of particular interest that virions containing the Foprecursor are not infective and gain infectivity upon such proteolytic cleavage. 8. Functions of Viral Glycoproteins

1. ADSORPTIONTO RECEPTORS In the case of enveloped viruses, glycoproteins located on the surface of the virion are the components involved in adsorption to cellular receptors, which may or may not be host cell glycoproteins. Hemagglutination by influenza virus has been studied as a model system for the adsorption of a virus to receptors, and considerable information has been obtained. Adsorption to the erythrocyte occurs by the HA spike binding to sialic acid-containing components on the cell surface. The receptor molecule has been isolated from chick red blood cells, and it is a major glycoprotein containing M and N blood group anti-

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

25 1

gens. A detailed summary of the properties of this receptor is given by Schulze (1975), as well as by Tanner in this volume. Removal of sialic acid from the receptor molecule by NA prevents the agglutination of erythrocytes by influenza virus. Hemagglutination can also be inhibited by pretreating the virus with specific antibodies. While the model for adsorption of influenza virus to erythrocytes seems straightforward, its applicability to the adsorption of viruses to other cell types is uncertain, and little information is available on the nature of receptors for viruses. Liposomes containing gangliosides may be capable of acting as receptors for Sendai virus, a paramyxovirus (Haywood, 1974, 1975). Sialyoglycoproteins inserted in liposomes can also act as receptors, but the fact that gangliosides can act as receptors raises the possibility that glycolipids may be involved in viral attachment. Further evidence that viral glycoproteins specify virus-host cell interactions has been obtained from the studies of Bishop et al. (1975). They produced spikeless particles of the Indiana serotype of VSV b y treating the virus with bromelain or pronase. These particles were shown to be noninfectious but, when they were reconstituted with purified G protein isolated from the same strain or from the New Jersey serotype, infectivity was restored. Antibody directed against the homologous G protein used for reconstitution effectively neutralized the virus, but antibody directed against the serotype of the spikeless particle was ineffective in neutralization when glycoproteins from a different serotype were used for reconstitution. The mechanism of adsorption of other enveloped viruses has not been studied to the same degree as that of ortho- and paramyxoviruses. Herpes viruses do not exhibit hemagglutinating activity, but they possess at least 13 glycoproteins which are asymmetrically located on the external surface of the envelope (Roizman and Furlong, 1974; O’Callaghan and Randall, 1976). Removal of the envelope by nonionic detergents irreversibly alters the infectivity of these viruses (Abodeely et al., 1970). Oncornavinis glycoproteins react specifically with host cell receptors and in the case of avian leukosis viruses they define the host range as well as the classification into subgroups based on interference and neutralization properties (Vogt and Ishizaki, 1966; Ishizaki and Vogt, 1966; Duff and Vogt, 1969). In addition, Tozawa et al. (1970) showed that the viral glycoproteins absorbed homologous neutralizing antibody but not antisera prepared against heterologous virions. The purified glycoproteins were shown to interfere with the early steps in infection by the homologous virus. The determination of host range by the glycoproteins of avian leukosis viruses was further demonstrated

252

R. W. COMPANS A N D M. C. KEMP

by phenotypic mixing experiments. Defective Rous sarcoma virions that lack the envelope gene cannot infect cells. However, when grown in the presence of an avian leukosis virus, they acquire the glycoproteins of that virus, and the host range of the sarcoma virus reflects that of the leukosis virus (Hanafusa, 1965; Vogt, 1965; Kawai and Hanafusa, 1973). The major glycoprotein of Rauscher murine leukemia virus (gp71)binds specifically to receptor molecules found on murine cells but not on other mammalian cells (DeLarco and Todaro, 1976).

2. CELLFUSION The phenomenon of cell fusion may be caused by members of several groups of enveloped viruses of which paramyxoviruses are the best studied. The properties of hemolysis and cell fusion are associated with the F glycoprotein as described above. Virus-induced cell fusion may occur in the absence of virus replication, in the presence of high concentrations of virus particles. Both infectious and noninfectious viruses are equally adept at causing such cell fusion which sometimes has been termed fusion from without (Bratt and Gallaher, 1972). In contrast, fusion with low multiplicities of virus has been called fusion from within and is dependent upon replication of the virus. It is likely that the F glycoprotein is involved in both types of fusion phenomena. The process of fusion through the direct action of concentrated virus has been studied intensively since it was first observed by Okada (1958). Sendai virus-induced fusion involves several separate identifiable steps, as indicated by Maeda et al. (1977): (1) adsorption of the virus to the cell, which seems to occur at the tips of the spikes located on the virion surface; (2) aggregation of cells; (3) fusion of the viral envelope with the cell membrane and finally fusion of the cells. These studies suggest that the envelope bilayer of the virion and the membrane of the target cell are brought into contact by the action of the viral HN glycoprotein. The F protein then causes destabilization of the lipids, and an intermixing of lipids occurs. The mechanism of action of the F protein remains to be determined.

3. NEURAMINIDASE ACTIVITY A neuraminidase activity has been shown to be associated with orthomyxoviruses and paramyxoviruses. The functional role of this enzyme was uncertain for many years, and there was even doubt at some point that it was a viral gene product, since similar enzyme ac-

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

253

tivity is present in normal cells (White, 1974); however, these doubts have been resolved by studies of the biochemical, genetic, and antigenic properties of viral neuraminidase (see review in Bucher and Palese, 1975). Several distinct functions have been proposed for this enzyme. It is postulated that neuraminidase-containing virus lodges in the upper respiratory tract and binds to mucin via HA. The glycosyl residues of muchin are terminated by sialic acid (N-acetylneuraminic acid) to which HA binds, and one role of the neuraminidase may be to cleave the sialoglycoprotein bond, freeing the bound virion. It is presumed that in this way the virus is released, and that underlying cell receptors are exposed to which HA can attach (Davenport, 1976). It has also been postulated that neuraminidase is involved in an early event such as penetration, but this is unlikely because virions remain infectious after inhibition of neuraminidase activity by specific antibody (Bucher and Palese, 1975). However, the possibility that neuraminidase participates in release of the budding virion from the infected cell surface has gained support from several types of experiments. By using influenza strains with different levels of enzyme activity, it was shown that virus strains having low activity were released from cells more slowly than strains with higher enzyme activity (Palese and Schulman, 1974). Further, in the presence of antibody to viral neuraminidase, which inhibited enzyme activity, virions were formed but release of virus into culture media was inhibited (Seto and Rott, 1966; Compans et al., 1969; Webster, 1970).When bacterial neuraminidase was added to the cultures, infectious virus was released from the cells. These studies, however, are complicated by the fact that bivalent antibody can cause cross-linking of virions to viral antigens on cell surfaces, and Becht et al. (1971)reported that monovalent Fab fragments inhibited neuraminidase activity without affecting virus release. More conclusive information on the function of the enzyme has been obtained with temperature-sensitive mutants of influenza virus which are defective in neuraminidase activity (Palese et al., 1974). At the nonpermissive temperature, no neuraminidase activity is detected; virus particles are produced by cells, despite the fact that infectivity titers are markedly reduced. However, the virus particles form large aggregates, and in contrast to wild-type virions these particles contain sialic acid as shown by colloidal iron hydroxide staining. Since influenza virions bind to sialic acid residues, these results indicate that the mutant virus particles aggregate to each other, because sialic acid is added to viral carbohydrates, and that the essential function of viral neuraminidase is to remove or prevent the addition of such sialic acid. In support of this conclusion, the addition of bacterial neur-

254

R. W. COMPANS A N D M. C. KEMP

aminidase to cells infected with these mutants cause a marked enhancement of virus release. Further evidence supporting this role for the neuraminidase was obtained with a neuraminidase inhibitor, 2-deoxy-2,3-dehydro-Ntrifluoracetylneuraminic acid (FANA). Influenza virus grown in the presence of FANA contains neuraminic acid on its envelope, and the particles undergo extensive aggregation (Palese and Compans, 1976). A marked reduction in virus yield is observed because of this aggregation, and treatment with purified neuraminidase results in a marked enhancement in progeny virus yields, apparently through disaggregation of virus. Thus viral neuraminidase is not required for assembly of progeny virions but appears to be essential for the removal of sialic acid from the surface of the virion itself.

4. GLYCOPROTEINS AS ANTIGENS From the studies described above it is evident that the membranes of enveloped viruses are asymmetrically constructed, with the glycoproteins that comprise the spikes or surface projections physically located on the exterior of the viral envelope. Thus the glycoproteins are exposed to the immune surveillance system of the host and, being good immunogens, may elicit humoral and cellular immune responses (Evans, 1976). The classification of virus isolates into specific strains depends largely upon serological procedures, and for enveloped viruses, surface glycoproteins are usually the relevant antigens in such tests. Neutralizing antibodies are usually directed against the viral proteins involved in attachment to receptors, e.g., HA glycoprotein of influenza virus (Webster and Laver, 1975), G protein of VSV (Wagner, 1975), gp69/71 of murine leukemia viruses (Fischinger et aZ., 1976; Strand and August, 1976), gp85 of avian leukosis viruses (Bolognesi, 1974), and E, of Sindbis virus (Dalrymple et d.,1976). High concentrations of antibody can prevent attachment of the virus to receptors, but low concentrations can also neutralize infectivity by a mechanism that is not understood. The specific determinants of viral glycoproteins recognized by virus-neutralizing antibodies have not been chemically characterized. It has been postulated that antibody molecules may recognize only determinants on the tip of the HA spike of the influenza virion (White, 1974). In reaching these conclusions, it has been assumed that the size of the antibody molecule precludes the possibility that it could make contact with any other part of the spike, since the spaces between the spikes are too small for the immu'noglobulin to interact with other regions. Electron microscopic observations ( L d e r t y and

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

255

Oertilis, 1963) indicate that antibody molecules interact with the tips of surface spikes, and analysis of the tryptic peptides of HA molecules isolated from *closely related influenza strains has shown that they rarely differ b y more than one or two peptides (Laver and Webster, 1972; Webster and Laver, 1972), suggesting that the strain differences are indeed restricted to small regions or determinants of the HA spike. The antigenic character of many viral glycoproteins is stable, but the determinants exhibited by the glycoproteins of influenza virions are characteristically variable, as evidenced by the many different strains of type-A and -B influenza. The HA and NA glycoproteins of influenza viruses are antigenically distinct, and they undergo antigenic changes independently of each other. The antigenic changes that occur may be gradual, in which case the different virus strains are clearly related to each other with respect to both surface antigens. Antigenic changes of this nature are termed antigenic drift, and they result from the interplay of viral mutability and immunological selection (Webster and Laver, 1975). The presence of antibody of low avidity may select for single-step mutants, which have an altered amino acid in the key area of the antigenic determinant, giving rise to a new viral strain. At intervals of 10-15 years sudden and complete changes in the determinants of type-A influenza glycoproteins occur; the changes are such that the viruses that arise possess glycoproteins that appear completely distinct on peptide mapping (Laver and Webster, 1972). Dramatic changes in the antigenic determinants of viral glycoproteins are termed antigenic shifts, and it is these new viruses that cause worldwide influenza pandemics. It has been postulated that antigenic shift may occur because type-A strains of human origin may undergo recombination with type-A strains of avian and animal origin. The antigenic determinants of the new HA or NA are sufficiently different that the human host does not possess immunity, and the virus gains a selective advantage. Interestingly type-B influenza viruses do not undergo antigenic shift. The reason for this may lie in the fact that type-B influenza viruses have not been isolated from other animal species; thus it is probable that the type-B viruses can not undergo similar recombination events (Webster and Laver, 1975). C. Function of Carbohydrates in Viral Glycoproteins

Viral glycoproteins provide excellent systems for analysis of the function of carbohydrates in membrane glycoproteins. Several approaches have been used to modify the carbohydrates, including treatment of virions with specific glycosidases, growth of virus in cells

256

R. W. COMPANS A N D M. C. KEMP

with specific sugar transferase defects, and treatment of virus-infected cells with inhibitors of glycosylation. Removal of sialic acid from the G protein of VSV has been reported to reduce the infectivity of VSV virions (Schloemer and Wagner, 1974),whereas the infectivity of SFV (Kennedy, 1974) and Friend leukemia virus (Schafer et al., 1977) were reportedly unaltered by such treatment. Moreover, enzymic addition of sialic acid to the glycoproteins of Sindbis virus did not alter the infectivity of this virus (Stollar et al., 1976), while enhancing the infectivity of influenza virus (Schulze, 1975) and restoring infectivity to neuraminidase-treated VSV (Schloemer and Wagner, 1974). Treatment of influenza virions with glycosidases alter hemagglutinating activity, but neuraminidase activity was unaffected; however, similar treatment of NDV, a paramyxovirus, resulted in no alteration in hemagglutinating or neuraminidase activities (Bike1and Knight, 1972). Schiifer et al. (1977)showed that the indirect hemagglutinating activity of Friend leukemia virus was inhibited by glycosidase treatment but that viral infectivity and determinants involved in viral interference and absorption of neutralizing antibody were unaltered. Schlesinger et a2. (1976)utilized a cell line deficient in the enzyme N-acetylglucosaminyltransferase to study the effects of alterations in carbohydrates of enveloped viruses. This enzyme deficiency results in the synthesis of membrane glycoproteins with decreased amounts of N-acetylglucosamine, galactose, and sialic acid. When VSV and Sindbis virus were grown in these cells, the apparent MWs of their glycoproteins were lower. The infectivity of VSV and Sindbis virus grown in these cells was not altered from that of fully glycosylated virions. However, glycosidase treatment of SFV was shown to decrease the infectivity of this virus (Kennedy, 1974).The apparent difference between these results remains to be resolved. It is possible that the core of the glycosyl side chain is added to the Sindbis virus glycoproteins in the enzyme-deficient cells, whereas the glycosidase treatment used by Kennedy may have removed more of the glycosyl moieties from the SFV glycoproteins. Inhibitors of glycosylation have also been used to assess the role of glycosyl side chains of glycoproteins. Primarily, three different inhibitors have been used for this purpose: 2-deoxy-~-glucose(2-dG),an analog of glucose which substitutes for mannose, preventing the further addition of monosaccharides to the glycosyl side chain; Dglucosamine, which is thought to inhibit glycosylation at high concentrations by decreasing UTP pools in the cell and subsequent activation of other sugars (Scholtissek, 1971); and tunicamycin (TM), a glucosamine-containing antibiotic that inhibits the formation of N-acetyl-

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

25 7

glucosamine-lipid intermediates which serve as donors for the synthesis of the oligosaccharide side chains of glycoproteins (Tkacz and Lampen, 1975). Kilbourne (1959) and Kaluza et al. (1972) were among the first investigators to utilize inhibitors of glycosylation to study the role of carbohydrates in influenza virions. They showed that 2-dG and D-glucosamine inhibited the biosynthesis of active HA, NA, and mature infectious influenza virions. Subsequent biochemical studies revealed that high concentrations of 2-dG or Dglucosamine prevented the synthesis of influenza virus glycoproteins (Klenk et al., 1972b, 1974; Compans et al., 1974; Nakamura and Compans, 1978a).Instead, an unglycosylated or incompletely glycosylated hemagglutinin precursor HA,, was detected. Once synthesized, HAowas found associated with cytoplasmic membranes, as is the case with the normal glycoprotein. HAo glycoprotein in cells infected with the fowl plague strain (FPV) was cleaved by cellular proteases to yield a heterogeneous product (Klenk et al., 1974). However, the comparable protein synthesized in cells infected with the WSN strain was processed and incorporated into virions (Nakamura and Compans, 1978a). Because 2-dG and D-ghcosamine interfere with metabolic reactions other than the glycosylation of glycoproteins, they may cause side effects that can affect virus replication. Therefore TM, a compound that seems to affect only the glycosylation of glycoproteins, has been employed to extend these studies. T M inhibited virion formation in FPVinfected cells and the unglycosylated glycoprotein HA,, appeared to be degraded by cellular proteases (R. T. Schwarz et al., 1976); HA,, of the WSN strain synthesized in the presence of TM also appeared to be completely unglycosylated, whereas even at high concentrations of 2-dG and Dglucosamine some glycosylation occurred (Nakamura and Compans, 1978a). Nonetheless, TM did not inhibit virion formation to the same extent as 2-dG (Nakamura and Compans, 1978).The surface spike layer of WSN virions produced in the presence of 2-dG, D-glucosamine, and TM was altered morphologically, and the hemagglutinating activity of the virions was significantly reduced. These results suggest that glycosylation of virion glycoproteins is not required for influenza virion formation but is needed for biological activity of viral glycoproteins. These glycosylation inhibitors have also been used with several other enveloped viruses. When herpes viruses are grown in the presence of 2-dG, the infectious virus yield is decreased by greater than 95%, but the yield of viral particles is not reduced (Courtney et al., 1973).The reduction in infectivity was attributed to an inability of the virions to attach to the host cell and penetrate, implying that oligosac-

25 8

R. W. COMPANS AND M. C. KEMP

charide residues of herpes virus glycoproteins play a role in the attachment and recognition of host cell receptors. The glycoproteins of VSV, Sindbis virus, and SFV are not glycosylated in the presence of T M (R. T. Schwarz et al., 1976; Leavittet al., 1977). Mature VSV and Sindbis virions are not released from TMtreated cells, but nonglycosylated precursors are synthesized and seem to be stable within the cell (Leavitt et al., 1977).Similarly, R. T. Schwarz et al. (1976) showed that unglycosylated SFV glycoprotein precursors were synthesized in TM-treated cells. Moreover, they were not degraded by host cell proteases and virion assembly was completely inhibited. It is interesting that, as noted above, HA, of FPV is degraded when it is grown in TM-treated chicken embryo fibroblasts, but the glycoprotein precursors of SFV synthesized in the same cells are stable. These results may be due to greater release of protease b y FPV infection or, alternatively, the polypeptide backbone of SFV glycoproteins may be less susceptible to proteolytic degradation. High concentrations of glucosamine rapidly shut off the production of infectious avian sarcoma virus particles (Hunter et al., 1974). However, avian sarcoma virus particles were assembled and released at about 60% of the control level in TM-treated cells, and such particles appeared to lack glycoproteins (R. T. Schwarz et al., 1976).The infectivity titer of the virus produced in TM-treated cells decreased by only

15%. These results indicate that the effects of glycosylation inhibitors may vary with the virus and host cell, as well as with the specific inhibitor used. Side effects of some inhibitors may have marked effects on virus replication. However, the fact that in some systems virus particles are produced with unglycosylated or incompletely glycosylated glycoproteins clearly demonstrates that the complete glycosylation process is not essential for intracellular migration of glycoproteins or their incorporation into the plasma membrane and subsequently into virus particles. The biological activities of some glycoproteins such as those of influenza virus and SFV appear to require glycosyl moieties, whereas for other viruses, such as murine leukemia virus, this may not be the case.

D. Effects of Viral Proteins on lipid Bilayer Structure

The membranes of enveloped viruses, particularly those of singlestranded RNA viruses, are unique in their simplicity of construction

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

259

and are useful systems for studying the interactions of specific proteins with the lipid bilayer. Hence comparative studies of the effects of viral proteins of the lipid bilayer structure have been made using the techniques of electron spin resonance (ESR), nuclear magnetic resonance (NMR), and fluorescence polarization (FP). ESR studies have provided evidence that the lipids of influenza virus, SV5, Rauscher leukemia virus, VSV, and SFV (Landsberger et uZ., 1971, 1972, 1973; Sefton and Gaffney, 1974) are bilayer structures with fluid lipid phases similar to those observed for other biological membranes, but all viral membranes were shown to be substantially more rigid than the corresponding host cell plasma membrane. Proteolytic removal of the glycoprotein spikes from the surface of SV5 and influenza virions did not appreciably alter the fluidity of the lipid bilayer of these viruses. Lenard et al. (1976)have provided further evidence that the glycoproteins of the influenza virion contribute little to the rigidity of the lipid bilayer. The phospholipid composition of standard influenza particles, and “incomplete” virus produced upon serial undiluted passage, were compared and found to be indistinguishable, as were the ESR spectra of the two types of particles. The incomplete particles were shown to contain approximately twice the amount of glycoproteins relative to the complete particles. Thus it was concluded that the rigidity of viral membranes may be determined b y the M protein and not by the viral glycoproteins. This may not be entirely the case for VSV and SFV, since Sefton and Gaffney (1974)and Landsberger and Compans (1976) observed that, when the glycoproteins of these viruses were removed by proteases, the envelope became more fluid, indicating that the viral glycoproteins may contribute to the rigidity of the envelope. However, Landsberger and Compans (1976) postulated that the major effect on VSV bilayer fluidity was exerted b y the M protein, since the fluidity of the lipid bilayer was altered only slightly when the G protein was removed by protease, whereas vesicles prepared from extracted viral lipids were much more fluid than lipids in virions. Using the technique of fluorescence depolarization, Moore et aZ. (1976) and Barenholz et al. (1976) showed that the envelope of SFV, Sindbis virus, and VSV has a higher microviscosity than that of the plasma membranes from which the virions budded. The increased microviscosity was attributed in part to insertion of the hydrophobic regions of the glycoproteins into the envelope bilayer. Stoffel and Bister (1975),using N M R spectra of ‘3C-labeledlipids, also demonstrated that the envelope lipids of VSV are highly rigid, as a result of either lipid-lipid or lipid-protein interactions. In general it may be concluded therefore that the lipid bilayer of enveloped virions is more

260

R. W. COMPANS AND M. C. KEMP

rigid than the host cell membrane from which the virion buds. While virion glycoproteins do interact with the lipid bilayer and may affect the rigidity of the membrane to some extent, the internal membrane protein(s) may also be of equal or greater importance in determining membrane rigidity.

V.

ASSEMBLY OF VIRAL MEMBRANES

A. Cellular Sites of Maturation

Most enveloped viruses form by a process of budding at the plasma membrane, with little or no participation of other membrane structures in the final steps of maturation. However, there are important exceptions in the case of certain virus groups. The capsids of herpes viruses are assembled in the nucleoplasm and are observed to bud through the inner nuclear membrane, acquiring their envelopes in the process (Roizman and Furlong, 1974; O’Callaghan and Randall, 1976). They appear to be transported as enveloped particles through cytoplasmic channels to the cell surface. Bunyaviruses (Murphy et aZ., 1973) and coronaviruses (Oshiro, 1973) appear to form primarily by budding into cytoplasmic cisternae, and extracellular virions are observed associated with the plasma membrane. For the rhabdovirus group, several modes of maturation have been reported. VSV, the most widely studied member, usually forms by budding at the cell surface, but maturation at intracellular membranes has been observed. The New Jersey strain of VSV was observed to bud primarily from plasma membranes of L or Vero cells, and almost entirely at intracytoplasmic membranes of pig kidney cells (Zee et al., 1970). These reports indicate that the site of maturation of a specific virus type may vary depending on the host cell. Rabies virus, which resembles VSV morphologically and biochemically, is exceptional in that assembly of the virion appears to occur through a process involving de n o w formation of membranes in the cytoplasmic matrix (Hummeler et al., 1967).De novo formation of membranes is the usual process of assembly for members of the pox virus group (Dales and Mosbach,

1968).

The appearance of viral proteins on the cell surface, as well as the cellular site of virus assembly, can be modified by external agents including antibodies and lectins, The phenomenon of antigenic modulation involves altering the expression of cell surface antigens by specific antibody, and such antigens may be of viral origin (Lampert et d . ,

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

261

1975). This undoubtedly has an effect on viral maturation, although the precise effects have not been determined. Exposure of influenza virus-infected cells to convanavalin A appears to alter the site of virus maturation (Stitz et al., 1977). In the presence of this lectin, normal maturation at the plasma membrane is not observed, but instead large numbers of virions are observed in intracellular vacuoles. It is evident therefore that the maturation site for enveloped viruses can vary with the virus type as well as the host cell, and can be altered in response to specific stimuli. The process is likely to be determined as a result of interactions between virus-specific proteins and host cell membranes. I t is remarkable that in most instances only a single type of cellular membrane is selected as the site of virus assembly in a particular virus-infected cell. An understanding of the mechanisms which govern the selection of the assembly site may provide new insights into the assembly of cellular membrane components and organelles, since similar interactions are likely to be involved in determining the location of subsets of cellular proteins in specific cellular organelles.

B.

Synthesis and lntracellular localization of Envelope Proteins

The synthesis and assembly of viral membrane components have been analyzed in cells in which host cell synthesis is inhibited as a result of virus infection. Similar conclusions have been made for several virus types. Viral glycoproteins appear to be synthesized on membrane-bound polyribosomes and remain associated with various cellular membranes (Spear and Roizman, 1970; Compans, 1973a,b; Stanley et al., 1973; Klenket al., 1974; David, 1973; Hay, 1974; Atkinsonet al., 1976; Nagai e t al., 1976; Knipe et al., 1977). Glycosylation occurs in association with cytoplasmic membranes. Glycoproteins appear to migrate from rough endoplasmic reticulum to smooth or Golgi complex membranes to the cell surface and are then incorporated into virions; at no time are they found as ‘‘soluble’’ cytoplasmic components. This general scheme for the synthesis and migration of proteins through intracellular membranes to the cell surface has been termed membrane flow. Although the precise intracellular location of viral glycoproteins remains to be established, a scheme may be envisaged in which these components remain associated with the cisternal side of membranes of the endoplasmic reticulum and Golgi complex after synthesis. Incorporation into the plasma membrane by a process of vesicle fusion would then result in the correct orientation on the cell surface for viral as-

262

R. W. COMPANS AND M.

C. KEMP

sembly. A similar scheme has been suggested for the incorporation of glycoproteins into the plasma membrane, based on the distribution of oligosaccharides in different membrane fractions (Hirano et al., 1972). Viral systems allowed investigators to follow the intracellular migration and glycosylation of specific glycoprotein species for the first time, whereas previous studies with cellular glycoproteins Kad dealt with cell fractions essentially uncharacterized as to the specific proteins present. Recently, Katz et al. (1977) and Rothman and Lodish (1977) have described a new model system for studies of the incorporation of viral glycoproteins into membranes. Katz et al. (1977) have shown that the mRNA for VSV G-protein translated in vitro by wheat germ extracts in the presence of dog pancreas rough endoplasmic reticulum is associated with the membrane. Furthermore, the newly synthesized G protein was shown to span the membrane with the amino-terminal asymmetrically located. In addition, the polypeptide portion which penetrated the membrane was shown to be glycosylated. Glycosylation did not occur in the absence of membranes. These studies were extended b y Rothman and Lodish (1977),who showed that the insertion of G-protein into the membrane begins when 80 or fewer amino acid residues are polymerized. Evidence was also obtained that the nascent chain is glycosylated while still attached to the ribosomes on the cytoplasmic side of the endoplasmic reticulum vesicle. Since the mechanism by which viral glycoproteins are inserted into membranes is in all probability similar to that of the insertion of host cell glycoproteins, further studies using viral systems may provide more insight into the mechanism of insertion of glycoproteins into membranes. The process of glycosylation does not appear to play an important role in determining the intracellular migration of glycoproteins. Although this has been suggested as a possible function for carbohydrate components of glycoproteins, the available data using inhibitors of glycosylation suggest that it is possible to inhibit or extensively modify the gl ycosylation process without preventing the migration of viral glycoproteins to the cell surface (Courtney et al., 1973; Compans et al., 1974; Nakamura and Compans, 1978a). The incorporation of carbohydrate-free M proteins into membranes appears to involve a distinct pathway in which cytoplasmic synthesis is followed by rapid association with the plasma membrane (Lazarowitz et d., 1971; Meier-Ewert and Compans, 1974; Hay, 1974; Nagai et al., 1976; Knipe et al., 1977). No evidence for migration through cytoplasmic membrane structures has been obtained' for these components, aad it has been suggested that they are inserted directly

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

263

into membranes after synthesis. Cytoplasmic synthesis followed by direct insertion into membranes has also been proposed for some classes of cellular membrane proteins (Lodish and Small, 1975). C. Sequence of Events in Viral Assembly

Although there are several unanswered questions concerning the precise steps in assembly even for the best studied enveloped viruses, the available information from electron microscope studies as well as the biochemical approaches described above suggest a scheme like that depicted in Fig. 2 for influenza virus. Glycoproteins appear to be inserted into the plasma membrane as the first step in assembly, after migration through cytoplasmic membranes. After they are inserted into the plasma membrane, initial random distribution may occur in which the proteins are free to undergo lateral diffusion in the plane of the membrane. Such random distribution is illustrated for the glycoproteins of parainfluenza virus in Fig. 3, and similar observations have been reported for other viruses (Birdwell and Strauss, 1974; H. Schwarz et al., 1976).Random distribution of viral antigens (Fig. 3 ) is observed only when ferritin-antibody conjugates are applied to cells after glutaraldehyde fixation. In previous studies of the distribution of antigen, in which unfixed cells were used (Compans and Choppin, 1971),antigens were observed in discrete patches. It is likely that under these conditions they undergo lateral redistribution and aggregation into patches as a result of bivalent antibody.

RNP

FIG.2. Schematic diagram of the assembly process of an influenza virion. Clycoproteins are thought to be inserted into the plasma membrane by a process called memhrane flow and are initially found randomly dispersed in the membrane. Following in-sertion of the M protein, glycoproteins are thought to accumulate in discrete regions from which host cell membrane proteins are excluded. The association of the ribonucleoprotein (RNP) with such regions of modified membrane is followed by budding and release of the completed virion.

FIG,3. Labeling of the surface of an infected cell with ferritin-conjugated antibody to the parainfluenza virus SV5. The cell was fixed with glutaraldehyde prior to the application of antibody. Under these conditions, ferritin molecules appear randomly dispersed on the cell surface. x 100,000.

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

265

In the case of influenza virus, the M protein may form a domain on the internal surface of the plasma membrane, stabilized by proteinprotein interactions. Specific recognition of the M protein by the glycoproteins could then produce an accumulation of glycoproteins in a circumscribed region of the cell surface. Alternatively, it is possible that M protein monomers associate with glycoprotein monomers at the plasma membrane, and that these complexes undergo lateral diffusion and aggregation into domains. In either case the lack of host cell membrane proteins in the viral envelope indicates that cellular proteins are efficiently excluded from the region of the plasma membrane which becomes the viral envelope. Association of the nucleocapsid with regions of the cell surface containing viral envelope protein may stimulate the process of budding. The formation of virions involves outfolding of the cell membrane and envelopment of the nucleocapsid by the modified membrane. Some examples of the final stages in assembly of the parainfluenza virus SV5 are depicted in Figs. 4 and 5. Regions of the cell membrane, with underlying helical nucleocapsids, are shown in Fig. 4.The presence of virus-specific glycoproteins on the external surface is indicated by tagging with ferritin-conjugated antibody. In Fig. 5, the emerging virus particles tagged with ferritin-antibody are shown at higher magnifica-

FIG.4. A region of the surface of an MDBK bovine kidney cell infected with the parainfluenza virus SV5. The helical nucleocapsids of the virus are aligned under the cell membrane, and the external surface of the cell in these regions is tagged with ferritin-labeled antiviral antibody. x 35,000.

2 66

R. W. COMPANS AND M. C. KEMP

FIG. 5. Filamentous SV5 virions emerging from the surface of an infected MDBK cell. The virions are clearly tagged with ferritin-antibody, whereas adjacent areas of the cell surface are devoid of viral antigen. The helical nucleocapsids, in cross section or longitudinal section, are seen clearly in the emerging virus particles. x 80,000,

tion; most particles are long filamentous virions. The specificity of the ferritin labeling is evident from the tagging of virions and absence of ferritin on adjacent cell membranes. For viruses with icosahedral nucleocapsids and no M protein, a similar mechanism for virus assembly has been proposed (Garoff and Simons, 1974) in which the nucleocapsid itself binds to glycoproteincontaining membranes, with subsequent lateral diffusion and clustering of the glycoproteins in this region. D. Macromolecular Interactions in Membrane Assembly

Although information is accumulating about the pathways by which viral membrane proteins are incorporated into plasma membranes, there is little direct evidence concerning the precise interactions which lead to the formation of domains on the cell surface which contain virus-specific proteins and lack host cell proteins. The lack of significant amounts of host cell protein in the virion indicates that such

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

267

domains must he intermediates in assembly. Further, the available data on viral protein composition (Table 11) and morphology suggest that assembly interactions may differ for various virus groups. I n all cases it is likely that protein-protein interaction serves to create a patch of virus-specific proteins in a cellular membrane, but this interaction may occur on the external or internal surface of the lipid bilayer. As discussed above, for viruses that contain M proteins or welldefined icosahedral nucleocapsids, some evidence has been obtained for the initial random distribution of glycoproteins on the cell surface, which may be followed by lateral diffusion and accumulation of glycoproteins in juxtaposition to the M protein or icosahedral nucleocapsid, with transmembrane interactions between the external and internal proteins serving to anchor the viral glycoproteins in place. With other virus types that contain a large amount of glycoprotein and no obvious M protein or well-defined icosahedral capsid, it is possible that lateral interactions between the glycoproteins are important in assembly. The glycoproteins of togaviruses and bunyaviruses have been observed in a regular surface arrangement (von Bonsdorff and Harrison, 1975; von Bonsdorff and Pettersson, 1975).The latter viruses lack an M protein or an icosahedral internal component, and it has been suggested that direct interactions between glycoproteins may be involved in assembly and maintenance of the viral structure (von Bonsdorff and Pettersson, 1975). Similar interactions may be important in other virus groups in which glycoproteins are major protein constituents of the virion and there is no known internal membrane protein, such as B-type oncomaviruses and coronaviruses. Pox viruses are unique in that formation of their lipid-containing membrane occurs de novo in the cytoplasm, rather than on a preexisting membrane structure. These viruses thus provide an unusual system for investigation of the molecular interactions involved in the formation of a highly organized structure within the cytoplasmic matrix (Dales and Mosbach, 1968). However, these viruses are structurally very complex, and limited information has been obtained on their molecular organization. Shape and size determination may also be controlled by viral proteins at various levels; nucleocapsids, M proteins, or glycoproteins could be the determining factor for different virus groups. The phenomenon of phenotypic mixing of envelope glycoproteins in cells doubly infected with VSV and the parainfluenza virus SV5 clearly demonstrates that the internal proteins and not the glycoproteins determine the particle size and shape of rhabdoviruses (Choppin and Compans, 1970; McSharry et al., 1971). This conclusion is supported

268

R. W. COMPANS AND M. C. KEMP

by the observation that particles with the internal proteins of VSV possess the characteristic bullet shape of VSV while containing mixtures of envelope glycoproteins derived from VSV and SV5. A similar analysis of phenotypically mixed particles produced by dual infections with viruses of other major groups may provide further insights into the macromolecular interactions involved in virion assembly. The fixed shape and size of many lipid-containing viruses stands in contrast to the situation in membranous cellular organelles, which generally exhibit pleomorphism. Similar pleomorphism is observed in some enveloped viruses. The assembly of such membranes of organelles and pleomorphic viruses may be regulated more by the production of materials than by precise constraints imposed by macromolecular interactions. ACKNOWLEDGMENTS Research by the authors was supported by Grants No. A1 12680 and CA 18611 from the USPHS, PCM76-09711 from the National Science Foundation, and VC 149B from the American Cancer Society. M.C.K. was supported b y a fellowship from the Anna Fuller Fund.

REFERENCES Abodeely, R. A,, Lawson, L. A., and Randall, C. C. (1970).Morphology and entry of enveloped and deenveloped equine abortion (herpes) virus. ]. Virol. 5,513-523. Allen, A. K., Skehel, J. J., and Yuferov, V. (1977).The amino acid and carbohydrate composition of the neuraminidase of B/Lee/40 influenza virus. J . Cen. Virol. 37,625-628. Atkinson, P. H., Moyer, S. A., and Summers, D. F. (1976). Assembly of vesicular stomatitis virus glycoprotein and matrix protein into HeLa cell plasma membranes. J . Mol. Biol. 102,613-631. Bachi, T., Gerhard, W., Lindenmann, J., and Muhlethaler, K. (1969). Morphogenesis of influenza A virus in Ehrlich ascites tumor cells as revealed by thin-sectioning and freeze-etching. J . Virol. 4,769-776. Barenholz, Y.,Moore, N. F., and Wagner, R. R. (1976). Enveloped viruses as model membrane systems: Uicroviscosity of vesicular stomatitis virus and host cell membranes. Biochemistry 15,3563-3570. Becht, H., Hammerling, U., and Rott, R. (1971). Undisturbed release of influenza virus in the presence of univalent antineuraminidase antibodies. Virology 46,337-343. Bikel, I., and Knight, C. A. (1972). Differential action ofAspergillus glycosidase activities of influenza and Newcastle disease viruses. Virology 49,326-332. Birdwell, C. R., and Straws, J. H. (1974). Replication of Sindbis virus. IV. Electron microscopic study of the insertion of viral glycoproteins into the surface of infected cells. J . Virol. 14,366-374. Bishop, D. H. L., and Roy, P. (1972). Dissociation of vesicular stomatitis virus and relation of the virion proteins to the viral transcriptase. J . Virol. 10,234-243. Bishop, D. H. L., Repik, P., Obijeski, J. F., Moore, N. F., and Wagner, R. R. (1975). Res-

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

269

titution of infectivity to spikeless vesicular stomatitis virus by solubilized viral components. /. Virol. 16,75-84. Blough, H. A., and Tiffany, J. M. (1975).Theoretical aspects of structure and assembly of viral envelopes. Cum. Top. Microbiol. Imtnunol. 70, 1-30. Bolognesi, D. P. (1974). Structural components of RNA tumor viruses. Adu. Virus Res. 19,315-359. Brand, C. M., and Skehel, J. J. (1972). Crystalline antigen from the influenza virus envelope. Nature (London),New Biol. 238, 145-147. Bratt, M. A,, and Gallaher, W. M. (1972). Biological parameters of fusion from without. In “Membrane Research” (C. F. Fox, ed.), pp. 383-406. Academic Press, New York. Brown, F., Smale, C. J., and Horzinek, M. C. (1974). Lipid and protein organization in vesicular stomatitis and Sindbis viruses.]. Gen. Virol. 22,455-458. Bucher, D. J., and Kilbourne, E. D. (1972). A,(N,) neuraminidase of the X-7 influenza virus recombinant: Determination of molecular size and subunit composition of the active unit.]. Virol. 10,60-66. Bucher, D., and Palese, P. (1975). The biologically active proteins of influenza virus: Neurarninidase. I n “The Influenza Viruses and Influenza” (E. D. Kilbourne, ed.), pp. 83-123. Academic Press, New York. Burge, B. W., and Huang, A. S. (1970). Comparison of membrane protein glycopeptides of Sindbis virus and vesicular stomatitis virus.]. Virol. 6, 176-182. Camright, B., Smale, C. J., Brown, F., and Hull, R. (1972).Model for vesicular stomatitis virus.]. Virol. 10, 256-260. Choppin, P. W., and Compans, R. W. (1970). Phenotypic mixing of envelope proteins of the parainfluenza virus SV5 and vesicular stomatitis virus. J. Virol. 5,609-616. Choppin, P. W., and Compans, R. W. (1975). Reproduction of paramyxoviruses. I n “Comprehensive Virology” (H. Fraenkel-Conrat and R. R. Wagner, eds.), Vol. 4, pp. 95-178. Plenum, New York. Choppin, P. W., Lazarowitz, S. G., and Goldberg, A. R. (1975). Studies on proteolytic cleavage and glycosylation of the hemagglutinin of influenza A and B viruses. I n “Negative Strand Viruses” (R. D. Barry and B. W. J. Mahy, eds.), pp. 105-120. Academic Press, New York. Compans, R. W. (1973a). Influenza virus proteins. 11. Association with components of the cytoplasm. Virology 51, 56-70. Compans, R. W. ( 1973b). Distinct carbohydrate components of influenza virus glycoproteins in smooth and rough cytoplasmic membranes. Virology 55,541-545. Compans, R. W., and Choppin, P. W. (1971).The structure and assembly of influenza and parainfluenza viruses. In “Comparative Virology” (K. Maramorosch and F. Kurstak, eds.), pp. 407-432. Academic Press, New York. Compans, R. W., and Choppin, P. W. (1975). Reproduction of myxoviruses. I n “Comprehensive Virology” (H. Fraenkel-Conrat and R. R. Wagner, eds.), pp. 179-252. Plenum, New York. Compans, R. W., and Pinter, A. (1975).Incorporation of sulfate into influenza virus glycoproteins. Virology 66, 151-160. Cornpans, R. W., Holmes, K. V., Dales, S., and Choppin, P. W. (1966). An electron microscopic study of moderate and virulent virus-cell interactions of the parainfluenza virus SV5. Virolog!j 30,411-426. Compans, R. W., Dimmock, N. J., and Meier-Ewert, H. (1969). Effect of antibody to neuraminidase on the maturation and hemagglutinating activity of an influenza A2 virus.]. Virol. 4,528-534.

2 70

R. W. COMPANS A N D M. C. KEMP

Compans, R. W., Klenk, H.-D., Caliguiri, L. A., and Choppin, P. W. (1970). Influenza virus proteins. I. Analysis of polypeptides of the virion and identification of spike glycoproteins. Virology 42,880-889. Compans, R. W., Content, J., and Duesburg, P. H. (1972). Structure of the ribonucleoprotein of influenza virus. J. Virol. 10,795-800. Compans, R. W., Meier-Ewert, H., and Palese, P. (1974).Assembly of lipid-containing viruses. J . Supramol. Struct. 2,496-511. Courtney, R. J., Steiner, S. M., and Benyesh-Melnick, M. (1973). Effects of 2-deoxy-Dglucose on herpes simplex virus replication. Virology 52,447-445. Dales, S . , and Mosbach, E. H. (1968). Vaccinia as a model for membrane biogenesis. Virology 35,564-583. Dalrymple, J . M., Schlesinger, S., and Russell, P. K. (1976).Antigenic characterization of two Sindbis envelope glycoproteins separated by isoelectric focusing. Virology 69, 93 -103. Davenport, F. M. (1976).Influenza virus. In “Viral Infections of Humans” (A. S. Evans, ed.), pp. 273-296. Plenum, New York. David, A. E. (1973).Assembly of the vesicular stomatitis virus envelope: Incorporation of viral polypeptides into the host plasma membrane. J . Mol. Biol. 76, 135-148. DeLarco, J., and Todaro, G. J. (1976).Membrane receptors for murine leukemia viruses: Characterization using the purified viral envelope glycoprotein, gp71. Cell 8,365371. Duff, R. G., and Vogt, P. K. (1969). Characteristics of two new avian tumor virus subgroups. Virology 39, 18-30. Etchison, J. R., and Holland, J. J. (1974a). Carbohydrate composition of the membrane glycoprotein of vesicular stomatitis virus grown in mammalian cell lines. Proc. Natl. Acad. Sci. U.S.A. 71,4011-4014. Etchison, J. R., and Holland, J. J. (1974b). Carbohydrate composition of the membrane glycoprotein of vesicular stomatitis virus. Virology 60,217-229. Etchison, J. R., Robertson, J. S., and Summers, D. F. (1977). Partial structural analysis of the oligosaccharide moieties of the vesicular stomatitis virus glycoprotein by sequential chemical and enzymatic degradation. Virology 78,375-392. Evans, A. S. (1976). Epidemiological concepts and methods. In “Viral Infections of Humans” (A. S. Evans, ed.), pp. 1-32. Plenum, New York. Famulari, N. G., Buchhagen, D. L., Klenk, H.-D., and Fleissner, E. (1976). Presence of murine leukemia virus envelope proteins gp70 and p15(E) in a common polyprotein of infected cells. J . Virol. 20, 501-508. Fischinger, P. J., Schafer, W., and Bolognesi, D. P. (1976). Neutralization of homologous and heterologous oncornaviruses by antisera against the p 15(E) and gp7l polypeptides of Friend murine leukemia virus. Virology 71, 169-184. Gahmberg, C. G., Uterman, G., and Simons, K. (1972).The membrane proteins of Semliki Forest virus have a hydrophobic part attached to the viral membrane. FEBS Lett. 28, 179-182. Gard, G. P., Vezza, A. L., Bishop, D. H. L., and Compans, R. W. (1977). Structural proteins of Tacaribe and Tamiami virions. Virology 83,84-95. Garoff, H. (1974). Cross-linking of the spike glycoproteins in Semliki Forest virus with dimethyl suberimidate. Virology 62,385-392. Garoff, H., and Simons, K. (1974). Location of the spike glycoproteins in the Semliki Forest virus membrane. Proc. Natl. Acad. Sci. U.S.A. 71, 3988-3992. Garoff, H., Simons, K., and Renkonen, 0. (1974). Isolation and characterization of the membrane proteins of Semliki Forest virus. Virology 61,493-504.

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

27 1

Hanafusa, H. (1965).Analysis of the defectiveness of Rous sarcoma virus. 111. Determining influence of helper virus on the host range and susceptibility to interference of RSV. Virology 25,248-255. Harrison, S. C., David,A., Jumblatt, J., and Darnell, J. E. (1971). Lipid and protein organization in Sindbis virus. J. Mol. Biol. 60,523-528. Haslam, E. A., Hampson, A, W., Egan, J. A., and White, D. 0. (1970).The polypeptides of influenza virus. 11. Interpretation ofpolyacrylamide gel electrophoresis patterns. Virology 42,555-565. Hay, A. J. (1974).Studies on the formation of the influenza virus envelope. Virology 60, 348-418. Haywood, A. M. (1974).Characteristics of Sendai virus receptors in a model membrane. J . Mol. Biol. 83,427-436. Haywood, A. M. (1975). Model membranes and Sendai virus: Surface-surface interactions. I n “Negative Strand Viruses” (R. Barry and B. Mahy, eds.), pp. 923-928. Academic Press, New York. Heine, J. W., Honess, R. W., Cassai, E., and Roizman, B. (1974). Proteins specified by herpes simplex virus. XII. The virion polypeptides of type 1 strains. J. Virol. 14, 640-651. Hierholzer, J. C., Palmer, E. L., Whitfield, S. G., Kaye, H. S., and Dowdle, W. R. (1972). Protein composition of coronavirus OC-43. Virology 48, 516-527. Hirano, H., Parkhouse, B., Nicolson, G. C., Lennox, E. S., and Singer, S. J. (1972). Distribution of saccharide residues on membrane fragments from a myeloma-cell homogenate: Its implications for membrane biogenesis. Proc. Natl. Acad. Sci. U.S.A. 69,2945-2949. Homma, M., and Ohuchi, M. (1973). Trypsin action on the growth of Sendai virus in tissue culture cells. 111. Structural difference of Sendai viruses grown in eggs and tissue culture cells. J . Virol. 12, 1457-1465. Honess, R. W., and Roizman, B. (1974). Regulation of herpes virus macromolecular synthesis. I. Cascade regulation of the synthesis of three groups of viral proteins. J . Virol. 14,s-19. Hummeler, K., Koprowski, H., and Wiktor, T. J. (1967). Structure and development of rabies virus in tissue culture. J . Virol. 1, 152. Hunt, L. A,, and Summers, D. F. (1976). Glycosylation of vesicular stomatitis virus glycoprotein in virus-infected HeLa cells. J. Virol. 20, 646. Hunter, E., Friis, R. R., and Vogt, P. K. (1974). Inhibition of avian sarcoma virus replication by glucosamine. Virology 58,449-456. Ishizaki, R., and Vogt, W. K. (1966). Immunological relationships among envelope antigens of avian tumor viruses. Virology 30,375-387. Johnson, I., and Clamp, J. R. (1971).T h e oligosaccharide units of human type L immunoglobulin M (macroglobulin). Biochem. J . 123,739-745. Jones, K. J., Scupham, R. K., Pfeil, J. A., Wan, K., Sagik, P., and Bose, H. R. (1977).Interaction of Sindbis virus glycoproteins during morphogenesis. ]. Virol. 21,778-787. Kaluza, G., Scholtissek, C., and Rott, R. (1972). Inhibition of the multiplication of enveloped RNA-viruses by glucosamine and 2-deoxy-D-glucose. J . Gen. Virol. 14, 251-259. Kaplan, A. S., and Ben-Porat, T. (1976). Synthesis of‘proteins in cells infected with herpes virus. XI. Sulfated structural proteins. Virology 70, 561-563. Katz, F. N., Rothman, J. E., Lingappa,V. R., Blobel, G., and Lodish, H. F. (1977). Membrane assembly In Vitro: Synthesis, glycosylation, and asymmetric insertion of a transmembrane protein. Proc. Natl. Acad. Sci. U.S.A. 74,3278-3282.

272

R . W. COMPANS AND M. C. KEMP

Kawai, S., and Hanafusa, H. (1973).Isolation of defective mutant of avian sarcoma virus. Proc. Natl. Acad. Sci. U.S.A. 70,3493-3497. Keegstra, K., Sefton, B., and Burke, D. (1975).Sindbis virus glycoproteins: Effect of the host cell on the oligosaccharides. J . Virol. 16,613-620. Kennedy, S . I. T. (1974).The effects of enzymes on structural and biological properties of Semliki Forest virus.]. Cen. Virol. 23, 129-143. Kilboume, E. D. (1959).Inhibition of influenza virus multiplication with a glucose antimetabolite (2-deoxy-D-glucose).Nature (London) 183,271-272. Klenk, H.-D. (1974). Viral envelopes and their relationship to cellular membranes. Curr.Top. Microbiol. Immunol. 68,2948. Klenk, H.-D., and Choppin, P. W. (1969). Lipids of plasma membranes of monkey and hamster kidney cells and of parainfluenza virions grown in these cells. Virology 38, 255-268. Klenk, H.-D., and Choppin, P. W. (1970). Glycosphingolipids of plasma membranes of cultured cells and an enveloped virus (SV5) grown in these cells. Proc. Natl. Acad. Sci. U.S.A. 66,57-64. Klenk, H.-D., Caliguiri, L. A., and Choppin, P. W. (1970a). The proteins of the parainfluenza virus SV5 11. The carbohydrate content and glycoproteins of the virion. Virology 42,473-481. Klenk, H.-D., Compans, R. W., and Choppin, P. W. (1970b). An electron microscope study of the presence or absence of neuraminic acid in enveloped viruses. Virology 42,1158-1162. Klenk, H.-D., Rott, R., and Becht, H. (1972a). On the structure of the influenza virus envelope. Virology 47,579-597. Klenk, H.-D., Scholtissek, C., and Rott, R. (1972b). Inhibition of glycoprotein biosynthesis of influenza virus by D-glucosamine and 2-deoxy-~-glucose.Virology 49, 723-734. Klenk, H.-D., Wollert, W., Rott, R., and Scholtissek, C. (1974). Association of influenza virus proteins with cytoplasmic fractions. Virology 57,28-41. Klenk, H.-D., Rott, R., Orlich, M., and BIodorn, J. (1975). Activation of influenza A viruses by trypsin treatment. Virology 68,426-439. Knight, C. A. (1944). A sedimentable component of allantoic fluid and.its relationship to influenza viruses.]. E r p . Med. 80,83-99. Knipe, D. M., Baltimore, D., and Lodish, H. F. (1977). Separate pathways of maturation ofthe major structural proteins of vesicular stomatitis virus.]. Virol. 21,1128-1139. Lafferty, K. J., and Oertilis, S. (1963).The interaction between virus and antibody. HI. Examination of virus-antibody complexes with the electron microscope. Virology 21,91-99. Lamb, R. A., and Mahy, B. W. J. (1975). The polypeptides and RNA of Sendai virus. In “Negative Strand Viruses” (B. W. J. Mahy and R. D. Barry, eds.), pp. 65-87. Academic Press, New York. Lampert, P. W., Joseph, B. S., and Oldstone, M. B. A. (1975).Antibody-induced capping of measles virus antigens on plasma membrane studied by electron microscopy. J . Virol. 15, 1248-1255. Landsberger, F. R., and Compans, R. W. (1976).Effect of membrane protein on the lipid bilayer structure: A spin label ESR study of vesicular stomatitis virus. Biochemistry 15,2356-2360. Landsberger, F. R., Lenard, J., Paxton, J., and Compans, R. W. (1971). Spin label ESR study of the lipid-containing membrane of influenza virus. Proc. Natl. Acad. Sci. U.S.A. 68,2579-2583.

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

2 73

Landsberger, F. R., Compans, R. W., Paxton, J., and Lenard, J. (1972). Structure of the lipid phase of Rauscher murine leukemia virus.J. Suprurnol. Struct. 1,M-S. Landsberger, F. R., Compans, R. W., Choppin, P. W., and Lenard, J. L. (1973). Organization of the lipid phase in viral membranes: Effects of independent variation of the lipid and the protein composition. Biochemistqt 12,4498-4502. Laver, W. G. (1971). Separation of two polypeptide chains from the hemagglutinin subunit of influenza virus. Virology 45,275-288. Laver, W. G., and Baker, N. (1972). Amino acid composition of polypeptides from influenza virus particles. J . Gen. Virol. 17,61-67. Laver, W. G., and Valentine, R. C. (1969). Morphology of the isolated hemagglutinin and neuraminidase subunits of influenza virus. Virology 38, 105-119. Laver, W. G., and Webster, R. C.(1972). Studies on the origin of pandemic influenza. 11. Peptide maps of the light and heavy polypeptide chains from the hemagglutinin subunits of A, influenza viruses isolated before and after the appearance of Hong Kong influenza. Virology 48,445-455. 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. Lazarowitz, S. G., Compans, R. W., and Choppin, P. W. (1971). Influenza virus structural and nonstructural proteins in infected cells and their plasma membranes. Virology 46,830-843. Lazarowitz, S. G., 6ompans, R. W., and Choppin, P. W. (1973). Proteolytic cleavage of the hemagglutinin polypeptide of influenza virus. Function of the uncleaved polypeptide HA. Virology 52, 199-212. Lazdins, I., Haslam, E. A., and White, D. 0. (1972). The polypeptide of influenza virus. VI. Composition of the neuraminidase. Virology 49,758-765. Leavitt, R., Schlesinger, S., and Kornfeld (1977). Tunicamycin inhibits glycosylation and multiplication of Sindbis and vesicular stomatitis viruses. J . Virol. 21, 375-385.

Lenard, J., and Compans, R. W. (1974). The membrane structure of lipid-containing viruses. Biochim. Biophys. Actu 344,51-94. Lenard, J., Tsai, D., Compans, R. W., and Landsberger, F. R. (1976). Organization of the membrane of standard and incomplete influenza virus. Virology 71,389-394. Lodish, H. F., and Small, B. (1975). Membrane proteins synthesized by rabbit reticulocytes. J . Cell Biol. 65,51-64. McSharry, J. J., and Wagner, R. R. (1971). Carbohydrate composition of vesicular stomatitis virus. J. Virol. 7,412-415. McShany, J. J., Compans, R. W., and Choppin, P. W. (1971). Proteins ofvesicular stomatitis virus and of phenotypically mixed vesicular stomatitis virus-simian virus 5 virions. J . Virol. 8,722-729. Maeda, T., Asano, A., Okada, Y., and Ohnishi, S.-I. (1977). Transmembrane phospholipid motions induced by F glycoprotein in hemagglutinating virus of Japan.]. Virol. 21,232-241.

Mattila, K., Luukkonen, A., and Renkonen, 0. (1976). Protein-bound oligosaccharides of Semliki Forest virus. Biochim. Biophys. Actu 419,435-444. Meier-Ewert, H., and Compans, R. W. (1974). Time course of synthesis and assembly of influenza virus proteins. ]. Virol. 14, 1083-1091. Moore, N. F., Barenholz, Y.,and Wagner, R. R. (1976). Microviscosity of togavirus membranes studied by fluorescence depolarization: Influence of envelope proteins and the host cell.]. Virol. 19, 126-135.

274

R. W. COMPANS A N D M. C. KEMP

Moyer, S. A., and Summers (1974). Vesicular stomatitis virus envelope glycoprotein a]terations induced b y host cell transformation. Cell 2,63-70. Moyer, S. A., Tsang, J. M., Atkinson, P. H., and Summers, D. F. (1976). Oligosaccharide moieties of the glycoprotein of Vesicular Stomatitis virus. J. Virol. 18, 167-175. Mudd, J. A. (1974). Glycoprotein fragment associated with vesicular stomatitis virus after proteolytic digestion. Virology 62, 573-577. Murphy, F. A., Harrison, A. K., and Whiffield, S. G. (1973) Bunyaviridae: Morphologic and morphogenetic similarities of Bunyamwera serologic supergroup viruses and several other arthropod-borne viruses. Zntervirology 1,297-316. Nagai, Y., and Klenk, H.-D. (1977). Activation of precursor to both glycoproteins of Newcastle disease virus by proteolytic cleavage. Virology 77,125-134. Nagai, Y.,Ogura, H., and Klenk, H.-D. (1976a). Studies on the assembly of the envelope of Newcastle disease virus. Virology 69,523-538. Nagai, Y.,Klenk, H.-D. and Rott,R. (1976b). Proteolytic cleavage of the viral glycoproteins and its significance for the virulence of Newcastle disease virus. Virology 72, 494-508. Nakamura, K., and Compans, R. W. (1977). The cellular site of sulfation of influenza virus glycoproteins. Virology 79,381-392. Nakamura, K., and Compans, R. W. (1978a). Effects of inhibitors on glycosylation sulfation, and assembly of influenza virus glycoproteins. Virology 84 : 303-319. Nakamura, K., and Compans, R. W. (1978b). Glycopeptide components of influenza viral glycoproteins. Virology (in press). Nakamura, K., and Compans, R. W. (19784. Host cell dependent glycosylation of influenza virus glycoproteins. Submitted for publication. Nakashima, Y.,Wiseman, €3.L., Kongburg, W., and Marvin, D. A. (1975). Primary structure and side chain interactions of PF, filamentous bacterial virus coat protein. Nature (London) 253,68-71. Obijeski, J. F., Bishop, D. H. L., Murphy, F. A,, and Palmer, E. L. (1976). The structural proteins of La Crosse virus.]. Virol. 19,985-997. O’Callaghan, D. J., and Randall, C. C. (1976). Molecular anatomy of herpes viruses: Recent studies. Prog. hled. Virol. 22, 152-210. Okada, Y. (1958). The fusion of Ehrlich’s tumor cells caused by HVJ virus in uitro. Biken]. 1, 103-110. Oshiro, L. S. (1973). Coronaviruses. In “Ultrastructure of Animal Viruses and Bacteriophages” (A. J. Dalton and F. Haguenau, eds.), pp. 331-343. Academic Press, New York. Palese, P., and Compans, R. W. (1976). Inhibition of influenza virus replication in tissue acid. (FANA): Mechaculture by 2-deoxy-2,3-dehydro-N-trifluoracetylneuraminic nism of action. J . Gen. ViroE. 33, 159-163. Palese, P., and Schulman, J. L. (1974). Isolation and characterization of influenza virus recombinants with high and low neuraminidase activity: Use of 2-(3‘-methoxypheny1)-N-acetylneuraminicacid to identify cloned populations. Virology 57,227237. Palese, P., Tobita, K., Ueda, M., and Compans, R. W. (1974). Characterization of temperature sensitive influenza virus mutants defective in neuraminidase. Virology 61, 397-410. Pinter, A., and Compans, R. W. (1975). Sulfated glycoproteins and polysaccharides of enveloped viruses. J. Virol. 16,859-866. Quigley, J. P., Rifkin, D. B., and Reich, E. (1971). Phospholipid composition of Rous sarcoma virus, host cell membranes, and other enveloped viruses. Virology 46, 106-116.

M E M B R A N E GLYCOPROTEINS OF ENVELOPED VIRUSES

2 75

Roizman, B., and Furlong, D. (1974).The replication of herpes viruses. In “Comprehensive Virology” (H. Fraenkel-Conrat and R. R. Wagner, eds.), Vol. 3,pp. 229403.Plenum, New York. Rothman, J. E., Tsai, D. K., Dawidowicz, E. A., and Lenard, J. (1976).Transbilayer phospholipid asymmetry and its maintenance in the membrane of influenza virus. Biochemistry 15,2361-2370. Schafer, W., Fischinger, P. J., Collins, J. J., and Bolognesi, D. P. (1977).Role of carbohydrate in biological functions of Friend murine leukemia virus gp7lJ.Virol. 21,35-

40. Scheid, A.,and Choppin, P. W. (1973).Isolation and purification of the envelope proteins of Newcastle disease virus.J. Virol. 11,263-271. Scheid, A., and Choppin, P. W. (1974).Identification of biological activity of paramyxovirus glycoproteins: Activation of cell fusion, hemolysis and infectivity by proteolytic cleavage of an inactive precursor protein of Sendai virus. Virology 57,475-

490. Scheid, A., Caliguiri, L. A., Compans, R. W., and Choppin, P. W. (1972).Isolation of paramyxovirus glycoproteins: Association of both hemagglutinating and neuraminidase activities with the larger SV5 glycoprotein. Virology 50, 640-652. Schlesinger, M. J., Schlesinger, S., and Burge, B. W. (1972).Identification of a second glycoprotein in Sindbis virus. Virology 47,539-541. Schlesinger, S., Gottlieb, C., Feil, D., Gelb, N., and Kornfeld, S. (1976).Grbwth of enveloped RNA viruses in a line of Chinese hamster ovary cells with deficientN-acetylglucosaminyl transferase activity. J. Virol. 17,239-246. Schloemer, R. H., and Wagner, R. R. (1974).Sialoglycoprotein of vesicular stomatitis virus: Role of the neuraminic acid in infecti0n.J. Virol. 14, 270-281. Schloemer, R. H., and Wagner, R. R. (1975a).Mosquito cells infected with VSV yield unsialylated virions of low infectivity. J . Virol. 15, 1029-1032. Schloemer, R. H., and Wagner, R. R. (197513).Association of vesicular stomatitis virus glycoproteins with virion membrane: Characterization of the lipophilic tail fragment. J . Virol. 16,237-2A9. Scholtissek, C. (1971).Detection of an unstable RNA in chick fibroblasts after reduction of the UTP pool by glucosamine. Eur. J. Biochem. 24,358-365. Schulze, I. T.(1970).The struchire of influenza virus. I. The polypeptides of the virion. Virology 42,890-904. Schulze, I. T. (1975).The biologically active proteins of influenza virus: The hemagglutinin. In “The,Influenza Viruses and Influenza” (E. D. Kilbourne, ed), pp. 53-82. Academic Press, New York. Schwarz, H., Hunsmann, G., Moenning, V., and Schafer, W. (1976).Properties of mouse leukemia virus. XI. Immunoelectron microscopic studies on viral antigens on the cell surface. Virology 69, 169-178. Schwarz, R. T.,and Klenk, H.-D. (1974).Inhibition of glycosylation of influenza virus hemagg1utinin.J. Virol. 14, 1023-1034. Schwarz, R. T., Schmidt, M. F. G., Anwer, U., and Klenk, H.-D. (1977).Carbohydrates of influenza virus I. Glycopeptides derived from viral glycoproteins after labeling with radioactive sugars. J. Virol. 23, 217-226. Schwarz, R. T., Rohrschneider, J. M., and Schmidt, M. F. G. (1976).Suppression of glycoprotein formation of Semliki Forest, influenza, and avian sarcoma virus by tunicamycin.J. Virol. 19,782-791. Sefton, B. M., and Gdfney, B. J. (1974). Effect of the viral proteins on the fluidity of the membrane lipids in Sindbis virus.J. Mol. Biol. 90,343-358. Segrest, J. P., Jackson, R. L., and Marchesi, V. T. (1972). Red cell membrane glycopro-

276 tein: Amino acid sequence of an intramembranous region. Biochem. Biophys. Res. Commun. 49,964-969. Seto, J. T., and Rott, R. (1966). Functional significance of sialidase during influenza virus multiplication. Virology 30, 731-737. Shimizu, K., Shimizu, Y.K., Kohama, T., and Ishida, N. (1974). Isolation and characterization of two distinct types of HVJ (Sendai virus) spikes. Virology 62,90-101. Singer, S. J., and Nicolson, G. L. (1972). The fluid mosaic model of the structure of cell membranes. Science 175,720-731. Skehel, J. J., and Waterfield, M. D. (1975). Studies on the primary structure of the influenza virus hemagglutinin. Proc. N a t l . Acad. Sci. U.S.A. 72, 93-97. Spear, P. H., and Roizman, B. (1970).The proteins specified by herpes simplex virus. IV. The site of glycosylation and accumulation of viral membrane proteins. Proc. N a t l . Acad. Sci. U.S.A. 66,730-737. Stanley, P., Gandhi, S. S., and White, D. 0. (1973).The polypeptides of influenza virus. VII. Synthesis of the hemagglutinin. Virology 53,92-106. S t i k , L., Reinacher, M., and Becht, H. (1977).Studies on the inhibitory effect of lectins on myxovirus re1ease.J. Gen. Virol. 34,523-530. Stoffel, W., and Bister, K. (1975). ‘SC nuclear magnetic resonance studies on the lipid organization of enveloped virions (vesicular stomatitis virus). Biochemistry 14, 2841-2847. Stollar, V., Stollar, D., Koo, R., Harrep, K. A,, and Schlesinger, R. W. (1976). Sialic acid content of Sindbis virus from vertebrate and mosquito cells. Virology 69, 104-115. Strand, M., and August, J. T. (1976). Structural proteins of ribonucleic acid tumor viruses.J. Biol. Chem. 251,559-564. Strauss, J. H., Jr.. Burge, B. W., and Darnell, J. E. (1970). Carbohydrate content of membrane protein of Sindbis virus. J . Mol. Biol. 47,437-438. Tkacz, J. S., and Lampens, J. 0. (1975). Tunicamycin inhibition of polyisoprenyl N-acetylglucosaminyl pyrophosphate formation in calf liver microsomes. Biochem. Biophys. Res. Commun. 65,248-257. Tozawa, H., Bauer, H., Graf, T., and Gelderblorn, H. (1970). Strain-specific antigen of the avian leukosis sarcoma virus group. Virology 40,530-539. Tsai, K. H., and Lenard, J. (1975). Asymmetry of influenza virus membrane bilayer demonstrated with phospholipase C. Nature (London)255,554-555. Utermann, G . , and Simons, K. (1974). Studies on the amphipathic nature of the membrane proteins in Semliki Forest virus.J. Mol. Biol. 85, 569-587. Vezza, A. C., Card, C. P., Compans, R. W., and Bishop, D. H. L. (1977). Structural components of the arenavirus Pichinde. J . Virol. 23,776-786. Vogt, P. K. (1965). A heterogeneity of Rous sarcoma virus revealed by selectively resistant chick embryo cells. Virology 25,237-247. Vogt, P. K., and Ishizaki, R. (1966). Patterns of viral interference in the avian leukosis and sarcoma complex. Virology 30,368-374. von Bonsdorff, C. H., and Harrison, S. C. (1975). Sindbis virus glycoproteins form a regular surface lattice.]. Virol. 16, 141-145. von Bonsdorff, C. H., and Pettersson, R. (1975). Surface structure of Uukuniemi virus.J. Virol. 16, 1296-1307. Wagner, R. W. (1975). Reproduction of rhabdoviruses. In “Comprehensive Virology” (H. Fraenkal-Conrat and R. R. Wagner, eds.), Vol. 4, pp. 1-93. Plenum, New York. Webster, R. G. (1970). Estimation of the molecular weights of the polypeptide chains from the isolated hemagglutinin and neuraminidase subunits of influenza viruses. Virology 40,643-654.

MEMBRANE GLYCOPROTEINS OF ENVELOPED VIRUSES

277

Webster, R. G., and Laver, W. G. (1972).Studies on the origin of pandemic influenza. I. Antigenic analysis of A2 influenza viruses isolated before and after the appearance of Hong Kong influenza using antisera to the isolated hemagglutinin subunits. Virology 48,433-444. Webster, R. G., and Laver, W. G. (1975). Antigenic variation of influenza viruses. In “The Influenza Viruses and Influenza” (E. D. Kilbourne, ed.), pp. 270-314. Academic Press, New York. White, D. 0. (1974). Influenza viral proteins: Identification and synthesis. Curr. Top. Microbiol. Immunol. 63,2-48. Wiley, D. C., Skehel, J. J., and Waterfield, M. (1977). Evidence from studies with a cross-linking reagent that the haemagglutinin of influenza virus is a trimer. Virolog!/ 79,446-448. Wrigley, N. G., Skehel, J. J., Charlwood, P. A., and Brand, C. M. (1973).The size and shape of influenza virus neuraminidase. Virology 51, 525-529. Yagi, M. J., and Compans, R. W. (1977). Structural components of mouse mammary tumor virus. I. Polypeptides of the virion. Virology 76, 751-766. Zee, Y. C., Hackett, A. J., and Talens, L. (1970).Vesicular stomatitis virus maturation sites in six different host cells./. Gen. Virol. 7,95-102.

This Page Intentionally Left Blank

CURRENT TOPICS IN MEMBRANES AND TRANSPORT. VOLUME

11

Erythrocyte Glycoproteins MZCHAEL J . A. T A N N E R Department of Biochemistry University of Bristol Bristol, United Kingdom

I. Introduction . . . . . . . . . . . . . . . . . . . 11. The Origin and Turnover of the Erythrocyte . . . . . . . . . 111. The Glycoproteins o f t h e Erythrocyte Membrane . . . . . . . . A. Periodate-Stainable Glycoproteins . . . . . . . . . . . B. Non-Periodate-Stainable Glycoproteins . . . . . . . . . IV. Organization of the Glycoproteins in the Erythrocyte Membrane . . . V. Structure of the Glycoproteins . . . . . . . . . . . . . A. The Major Human Sialoglycoprotein (Glycophorin A) . . . . . B. The Minor Periodate-Stainable Proteins of the Human Erythrocyte . C. Polypeptide3 . . . . . . . . . . . . . . . . . VI. Functions of Glycoproteins . . . . . . . . . . . . . . A. Polypeptide3 . . . . . . . . . . . . . . . . . B. Periodate-Stainable Glycoproteins . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . .

1.

279 280 281 282 284 285 288 288 295 297 304 304 306 316

INTRODUCTION

Erythrocyte membrane proteins have been the subject of several recent reviews (Zwaal et al., 1973; Juliano, 1973; Steck, 1974; Marchesi et al., 1976) which reflect the considerable progress made in recent years in characterizing these proteins and understanding their organization within the membrane. Only membrane glycoproteins are discussed in this chapter, and an attempt is made to explore the functional significance of these components with an emphasis on functional attributes which might depend on the carbohydrates they contain. It is sometimes felt that the erythrocyte and its membrane are too atypical and specialized to be useful as a representative model for study of the involvement of plasma membranes in many of the complex phenomena which occur in the mammalian organism. It is true that the erythrocyte is highly specialized for the transport of oxygen 279 Copyright 0 1978 by Academic Press. Inc. All rights of reproduction in any form reserved. ISBN 0-12-153311-5

280

MICHAEL J. A. TANNER

and bicarbonate ions and lacks intracellular organelles and some features, such as hormone sensitivity, found in many other cells. This degree of specialization is reflected in the properties of the erythrocyte membrane. However, it should be recognized that, in adult mammals, most tissue cells are equally highly differentiated and specialized and that their plasma membranes often have equally unique morphological and biochemical characteristics which give rise to the particular features of different tissues. The technical advantages which result from the ability to obtain large quantities of homogeneous plasma membranes from the erythrocyte are well known. In addition, the mode of generation of erythrocytes provides several useful features in studying the involvement of plasma membrane components in cellular differentiation and cellular interactions. The erythropoietic system is the largest “organ” in the adult animal and is dedicated to the production of a single cell type. The high rates of turnover of this regenerating system (particularly in the anemic condition) make it possihle to isolate erythrocytes in the intermediate stages of maturation in quantities which allow biochemical studies of the plasma membranes. These intermediate stages are well defined with regard to both morphology and histochemistry. Finally, erythroid cells undergo an unusual, defined transition from a tissue-bound phase in the bone marrow to a circulating phase in the bloodstream. The cells in the latter phase undergo few interactions with other cells, and this stage provides a particularly useful experimental system for study of the function of cell surface glycoproteins.

II. THE ORIGIN AND TURNOVER OF THE ERYTHROCYTE

Before considering the functional role of the glycoproteins present in the erythrocyte membrane, it is useful to discuss-the life history of the erythrocyte, since in addition to having certain functions during the circulating phase of the life of the erythrocyte, these membrane components may also be involved during its generative and degradative stages. A few pertinent features of the events involved in the maturation and turnover of the erythrocyte are summarized in the following discussion. An excellent and most detailed review of this subject is available (Wickramasinghe, 1975). The early nucleated stages of erythroid cells occur in an organized system in the sinusoids of the bone marrow. Erythroid cells are derived from erythropoietin-sensitive stem cells which are not well characterized. The earliest recognizable erythroid cell is a large, dividing, nucleated cell designated the pronormoblast. The successively more mature stages of this cell,

ERYTHROCYTE GLYCOPROTEINS

28 1

the basophilic normoblast, polychromatic normoblast, and orthochromatic normoblast, have distinctive histochemical characteristics. At each of these nucleated stages except the last, the cell undergoes mitosis, but the final nucleated stage, the orthochromatic normoblast, is not capable of mitosis and DNA synthesis. Thus a single pronormoblast yields 16 orthochromatic normoblasts. During this progression the cells decrease in diameter from approximately 18 p m to 10p m and increase in hemoglobin content. A large proportion of the hemoglobin of the mature erythrocyte has been synthesized by the time the orthochromatic normoblast stage of the cell is reached. The final stage of erythroid cell maturation in the bone marrow is characterized by loss of the nucleus from the orthochromatic normoblast to yield the reticulocyte. The nucleus is extruded from this cell in a form which is surrounded by a layer of plasma membrane containing about 5% of the cytoplasm of the cell. Some investigators suggest that this process of nuclear extrusion is concurrent with, and indeed allows, release of the reticulocyte from the sinusoids of the bone marrow into the circulation (Tavassoli and Crosby, 1973),while others believe that the most immature form of the reticulocyte resides in the bone marrow for a short time before its release into the circulation. It takes about 7 days for the human pronormoblast to become a reticulocyte, and about onehalf this time is spent in cell division. The final maturation phase, reticulocyte to erythrocyte, results in the loss of protein-synthesizing ability and intracellular membranous organelles from the cell and is accompanied by a further decrease in the size of the cell from about 9 pm to 7 pm. The mature cell has an average circulating life span of 115 days in the human and is probably destroyed by the reticuloendothelial organs (liver, spleen, and bone marrow). There is some evidence which suggests that carbohydrates on the erythrocyte surface are involved in the removal of erythrocytes from the circulation. Neuraminidase-treated erythrocytes have a much reduced survival time in the rabbit and rat, and it has been suggested that this decreased survival time is due to the exposure of galactose residues subterminal to sialic acid residues (Gattegno et al., 1974; Durocher et al., 1975). This system appears to be analagous to that found for clearance from the circulation of many plasma glycoproteins (Morel1 et al., 1968, 1971). 111.

THE GLYCOPROTEINS OF THE ERYTHROCYTE MEMBRANE

The membrane of the human erythrocyte, the most intensively studied red cell membrane, contains 8-10% carbohydrate, and a large pro-

282

MICHAEL J. A. TANNER

portion of this carbohydrate is bound to membrane proteins (Winzler, 1969, 1971).

A. Periodate-Stainable Glycoproteins

1. HUMANERYTHROCYTE MEMBRANE When the proteins of human erythrocyte membranes are separated

by sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis (Fairbanks et al., 1971), only one protein (polypeptide 3)which can be visualized with coomassie blue is a glycoprotein (see Section V,C). If the gel is treated with the periodic acid-Schiff (PAS) stain, one major and several minor bands are observed. The distribution and resolution of the minor PAS staining depends markedly on the conditions of gel electrophoresis. The clearest resolution of the minor bands is obtained with the discontinuous SDS gel electrophoresis system of Laemmeli (1970), see Dahr et al., (1975b). Figure 1 shows the nomenclature used here to designate these periodate staining bands. These bands are not all unique glycoproteins. The relative proportions of PAS-1 and PAS-2 bands obtained depend on the conditions used to dissolve the glycoprotein preparations in the detdrgent for SDS gel electrophoresis (Marton and Garvin, 1973; Tuech and Morrison, 1974). If dissolution is carried out in the presence of phosphate buffers, the PAS-1 form predominates, while heating and the use of tris-containing buffers results in the formation of more of the PAS-2 form. These two bands are interconvertible forms of the major sialoglycoprotein (glycophorin A, Tomita and Marchesi, 1975). Studies with an erythrocyte variant lacking the major sialoglycoprotein confirm this conclusion (Tanner and Anstee, 1976b). It appears that the PAS-1 and PAS-2 bands have a dimer-monomer relationship, and that the site of association of the dimer is the hydrophobic domain in the polypeptide chain of the major sialoglycoprotein (Furthmayr and Marchesi, 1976). The PAS4 band also appears to be a complex, in this case a heterocomplex, consisting of the major sialoglycoprotein and PAS-3. Membranes from En (a-) cells, which lack the major sialoglycoprotein, do not yield the PAS-4 band (Tanner and Anstee, 1976b), and S - s human erythrocytes, which have an altered PAS-3 glycoprotein, also show alterations in the PAS4 band (Dahr et al., 1975c; Tanner et al., 1977). Presently available evidence suggests that polypeptide 3, the major sialoglycoprotein, PAS-S', and PAS-3 are all distinct entities. It

283

ERYTHROCYTE GLYCOPROTEINS

PAS-].

0.l!

0.1c

O.D.

POLYPEPTIDE

0.05

0

,i” 1

1

3

2

CM.

AIDNC

4

5

6

GEL

FIG. 1. Periodate-stainable glycoproteins of the human erythrocyte membrane. Human erythrocyte ghosts were separated by SDS gel electrophoresis on a gel containing 10%acrylamide with an overlay containing 4% acrylamide, using the buffer system described by Fairbanks et al. (1971). The gels were stained with the PAS stain and scanned at 560 nm. 0. D., Optical density.

is not known whether the remaining minor components are unique glycoprotein species or complexes of other glycoproteins. The apparent MWs obtained for these glycoproteins by the SDS gel electrophoresis method are unreliable. The major bands all yield apparent MWs which change with the acrylamide concentration used in the gel electrophoresis system (Bretscher, 1971b). Thus the monomer of the major sialoglycoprotein, the PAS-2 band, yields apparent MWs of 53,000 and 40,000 from 5% and 8% acrylamide gels, respectively (Tanner and Boxer, 1972), but the amino acid sequence and carbohydrate content of the protein suggest that it has a true MW of 31,000 (Tomita and Marchesi, 1975). These anomalies have been attributed to different extents of binding of detergent to the oligosaccharide-rich and the hydrophobic regions of these substantially gl ycosylated membrane proteins, compared with the binding of detergent to the polypeptide chain of the nonmembrane proteins used to calibrate MWs in the gel electrophoresis system (Grefrath and Reynolds, 1974; Tanford and Reynolds, 1976).

2 84

MICHAEL J. A. TANNER

2. ERYTHROCYTE MEMBRANESOF OTHER SPECIES

The periodate-stainable glycoproteins of erythrocytes from several other species have also been studied. The major periodate-stainable components of different species (presumably analogous to the human major sialoglycoprotein) show considerable variation in their apparent MWs on SDS gel electrophoresis (Lenard, 1970; Kobylka et al., 1972; Capaldi, 1973; Carraway et al., 1975; Ralston, 1975). It is not clear how much this variability in electrophoretic mobility reflects changes in carbohydrate content rather than MW. The rabbit erythrocyte is unusual in that it lacks any major periodate-stainable glycoprotein equivalent to the human sialoglycoprotein (Lodish and Small, 1975; Light and Tanner, 1977). Polypeptide 3 shows relatively little variation among species, although it has a somewhat higher MW in the camel erythrocyte (Ralston, 1975). 0. Non-Periodate-btoinobleGlycoproteins

Staining of glycoproteins with the PAS stain depends almost entirely on the presence of sialic acids in these glycoproteins (Dahr et aZ., 1974, 1976).The staining method used by Liao et aZ. (1973), utilizing mild periodate oxidation followed by reduction with radioactive borohydride, is also dependent on sialic acid and gives patterns which closely resemble those obtained with the PAS stain. This restricted specificity of the PAS staining technique has become increasingly apparent on examining the erythrocyte membrane with probes having a specificity for sugars other than sialic acid. When erythrocyte membranes are examined for galactose and N-acetylgalactosamine-containing oligosaccharides by oxidation with galactose oxidase followed by reduction with radioactive borohydride, a heterogeneous mixture of components is labeled in the region between the PAS-1 and PAS-2 bands (Steck, 1972a; Gahmberg and Hakomori, 1973; Steck and Dawson, 1974; Gahmberg, 1976). Many of these bands do not correspond to any of the bands detected with the protein or the PAS stain. Similar results are obtained when galactose and N-acetylgalactosaminespecific lectins (such as those from Ricinis communis and Phaseolus vulgaris) are used as probes (Tanner and Anstee, 1976a). It is generally assumed that these are glycoproteins, but some of them may be complex glycolipids containing large oligosaccharides of the type mentioned by Gardas and Koscielak (1974a,b). No estimates of the abundance of these components in the membrane are available, but they are generally considered minor components.

ERYTHROCYTE GLYCOPROTEINS

IV.

285

ORGANIZATION OF THE GLYCOPROTEINS IN THE ERYTHROCYTE MEMBRANE

Several reviews are available which cover the general subject of the organization of proteins in the erythrocyte membrane (Steck, 1974; Zwaal et al., 1973; Marchesi et al., 1976). As a group, the glycoproteins of the erythrocyte membrane all appear to b e integral membrane proteins and are hydrophobically bound to the lipid bilayer of the membrane. Thus these proteins remain associated with the lipid of the membrane when simple extraction procedures are used on erythrocyte ghosts (see for example, Hoogeveen et al., 1970; Fairbanks et al., 1971; Tanner and Boxer, 1972; Steck and Yu, 1973) but can be solubilized from the membrane by detergents (Yu et al., 1973), chaotropic agents (Winzler, 1969; Marchesi and Andrews, 1971), and certain organic solvents (Blumenfeld, 1968; Hamaguchi and Cleve, 1972; Anstee and Tanner, 1974a). A wide variety of impermeable protein-modifying probes has been used to define the location of these glycoproteins with respect to the lipid bilayer of the membrane (see Hubbard and Cohn, 1976, for a detailed review). The experimental approach generally follows that used by Bretscher (19714, Bender et al. (1971),and Phillips and Morrison (1971). Intact erythrocytes and various membrane preparations which are permeable or impermeable to the probe are exposed to the reagent. The proteins accessible to the probe in each of these preparations are then identified. Given a knowledge of the permeability of the membrane to the probe, and of the orientation or sidedness of the membrane preparations, it is possible to infer the location of the membrane proteins relative to the permeability barrier of the membrane. An interesting variation on this procedure is to compare the labeling patterns obtained b y using chemically analogous permeant and impermeant labeling agents on the intact erythrocyte (Whitely and Berg,

1974). Several major assumptions are implicitly made when using these types of chemical approaches to locate proteins. It is assumed that (1) the modification procedure does not itself alter the structure of the membrane, (2) the proteins have the same orientation in each of the membrane preparations being compared, and (3) the membrane is impermeable to the probe. In practice, it has been difficult to show rigorously that all these criteria have been met in many of the experimental systems used. Nevertheless, the use of a wide variety of protein-specific probes and of a carbohydrate-specific probe leads to the conclusion that all the erythrocyte glycoproteins are accessible at the

286

MICHAEL J. A. TANNER

external surface of the cell. The reagents used in this way include several low-MW compounds (Bender et al., 1971; Bretscher, 1971a,b,c; Whiteley and Berg, 1974; Staros and Richards, 1974; Cabantchik et al., 1975), proteases (Bender et al., 1971; Steck, 1972a), and lactoperoxidase (Phillips and Morrison, 1971). While all these probes are protein-specific, galactose oxidase has been used as a carbohydratespecific probe (Steck, 1972a; Gahmberg and Hakomori, 1973; Steck and Dawson, 1974). Ambiguities of interpretation which arise when these methods are used, in the case of proteins that span the membrane and are exposed at both membrane surfaces, have been overcome by showing that unique regions of polypeptide chain are accessible at each membrane surface (Bretscher, 1971b,c; Boxer et al., 1974), and by using dual isotope labeling of different membrane preparations (Reichstein and Blostein, 1973, 1975; Mueller and Morrison, 1974; Shin and Carraway, 1974). Polypeptide 3 and the major sialoglycoprotein have both been shown to penetrate right through the membrane and span it in this way. There is evidence that the minor glycoproteins (PAS-2' and PAS-3) also span the membrane (Mueller and Morrison, 1974), but Marchesi et al. (1976) suggest that these minor glycoproteins are bound to the lipid bilayer but do not extend beyond the cytoplasmic surface of the membrane. The use of the carbohydrate-specific probe galactose oxidase in a similar experimental approach (Steck, 1972a; Gahmberg and Hakomori, 1973; Steck and Dawson, 1974) has led to the important conclusion that all the carbohydrate of the erythrocyte is present at the external surface of the cell, confirming the conclusions of morphological studies (Winzler, 1971). It has proved difficult to define unambiguously the organization and associations of the membrane glycoproteins in the plane of the membrane. The associations of the proteins in the membrane have been examined using a variety of chemical cross-linking procedures. These experiments suggest that polypeptide 3 may exist in a dimeric form in the membrane (Steck, 1972b; Wang and Richards, 1974), a possibility strengthened by the observation that the isolated protein assumes a dimeric form in the presence of a nonionic detergent (Yu and Steck, 197513). The remaining glycoproteins are markedly unreactive to cross-linking agents, but in this case the large amounts of carbohydrate associated with these proteins may impose steric restrictions on their ability to become cross-linked. The persistence of a dimeric form of the major sialoglycoprotein in the presence of SDS has led to the suggestion that this protein may also be in a dimeric state in the mem-

ERYTHROCYTE GLYCOPROTEINS

287

brane (Marchesi et al., 1976). The presence of an aggregate of the major sialoglycoprotein and PAS-3 (the PAS-4 band) under similar conditions could lead one to suppose that a complex of the major sialoglycoprotein and PAS-3 also exists in the membrane. However, there is no other evidence either supporting or refuting these two suggestions at the present time. Our knowledge of the longer-range organization of these glycoproteins comes mainly from morphological studies utilizing the electron microscope. These have shown that a variety of erythrocyte antigens and receptors is present in a random array at the surface of the erythrocyte (Pinto da Silva et al., 1971; Tillack et al., 1972; Nicolson, 1973; Pinto da Silva and Nicolson, 1974). Freeze-cleavage of erythrocyte membranes exposes particles (intramembranal particles) which are probably proteinaceous and are embedded in the lipid bilayer (Pinto da Silva and Branton, 1970; Pinto da Silva and Nicolson, 1974). A large number of particles is exposed on the cleavage faces retaining the cytoplasmic half of the bimolecular lipid leaflet (the A or P F face), and these appear to be relatively homogeneous in size. The cleavage faces retaining the extracellular half of the bimolecular lipid leaflet (the B or E F face) are less densely populated with particles, and these vary a great deal in size. Most of the studies reviewed here relate to the particles on the A fracture face. The abundance, properties, and location of the major sialoglycoprotein and polypeptide 3 suggest that these glycoproteins might be components of the intramembranal particles. Studies of the distribution on the membrane of various receptors known to be associated with the major sialoglycoprotein and polypeptide 3 have shown them to be correlated with the intramembranal particles and have led to the suggestion that both these glycoproteins are components of the intramembranal particles (Pinto da Silva and Nicolson, 1974; Nicolson, 1976). However, there are some ambiguities in this interpretation, particularly with regard to the major sialoglycoprotein, since non of these receptors or markers have been shown to be exclusively associated with the major sialoglycoprotein. The markers used for the major sialoglycoprotein include blood-group-A antigenic activity (Pinto da Silva et al., 1971), phytohemagglutinin receptors (Tillack et al., 1972), and anionic sites (Pinto da Silva et al., 1973). However, recent evidence suggests that ABO(H) blood group antigens are not present on the major sialoglycoprotein (Hamaguchi and Cleve, 1972; Brennessel and Goldstein, 1974; Anstee and Tanner, 1974a,b, 1975), while phytohemagglutinin binds to polypeptide 3 and other membrane components in addition to the major sialoglycoprotein (Tanner and Anstee, 1976a).

288

MICHAEL J. A. TANNER

Sialoglycolipids and the minor periodate-stainable glycoproteins also provide anionic sites which are available at the cell surface (Swee'ley and Dawson, 1969; Marchesi et al., 1976). Bachi et al. (1977) showed that there are no significant differences in the distribution and density of intramembranal particles in membranes from normal erythrocytes and those from erythrocytes which lack the major sialoglycoprotein. A proportion of polypeptide-3 molecules appears to carry concanavalin-A (Con-A) receptors (Findlay, 1974; Tanner and Anstee, 1976a; Jenkins and Tanner, 1977b), and the results of experiments in which the binding of concavavalin A-ferritin conjugates was studied suggest that polypeptide 3 is a component of the intramembranal particles (Pinto da Silva and Nicolson, 1974). However, there have recently been suggestions that polypeptide 3 may not be entirely located in the intramembranal particles (Howe & Bachi, 1973; Bachi & Schnebli, 1975). Many problems remain in transferring the information from freezecleavage replicas to a detailed model of the architecture of the erythrocyte membrane. It is not known whether the intramembranal particles represent intact proteins or only the hydrophobic regions of proteins (Bretscher and Raff, 1975). Nor is it possible to assess the extent of heterogeneity of the particles in the A fracture face and their physical and chemical relationship to the particles in the B fracture face. V.

STRUCTURE

OF THE GLYCOPROTEINS

A. The Major Human Sialoglycoprotein (Glycophorin A)

1. ISOLATIONAND STRUCTUREOF THE HUMAN SIALOGLYCOPROTEIN

This glycoprotein carries the human erythrocyte blood group-M and -N antigens and, because of its immunochemical interest, has been the subject of study for many years. The earliest preparations were obtained by methods based on phenol extraction (Klenk and Uhlenbruck, 1960; Kathan et al., 1961; Springer et al., 1966; Winzler, 1969). More recently developed isolation methods include the extraction of erythrocyte ghosts with an organic solvent such as pyridine (Blumenfeld, 1968; Zvilichovsky et al., 1971),chloroform-methanol (Hamaguchi and Cleve, 1972), or butanol (Anstee and Tanner, 1974a). In each case, the major sialoglycoprotein is extracted in a water-soluble form,

ERYTHROCYTE G LYCOPROTEI NS

289

while the remaining erythrocyte ghost protein is precipitated, and the bulk of the lipid is extracted into the organic phase or remains associated with the insoluble protein. The sialoglycoprotein preparations used by Marchesi and co-workers for structural studies were prepared by a procedure based on the use of lithium diiodosalicylate (Marchesi and Andrews, 1971). The minor PAS-stainable components behave in much the same way as the sialoglycoprotein during all these different isolation procedures and this, together with the tendency of the sialoglycoprotein to form aggregates during preparation, has complicated its purification. Nevertheless, by applying suitable further purification steps after the extraction, homogeneous preparations of the sialoglycoprotein can be obtained (Zvilichovsky et al., 1971; Tanner and Boxer, 1972; Hamaguchi and Cleve, 1972; Anstee and Tanner, 1974a; Furthmayr et al., 1975). The purified human major sialoglycoprotein contains about 60% carbohydrate by weight (Winzler, 1969) and is particularly rich in sialic acid (approximately 25% by weight). This protein is the major carrier of cell surface carbohydrate in the erythrocyte. The subunit MW of the protein has proved difficult to ascertain by physical methods, because of the anomalous detergent-binding effects and the presence of different aggregation states. [Marchesi et al. (1976) discuss this aspect more fully.] But the MW derived from amino acid sequence data and the carbohydrate content is 31,000 (Tomita and Marchesi, 1975). The amphiphilic nature of the major sialoglycoprotein was first recognized by Morawiecki (1964), and this property was confirmed and its significance elaborated on by Winzler (1969). Treatment of watersoluble preparations of the sialoglycoprotein with trypsin yields an insoluble peptide which contains little or no carbohydrate and a soluble sialic acid-rich glycopeptide. A similar sialic acid-rich glycopeptide was obtained by treatment of intact erythrocytes with trypsin. Winzler (1969) suggested that oligosaccharide chains were present only in the extracellular region of the protein (the N-terminal segment), while the insoluble C-terminal region was hydrophobic and interacted with the lipid bilayer of the membrane. This orientation of the major sialoglycoprotein (Fig. 2 ) has been confirmed using various labeling techniques (Bretscher, 1971b; Segrest et al., 1973), and it has also been shown that the C-terminus of the protein is exposed at the cytoplasmic face of the membrane (Bretscher, 1975; Mueller and Morrison, 1974). The complete amino acid sequence of the major sialoglycoprotein has been recently determined by Marchesi and co-workers (Tomita and Marchesi, 1975).Figure 3 shows this sequence and also shows the

290

MICHAEL J. A. TANNER

MEMBRANE

I

FIG.2. Orientation of the major sialoglycoprotein in the human erythrocyte membrane. The continuous line represents the polypeptide backbone and the solid circles attached to this line indicate the oligosaccharide chains.

distribution of oligosaccharides on the amino acid chain. The amphiphilic nature of the protein is evident, the N- and C-terminal portions of the molecule being separated by a segment of about 20 nonpolar amino acids which probably penetrates the lipid bilayer of the membrane (Segrest et al., 1972). The adjacent regions on the N- and C-terminal sides of this segment contain clusters of charged residues which may interact with hydrophilic components at the surfaces of the membrane (Marchesi et al., 1976). The asymmetry in the distribution of oligosaccharides on this molecule is striking. The protein contains 16 oligosaccharide units. Fifteen are relatively small sialic acid-rich units which are O-glycosidically linked to serine or threonine residues (sialotetrasaccharides). The remaining carbohydrate is present in a larger and more complex mannose- and N-acetylglucosamine-rich oligosaccharide which is N-glycosidically linked to the protein via an asparagine residue. The polypeptide is particularly densely substituted with sialotetrasaccharides at the extreme N-terminus of the sialoglycoprotein where each of the residues from positions 10 to 15 carries one of these oligosaccharide units. The majority of the O-glycosidically linked oligosaccharides appear to be tetrasaccharides of the type shown in Fig. 4A (Thomas and Winder, 1969), but incomplete forms of this oligosaccharide are probably also present. It is possible that some of the oligosaccharides may have the related structure shown in Fig. 4B (Springer and Desai, 1975). The structure of the large N-glycosidically linked oligosaccharide remains uncertain. Two structures have been proposed for

291

ERYTHROCYTE GLYCOPROTEINS

I LEU

Ism 1

10

20

30

40

50

60

90

100

110

I20

I30

FIG.3. Amino acid sequence and carbohydrate distribution on the major human sialoglycoprotein. The solid circles indicate the residues in the hydrophobic segment which probably passes through the lipid bilayer. The attached diamond shapes indicate the positions of the O-glycosidically linked oligosaccharides, while the position of the N-glycosidically linked oligosaccharide is shown by the hexagonal shape. (From Marchesi e t al., 1976, reproduced with permission.)

A -

NANA %GAL

B -

NANA

&GAL

3@

GALNA~

% SENTHR)

@

FIG.4. Proposed structures of O-glycosidically linked oligossaccharides. (A) Structure suggested by Thomas and Winzler (1969).(B)Structure proposed by Springer and Desai (1975).

292

MICHAEL J. A. TANNER NAM

&a

2-6

GAL

GAL

&B

J, B 1-3(4) CLCNAc

1-3(4)

GLCNAc

ASN

FIG.5. Structure proposed by Kornfeld and Kornfeld (1970)for the N-glycosidically linked oligosaccharide of the major sialoglycoprotein.

this unit, one (Fig. 5) by Kornfeld and Kornfeld (1970) and a second (Fig. 6) b y Thomas and Winzler (1971).Marchesi et al. (1976) suggest that this oligosaccharide is larger and contains more N-acetylglucosamine than either of these proposed structures. Interpretation of the results of the earlier workers is complicated by the possibility that their sialoglycoprotein preparations were contaminated by other glycoprotein components. Marchesi et al. (1976)also suggest that the microheterogeneity of their sialoglycoprotein preparations could give rise to misleading results with regard to its composition.

2. ANTIGENS AND RECEPTORS PRESENTON THE HUMANERYTHROCYTE SIALOGLYCOPROTEIN

The blood group-M and -N antigens of the human erythrocyte are located on the major sialoglycoprotein. The isolated protein has potent M and N antigenic activity, and antigenically active glycopeptides can be obtained after proteolytic digestion (Lisowska and Jeanloz, 1973). However, when the 0-glycosidically linked oligosaccharides are released from the polypeptide by alkali treatment, neither the released oligosaccharides nor the remaining polypeptide have antigen activity (Lisowska, 1969).Treatment of the glycoprotein with amino group-blocking reagents (Lisowska and Morawiecki, 1967; Lisowska and Duk, 1975)results in the loss of antiFUC JCY

1-2(6)

GAL

.le

GLCNAc

-&

GAL

NANA

GLCNAc

GAL

&B J B

(MAN)3&

4 3.8 GLCNAc

4

GLCNAC

.l+B

ASN

r l G . 6. Structure proposed by Thomas and Winzler (1971) for the N-glycosidically linked oligosaccharide of the major sialoglycoprotein.

ERYTHROCYTE GLYCOPROTEINS

293

genic activity, suggesting that residues of the polypeptide chain as well as carbohydrate moieties are part of the antigenic determinant. Studies using periodate oxidation and a variety of M- and N-specific lectins suggest that one of the sialotetrasaccharides of the type shown in Fig. 4A is part of the M antigenic determinant (Dahr et al., 1975b). However, Springer and co-workers (Springer et aZ., 1966; Springer and Desai, 1974) suggest that the oligosaccharide shown in Fig. 4B is part of the antigenic determinant. The structural basis for the difference between the allelic M and N antigens has been the subject of some controversy. No difference has been found in the composition of the 0- and N-glycosidically linked oligosaccharides from blood-group-M and -N cells (Thomas and Winzler, 1969, 1971; Adamany and Kathan, 1969; Kornfeld and Kornfeld, 1970),but it seems unlikely that the presence or absence of a single residue in a large number of oligosaccharide chains would have been detected by these workers. It has been suggested (Springer and Desai, 1974)that the N antigen contains an oligosaccharide of the type shown in Fig. 4B, but deficient in one terminal sialic acid, and that this structure is the substrate for a sialyltransferase dependent on the blood group-M gene. However, other studies do not support this view and suggest that the difference between the M and N antigens does not depend on sialic acid but reflects changes in the amino acid sequence of the polypeptide chain of the major sialoglycoprotein (Dahr et al., 1975b; Anstee et aZ., 1977).Recent studies have shown that the two amino acids found by Tomita and Marchesi (1975) at residues 1 and 5 in the amino acid sequence of the major sialoglycoprotein reflect differences in the amino acid sequence of the blood group M and N-active sialoglycoproteins in their pooled preparations (see Fig. 3 ) . The M glycoprotein contains serine and glycine at residues 1 and 5 while the N glycoprotein contains leucine and glutamic acid at these positions (Wasniouska et ul., 1977; Dahr et al., 1977).It seems likely that these amino acids are involved, at least in part, in distinguishing the M and N antigenic determinant. It has also been shown that in certain situations blood-group-M and -N antigenic activity can be found on membrane components other than the major sialoglycoprotein. Thus the traces of N activity found on homozygous M erythrocytes appears to be present on the PAS-3 glycoprotein (Hamaguchi and Cleve, 1972; Dahr et ul., 1975b), and some individuals whose erythrocytes lack the major sialoglycoprotein carry M antigen on membrane components other than the major sialoglycoprotein (Anstee et al., 1977). Several reports have associated blood-group-ABH and -I activity

2 94

MICHAEL J. A. TANNER

with the major sialoglycoprotein (Springer et al., 1966; Marchesi and Andrews, 1971; Fukuda and Osawa, 1973; Liao et al., 1973).These antigenic activities appear to be due to the presence of tightly bound contaminating glycolipids in these sialoglycoprotein preparations (Hamaguchi and Cleve, 1972; Brennessel and Goldstein, 1974; Anstee and Tanner, 1974a,b, 1975; Gardas and Koscielak, 1974a,b). The major sialoglycoprotein contains receptors for influenza virus (Winzler, 1971; Jacksonet al., 1973). In addition, since the protein carries both 0- and N-glycosidically linked oligosaccharides, receptors for a wide variety of lectins are present on the molecule. The oligosaccharide and carbohydrate residues which are binding sites for some of these lectins are summarized in Table I.

3. SIALOGLYCOPROTEINS OF OTHERSPECIES In the past, interest has been almost entirely directed toward the human sialoglycoprotein. However, the number of studies being done on analogous glycoproteins from other species is increasing (see Section III,A,2). Bovine glycoprotein (Capaldi, 1973; Emerson and Kornfeld, TABLE I

LECTINRECEPTORS ON THE MAJOR SIALOGLYCOPROTEIN~

SOME OLIGOSACCHARIDE

0-Glycosidically linked

N-glycosidically linked

Maclura aurantiaca (GalNAc)* Bauhinia purpurea (GalNAc)* Arachis hypogea (Cal)c

Ricinis communis (Gal)" Robinia pseudoaccncio (Gal)" Triticurn uulgaris (WGA, GlcNAcY Phaseolus uulgaris (Gal + Man, complexp Lens culinaris (GlcNAc + Man, complex)h

The sugar residues in parentheses are probably the major determinants for the binding of each lectin to these particular oligosaccharides. These assignments are mainly based on inhibition studies with monosaccharides. GalNAc, N-acetylgalactosamine; Gal, galactose; WGA, wheat germ agglutinin; GlcNAc,N-acetylglucosamine, Man, mannose. * Dahr et al. (1975a). Reacts after removal of terminal sialic acid (Lotan et al., 1975; Terao et al., 1975). Fukuda and Osawa (1973). Leseney et al. (1972). Jackson et al. (1973). Komfeld and Kornfeld (1970). Kornfeld et al. (1971). f

ERYTHROCYTE GLYCOPROTEINS

295

1976) contains 80% carbohydrate, and this is mainly present in the form of O-glycosidically linked oligosaccharides. However, the majority of these oligosaccharides differ from human sialotetrasaccharides in that they contain N-acetylglucosamine and have a lower content of sialic acid. Glockner et al. (1976) examined the carbohydrate content of the glycoproteins from a variety of species and found sialotetrasaccharides similar to human ones in many species. However, bovine and porcine glycoproteins contained much lower amounts of these components than the glycoproteins of the other animals examined. 6. The Minor Periodate-Stainable Proteins of the Human Erythrocyte

1. PAS-2’ This glycoprotein component was not detected until recently because it has a mobility on SDS gel electrophoresis similar to that of the PAS-2 form of the major sialoglycoprotein. Mueller and Morrison (1974) recognized this component because, on lactoperoxidase radioiodination, it labeled more strongly from the cytoplasmic side of the erythrocyte membrane than from the extracellular side. The converse is true for the PAS-1 and PAS-2 forms of the major sialoglycoprotein. The uniqueness of this component has been confirmed by studies on the glycoproteins of cells which lack the major sialoglycoprotein (Tanner and Anstee, 1976b) and cells which are defective in the PAS-3 glycoprotein (Dahr et al., 1975c; Tanner et al., 1977). PAS-2’ takes up about 10% as much of the periodate stain as the major sialoglycoprotein. The fact that it (PAS-2’) contains receptors for lectins from Maclura aurantiaca and Arachis hypogea (Tanner and Anstee, 1976a; Anstee et al., 1977),as well as sialic acid as demonstrated by PAS staining, suggests that O-glycosidically linked sialic acid-rich oligosaccharides, similar to the sialotetrasaccharides found in the major sialoglycoprotein, are also present in this molecule. In addition, it binds P . vulgaris phytohemagglutinin and probably also contains Nglycosidically linked carbohydrates. The studies of Dahr et al. (1975d) on erythrocytes from individuals with the Tn syndrome are consistent with this conclusion. No data are available on the abundance or composition of PASS’, but it is usually regarded as a minor component of the erythrocyte membrane. Although Fujita and Cleve (1975)isolated a sialic acid-rich glycoprotein with an electrophoretic mobility similar to that of PAS-2’, this preparation did not contain the N-acetylga-

2 96

MICHAEL J. A. TANNER

lactosamine which might be expected from the reactivity of PAS-2' with lectins. 2. PAS3 There are conflicting reports about the disposition of this protein in the erythrocyte membrane. It is certainly accessible at the extracellular surface of the membrane, and Mueller and Morrison (1974) also reported that it could be labeled by the lactoperoxidase iodination technique from the cytoplasmic side of the membrane and thus spans the membrane. However, Marchesi et al. (1976) suggest that the protein does not project through the cytoplasmic surface of the membrane. The protein stains with about 10% of the intensity of the major sialoglycoprotein on PAS staining and has receptors for M . aurantiaca and A. hypogea lectins as well as wheat germ agglutinin, but no receptors for P. vulgaris and €3. communis lectins (Robinson et al., 1975; Tanner and Anstee, 1976a; Anstee et al., 1977). Thus 0-glycosidically linked sialic acid-rich sialotetrasaccharides of the type found in the major sialoglycoprotein are probably present, while N-glycosidically linked oligosaccharides, if present, may be different from those found on the major sialoglycoprotein. Furthmayr et al. (1975) have published analytical data on a crude preparation of PAS-3. The carbohydrate content of their preparation is consistent with the above interpretation, and their results suggest that PAS-3 represents 5-10% by weight of the total periodate-stainable glycoproteins. Fujita and Cleve (1975) also isolated a glycoprotein similar in electrophoretic mobility to PAS-3. However, their glycoprotein lacks N-acetylgalactosamine, while the data of Furthmayr et al. (1975) and the lectin-binding properties of PAS-3 suggest that this sugar is present. A similar inconsistency has been noted in the carbohydrate content found by Fujita and Cleve (1975) for a glycoprotein similar to PAS-2' (see Section V,B,l). It has been suggested that PAS-3 has a carbohydrate composition similar to that of the major sialoglycoprotein, but that the polypeptide portions differ. Marchesi et al. (1976) also report that PAS-3 is similar to the sialoglycoprotein in having a glycosylated N-terminal region and a hydrophobic polypeptide segment, but it lacks the hydrophilic intracellular C-terminal segment found in the major sialoglycoprotein. In addition to lectin receptors, PAS-3 probably carries the erythrocyte Ss antigens (Hamaguchi and Cleve, 1972; Fujita and Cleve, 1975; Anstee and Tanner, 1975). Cells lacking these antigens have an altered PAS-3 (Dahr et al., 1975c; Tanner et al., 1977). There is some evidence that the small amounts of N antigen found in homozygous M

ERYTHROCYTE GLYCOPROTEINS

297

erythrocytes are also present on this component (Dahr et al., 1975b; Anstee and Tanner, 1975). However, it seems likely that the ABH and I antigenic activities found associated with this glycoprotein (Hamaguchi and Cleve, 1972; Fujita and Cleve, 1975)are the result of tightly bound contaminating glycolipids (Gardas and Koscielak, 1974a,b; Anstee and Tanner, 1975). In the intact erythrocyte PAS-3 resists trypsin treatment but can be degraded with chymotrypsin (Dahr et al., 197513). The erythrocyte Duffy antigens (Fy” and Fyb) have similar properties (Miller et al., 1975), and the F y a antigen has chromatographic properties similar to those of PAS-3, although so far there is no evidence to show that it is the same as PAS-3 (Anstee and Tanner, 1975).This is of interest, since it has been shown that erythrocytes lacking Duffy antigens are resistant to infection by Plasmodium uivax, a human malaria (Miller et al., 1975). C. Polypeptide 3

Polypeptide 3 is the most abundant of the erythrocyte membrane proteins, comprising about lo6copies per erythrocyte (25% of the coomassie blue-stainable proteins, Steck, 1974). This glycoprotein migrates as a characteristically diffuse band on SDS gel electrophoresis, the front edge of the band having an apparent MW of 86,000-90,000, and it stains very weakly with the PAS stain (Fig. 1).Much evidence suggests that this protein spans the erythrocyte membrane (see Section IV) and that it is an integral membrane protein. Unlike the major sialoglycoprotein, polypeptide 3 has not yet been obtained in a water-soluble form in the absence of detergents, and this has hindered purification of the protein. However, several methods for purifying polypeptide 3 have recently been reported. Most of these methods utilize various types of preliminary extraction procedures performed on erythrocyte ghosts to remove the bulk of the extrinsic membrane proteins, followed by solubilization in a detergent such as Triton X-100 (Yu et al., 1973; Yu and Steck, 1975a; Furthmayr et d.,1976) or SDS (Tanner and Boxer, 1972; Ho and Guidotti, 1975; Tanner et al., 1976). Although apparently native protein is isolated by solubilization in nonionic detergents (Yu and Steck, 1975a; Drickamer, 1976), solubilization in SDS has proven convenient for obtaining large amounts of protein for structural studies (Tanner et al., 1976; Jenkins and Tanner, 1977b). Affinity chromatography methods have also been used to purify polypeptide 3 (Findlay, 1974; Adair and Kornfeld, 1974).

298

MICHAEL J. A. TANNER

The isolated protein is rich in leucine and glutamic acid and/or glutamine and, although the amino acid composition is more hydrophobic than that of typical water-soluble proteins, it is less hydrophobic than that of some other integral membrane proteins (Yu and Steck, 1975a). Unlike the major sialoglycoprotein, which has no sulfhydryl groups, polypeptide 3 contains about six half-cystine residues (Jenkins and Tanner, 1977b), at least one of which is in the reduced form (Steck, 197213).There is no evidence for the presence of any form of covalent interchain cross-links in the protein (Bailey et al., 1976; Jenkins and Tanner, 1977b). The protein is a glycoprotein (Ho and Guidotti, 1975; Yu and Steck, 1975a; Tanner et al., 1976; Gahmberg et al., 1976) with a distribution of sugars quite different from that found in the major sialoglycoprotein (Table 11). The small amounts ofN-acetylgalactosamine and sialic acid suggest that few, if any, sialic acid-rich 0-glycosidically linked oligosaccharides are present and that most of the carbohydrate is linked N-glycosidically to the protein via asparagine residues. The protein contains 15% carbohydrate, and it has been estimated that it carries about 18% of the surface carbohydrate of the erythrocyte (Tanner et al., 1976; Jenkins and Tanner, 1977b). Thus it contributes about one-third as much carbohydrate as the major sialoglycoprotein to the total cell surface carbohydrate. The protein contains receptors for Con A and lectins from P . uulgaris and R. communis (Findlay, 1974; Adair and Kornfeld, 1974; Tanner and Anstee, 1976a) but does not have receptors for any of the lecTABLE I1

CARBOHYDRATE CONTENTSOF POLYPEPTIDE 3 MAJOR SIALOGLYCOPROTEIN

AND THE

Mole carbohydrate per mole protein Component

Major sialoglycoprotein"

Polypeptide 3b

Fucose Mannose Galactose N-Acetylglucosamine N-Acetylgalactosamine Sialic acid

2 5 18 9 15 18

4 7

a

Based on data from Furthmayr et al. (1975).

* Tanner et al. (1976).

2A 25 4 5

ERYTHROCYTE GLYCOPROTEINS

299

tins which bind to-O-glycosidically linked oligosaccharides such as those from M . uurantiaca and A. hypogea (Tanner and Anstee, 1976a; Anstee et al., 1977).This confirms the absence of the O-glycosidically linked type of oligosaccharide from the molecule. It can also be labeled using the galactose oxidase technique (Gahmberg and Hakomori, 1973; Steck and Dawson, 1974; Yu and Steck, 1975a). The apparent MW of polypeptide 3 obtained from SDS gel electrophoresis may be unreliable, not only because it is glycosylated but also because it is an intrinsic membrane protein. It has been shown that both the heavily glycosylated segments and the hydrophobic segments of membrane proteins behave anomalously in binding SDS (Grefrath and Reynolds, 1974; Tanford and Reynolds, 1976)compared with the water-soluble proteins used as MW standards in this system. The broadness of the polypeptide3 band further complicates MW estimation and has led to suggestions that polypeptide 3 may be heterogeneous, with regard to either its polypeptide chains or its carbohydrates. It is true that several other erythrocyte membrane proteins migrate in the same region as polypeptide 3. These include acetylcholinesterase (Bellhorn et a1., 1970),the phosphorylated intermediate of Na+,K+-activatedATPase (Avruch and Fairbanks, 1972), and a membrane-penetrating protein of MW 90,000 (Reichstein and Blostein, 1975; Jenkins and Tanner, 1977a). However, the amounts of these components are sufficiently small that it is unlikely that they represent significant contamination of polypeptide3 preparations in a protein chemical sense. The behavior of the protein on proteolysis (Bender et ul., 1971), oxidative dimerization (Steck, 1972b), and cleavage with a variety of reagents (Steck et al., 1976; Drickamer, 1976; Jenkins and Tanner, 1977a) suggests that the polypeptide is homogeneous, or else that the polypeptide3 band contains a family of very closely related proteins. If any such heterogeneity exists, studies on fragments of the protein suggest that it is likely to be present in the C-terminal region of the molecule (Jenkins and Tanner, 197713). An increasing amount of evidence suggests that there is heterogeneity in the oligosaccharides present on polypeptide 3 and that the diffuseness of the band obtained on SDS gel electrophoresis is associated with this heterogeneity. The leading and trailing edges of the band differ in reactivity toward galactose oxidase (Yu and Steck, 1975a) and ability to bind Con A (Tanner and Anstee, 1976a). Proteolytic removal of the region of the protein most heavily glycosylated yields a large protein fragment which migrates as a much sharper band on SDS gel electrophoresis. A C-terminal fragment of polypeptide 3 which retains this glycosylated region yields a more diffuse

300

MICHAEL J. A. TANNER

band than intact polypeptide 3, and the components in this fragment preparation differ in their ability to bind Con A and lectins from P . uulgaris and R. communis (Jenkins and Tanner, 1977b).This carbohydrate heterogeneity appears to be restricted to the C-terminal region of the protein. Polypeptide 3 in the intact erythrocyte is much more resistant to proteolysis than the major sialoglycoprotein, Thus it is not cleaved by trypsin (Triplett and Carraway, 1972; Boxer et al., 1974), while treatment of erythrocytes with a variety of other proteases (Bender et al., 1971; Bretscher, 1971a; Boxer et al., 1974; Jenkins and Tanner, 1975) yields a fragment which remains reactive toward extracellularly applied reagents. The resistance of both the extracellular and intracellular regions of the protein to proteolysis depends on the ionic strength of the medium. Both regions are much more resistant to trypsin at isoosmotic ionic strengths than at low ionic strengths. This suggests that the structure of the protein depends on the ionic strength of the medium (Jenkins and Tanner, 1977a). Changes in the accessibility of tyrosine residues of the protein to lactoperoxidase radioiodination support this conclusion. The protein probably has a tighter structure at ionic strengths similar to those found in vivo than under the lowionic-strength conditions usually used for preparing and manipulating erythrocyte ghosts. It is possible that some of the functional properties associated with polypeptide 3 depend on the ionic strength of the medium in which they are studied. Since polypeptide 3 is involved in erythrocyte anion transport, the way in which the polypeptide is folded in the membrane has been of interest. Some intrinsic membrane proteins such as the major erythrocyte sialoglycoprotein and microsomal cytochrome b, appear to contain a singIe hydrophobic polypeptide segment which interacts with the lipid bilayer of the membrane (Segrest et al., 1972; Tomita and Marchesi, 1975; Spatz and Strittmatter, 1971; Visser et al., 1975). However, there is no reason to suppose that the purpose of the hydrophobic segment in this class of protein is other than to lock the protein into the membrane. Membrane proteins involved in transport are likely to have a much greater proportion of their polypeptide within the lipid bilayer, since the structures which allow penetration of the substrates through the membrane must be constructed from these intramembranous regions. Thus a protein of this class, the light-dependent proton pump bacteriorhodopsin, contains seven helical rods which traverse the membrane in a porelike structure (Henderson and Unwin, 1975).This protein has a MW of 25,000 (Bridgen and Walker, 1976) and is almost entirely embedded in the membrane. An a-helical

ERMHROCME GLYCOPROTEINS

301

rod needs to be only 20 to 30 amino acids long to traverse a membrane (Tanford and Reynolds, 1976). Polypeptide 3 contains about 750 amino acid residues, and a number of traverses of the polypeptide chain similar to that found in bacteriorhodopsin would not be unexpected and would require only about one-fourth of the polypeptide to be embedded in the membrane. We studied the folding of polypeptide 3 in the erythrocyte membrane by utilizing the tyrosine sites on the protein which can be radioiodinated with lactoperoxidase as markers for the location of particular portions of the polypeptide with respect to the membrane. These tyrosine-containing sites were distinguished by their mobility on peptide maps after thermolysin digestion of the radioiodinated protein. Unique sets of these sites are present in the extracellular and intracellular regions of the protein (Boxer et al., 1974). Two distinct labeled peptides were obtained from polypeptide 3 on proteolysis of erythrocyte membranes, and each of these peptides contained a set of extracellular labeled sites linked to a different and nonoverlapping set of intracellular labeled sites (Jenkins and Tanner, 1975, 1977a).Thus the polypeptide traverses the membrane at least twice. It should be noted that this type of technique can only yield a minimum figure for the number of traverses of the polypeptide chain, for two main reasons. The method used for determining the sidedness of individual peptide segments may not detect all the different polypeptide sections exposed at the membrane surface, either because of limited accessibility of potential labeled sites or because appropriate residues of the reactive amino acid are not randomly distributed throughout the length of the polypeptide. Similarly, it is not possible to establish, a priori, that peptides containing only a single traverse are obtainable from all the traverses under any given set of cleavage conditions. Thus in practice it is very difficult to show that a transmembrane segment of a protein does not cross the membrane more than once without the aid of detailed amino acid sequence data. Our studies also suggest that the tyrosine sites in the extracellular region of polypeptides are duplicated. Each of the regions containing a set of these duplicated extracellular sites is separated by an intracelMar region of the polypeptide chain (Jenkins and Tanner, 1975, 1977a).It is difficult to be certain that this interpretation has not been made in error as a result of the presence of overlapping partial digestion products. However, the two sets of sites behave differently on radioiodination at isoosmotic and low ionic strength. The presence of two sets of sites is consistent with the earlier results of Bretscher (19714, who used a labeling reagent with a different amino acid spec-

302

MICHAEL J. A. TANNER

ificity. The extent and significance of the amino acid sequence duplication these sites represent are unknown, but it may reflect some symmetry in the molecule, possibly at the anion-binding site of the protein. The relatively low inhibitory potency of sulfanilic acid derivatives toward erythrocyte anion transport compared with that of their " dimeric" forms-diaminostilbene disulfonic acid derivatives-lends some weight to this suggestion (Ho and Guidotti, 1975; Cabantchik and Rothstein, 1972, 1974a). The composition and terminal group analysis of the isolated protein, and fragments of it, suggest that the protein has a blocked Nterminus and that the termini are oriented in the membrane as shown in Fig. 7. Oligosaccharides are located in three regions of the protein. The most highly glycosylated portion, which contains lectin receptors, is present in the C-terminal region of the protein (Jenkins and Tanner, 1977b). Drickamer (1976) suggests that the protein may have valine at the C-terminus, while Jenkins and Tanner (197713) believe that C-terminal leucine is present. However, rather low yields of Cterminal amino acids were obtained in both cases. Several other laboratories have recently reported the results of other structural studies on polypeptide 3 (Steck et al., 1976; Drickamer, 1976). These reports suggest that the N-terminus of the protein is on the cytoplasmic side of the membrane rather than extracellular, as shown in Fig. 7, and that the polypeptide traverses the membrane only once. We feel that the content and known sidedness of lactoperoxidase-labeled sites in fragments obtained from this part of the protein unambiguously show that it is a transmembrane segment of the

FIG.7. Structure of polypeptide 3 in the human erythrocyte membrane. The continuous line represents the polypeptide backbone, while the solid circular shapes represent the attached carbohydrates. These solid shapes are intended to indicate the relative amounts of carbohydrate in each region and not exact numbers of oligosaccharide chains. The general location of several binding sites is also shown.

ERYTHROCYTE GLYCOPROTEINS

303

protein which is distinct from the other membrane-penetrating segment. The model shown in Fig. 7 is the simplest one consistent with our results. The N-terminal part of the chain could of course penetrate the membrane again and protrude from the cytoplasmic side of the membrane, leaving lactoperoxidase-labeled sites in an extracellular loop. Drickamer (1976) was unable to detect any labeling by lactoperoxidase in this region of the protein. This may be a result of the lower concentrations of lactoperoxidase used during labeling in his study, as even under our labeling conditions this N-terminal region was only weakly labeled. A situation similar to this exists with respect to lactoperoxidase labeling of the tyrosine residue of the major sialoglycoprotein which is situated adjacent to the hydrophobic segment in the cytoplasmic region of the protein. This tyrosine residue is relatively unreactive toward labeling, and it is possible that the labeling obtained is due to the generation of excess iodine radicals (Marchesi et al., 1976). Nevertheless this tyrosine residue appears to be located on the cytoplasmic side of the membrane permeability barrier. Even if labeling by excess iodine radicals occurring in polypeptide 3 under our labeling conditions, we showed that under these conditions the sites iodinated by extracellularly located lactoperoxidase are distinct from and do not overlap with the sites labeled by intracellularly located lactoperoxidase (Boxer et al., 1974). Thus the sidedness of labeling is retained under these conditions. The most reasonable interpretation of the distinct domains of labeling observed is that this reflects labeling at each surface of the membrane. In this connection it is interesting that the same tyrosine sites are labeled in polypeptide 3 when intact erythrocytes are labeled either with '251Cl or lactoperoxidase, although some additional sites are labeled by '251Clwhich are not labeled by lactoperoxidase in any membrane preparation. When IC1 is used to label erythrocyte ghosts, new sites of labeling occur in polypeptide 3 , and among these new sites is a group of peptides labeled b y lactoperoxidase only from the cytoplasmic side of the membrane (Tanner, unpublished observations). The apparent sidedness of IC1 labeling of the intact erythrocyte may result in part from the destruction of any intracellular labeling reagent b y the reducing environment inside the erythrocyte, and perhaps also from the presence of a "sink" of intracellular protein which can compete with the polypeptide cytoplasmic region for the reagent. In this case, reduced glutathione, a major source of cellular reducing power, may define the apparent permeability barrier by its ready accessibility to the cytoplasmic surface of the membrane. It has not been possible to identify positively any intramembranously located tyrosine resi-

304

MICHAEL J. A. TANNER

dues using lZ5ICllabeling. Hubbard and Cohn (1976) have discussed

the general problem of labeling studies with permeant reagents in more detail. (See also Juliano, this volume.) The mechanism of biosynthesis of a protein such as polypeptide 3 is obscure. This problem also applies to other intrinsic membrane proteins of this type, such as bacteriorhodopsin, in which the peptide backbone crosses the membrane more than once. Many of the current concepts of the biogenesis of membrane proteins are based on studies of the transfer of secreted proteins across membranes (see Sturgess et al., this volume). No experimental data are available on the biosynthesis of intrinsic membrane proteins which contain multiple traverses of the polypeptide across the membrane. Yu and Steck (1975b)have reported the results of some interesting studies which suggest that isolated, nondenatured polypeptide 3 behaves as a dimer in nonionic detergent solutions. The dimeric polypeptide-3 complexes were able to bind two tetrameric erythrocyte glyceraldehyde-3-phosphate dehydrogenase molecules. A tryptic peptide of MW 22,000, derived from the cytoplasmic region of polypeptide 3, was also able to bind tetrameric glyceraldehyde-3-phosphate. They also provided evidence which suggests that polypeptide 3 remains associated with band 4.2 in nonionic detergent solutions. The latter component (4.2) is probably itself tetrameric (Steck, 1972b).

VI.

FUNCTIONS OF GLYCOPROTEINS

Apart from polypeptide 3, there is no evidence to suggest that other periodate-stainable glycoproteins have a role in any of the enzymic or pseudoenzymic (e.g., transport) processes which occur in the erythrocyte. A. Polypeptide 3

Erythrocyte anion transport is an exchange transport system proba-

bly utilized in vioo for the exchange of C1- and HC0,- across the erythrocyte membrane: The system has a broad specificity and carries both mono- and divalent anions, but equilibrates small monovalent anions such as C1- particularly quickly (Tosteson, 1959; Passow and Wood, 1974). Even large organic anions appear to penetrate the membrane through this system, albeit slowly (Cabantchik et al., 1975, 1976; Staros et al., 1975).

ERYTHROCYTE GLYCOPROTEINS

305

Evidence suggesting an involvement of polypeptide 3 in the erythrocyte anion transport system is based mainly on the use of inhibitors of transport which covalently react with erythrocyte membrane proteins. One of the most potent inhibitors of erythrocyte anion disulfonate (DIDS) (Catransport is 4,4’-diisothiocyano-2,2’-stilbene bantchik and Rothstein, 1972). The binding of a radioactively labeled reduced form of DIDS to polypeptide 3 correlates well with the inhibition of anion transport, polypeptide 3 being essentially the only component bound by the compound (Cabantchik and Rothstein, 1974a,b; Lepke et al., 1976). This compound does not penetrate the membrane, and it binds to an extracellular region of polypeptide 3 between the cleavage site for extracellularly applied pronase and intracellularly applied trypsin (see Fig. 7; Cabantchik and Rothstein, 1974b; Lepke and Passow, 1976). About 1 x lo6 sites per cell are labeled with DIDS, that is, approximately one molecule of inhibitor per polypeptide-3 monomer (Lepke et al., 1976; Ship et al., 1977). The conclusion that polypeptide 3 is involved in erythrocyte anion transport has been confirmed in studies using a related, less potent inhibitor, 1-isothiocyano4benzenesulfonate (Ho and Guidotti, 1975). Vesicles containing polypeptide 3 as a major component are also able to transport anions (Rothstein et al., 1975; Cabantchik et al., 1977)and recently Ross and McConnell(l977) reconstituted the anion transport system by incorporating purified polypeptide 3 into liposomes. The possibility that the major sialoglycoprotein is involved in this transport process has been ruled out. Specific inhibitors of anion transport show little binding to this protein under conditions which completely abolish transport (Cabantchik and Rothstein, 1974a; Ho and Guidotti, 1975; Lepke et al., 1976; Ship et al., 1977). In addition, erythrocytes which lack the major sialoglycoprotein have normal anion permeability (Tanner et al., 1976). Initial attempts to identify the protein involved in erythrocyte Dglucose transport led to the suggestion that polypeptide 3 might also be involved in this process. This was based on studies of the binding sites of cytochalasin B, a noncovalently bound inhibitor of &glucose transport (Taverna and Langdon, 1973a; Lin and Spudich, 1974), affinity labeling studies employing D-glucosylisothiocyanate (Taverna and Langdon, 1973b), selective extraction (Kahlenberg, 1976),and experiments using reconstituted vesicles (Kasahara and Hinkle, 1976). Recent results suggest that this interpretation is incorrect. Studies using impermeant maleimides under selective conditions (Batt et al., 1976) and further reconstitution studies (Kasahara and Hinkle, 1977)suggest that the component involved in D-glucose transport is a minor protein

306

MICHAEL J. A. TANNER

(band 4.5) which is relatively difficult to detect and copurifies with polypeptide 3 during selective extraction procedures. A large fraction of the total erythrocyte glyceraldehyde-3-phosphate dehydrogenase remains associated with the membrane after the hypotonic lysis of erythrocytes (Mitchell et al., 1965; Tanner and Gray, 1971). The enzyme is specifically and reversibly bound to the membrane and can be dissociated from it at high ionic strengths (Kant and Steck, 1973; McDaniel et al., 1974). Detergent-solubilized polypeptide 3 binds the enzyme, and the resulting complex can also be dissociated at high ionic strengths (Yu and Steck, 1975b). The association of the enzyme with both erythrocyte ghosts and with isolated polypeptide 3 is also influenced by both oxidized and reduced pyridine nucleotide coenzymes (Kant and Steck, 1973; McDaniel and Kirtley, 1974; Yu and Steck, 1975b). It is not known whether there is a relationship between the ionic strength-dependent structural changes which occur in polypeptide 3 (Jenkins and Tanner, 1977a) and the ability of polypeptide 3 to bind with glyceraldehyde-3-phosphate dehydrogenase. The significance of the association of these two proteins in vivo is not understood, but it seems likely that, if the binding of glyceraldehyde-3-phosphate dehydrogenase to the erythrocyte membrane occurs in the intact erythrocyte, its extent will be dependent on the metabolic state of the cell. Fossel and Solomon (1977) report that the glycolytic enzymes phosphoglycerate kinase and monophosphogl ycerate mutase also interact with glyceraldehyde-3-phosphate dehydrogenase and suggest that these enzymes provide a link between metabolism and cation transport in the erythrocyte. Erythrocyte ghosts have also been reported to bind another glycolytic enzyme, aldolase (Strapazon and Steck, 1976). B. Periodate-Stainable Glycoproteins

The functional role of periodate-stainable glycoproteins is poorly understood. The structure of the major sialoglycoprotein with its heavily glycosylated N-terminal region, high local concentration of charged sialic acid residues and carbohydrates, and single-membranepenetrating polypeptide chain segment suggest an extended shape rather than the globular shapes usually found for enzymes. No enzymic activity is known to be associated with this protein. Insofar as information is available on the structure of PAS-2' and PAS-3 they appear to be similar to the major sialoglycoprotein (Marchesi et al., 1976). These glycoproteins appear to share the common characteristic of being heavily glycosylated, and it seems reasonable to suppose that

ERYTHROCYTE GLYCOPROTEINS

307

the carbohydrates associated with them play a major part in their functional role, while the peptide portions may act as a framework for the distribution and orientation of these oligosaccharides.

1. ERYTHROCYTE GLYCOPROTEIN VARIANTS One approach to studying the function of erythrocyte membrane glycoproteins has been to investigate erythrocyte variants which are defective in these components with a view toward correlating these defects with the hematology of the individuals carrying these cells. This approach is immensely simplified in the case of the human erythrocyte, because of the importance of erythrocyte genetics in blood transfusion. Thus much information has been accumulated by blood group serologists covering a variety of human populations, SO that genetically typed cells, even those which are quite rare in human populations, are available. Although only a small number of studies on human erythrocyte variants so far has been done, a few erythrocytes with interesting glycoprotein defects have already been found. Erythrocytes homozygous for the type En(a-) have been shown to lack the major sialoglycoprotein and also to have alterations in the carbohydrate content of their polypeptide 3 (Tanner and Anstee, 1976b; Tanner et ul., 1976; Gahmberg et al., 1976; Anstee et al., 1977).These cells lack a normal erythrocyte antigen, Ena, and have a low sialic acid content, weak M and N antigens, and lowered mobility on cell electrophoresis (Darnborough et al., 1969; Furuhjelm et al., 1969, 1973).The location of the Ena antigen is unknown, but it is not present on the major sialoglycoprotein (Tanner and Anstee, 197613). The nomenclature for these cells is somewhat confusing, and we have suggested that it might be more appropriate to describe them phenotypically as sialoglycoprotein negative (SGP -) erythrocytes (Anstee et al., 1977). Loss of the major sialoglycoprotein in En(a-) cells must result in drastic changes at the surface of these cells, since this protein carries nearly half the total carbohydrate and an even larger proportion of the sialic acid present at the surface of the normal erythrocyte. The increased carbohydrate content of the polypeptide 3 probably compensates to some extent for this loss, though not in a gross way (Tanner et al., 1976; Gahmberg et al., 1976). It is particularly interesting to note that, although the major source of O-glycosidically linked sialic acidrich oligosaccharides found in the normal cell is lost in these cells, these oligosaccharides units do not appear in the polypeptide 3 of En(a -) cells. However, these oligosaccharides are probably present on the PAS-2’ and PAS-3 which remain in En(a-) cells.

308

MICHAEL J. A. TANNER

The loss of the sialoglycoprotein does not appear to have a detrimental effect on individuals with homozygous En(a -) erythrocytes. However, while no clinically significant abnormalities appear to be associated with the En(a-) condition, there have been no studies to determine whether or not there are less dramatic alterations in the physiological properties of these cells. The influence of the sialoglycoprotein deficiency on the in vivo lifetime of human erythrocytes, their fragility, and their flow properties, will be of particular interest. The rarity of the sialoglycoprotein-deficient condition in human populations suggests that the sialoglycoprotein confers some advantages. Several other erythrocytes have characteristics suggesting the presence of defects in the major sialoglycoprotein. These include the Mk, Mg,and Miltenberger antigens (Race and Sanger, 1975).However, in most of these cases the homozygous condition is unknown or very rare. We have found that it is difficult to pin down the nature of these defects unambigously unless a homozygous defective cell is available (Anstee and Tanner, unpublished observations). Erythrocytes lacking Ss antigens (S - s -) have also been studied (Dahr et al., 1975c; Tanner et al., 1977).This type is unknown in Caucasians but is relatively common among African tribal groups (Race could not detect any PAS-3 in and Sanger, 1975). Dahr et al. (1975~) these cells, but our results suggest that the polypeptide chain of PAS-3 remains but is defectively glycosylated, probably having lost the 0glycosidically linked sialic acid-rich units characteristic of normal PAS3 (Tanner et al., 1977). Again, this defect is not associated with any abnormality of clinical significance. These are two examples of viable erythrocytes which are defective in one of the periodate-positive glycoproteins. A cursory glance at the authoritative summary of information on human erythrocyte antigens (Race and Sanger, 1975) suggests that many other examples must exist. The examples discussed above are of inherited defects. T and Tn erythrocytes provide an interesting group of acquired changes in the erythrocyte surface (Race and Sanger, 1975). In both cases, erythrocytes become polyagglutinable, that is, become agglutinable by most normal human sera. T polyagglutinability is a transient phenomenon, often associated with infection by microorganisms. These erythrocytes carry a new antigen (the T antigen) which reacts with A. hypogea lectin (Bird, 1964).In vitro treatment of normal erythrocytes with bacterial neuraminidases also exposes the T antigen (Prokop and Uhlenbruck, 1969). The antigenic determinant appears to consist of galactose residues in the O-glycosidically linked sialotetrasaccharides of

ERYTHROCYTE GLYCOPROTEINS

309

the membrane which are exposed by the removal of sialic acid (Dahr

et al., 1975a). The Tn syndrome is a related, acquired, but more persistent condition. It is interesting that in this case the disorder is associated with hematological abnormalities (Bird et al., 1971; Race and Sanger, 1975). Blood from individuals with this syndrome contains varying proportions of defective cells. Studies on the defective cells (Dahr et al., 1975c) have shown that the O-glycosidically linked oligosaccharides of the major sialoglycoprotein and PAS3 (and possibly also PAS-2’) are probably incomplete and consist solely ofN-acetylgalactosamine linked to the polypeptide chain.

2. APPEARANCE OF SURFACE GLYCOPROTEINS DURING MATURATIONOF THE ERYTHROCYTE Another experimental approach to obtaining information on the functional role of the erythrocyte glycoproteins is to correlate the appearance of the glycoproteins of the mature erythrocyte with the physiological requirements imposed on the cell during the period when these components are present. The rabbit erythropoietic system is a convenient one to study, particularly because the rabbit erythrocyte lacks the major sialoglycoprotein found in humans and most other species (Lodish and Small, 1975; Light and Tanner, 1977).Polypeptide 3 is the major surface glycoprotein of the rabbit erythrocyte, and three other minor glycoproteins are also present. This lack of a major sialoglycoprotein is not entirely unexpected, since the rabbit erythrocyte has been shown to have an unusually low sialic acid content, surface charge, and mobility on cell electrophoresis (Walter et al., 1967; Durocher et d., 1975). It is interesting to note that no O-glycosidically linked oligosaccharides appear to be present on the surface of rabbit erythrocytes since none of the glycoproteins bind to M . auruntica lectin (Light and Tanner, 1977). We studied the changes in surface glycoproteins which take place during the differentiation of rabbit erythroid cells in the bone marrow and during the release of nonnucleated erythroid cells into the circulation (Light and Tanner, 1977). Nucleated bone marrow-bound erythroid cells and circulating erythroid cells were found to have unique and mutually exclusive groups of surface membrane proteins. The two types of membranes were shown to give distinct protein patterns on gel electrophoresis and after lactoperoxidase radioiodination. They also had different distributions of lectin-binding components when tested with four lectins of differing specificity. Bone marrowbound erythroid cells can be separated according to their age by ve-

310

MICHAEL J. A. TANNER

locity sedimentation using methods similar to those of Miller and Phillips (1969) and Denton and Arnstein (1973). The plasma membranes of different age groups of nucleated erythroid cells were found to have the same surface components. We can infer from these results that neither polypeptide 3 nor the other minor glycoproteins found in circulating erythroid cells have a functional role during any of the earlier nucleated erythroid cell stages of the erythrocyte. This makes it possible to narrow down considerably the events in the life history of the erythrocyte in which the carbohydrates bound to polypeptide 3 might play a part. These stages are (1) the dissociation of the reticulocyte from the bone marrow, (2) the circulatory phase of the erythrocyte, and ( 3 )the turnover of senescent erythrocytes. The carbohydrate contents of rabbit reticulocyte and erythrocyte polypeptide 3 are shown in Table I11 (Light, 1976).The rabbit protein has a carbohydrate composition similar to that of the human protein and lacks N-acetylgalactosamine and sialic acid. Since the turnover of erythrocytes in the rabbit appears to involve the removal or loss of sialic acid residues (see Section 11),it is unlikely that the carbohydrates associated with polypeptide 3 play any part in this process. The source of the sialic acid residues involved in the process of erythrocyte turnover is not known, but they may be present on rabbit minor glycoproteins or on sialic acid-containing glycolipids. The latter may be a significant source, since Sweeley and Dawson (1969) estimate TABLE I11

RABBIT RETICULOCYTEAND ERYTHROCYTE POLYPEPTIDE3"

CARBOHYDRATE COMPOSITION OF

Nanomoles per milligram of protein in polypeptide 3 Component

Reticulocytes

Erythrocytes

Fucose Mannose Galactose N-Acetylglucosamine .N-Acetylgalactosamine Sialic acid

5 30 62 40

5 24 49

-

37

a Data from Light (1976). The protein was purified from erythrocytes and reticulocytes as described by Tanner et al. (1976).

ERYTHROCYTE GLYCOPROTEINS

31 1

that a single erythrocyte carries several million molecules of total glycolipid, although the relative proportions of the neutral and the sialic acid-containing-types vary considerably in different species. The functional role of the protein portion of polypeptide 3 in the equilibration of anions during the circulating phase of the erythrocyte has already been discussed (Section V1,A). Since the nucleated erythroid cells present in the bone marrow are unlikely to need to equilibrate anions at as a rate fast as the erythrocyte, the absence of polypeptide 3 from these cells is not altogether surprising. Light (1976) found that the anion permeability of bone marrow erythroid cells was much lower than that of erythrocytes. The correspondence between the dissociation of erythroid cells from the bone marrow and the appearance of polypeptide 3 suggests that the carboh.ydrates bound to this protein may be involved in the alteration in intercellular interactions which must take place during the transition from a tissue-bound erythroid cell to a freely circulating erythroid cell.

3.

SOME CONSIDERATIONS AND SPECULATIONS ON THE FUNCTION OF THE GLYCOPROTEIN-BOUND CARBOHYDRATES AT THE SURFACE OF THE ERYTHROCYTE

Although glycoproteins are found in a wide variety of situations in mammals and in many other organisms, in general we have only a minimal understanding of the attributes the bound carbohydrates bestow on these proteins. Even less is known about the detailed structural and conformational properties of oligosaccharides which cause them to be so widely distributed. Therefore the considerations presented in this section reflect the limited theoretical and experimental knowledge available at this time and contain a considerable element of speculation. The justification for including it at all is to highlight some unresolved questions which to some extent have not been a focus of attention for investigators interested in the biochemistry of erythrocyte membrane glycoproteins. Two types of functional attributes can be broadly distinguished for most biologically important compounds:

1. Functions involving specific chemical interactions. In this case, the requirement to be able to interact specifically with some other structure can impose considerable chemical and conformational limitations on the molecule. The absence of the molecule often results in the cessation of some sequence of biological reactions.

312

MICHAEL J. A. TANNER

2. Functions which may be broadly described as structural. Here, the function depends on the general properties of the particular class of compounds and does not impose such restrictive requirements on the chemical structure of the molecules involved. Many classes of compound can have both types of functional role. Thus the catalytic serine residue in serine proteases is an example of the first type, while the large number of amino acid residues which can be substituted for by a residue of a similar type without affecting biological activity in many proteins represents the second type. The requirement of many membrane-bound enzymes for particular phospholipids for enzymic activity, and the general and relatively nonspecific ability of phospholipids to form bimolecular leaflets in biological membranes, illustrate a similar dual role for this class of compounds. There is no reason to suppose that cell surface carbohydrates cannot have a similar duality in functional role. Several mechanisms have been suggested which involve cell surface carbohydrates in specific processes such as cell recognition and cell-cell interactions (see Wallach, 1975, for a recent review). We can ask whether or not the carbohydrates associated with the periodatestainable glycoproteins of the erythrocyte, and in particular the major sialoglycoprotein, have any involvement in this type of specific process. Possible functions of the carbohydrates associated with polypeptide 3 are considered in Section VI,B,2. The apparent absence of clinical abnormalities in En(a -) human erythrocytes, which lack the major sialoglycoprotein, suggest that the normal function and viability of the human erythrocyte are not critically dependent on the presence of this protein and the oligosaccharides present on it, and that loss of this carbohydrate can be compensated for by a relatively small increase in the carbohydrate content of polypeptide 3. The major source of the O-glycosidically linked sialotetrasaccharides of the normal human erythrocyte is absent in the En(a-) cell, and these sialotetrasaccharides do not appear in the new sugar units found on the altered polypeptide 3 (Tanner et al., 1976). Since no hematological disorders occur in this case, it seems likely that these O-glycosidically linked sialic acid-rich oligosaccharides of the sialoglycoprotein are not involved in interactions of the specific type. A similar argument. can be presented for the O-glycosidically linked oligosaccharides of PAS-3 which are absent in S - s - erythrocytes. The very large number of these O-glycosidically linked sialic acid-rich oligosaccharides present at the surface of the normal erythrocyte (approximately 8 x lo6 per erythrocyte) tends to support

ERYTHROCYTE GLYCOPROTEINS

31 3

this view. It is difficult to see why there would need to be so many of these if the oligosaccharide had a specific receptor function. For similar reasons it is even more difficult to envisage the necessity for 15 apparently similar oligosaccharides on a single sialoglycoprotein molecule, many on adjacent amino acid residues, when a single accessible oligosaccharide on each protein molecule might be expected to be adequate. In the rabbit erythrocyte the absence of glycoproteins analogous to the periodate-stainable glycoproteins of the human cell, together with the lack of O-glycosidically linked oligosaccharides, is also consistent with this suggestion. Nevertheless, the almost ubiquitous presence of the sialoglycoprotein in humans, and the presence of analagous glycoproteins in nearly all other animals, suggests that the carbohydrates associated with this protein do have a function. It seems probable that this is of the more general or structural type referred to at the beginning of this section. The compensatory change in carbohydrate content in polypeptide 3 of En(a-) erythrocytes is consistent with this, and it is interesting to note that the only other known sialoglycoprotein-deficient erythrocyte, the rabbit cell, is also unusual in containing a N-acetylglucosaminyl lipid as its major glycolipid (Sweeley and Dawson, 1969), which might take over the role of the sialoglycoprotein-like molecules in this case. The sialic acid on the erythrocyte (which is mainly associated with the sialoglycoprotein) has been thought to serve to keep erythrocytes separate (Klenk, 1956). Although this may be the case, neither rabbits nor individuals with En(a-) erythrocytes appear to suffer any detrimental effects because of increased agglutinability of their erythrocytes. It is unlikely that the sialic acid residues associated with erythrocyte sialoglycoproteins are involved in the turnover of erythrocytes (see Section VI,B,2). The plasma membrane of the erythrocyte is a remarkably stable entity. The cell remains intact and does not fuse with other cells or vesicularize durihg the long circulating phase of the cell (120 days at 37°C). During this period the cell is subjected to continuous and very severe mechanical stresses, particularly while in the microcirculation. Glycoprotein-bound carbohydrates may contribute to the stability of this membrane in some way. In particular, they may provide a hydrophilic and highly hydrated layer at the cell surface which could serve as a thermodynamic barrier to cell fusion. It has been suggested that the aggregation of intramembranal particles (which in the case of the erythrocyte might contain the membrane glycoproteins; see Section IV) is an important prerequisite to cell fusion (Poste and Allison, 1973), and that cell fusion involves the interaction of protein-free, bare, lipid bilayers (Maroudas, 1975; Ahkong et al., 1975; Vos et al.,

314

MICHAEL J. A. TANNER

1976).Thus it might be necessary to break the continuity of the carbohydrate layer in order for cell fusion to proceed. The effectiveness and continuity of such a barrier would be improved by extensive interactions between oligosaccharides on the same and different glycoprotein molecules, so as to form an extended network over the surfaceof the cell. This type of structure has been suggested by Bretscher (1973). It seems likely that the periodatestainable glycoproteins have extended structures, because of the high density of negatively charged sialic acid residues they contain (Winzler, 1971).An extended network could be built up between carbohydrate moieties both by interconnecting hydrogen bonds and by divalent cation bridges between sialic acid residues. Approximately 6 x lo6 Ca2+-bindingsites per cell are located on sialic acid residues on the erythrocyte surface (Seaman et al., 1969). Only 6 basic amino acid residues are available in the extracellular region of the human major sialoglycoprotein for intramolecular interactions with the 20 or 30 sialic acid molecules present on the protein, and it seems probable that most of the latter interact with divalent cations. This glycoprotein network need not be entirely continuous but may contain polypeptide islands which do not contain carbohydrates. These could be determined b y proteins like polypeptide 3 and form an access path for anions to the transport system and in general terms allow natural ligands to reach their receptors (Fig. 8).One of the functions of the relatively small proportion of carbohydrate found in polypeptide 3 may be to allow the protein to become integrated into a glycoprotein network. The presence of 15apparently similar O-glycosidically linked oligosaccharides in the major sialoglycoprotein, particularly the 6 adjacent oligosaccharides on residues 10 to 15, suggests that some sort of repetitive structure can be formed by these oligosaccharides. Strong selection pressures, presumably structural in nature, must operate on the polypeptide and the carbohydrate of this protein to preserve this large number of similar chains. It is striking that closely related oligosaccharides all having the core structure found in the sialotetrasaccharides, GalPl- 3GalNAc 4 Ser(Thr), are found in a great variety of vertebrate glycoproteins, very often in dense clusters on adjacent or nearly adjacent amino acid residues (Kornfeld and Kornfeld, 1976), again suggesting that groups of oligosaccharides contribute to some sort of a repeating structure. The similarities in the core structures of the generally more complex N-glycosidically linked oligosaccharides and the often repetitive sequences present in the outer chains of these oligosaccharides are also becoming evident (Kornfeld and Kornfeld,

ERYTHROCYTE GLYCOPROTEINS

315

FIG.8. Schematic representation of the carbohydrate barrier at the erythrocyte surface. The solid areas represent the glycoprotein network, while the hatched areas show regions of polypeptide. The periodate-stainable glycoproteins are indicated by extended polypeptides, and the globular protein represents polypeptide 3, the anion transport protein.

1976). When the relatively restricted variety of oligosaccharide sequences found is compared with the immense variety of theoretically possible sequences, the conclusion that groups of these sequences have special structural properties which has caused their selection and restricted their divergence seems a reasonable one. Since virtually nothing is known about the three-dimensional structure of these heterooligosaccharides, it is not possible to surmise the nature of these structures. However, it is conceivable that in many biological situations it is advantageous for proteins to possess localized or extended hydrogen-bonded networks which could provide hydrophilic, highly hydrated, impenetrable barriers in certain exposed regions of proteins or to cover particular surface areas. If a carbohydrate lattice is present at the surface of erythrocytes, the viability of En(a -) human erythrocytes and rabbit erythrocytes might be explained by the ability of the compensatory change in carbohydrate content of En(a-) erythrocyte polypeptide 3, and the unusual glycolipid and the carbohydrate on polypeptide 3 of rabbit erythrocytes, to provide a sufficiently continuous network of this type of yield a membrane with adequate stability. Whether or not a similar situation applies to the surface membrane of other mammalian cells cannot yet be predicted, since very little information is available on the structure of the oligosaccharides bound to the surface glycoproteins. However, in view of the general compositional similarities among these and other glycoproteins, it would not be surprising to find similar surface barriers in other cells, particularly in tissues where plasma membranes are in close juxtaposition and yet cells must retain their integrity. An interesting consequence of the presence of a restricting barrier of this type is that either surface receptors must be external to this barrier or ligands must penetrate or disrupt this network in order to interact with components in the cell

31 6

MICHAEL J. A. TANNER

membrane. The well-known and diverse effects of lectins on cellular events and on membrane structure (see, e.g., Lis and Sharon, 1973; Wallach, 1975; Nicholson, 1976) may be a consequence of the disruption of this lattice by these polyvalent, carbohydrate-specific ligands. It is evident that, in order to understand the function and importance of membrane glycoproteins, it will be necessary to understand the detailed three-dimensional structures and interactions of the heterooligosaccharides found in these molecules. The resolution of these structures remains a problem of major significance in current biology. ACKNOWLEDGMENTS

I thank my colleagues Drs. D. H. Boxer, R. E. Jenkins, and N. D. Light for their enthusiastic contributions to the work that has been done in our laboratory, and especially Dr. D. J. Anstee for his collaboration and the many hours of stimulating discussion we have had. I am indebted to Annual Reviews, Inc., and to Dr. V. T. Marchesi, for perniission to use Fig. 3. The work from our laboratory has been supported in part by grants from the Medical Research Council and the Wellcome Trust. REFERENCES Adair, W. L., and Kornfeld, S. (1974). Isolation of the receptors for wheat germ agglutinin and the Ricinis communis lectins from human erythrocytes using affinity chromatography. J . Biol. Chem. 249,4695-4704. Adamany, A. M., and Kathan, R. H. (1969). Isolation of a tetrasaccharide common to MM, NN and NN antigens. Biochem. Biophys, Res. Commun. 37,71-77. Ahkong, Q . F., Fisher, D., Tampion, W., and Lucy, J. A. (1975). Mechanisms of cell fusion. Nature (London)253, 141-195. Anstee, D. J., and Tanner, M. J. A. (1974a). Distribution of blood group antigens on butanol extraction of human erythrocyte membranes. Biochem. J . 138,381486. Anstee, D. J., and Tanner, M. J. A. (1974b). Blood group serology of fractions obtained from the human erythrocyte membrane. Eur. J . Biochem. 45,31-37. Anstee, D. J., and Tanner, M. J. A. (1975).Separation ofABH, I, Ss antigen activity from the MN active sialoglycoprotein of the human erythrocyte membrane. Vox Sang. 29,378-389. Anstee, D. J., Barker, D. M., Judson, P. A., and Tanner, M. J. A. (1977).Inherited sialoglycoprotein deficiencies in erythrocytes of type En(a-). Br.J . Haematol. 35,309320. Avruch, J., and Fairbanks, G. (1972).Demonstration of a phosphopeptide intermediate in the Mg2+dependent Na+ and K+ stimulated adenosine triphosphate reaction of the erythrocyte membrane. Proc. Natl. Acad. Sci. U S A . 69, 1215-1220. Bachi, T., and Schnebli, H. P. (1975). Reaction of lectins with human erythrocytes 11. Mapping of Con A receptors by freeze-etching electron microscopy. E x p . Cell Res. 91,285-295. Bachi, T., Whiting, K., Tanner, M. J. A., Metaxas, M. N., and Anstee, D. J. (1977). Freeze-fracture electron microscopy of human erythrocytes lacking the major membrane sialoglycoprotein. Biochim. Biophys. Acta 464,635-639. Bailey, A. J.. Robins, S. P., and Tanner, M. J. A. (1976). Reducible components in the proteins of the human erythrocyte membrane. Biochim. Biophys. Actu 434,51-57.

ERYTHROCYTE GLYCOPROTEINS

31 7

Batt, E. R., Abhott, R. E., and Schachter, D. (1976).Identification ofan exofacial component of the human erythrocyte hexose transport mechanism. J. Biol. Chem. 251, 7184-7190. Bellhorn, M., Blumenfeld, 0. O., and Gallop, P. M. (1970). Acetylcholinesterase of the human erythrocyte membrane. Biochem. Biophys. Res. Commun. 39,267-273. Bender, W. W., Garan, H., and Berg, H. C. (1971). Proteins of the human erythrocyte membrane as modified by pronase. J. Mol. B i d . 58,783-797. Bird, G . W. G. (1964).Anti T in peanuts. Vox Sung. 9,748-756. Bird, G . W. G., Shinton, N. K., and Wingham, J. (1971). Persistent mixed-field polyagglutination. Br. J . Haemutol. 21,443-453. Blumenfeld, 0.0. (1968). The proteins of the erythrocyte membrane obtained by solubilization with aqueous pyridine solution. Biochem. Biophys. Res. Commun. 30, 200-205. Boxer, D. H., Jenkins, R. E., and Tanner, M. J. A. (1974). The organisation of the major protein of the human erythrocyte membrane. Biochem. J. 137,531-534. Brennessel, B. A., and Goldstein, J. (1974).Separation of H activity from isolated glycoproteins of human 0 erythrocyte membranes. Vox Sung. 26,405-414. Bretscher, M. S. (1971a). Human erythrocyte membranes: Specific labelling of surface proteins. J . M o l . Biol. 58, 775-781. Bretscher, M. S. (19711)). Major human erythrocyte glycoprotein spans the cell membrane. Nature (London)231,229-232. Bretscher, M. S. ( 1 9 7 1 ~ )A. major protein which spans the erythrocyte membrane. J. Mol. Biol. 59,351-357. Bretscher, M. S. (1973). Membrane structure: Some general principles. Science 181, 622-629. Bretscher, M. S. (1975). C-Terminal region of the major erythrocyte sialoglycoprotein is on the cytoplasmic side of the membrane. J. Mol. B i d . 98,830-833. Bretscher, M. S., and Raff, M. C. (1975).Mammalian plasma membranes. Nature (London) 258,43-49. Bridgen, J., and Walker, I. D. (1976). Photoreceptor protein from the purple membrane offfalobacteriumhalobium: Molecular weight and retinal binding site. Biochemistr!/ 15, 792-801. Cabantchik, Z. I., and Rothstein, A. (1972). The nature of the membrane sites controlling anion permeability as determined by studies with disulphonic stilhene derivatives.]. Membr. Biol. 10,311-327. Cabantchik, Z. I., and Rothstein, A. (1974a). Membrane proteins related to anion permeability of human red blood cells. I. Localization of disulfonic stilbene binding sites in proteins involved in permeation.]. Membr. B i d . 15,207-226. Cabantchik, Z. I., and Rothstein, A. (1974b). Membrane proteins related to anion permeability of human red cells. 11. Effects of proteolytic enzymes on disulphonic stilbene sites of surface proteins. J. Membr. Biol. 15, 227-248. Cabantchik, Z. I., Balshin, M., Breuer, W., Markus, H., and Rothstein, A. (1975).A comparison of intact human red blood cells and resealed and leaky ghosts with respect to their interactions with surface labelling agents and proteolytic enzymes. Biochim. Biophys. Actu 382,621-633. Cabantchik, Z. I., Knauf, P. A., Ostwald, T., Markus, H., Davidson, L., Breuer, W., and Rothstein, A. (1976). Interaction of an anionic photoreactive probe with the anion transport system of the human red cell. Biochim. Biophys. Actu 455,526-537. Cabantchik, Z. I., Woloshin, J. M., Ginsburg, H., and Zemal, 0. (1977). Structural and functional properties of the anion transport system isolated from human erythrocyte

31 8

MICHAEL J. A. TANNER

membranes. In “Biochemistry of Membrane Transport” (G. Semenza and E. Carafoli, eds.), FEBS Symposium No. 42, pp. 328-345. Springer-Verlag, Berlin and New York. Capaldi, R. A. (1973). Properties of the major glycoprotein of the beef erythrocyte membrane. Biochim. Biophys. Acta 311,386-390. Carraway, K. L., Colton, D. G., Shin, B. C., and Triplett, R. B. (1975).Species variability in the modification of erythrocyte surface proteins by enzymatic probes. Biochim. Biophys. Acta 382, 181-192. Dahr, W., Uhlenbruck, G., and Bird, G. W. G. (1974). Cryptic A-like receptor sites in human erythrocyte glycoproteins: Proposed nature of Tn antigen. Vox Sang. 27,29 -42. Dahr, W., Uhlenbruck, G., and Bird, G. W. G. (1975a). Further characterization of some heterophile agglutinins reacting with the alkali-labile carbohydrate chains of human erythrocyte glycoproteins. Vox Sang. 28, 133- 148. Dahr, W., Uhlenbruck, G., and Knott, H. (197513). Immunochemical aspects of the MNSs-blood group system. J . Zmmunogenet. 2,87-100. Dahr, W., Uhlenbruck, G., Issitt, P. D., and Allen, F. H. ( 1 9 7 5 ~ )SDS-polyacrylamide . gel analysis of the membrane glycoproteins from S-s-U- erythrocytes.]. Immunogenet. 2,249-251. Dahr, W., Uhlenbruck, G., Gunson, H. H., and Van der Hart, M. (1975d). Molecular basis of Tn-polyagglutinability. Vox Sang. 29,36-50. Dahr, W., Uhlenbruck, G., Schmalisch, R., and Janssen, E . (1976). Ss blood group associated PAS-staining polymorphism of glycoprotein 3 from human erythrocyte membranes. Human Genetics 32, 121-135. Dahr, W., Uhlenbruck, G., Jannsen, E. and Schmalisch, R. (1977). Different N-terminal amino acids in the M,N glycoprotein from MM and NN erythrocytes. Human Genetics 35,335-343. Damborough, J., Dunsford, I., and Wallace, J. A. (1969).The Ena antigen and antibody. Vox Sang. 17,241-255. Denton, M. J., and Arnstein, H. €3. V. (1973). Characterization of developing adult mammalian erythroid cells separated by velocity sedimentation. Br. J . Haematol. 24, 7-17. Drickamer, L. K. (1976). Fragmentation of the 95,000-dalton transmembrane polypeptide in human erythrocyte membranes. J . Biol. Chem. 251, 5115-5123. Durocher, J. R., Payne, R. C., and Conrad, M. E. (1975).Role of sialic acid in erythrocyte survival. Blood 45, 11-20. Emerson, W. A., and Kornfeld, S. (1976).Characterization of the oligosaccharide units of the bovine erythrocyte membrane protein. Biochemistry 15, 1697-1703. Fairbanks, G., Steck, T. L., and Wallach, D. F. H. (1971). Electrophoretic analysis of the major polypeptides of the human erythrocyte membrane. Biochemistry 10,26062616. Findlay, J. B. C. (1974). The receptor proteins for concanavalin A and Lens culinaris phytohaemagglutinin in the membrane of the human erythrocyte. J . Biol. Chem. 249,4398-4403. Fossel, E. T., and Solomon, A. K. (1977). Membrane mediated link between ion transport and metabolism in human red cells. Biochim. Biophys. Acta 464,82-92. Fujita, S., and Cleve, H. (1975). Isolation and partial characterization of two minor glycoproteins from human erythrocyte membranes. Biochim. Biophys. Acta 382,172180. Fukuda, M., and Osawa, T. (1973).Isolation and characterization of a glycoprotein from human blood group 0 erythrocyte rnembranes.J. Biol. Chem. 248,5100-5105.

ERYTHROCYTE GLYCOPROTEINS

319

Furthmayr, H., and Marchesi, V. T. (1976).Subunit structure of human erythrocyte glycophorin A. Biochemistry 15, 1137-1144. Furthmayr, H., Tomita, M., and Marchesi, V. T. (1975). Fractionation of the major sialoglycopeptides of the human red blood cell membrane. Biochern. Biophys. Res. Commun. 65, 113-121. Furthmayr, H., Kahane, I., and Marchesi, V. T. (1976). Isolation of the major intrinsic transmembrane protein of the human erythrocyte membrane. J. Membr. Biol. 26, 173- 187. Furuhjelm, V. Myllyla, G. Nevanlinna, H. R., Nordling, S., Pirkola, A., Gavin, J., Gooch, L., Sanger, R., and Tippett, P. (1969).The red cell phenotype En(a-) and anti Ena. Vox Sang. 17,256-278. Fumhjelm, U., Nevanlinna, H. R., and Pirkola, A. (1973).A second Finnish En(a-) Propositus with anti-Ena. Vor Sang. 24,545-549. Gahmberg, C. G. (1976).External labelling of human erythrocyte glycoproteins.]. Biol. Chem. 251,510-515. Gahmberg, C. G., and Hakomori, S . (1973).External labelling of cell surface galactose and galactosamine in glycolipid and glycoprotein of human erythrocytes. J. Biol. Chem. 248,4311-4317. Gahmberg, C. C., Myllyla, G., Leikola, J., Pirkola, A., and Nordling, S . (1976).Absence of the major sialoglycoprotein in the membrane of human En(a-) erythrocytes and increased glycosylation of band 3.1. Biol. Chem. 251,6108-6116. Cardas, A., and Koscielak, J. (I974a). Megaloglycolipids-Unusually complex glycosphingolipids of the human erythrocyte membrane: A, B, H and I blood group specificity. FEBS Lett. 42,101-104. Cardas, A., and Koscielak, J. (1974b).I-Active antigen of human erythrocyte membrane. Vox Sang. 26,227-237. Gattegno, L., Bladier, D., and Cornillo, P. (1974).The role of sialic acid in the determination of the survival of rabbit erythrocytes in the circulation. Carbohydr. Res. 34, 361-369. Glockner, W. M., Newman, R. A., Dahr, W., and Uhlenbruck, G. (1976). Alkali-labile oligosaccharides from glycoproteins of different erythrocyte and milk fat globule membranes. Biochim. Biophys. Acta 443,402-413. Grefrath, S . P., and Reynolds, J. A. (1974).Molecular weight of the major glycoprotein from the human erythrocyte membrane. Proc. Natl. Acad. Sci. U.S.A. 71, 39133916. Hamaguchi, H., and Cleve, H. (1972). Solubilization of human erythrocyte glycoproteins and separation of the MN glycoprotein from a glycoprotein with I, S and A activity. Biochim. Biophys. Acta 278,271-280. Henderson, R., and Unwin, P. N. T. (1975).Three dimensional model of purple membrane obta’ined by electron microscopy. Nature (London)257,28-32. Ho, M. N., and Guidotti, G. (1975).A membrane protein from human erythrocytes involved in anion exchange. J. Biol. Chem. 250,675-683. Hoogeveen, J . P. T., Juliano, R. L., Coleman, J.. and Rothstein, A. (1970).Water soluble proteins of the human red cell membrane.J. Membr. Biol. 3, 156-172. Howe, C., and Bachi, T. (1973).Localization of erythrocyte membrane antigens by immune electron microscopy. Exp. Cell Res. 76,321-332. Hubbard, A. L., and Cohn, Z. A. (1976).Specific labels for cell surfaces. I n “Biochemical Analysis of Membranes” (A. H. Maddy, ed.), pp. 427501. Chapman & Hall, London. Jackson, R. L., Segrest, J. P., Kahane, I., and Marchesi, V. T. (1973).Studies on the major sialoglycoprotein of the human red cell membrane. Biochemistry 12,3131-3138.

320

MICHAEL J. A. TANNER

Jenkins, R. E., and Tanner, M. J. A. (1975). The major human erythrocyte membrane protein. Biochem. J . 147,393-399. Jenkins, R. E., and Tanner, M. J. A. (1977a). Ionic-strength-dependent changes in the structure of the major protein of the human erythrocyte membrane. Biochem. J. 161,131-138. Jenkins, R. E., and Tanner, M. J. A. (1977b). Structure of the major human erythrocyte membrane protein. Biochem. J. 161, 139-147. Juliano, R. L. (1973). The proteins of the erythrocyte membrane. Biochim. Biophys. Acta 300,341-378. Kahlenberg, A. (1976). Partial purification of a membrane protein from human erythrocytes involved in glucose transp0rt.J. Biol. Chem. 251, 1582-1590. Kant, J. A., and Steck, T. L. (1973). Specificity in the association of glyceraldehyde 3phosphate dehydrogenase with isolated human erythrocyte membranes. J. Biol. Chem. 248,8457-8464. Kasahara, M., and Hinkle, P. C. (1976). Reconstitution of D-glucose transport catalysed 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). D-Glucose transport in vesicles formed from an erythrocyte protein fraction and phospholipids. In “Biochemistry of Membrane Transport” (C. Semenza and E. Carafoli; eds.), FEBS Symposium No. 42, pp. 346351. Springer-Verlag, Berlin and New York. Kathan, R. H., Winder, R. J., and Johnson, C. A. (1961).Preparation of an inhibitor of viral haemagglutination from human erythrocytes. J. E x p . Med. 113,37-45. Klenk, E. (1956). Chemie and Biochemie der Neuraminsaure. Angew. Chem. 68,349354. Klenk, E., and Uhlenbruck, G. (1960). Uber neuraminsaurehaltige Mucoide aus menschener Erthrozytenstroma, ein Beitrag zur Chemie der Agglutinogene. Z. Physiol. Chem. 319,151-160. Kobylka, D., Khettry, A., and Shin, B. C., and Carraway, K. L. (1972).Proteins and glycoproteins of the erythrocyte membrane. Arch. Biochem. Biophys. 148,475-487. Komfeld, R., and Komfeld, S. (1970). The structure of a phytohaemagglutinin receptor site from human erythrocytes. J. Biol. Chem. 245,2536-2545. Komfeld, R.,and Komfeld, S. (1976). Comparative aspects of glycoprotein structure. Annu. Reo. Biochem. 45,217-237. Komfeld, S., Rogers, J., and Gregory, W. (1971).The nature of the cell surface receptor site for Lens culinaris phytohaemagglutinin. J . Biol. Chem. 246,6581-6586. Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature (London)227,680-685. Lenard, J. (1970). Protein components of erythrocyte membranes from different animal species. Biochemistry 9,5037-5040. Lepke, S., and Passow, H. (1976).Effects of incorporated trypsin on anion exchange and membrane proteins in human red blood cell ghosts. Biochim. Biophys. Acta 455, 353-370. Lepke, S., Fasold, H., Pring, M., and Passow, H. (1976). Study of the relationship between inhibition of anion exchange and binding to the red blood cell membrane of 4.4’-diisothiocyano stilbene-2,2’-disulphonic acid and its dihydro derivative. /. Membr. Biol. 29, 147-177. Leseney, A. M., Boumillon, R., and Kornfeld, S. (1972). The nature of the erythrocyte receptor site for Robinia pseudoaccacia phytohaemagglutinin. Arch. Biochem. Biophys. 153,831-836. Liao, T.,Gallop, P. N., and Blumenfeld, 0. 0. (1973). Modification of sialyl residues of

ERYTHROCYTE GLYCOPROTEINS

32 1

sialoglycoproteins of the human erythrocyte surface. J. Biol. Chem. 248, 82478253. Light, N. D. (1976). Studies on Plasma Membrane Proteins of Differentiating Erythroid Cells in the Rabbit. Ph.D. Thesis, Univ. Of Bristol, Bristol, England, Light, N. D., and Tanner, M. J. A. (1977). Changes in surface membrane components during the differentiation of rabbit erythroid cells. Biochem. J. 164, 565-578. Lin, S., and Spudich, J. A. (1974). Biochemical studies on the mode of action of cytochalasin B. J . Biol. Chem. 249,5778-5783. Lis, H., and Sharon, N. (1973). The biochemistry of plant 1ectins.Annu. Rev. Biochem. 42,541-574. Lisowska, E. (1969). The degradation of M and N blood group glycoproteins and glycopeptides with alkaline borohydride. Eur. J. Biochem. 10,574-579. Lisowska, E., and Duk, M. (1975). Modification of amino groups of human erythrocyte glycoproteins and a new concept on the structural basis of M and N blood group specificity. Eur. J . Biochern. 54,469-474. Lisowska, E., and Jeanloz, R. W. (1973).Composition and distribution of carbohydrate chains in glycoprotein of human erythrocyte membrane. Carbohydr. Res. 29, 181191. Lisowska, E., and Morawiecki, A. (1967). The role of free amino groups in the blood group activity of M and N mucoids. Eur. J. Biochem. 3,237-241. Lodish, H. F., and Small, B. (1975). Membrane proteins synthesized by rabbit reticulocytes. J. Cell Biol. 65,51-64. Lotan, R., Skutelsky, E., Danon, D., and Sharon, N. (1975).The purification, composition and specificity of the anti-T lectin from peanut (Arachis hypogaea). J. Biol. Chem. 250,8518-8525. McDaniel, C. F., and Kirtley, M. E. (1974).Cooperativity in the interaction of glyceraldehyde 3-phosphate dehydrogenase with erythrocyte membranes. Biochem. Biophys. Res. Commun. 65, 1196-1200. McDaniel, C. F., Kirtley, M. E., and Tanner, M. J. A. (1974). Interaction of glyceraldehyde %phosphate dehydrogenase with erythrocyte membranes. J . Biol. Chem. 249, 6478-6485. Marchesi, V. T., and Andrews, E. P. (1971). Glycoproteins: Isolation from cell membranes with lithium diiodosalicylate. Science 174, 1247-1248. Marchesi, V. T., Furthmayr, H., and Tomita, M. (1976). The red cell membrane.Annu. Reu. Biochem. 45,667-698. Maroudas, N. G . (1975). Polymer exclusion, cell adhesion and membrane fusion. Nature (London) 254,695-696. Marton, L. S. G., and Garvin, J. E. (1973).Subunit structure of the major human erythrocyte glycoprotein: Depolymerization by heating ghosts. Biochem. Biophys. Res. Commun. 52, 1457-1462. Miller, L. H., Mason, S. J., Dvorak, J. A., McGinniss, M. H., and Rothman, I. K. (1975). Erythrocyte receptors for malaria: Duffy blood group determinants. Science 189, 56 1-562. Miller, R. G., and Phillips, R. A. (1969).Separation of cells by velocity sedimentation. J. Cell Physiol. 73, 191-201. Mitchell, C. D., Mitchell, W. B., and Hanahan, D. (1965). Enzyme and haemoglobin retention in human erythrocyte stroma. Biochim. Biophys. Acta 104,348-358. Morawiecki, A. (1964).Dissociation of M and N group mucoproteins into subunits in detergent solutions. Biochim. Biophys. Acta 83,339-347. Morell, A. G . , Irvine, R. A., Sternlieb, I. H., Scheinberg, I. H., and Ashwell, G. (1968). Physical and chemical studies on ceruloplasmin. J. Biol. Chem. 243, 155-159.

322

MICHAEL J. A. TANNER

Morel], A. G., Gregoriadis, G., Scheinberg, I. H., Hickman, J., and Ashwell, G. (1971). The role of sialic acid in determining the survival of glycoproteins in the circulati0n.J. Biol. Chem. 246, 1461-1467. Mueller, T. J., and Morrison, M. (1974). The transmembrane proteins in the plasma membrane of n o F a l human erythrocytes. 1.Biol. Chem. 249,7568-7573. Nicolson, G. L. (1973). Anionic sites of human erythrocyte membranes. I. Effects of trypsin, phospholipase C, and pH on the topography of bound positively charged colloidal particles. J . Cell Biol. 57,373487. Nicolson, G. L. (1976). Transmembrane control of the receptors on normal and tumour cells. I. Cytoplasmic influence over cell surface components. Biochim. Biophys. Acta 457,57-108. Passow, H., and Wood, P. G. (1974).Current concepts of the mechanism of anion permeability. In “Drugs and Transport Processes” (B. A. Callingham ed.), pp. 149171. Macmillan, New York. Phillips, D. R., and Morrison, M. (1971).Exposed protein on the intact human erythrocyte. Biochemistry 10, 1766-1771. Pinto da Silva, P., and Branton, D. (1970). Membrane splitting in freeze-etching.]. Cell Biol. 45, 598-605. Pinto da Silva, P., and Nicholson, G. L. (1974).Freeze-etch localization of concanavalin A receptors to the membrane intercalated particles of human erythrocyte ghost membranes. Biochim. Biophys. Acta 363,311-319. Pinto da Silva, P., Branton, D., and Douglas, S. D. (1971). Localization ofA antigen sites on human erythrocyte ghosts. Nature (London)232, 194-196. Pinto da Silva, P., Moss, P. S., and Fudenberg, H. H. (1973).Anionic sites on the membrane intercalated particles of human erythrocyte ghosts. E x p . Cell Res. 81, 127138. Poste, G., and Allison, A. C. (1973). Membrane fusion. Biochim. Biophys. Acta 300, 42 1 -465. Prokop, O.,and Uhlenbmck, G. (1969).“Human Blood and Serum Groups.” Maclaren, London. Race, R. R., and Sanger, R. (1975). “Blood Groups in Man.” Blackwell, Oxford. Ralston, G. R. (1975). Proteins of the camel erythrocyte membrane. Biochim. Biophys. Acta 401,83-94. Reichstein, E., and Blostein, R. (1973). Asymmetric iodination of the human erythrocyte membrane. Biochem. Biophys. Res. Commun. 54,494-500. Reichstein, E., and Blostein, R. (1975). Arrangement of human erythrocyte membrane proteins, J . Biol. Chem. 250,6256-6263. Robinson, P. J., Bull, F. G., Anderton, B. H., and Roitt, I. (1975). Direct autoradiographic visualization in SDS-gels of lectin-binding components of the human erythrocyte membrane. FEBS Lett. 58,330-333. Ross, A. H., and McConnell, H. M. (1977). Reconstitution of band 3, the erythrocyte anion exchange protein. Biochem. Biophys. Res. Commun. 74,1318-1325. Rothstein, A., Cabantchik, Z. I., Balshin, M., and Juliano, R. (1975). Enhancement of anion permeability in lecithin vesicles by hydrophobic proteins extracted from red blood cell membranes. Biochem. Biophys. Res. Commun. 64, 144-150. Seaman, G. V. F., Vassar, P. S., and Kendall, M. J . (1969).Calcium binding to blood cell surfaces. Experientia 25,1259-1260. Segrest, J. P., Jackson, R. L., and Marchesi, V. T. (1972). Red cell membrane glycoprotein: amino acid sequence of an intramembranous region. Biochem. Biophys. Res. Commun. 49,964-969. Segrest, J. P., Kahane, I., Jackson, R. L., and Marchesi, V. T. (1973).Major glycoprotein

ERYTHROCYTE GLYCOPROTEINS

323

of the human erythrocyte membrane: Evidence for an ainphipathic molecular structure. Arch. Biochem. Biophys. 155, 167-183. Shin, B. C., and Carraway, K..L. (1974). Lactoperoxidase labelling of erythrocyte membranes from the inside and outside. Biochim. Biophys. Actu 345, 141-153. Ship, S., Shami, Y., Breuer, W., and Rothstein, A. Synthesis of tritiated 4,4’-Diisothiocyano-2,2’-stilbene disulphonic acid ([3HlDIDS) and its covalent reaction with sites related to anion transport in human red blood ce1ls.J. Membr. Biol. 33,311323. Spatz, L., and Stritmatter, P. (1971). A form of cytochroine b, that contains an additional hydrophobic sequence of40 residues. Proc. N u t l . Acud. Sci. U.S.A. 68,1@42-1046. Springer, G. F., and Desai, P. R. (1974). Common precursors of human blood group MN specificities. Biochem. Biophys. Res. Commun. 61,470-475. Springer, G . F., and Desai, P. R. (1975).Human blood group MN and precursor specificities. Carbohydr. Res. 40, 183-192. Springer, G. F., Nagai, Y., and Tegtmeyer, H. (1966).Isolation and properties of human blood group NN and meconium-Vg antigens. Biochemistry 5,3254-3272. Staros, J. V., and Richards, F. M. (1974). Photochemical labelling of the surface proteins of human erythrocytes. Biochemistry 13,2720-2726. Staros, J. V., Richards, F. M., and Haley, B. E. (1975). Photochemical labelling of the cytoplasmic surface of the membranes of intact human erythr0cytes.J. Biol. Chem. 250,8174-8178. Steck, T. L. (1972a). The organisation of proteins i n human erythrocyte membranes. I n “Membrane Research” (C. F. Fox, ed.), pp. 71-93. Academic Press, New York. Steck, T. L. (197213) Crosslinking of the major proteins of the isolated erythrocyte membrane. J . Mol. Biol. 66,295-305. Steck, T. L. (1974). The organisation of the proteins in the human red blood cell memi1rane.J. Cell Biol. 62, 1-19. Steck, T. L., and Dawson, G. (1974). Topographical distribution of complex carbohydrates in the erythrocyte membrane. J. Biol. Chem. 249,2135-2142. Steck, T. L., and Yu, J. (1973). Selective solubilization of proteins from red blood cell membranes by protein perterbants. J. Suprumol. Struct. 1,220-232. Steck, T. L., Ramos, B., and Strapazon, E. (1976). Proteolytic dissection of band 3, the predominant transinembrane polypeptide of the human erythrocyte membrane. Biochemistry 15, 1154-1161. Strapazon, E., and Steck, T. L. (1976).The binding of rabbit aldolase to band 3, the-predominant polypeptide of the human erythrocyte membrane. Biochemistry 15, 1421-1424. Sweeley, C. C., and Dawson, G. (1969). Lipids of the erythrocyte. In “Red Cell Membrane” (G. A. Jamieson and T. J. Greenwalt, eds.), pp. 172-227. Lippincott, Philadelphia, Pennsylvania. Tanford, C., and Reynolds, J. A. (1976).Characterization of membrane proteins in detergent solutions. Biochim. Biophys. Actu 457, 133-170. Tanner, M. J. A., and Anstee, D. J. (1976a). A method for the direct demonstration of the lectin-binding components of the human erythrocyte membrane. Biochem. J . 153, 265-270. Tanner, M. J. A., and Anstee, D. J. (1976b). The membrane change in En(a-) human erythrocytes. Biochem. J. 153, 271-277. Tanner, M. J. A., and Boxer, D. H. (1972). Separation and some properties of the major proteins of the human erythrocyte membrane. Biochem. J . 129,333-347. Tanner, M. J. A., and Gray, W. R. (1971). The isolation and functional identification of a protein from the human erythrocyte ghost. Biochem. J. 125, 1109-1117.

324

MICHAEL J. A. TANNER

Tanner, M. J. A., Jenkins, R. E., Anstee, D. J., and Clamp, J. R. (1976). Abnormal carbohydrate composition of the major penetrating membrane protein of En(a-) human erythrocytes. Biochem. J . 155,701-703. Tanner, M. J. A,, Anstee, D. J., and Judson, P. A. (1977). A carbohydrate deficient membrane glycoprotein in human erythrocytes of phenotype S-s-. Biochem. J . 165, 157-161. Tavassoli, M., and Crosby, W. H. (1973). Fate of the nucleus of the marrow erythroblast. Science 179,912-913. Taverna, R. D., and Langdon, R. G. (1973a). Reversible association of cytochalasin B with the human erythrocyte membrane. Biochim. Biophys. Acta 323,207-219. Taverna. R. D., and Langdon, R. G. (1973b). DGlucosyl isothiocyanate, an affinity label for the glucose transport proteins of the human erythrocyte. Biochem. Biophys. Res. Commun. 51,593-599. Terao, T., Irimura, T., and Osawa, T. (1975).Purification and characterization of a haemagglutinin from Arachis hypogaea. Hoppe-Seyler’s Z . Physiol. Chem. 356, 16851690. Thomas, D. B., and Winder, R. J. (1969).Structural studies on human erythrocyte glycoproteins: Alkali-labile oligosaccharides. J . Biol. Chem. 244, 5943-5946. Thomas, D. B., and Winzler, R. J. (1971). Structure of glycoproteins of human erythrocytes: Alkali-stable oligosaccharides. Biochem. J . 124, 55-59. Tillack, T. W., Scott, R. E., and Marchesi, V. T. (1972). The structure of erythrocyte membranes by freeze-etching. 11. Localization of receptors for phytohaemagglutinin and influenza virus to the intramembranous partic1es.J. Erp. Med. 135,12091221. Tomita, M., and Marchesi, V. T. (1975).Amino acid sequence and oligosaccharide attachment of human erythrocyte glycophorin. Proc. Natl. Acad. Sci. U S A . 72,29642968. Tosteson, D. (1959). Halide transport in red blood cells.Actu Physiol. Scand. 46,19-41. Triplett, R. B., and Carraway, K. L. (1972). Proteolytic digestion of erythrocytes, resealed ghosts and isolated membranes. Biochemistry 11, 2897-2903. Tuech, J. K., and Morrison, M. (1974).Human erythrocyte membrane glycoproteins: A study of interconversion. Biochem. Biophys. Res. Commun. 59,352-360. Visser, L., Robinson, N. C., and Tanford, C. (1975). The two domain structure of cytochrome b, in deoxycholate solution. Biochemistry 14, 1194-1199. Vos, J., Ahkong, Q. F., Boltham, G. M., Quirk, S . J., and Lucy, J. A. (1976). Changes in the distribution of intramembranous parttcles in hen erythrocytes during cell fusion induced by the bivalent cation ionophore A23187. Biochem.J. 158,651-653. Wallach, D. F. H. (1975). “Membrane Molecular Biology of Neoplastic Cells,” pp. 345408. Elsevier, Amsterdam. Walter, H., Selby, F. W., and Garza, R. (1967). On the countercurrent distribution of red blood cells. Biochim. Biophys. Acta 136, 148-150. Wang, K., and Richards, F. M. (1974). An approach to nearest neighbour analysis of membrane proteins. J . Biol. Chem. 249,8005-8018. Wasniouska, K., Drzenick, Z., and Lisowska, E. (1977). The amino acids of M and N blood group glycopeptides are different. Biochem. Biophys. Res. Commun. 76, 385390. Whitely, N. M., and Berg, H. C. (1974). Amidination of the outer and inner surfaces of the human erythrocyte membrane.J. Mol. B i d . 87,541-561. Wickramasinghe, S. N. (1975). “Human Bone Marrow.” Blackwell, Oxford. Winzler, R. J. (1969). A glycoprotein in human erythrocyte membranes. In “Red Cell

ERYTHROCYTE GLYCOPROTEINS

325

Membrane” (G. A. Jamieson and T. J. Greenwalt, eds.), pp. 157-171. Lippincott, Philadelphia, Pennsylvania. W i d e r , R. J. (1971). Carbohydrates in cell surfaces. Znt.Reo. Cytol. 29, 77-125. Yu, J., and Steck, T. L. (1975a). Isolation and characterization of band 3, the predominant polypeptide of the human erythrocyte membrane. J . BioE. Chem. 250,91709175. Yu, J., and Steck, T. L. (197%). Associations of band 3, the predominant polypeptide of the human erythrocyte membrane. J . Biol. Chem. 250,9176-9184. Yu, J., Fischman, D. A., and Steck, T. L. (1973).Selective solubilization ofproteins and phospholipids from red blood cell membranes by nonionic detergents.j. Suprumol. Struct. 1,233-247. Zvilichovsky, B., Gallop, P. M., and Blumenfeld, 0. 0. (1971). Isolation of a glycoprotein-glycolipid fraction from human erythrocyte membranes. Biochem. Biophys. Res. Commun. 44, 1234-1243. Zwaal, R. F. A., Roelofsen, B., and Colley, C. M. (1973).Localization of red cell membrane components. Biochim. Biophys. Actu 300, 1.59-182.

This Page Intentionally Left Blank

CURRENT TOPICS IN MEMBRANES AND TRANSPORT, VOLUME

11

Biochemical Determinants of Cell Adhesion LLOYD A. CULP Department of Microbiology School of Medicine Case Western Reserve University Cleoelund, Ohio

I. Introduction . . . . . . . . . . . . 11. Cell-Substrate Adhesion . . . . . . . . A. Role of Serum Macromolecules . . . . . B. A Cell Surface Adhesion Complex . . . . C. Adhesion to Collagen . . . . . . . . 111. Cell-Cell Adhesion . . . . . . . . . . A. Simple Eukaryotic Systems . . . . . . B. Avian and Mammalian Systems . . . . . C. Role ofcell Surface Glycosyltransferases(?) . IV. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . References

. . . . . . . 327 . . . . . . . 328

. . . . . . .

329

. . . . . . . 337 . . . . . . . 353

. . . . . . .

356

. . . . . . . 357

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

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

365 376 381 383

I. INTRODUCTION

During the past decade, a considerable amount of evidence has been found for the importance of cell surface glycoproteins and high MW proteoglycans as mediators of cell-to-cell and cell-to-substrate adhesion. This evidence is reviewed here for a wide variety of eukaryotic cell systems, from yeast and sponge cells as perhaps the simplest, to the much more complex avian and mammalian embryonic cell systems. The emphasis is on a critical analysis of the biochemistry of these determinants. For more thorough analyses of the theory and practice of measuring cell adhesion in uitro, the reader is referred to the reviews of Weiss (1970) and Curtis (1973). For analyses of the structure and synthesis of complex polysaccharides, complete reviews with somewhat differing emphases have been provided by Roseman (1970),Kraemer (1971), and Lennarz (1975);this information has now been integrated with more recent data by Sturgess et al. in this volume. This chapter does not survey lectin-mediated agglutination (see Nicolson, 1974),platelet-collagen interaction, sperm-egg adherence, 327 Copyright @ 1978 hv Academic Press. lnc. All rights of reproductir~nin any timn reserved. ISBN 0-12-153311-5

328

LLOYD A. CULP

lymphoctye-target cell adhesion, or the morphology of cellular adhesions (see reviews by McNutt and Weinstein, 1973; Revel, 1974; Overton, 1975) but focuses essentially on the biochemistry of cell surface components implicated as mediators of cell-cell or cell-substrate adhesion. (Unless noted otherwise, the term “substrate” generally refers to artificiai glass or plastic tissue culture substrates used for culturing avian or mammalian cells in uitro.) Initial considerations that cellular adhesion resulted from nonspecific interactions via neutralization of negative charge on cell surfaces or van der Waals’ forces acting over broad areas of surface membrane have now been replaced with models of cellular adhesion featuring the interaction of specific cell surface proteins and polysaccharides. These models, derived particularly from studies on cell-cell adhesion in simple eukaryotic cells or cell-substrate adhesion in mammalian and avian cells, are based on accumulating experimental evidence and emphasize a specificity of interaction analogous to that of enzyme-substrate, antigen-antibody, and lectin-polysaccharide interactions. There is also suggestive evidence that simple eukaryotic cells adhere by mechanisms that are biochemically simpler than those found in the much more complex avian and mammalian cells. Adhesion of the latter types of cells appears to be mediated by several classes of cell surface proteins and polysaccharides, which are probably quantitatively and/or qualitatively altered on the surfaces of specific populations of cells during conversion of a normal cell into a malignant cell.

II. CEU-SUBSTRATE ADHESION

In order for normal embryonic or adult cells from avian or mammalian organisms to survive and divide in uitro, they stringently require adherence to a specific type of tissue culture substrate-particular glasses or “activated” polystyrene plastic (e.g., see Martin and Rubin, 1974).Malignant cells, however, frequently lose this stringent anchorage dependence and can be adapted to grow as single cells in suspension. Cells transformed with oncogenic viruses or derived from tumors have been shown to move more aggressively across the substrate (Gail and Boone, 1972), to be more easily detachable from the substrate (Culp and Black, 1972a; Shields and Pollock, 1974), and to underlap readily neighboring cells during movement across the substrate (Di Pasquale and Bell, 1974). Martz et al. (1974) have proposed that the contact inhibitory phenomena displayed by normal cells rep-

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

329

resent a more effective formation of cell-substrate adhesions than that of cell-cell adhesions. The upper surface of cells in culture appears to lack adhesive components (Harris, 1973a; Di Pasquale and Bell, 1974; DiPasquale, 1975; Vasiliev et al., 1975), such that adhesion to other cells or to the substrate appears to be initiated at the leading lamella of the moving cell (Follett and Goldman, 1970; Abercrombie et al., 1971; Trinkaus et al., 1971; Witkowski and Brighton, 1971; Erickson and Trinkaus, 1976; Albrecht-Buehler, 1976). Therefore a detailed study of the mechanisms of the cell-substrate adhesive processes of normal and malignant cells should provide considerable insight into the different social behaviors displayed by these two cell types. Such analyses take advantage of the static nature of the artificial tissue culture substrate b y detaching cells with mild and nondegrading reagents, with the prospect of leaving a portion of cell surface material behind which mediates the adhesion process. Biochemical analyses of cell-substrate adhesion may eventually provide more effective methodologies and reagents for attacking the mechanism of cell -cell adhesion. A. Role of Serum Macromolecules

1. BACKGROUND AND PHENOMENOLOGICAL APPROACHES That mammalian cells do not adhere directly to a polystyrene or glass substrate but to an adsorbed layer of serum components was first shown by Rosenberg (1960), Taylor (1961), and Weiss (1961).The nature of serum interaction with substrates was studied more systematically by Takeichi (1971), who showed that trypsinized chick embryonic sclera cells adhered to naked substrates in such a fashion that they were not detachable b y conventional trypsin or (ethylenedinitri1o)tetraacetic acid (EDTA) treatments; the cells were, however, detachable after adherence to serum-preincubated substrates, indicating a stable association of serum components with the substrate. The importance of serum proteins in this process was demonstrated by treating preadsorbed substrates with trypsin, producing a loss of ability of freshly attached cells (no serum in the medium) to be detached b y trypsin or EDTA. Similar effects for substrate-adsorbed serum components were observed in erythrocytes (Bolund et al., 1970), HeLa cells (Bolund et al., 1970; Unhjem and Prydz, 1973), human fetal lung cells (Witkowski and Brighton, 1972); suspension-adapted baby hamster kidney (BHK) cells (Grinnell, 1976a), and normal, SV40-transformed, and revertant mouse 3T3 cells (Culp and Buniel, 1976). Imaginative

330

LLOYD A. CULP

methods for the processing of serum-adsorbed tissue culture substrates by Yaoi and Kanaseki (1972), Revel and Wolken (1973), Pegrum and Maroudas (1975),and Grinnell et al. (1976)for transmission electron microscopy studies indicated that the serum components were evenly spread across the substrate as a layer whose thinness (3-5 nm) suggested that it may be unimolecular in depth.

2. CHEMISTRY OF ADSORBEDSERUM COMPONENTS The glycoprotein nature of this adsorbed serum layer was demonstrated with experiments using the lectin concanavalin A (Con A). A variety of cells incubated with Con A during reattachment to fresh serum-adsorbed substrate adhered so tenaciously that they were no longer detachable by trypsinization or EDTA treatment (Grinnell, 1973; Mori et al., 1973; Sato and Takasawa-Nishizawa, 1974).This effect was reversed by adding a-methylmannoside to cultures and was mimicked b y pretreating the adsorbed serum layer with Con A before performing the attachment process in medium lacking Con A and serum, a direct indication that Con-A-binding glycoproteins were indeed components of the adsorbed serum layer (Grinnell, 1973) (although this is by no means proof that these specific components are mediators of the natural adhesion process). Since Con-A binding prevented EDTA-mediated removal of cells and since EDTA treatment probably does not dissociate outer cell surface adhesive components (to be discussed in Section II,B), it is quite likely that Con A cross-links serum-adsorbed glycoproteins with a wide variety of cell surface glycoproteins over much of the cell undersurface, and not just with cellular glycoproteins which are specifically adhesion components. This type of interaction would then prevent EDTA from causing cell detachment by its postulated mechanism (see evidence in Section 11,B). Consistent with this interpretation was the finding (Sato and Takasawa-Nishizawa, 1974) that Con-A-induced resistance to trypsinization could not be demonstrated below 15"C,while cell attachment occurred moderately well below 15°C-this temperature being an approximate temperature for inhibiting Con-A-mediated rearrangement of cell surface glycoproteins (Noonan and Burger, 1973; Nicolson, 1974) because of the fluid-to-gel transition of the lipid moieties in the bilayer (Oseroff et al., 1973; Wisnieski et al., 1974). Treatment of serum-adsorbed substrates with purified histones, which may bind to negatively charged glycoproteins and/or glycosaminoglycans (GAGS)in the serum layer, affected the morphology and packing behavior of HeLa cells (Bases et al., 1973). Treatment with

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

331

polylysine or other basic polymers stimulated the clonal growth of primary and normal cell lines (McKeehan and Ham, 1976). The importance of negatively charged moieties (polysaccharides?) in the adsorbed serum layer has also been suggested by the experiments of Grinnell (1976a); blockage of carboxyl groups by reaction of the adsorbed serum layer with a soluble carbodiimide in the presence of glycine methyl ester abolished the time lag during the initial attachment of BHK cells, whereas blockage of sulfhydryl or free amino groups had little effect. An alternative interpretation of these latter experiments is that nonspecific alteration of the conformation of one or many of the adsorbed serum proteins occurs because of the lack of specificity in using these reactive chemical reagents and that cell surface receptor sites do not interact with these components in a proper fashion. The addition of polycationic ferritin to serum-preadsorbed substrates resulted in attachment kinetics similar to those observed with naked substrate but also prevented cytoplasmic spreading over the substrate (Grinnell, 1976a). Early models of cell-substrate adhesion suggested that divalent cations mediate the adhesion process by cross-bridging negatively charged cell surface components with negatively charged adsorbed serum components (e.g., see Curtis, 1973; Culp, 1974; Grinnell, 1976a).Recent evidence, however suggests that this is probably oversimplistic and lacking in specificity. The attachment of chick embryonic sclera cells to artificial substrates was only partially dependent on the presence of a divalent cation (Takeichi and Okada, 1972). Centrifugation of cells against the substrate overcame the requirement for divalent cations with the formation of stable adhesive bonds as measured by sensitivity to detachment (Milam et al., 1973; Grinnell, 1974a); however, the spreading of cell processes over the substrate did not occur under these conditions. Rosen and Culp (1977) demonstrated that ethylenebis(oxyethylenenitri1o)tetraacetic acid (EGTA)-mediated detachment of cells occurred by pinching off of the footpads (Revel et al., 1974) by which cells adhere to the substrate, with minimal effects on the morphological organization of these footpads. Grinnell(1976a) measured the kinetics of attachment and extent of cytoplasmic spread over substrates pretreated with fetal calf serum in serumless medium with varied concentrations of Ca2+;the absence of Ca2+in the medium had only a slight effect on the attachment kinetics but completely prevented cytoplasmic spreading. These spreading processes are probably mediated by reorganization of the cytoskeleton, requiring Ca2+ and perhaps Mg”. These experiments strongly suggest that divalent cations, particularly Ca2+,may not be required

332

LLOYD A. CULP

for formation of the adhesive bonds per se between the cell surface and substrate-adsorbed serum components but may be required for organization of the subsurface cytoskeleton which is no doubt highly important in strengthening adhesion sites (see evidence in Section II,B) and in spreading the cytoplasm over the substrate. Divalent cations may also be stringently required for the recovery of protein biosynthesis for repair of cell surface damage caused by trypsinization (Takeichi and Okada, 1972; Grinnell, 1974a) and reorganization of disarrayed surface components to permit the initiation of adhesion. Adhesion to serum-adsorbed substrates is necessary to permit normal cellular behavior in uitro. Grinnell(1974b) showed that pretreatment of suspended BHK cells with trypsin, N-ethylmaleimide, or glutaraldehyde had no deleterious effect on adherence to naked substrate but inhibited attachment and cytoplasmic spreading over serumpreadsorbed substrate. Culp and Buniel(l976) showed that substrates preadsorbed with serum components induced (in the absence of serum in the medium) (1)a time lag before the initiation of stable adhesion, an indication that topographical rearrangement of cell surface and/or substrate-bound serium components is required to achieve complementary binding between cell and serum components; (2) a reduction in the maximum number of attached cells, perhaps by selecting for cells whose surface membrane organization was only reversibly damaged b y prior detachment with EGTA treatment; (3) the stable attachment of cells over a 2.4-hour period; and (4)pseudopodial spread of the cell over the substrate. The requirement for adsorption of specific serum macromolecules may explain the different behaviors of cells attaching to or moving onto various types of substrates. In several fibroblastic cell types Harris (197313) observed a preferential gradient of affinity for various substrates in medium containing serum: palladium-coated < cellulose acetate < glass < nonwettable polystyrene < Falcon tissue culture plastic. Similarly, Letourneau (1975) reported a gradient of preferential adherence of chick neuronal growth cones in serumcontaining medium, with adherence to glial cell surfaces and polyornithine-coated substrates being the greatest, followed by adherence to collagenous substrates, tissue culture plastic, and palladiumcoated plastic. Chipowsky et al. (1973) measured the attachment of various cell types in medium without serum to Sephadex beads derivatized with specific monosaccharides and observed preferential adherence of SV3T3 cells to D-galactose-derivatized beads, rather than beads linked with Dglucose or N-acetyl-D-glucosamine. Martin and Rubin (1974) observed very different patterns of attachment, spread-

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

333

ing, and growth of chick embryonic fibroblasts on bacteriological and tissue culture plastics. It is highly likely that these effects are mediated by the presence or absence of specific serum macromolecules which act as receptor sites for specific cell surface components, rather than differences in direct interaction between cell surface components and in the described substrates.

3. SERUMFRACTIONATION STUDIES What are the molecular complexity and chemical composition of the adsorbed serum proteins and polysaccharides? Two approaches have been taken in answering this question: (1)extraction of serum components adsorbed to substrates under different cell attachment and growth conditions, or (2) fractionation of whole serum to enrich for classes of macromolecules which permit normal attachment and spreading on the substrate. Culp and Buniel (1976) succeeded in extracting several substrate-adsorbed serum components by the use of sodium dodecyl sulfate (SDS) under extraction conditions which removed virtually all substrate-bound cellular proteins and polysaccharides using plastic substrates. The serum components were resolved into approximately 12 distinct bands as determined by coomassie blue staining after slab SDS polyacrylamide gel electrophoresis (SDS-PAGE) analysis; this is verification that specific serum proteins bind to tissue culture plastic and not a representative proportion of all the protein components of serum (which form a continuous “smear” of stainable protein throughout an SDS-PAGE gel). One of these components (S,) coelectrophoresed with bovine serum albumin. Since pretreatment of substrates with bovine serum albumin alone is insufficient to catalyze the adhesion changes noted above (Grinnell, 1976b; Culp, unpublished data), it is unlikely that S6 (if it indeed is albumin) plays an important receptor site role for cell surface components by itself. SDS extraction of serum-preadsorbed glass substrates and electrophoresis of the extracts b y SDS-PAGE have demonstrated that glass substrates bind most of the proteins which can be adsorbed by plastic substrates but in different relative proportions; in addition, glass substrates bind many other serum proteins which plastic substrates do not (Haas and Culp, unpublished data). The chemical composition of the substrate therefore is important in determining the array of serum components which can bind. It has not been established that the SDS extraction method, although quantitatively effective for the removal of cellular proteins and polysaccharides, is effective in stripping the substrate of all serum proteins. [Unfortunately,

334

LLOYD A. CULP

treatment with alkali, which appears to be quantitatively effective in removing adsorbed serum proteins (Grinnell, 1975), does not yield intact proteins which are bandable in SDS-PAGE analysis (Culp, unpublished data).] Two of the coomassie blue-stainable serum proteins, S, and S z , identified by Culp and Buniel (1976), were located at the tops of the stacking and separating gels where the high MW cellular proteoglycans (GAG-containing) band (Culp, 1976a). These cellular proteoglycans appear to play an important, as yet unknown, role in the adhesion process, and this raises a question as to a potential role for serum proteoglycans. However, it has not been established that S1 and S, bands have a high polysaccharide/protein ratio, although these two components are the principal moieties radioiodinated in the lactoperoxidasecatalyzed treatment of cell-free adsorbed serum layers (Chi and Culp, unpublished data), which is consistent with the large size and accessibility of a presumptive proteoglycan on the substrate. Although adsorbed serum layers are sensitive to trypsin digestion as far as loss of catalysis of physiologically compatible adhesion is concerned (Grinnell, 1975; Culp and Buniel, 1976), they are not very sensitive to treatment with hyaluronidase (Culp and Buniel, 1976); the latter digestions are complicated by persistent adsorption of the enzyme to the serum layer, lack of an irreversible inhibitor of the enzyme, and incomplete digestion of long polysaccharide chains. Further evidence will be required to determine (1)if S, and S, are homogeneous GAGcontaining proteoglycans and/or denatured proteins which collect at gel interfaces and (2) if they play a stringent role in the cell-substrate adhesion process. Which of the 12 protein bands identified in SDS-PAGE gels by Culp and Buniel(l976) are glycoproteins has not been determined. At least 10 of these adsorbed serum proteins (S, to S,,) persist after the growth of BALB/c 3T3 or SV40-transformed 3T3 cells to confluence (Culp, 1976a). This may indicate (1)that cellular adhesion and movement on the adsorbed layer do not result in extensive proteolytic degradation of these components, (2) that degradation can be balanced by replacement of these components from the medium pool, or (3) that the focal nature of cellular adhesion sites (Revel et al., 1974; Rosen and Culp, 1977) affects the integrity of only a small percentage of the substrate area. These alternatives have not been resolved. A second approach in analyzing the importance of substrate-adsorbed serum components has been the fractionation of whole serum to enrich for factors which catalyze the spreading of cells over the substrate. Grinnell (1976a) obtained a fraction of fetal calf serum which

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

335

eluted from DEAE-cellulose at a salt concentration of 0.3 M (perhaps enriched in negatively charged polysaccharides?) and which catalyzed the normal pseudopodial spread of cells over substrates pretreated with this material during the attachment of suspensionadapted BHK cells in medium containing 5% fetal calf serum. Pretreatment of the substrate with other fractions from the DEAEcellulose column, including the fetuin-enriched fraction, blocked the adsorption and catalytic effects of this serum spreading factor (SF). Preadsorption of SF alone was later shown to catalyze normal cell spreading in media lacking serum (Grinnell, 197613); preadsorption of bovine serum albumin or bovine gamma globulin to substrates permitted cell attachment but not cytoplasmic spreading-further evidence that cells can attach to many different types of substrate in an unstable and nonphysiological fashion and that specific serum components are required for proper attachment and eventual cytoplasmic spreading. The cytoplasmic spreading activity of whole serum or SF was shown (Grinnell, 1976b) to be inhibited by an antiserum prepared against SF, either b y the coaddition of SF (or serum) and anti-SF to the medium containing the attaching cells or by pretreating SF-preadsorbed (or serum-preadsorbed) substrates with anti-SF. Again, initial attachment processes were only partially inhibited, while cytoplasmic spreading was completely inhibited. SF activities similar to that found in fetal calf serum after DEAE-cellulose chromatography were also obtained from calf, porcine, human, rabbit, and chicken serum and were effective in catalyzing the spread of suspension-adapted BHK, polyoma-transformed BHK (PyBHK), mouse L929, HeLa, and Chinese hamster ovary (CHO) cells. When the immunoglobulin fraction of an antiserum prepared against BHK surface membrane was adsorbed to the substrate, BHK cells attached and spread normally (Grinnell, 1976b, and personal communication). This suggests that any ligand complementary to one or more cell surface determinants may be sufficient to permit cytoplasmic spreading. It will be interesting to determine which cell surface components generate antibodies which are effective in this assay. Similar effects were noted for the adsorption of Con A alone, suggesting that the “select” cell surface determinant(s) may be a Con-A-binding glycoprotein (Grinnell, 1976b; spreading was inhibited in the presence of a-methylmannoside), perhaps the Con-A-binding large external transformation-sensitive (LETS) glycoprotein (Burridge, 1976) which is a major component of cellular substrate-attached material (SAM) (Culp, 1976a,b; see further evidence in Section I1,B) and is

336

LLOYD A. CULP

present in modified form in serum as cold-insoluble globulin (CIG). SDS-PAGE analysis of the SF from fetal calf serum under reducing conditions identified approximately eight distinct coomassie bluestainable bands (Culp and Grinnell, unpublished data). Two of these bands in SF coelectrophoresed with the high-MW S, and S2 components extractable from serum-adsorbed substrates (Culp and Buniel, 1976). A major protein component in SF coelectrophoresed with the major cell surface large external transformation-sensitive (LETS) glycoprotein (Hynes, 1976) which has been implicated in the substrate adhesion process (see evidence in Section II,B), although this band in SF has not been definitively shown to be CIG. It will be interesting to determine if this component of SF is cold isoluble globulin (CIG) which is a prominent component of animal serum and shares many properties with the LETS glycoprotein. Highly purified CIG from sera mimicked the various activities of SF, suggesting that it may be an important serum-derived mediator of adhesion (Grinnell, personal communication) perhaps by binding to cell surface LETS glycoprotein. SF also contained additional proteins which coelectrophoresed with the S; and S6 components which are SDS-extractable from serum-adsorbed substrates (Culp, 1976a). SF has recently been shown to be approximately 10% carbohydrate (Grinnell, personal communication). Therefore fractionation of whole serum to enrich for cytoplasmic spreading activity has coincidentally enriched for several protein moieties which electrophorese similarly to some proteins SDS-extracted from serum-preadsorbed substrates. If cell surface receptor sites bind to specific adsorbed serum components, why aren’t these receptor sites blocked during the addition of suspended cells to serum-containing medium for reattachment? Perhaps the most likely reason why such a blocking phenomenon does not occur is that the SFs in the medium may not be reactive toward surface receptor sites until they bind to the substrate surface. This binding may perhaps change the conformation of the serum protein, resulting in reactivity toward cell surface receptors, or lead to proteolytic modification upon substrate binding with liberation of the active site, or perhaps result in association with another serum component (interaction of both of these components may be required to generate an active site). A precedent exists for this in the substrate-mediated activation of Hagemann factor XI1 (Austen, 1974). Another question arises concerning the mechanism by which a few cell types can adhere to artificial substrates and grow in the total absence of serum. Perhaps these unique cell types synthesize and secrete collagen and/or serumlike factors which autocatalytically promote adhesion

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

337

and growth (no attempt has been made to examine the substrate adhesion mechanisms of these cell types as far as we are aware). The limited number of serum proteins in the substrate-bound fraction (Culp and Buniel, 1976) or in SF (Grinnell, 197613) offers considerable encouragement for the eventual identification of the mechanism by which one or more of these proteins, glycoproteins (CIG in particular), and/or GAGS mediate the adhesion process. In summary, proper cell adhesion to the substrate during initial attachment of suspended cells, followed by cytoplasmic spread over the substrate, appears to be mediated by specific serum components which may be glycoproteins (CIG?) and/or GAG-containing proteoglycans and which appear to be activated upon binding to the substrate. Evidence is accumulating that divalent cations may not act as simple cross-bridging agents between negatively charged serum and cell surface components but that the adhesive bonds involve highly complementary protein-protein and particularly protein-polysaccharide interactions.

6. A Cell Surface Adhesion Complex

1. BIOLOGYAND ORIGIN OF SUBSTRATE-ATTACHED MATERIAL

Identification and characterization studies on cell surface components involved in substrate adhesion in avian and mammalian cells were stimulated by the observation of a so-called microexudate deposited by cells onto the substrate and detected by an ellipsometric method as proteinaceous material (Rosenberg, 1960). The production of microexudates was later shown to be prevented by the addition of metabolic inhibitors to attaching cells, suggesting its origin as cell surfhce material (Poste et d.,1973).The microexudates left on the substrate after cell detachment were shown to affect cell attachment and cytoplasmic spread markedly (Weiss, 1961; Taylor, 1961; Bolund e t d.,1970) and to contain cell surface antigenic material (Weiss and Lachmann, 1964). These microexudates may be identical to the electron dense plaques observed at cell-substrate adhesion sites in electron micrographs of thin sections (Abercrombie et d.,1971; Yaoi and Kanaseki, 1972; Revel and Wolken, 1973; Pegrum and Maroudas, 1975).In the simple eukaryote Entamoeba histolytica, this material has been shown to contain Con-A-binding sites and material stainable with ruthenium red or alcian blue, consistent with its being negatively

338

LLOYD A. CULP

charged proteoglycan material (Pinto da Silva et al., 1975). Similarly, Grinnell et al. (1976) observed some polycationic ferritin penetration into the 9-nm gap between the lipid bilayer and serum layer of attaching BHK cells which may reflect binding to negatively charged macromolecules (polysaccharides?). Culp (1974)studied the effects of substrate-attached material (SAM), which remains firmly adherent to the substrate after EGTA-mediated removal of mouse 3T3 cells, upon the attachment and growth of several cell types. The attachment kinetics of 3T3 or revertant cells [selected from a population of SV40-transformed 3T3 cells by Con-A treatment (Culp and Black, 1972b)l were unaffected by 3T3 SAM; the initial rates and maximal levels of attachment of SVT2 cells were positively affected by 3T3 SAM, but much of this effect appears to result from modification of the adsorbed serum layer of SAM coatings (Culp and Buniel, 1976). [The term “coating” here does not indicate an evenly spread layer of material, since SAM has been shown to exist as focal pools of cellular material on the substrate (Culp, 1975; Rosen and Culp, 1977).] SV4O-transformed cells were uniquely affected in their growth and movement behavior on 3T3 SAM coatings (Culp, 1974). As colonies developed on SAM-coated surfaces, transformed cells became epithelioid in morphology, resisted crawling under neighboring cells, and resisted movement away from the edges of colonies. These effects could not be mimicked by adsorbed serum layers and were undoubtedly caused by cell surface materials left in SAM. No effects were noted on the growth of 3T3 or revertant cells, suggesting that the chemical components of this material (described in Sections II,B,2 and II,B,3) may be important mediators of the decreased adhesiveness of virus-transformed cells when compared to that of normal and revertant cells. Density-dependent inhibition of growth and movement were not induced in mass populations of transformed cells, perhaps because SAM,materials can be functionally inactivated at high cell densities. Based on the EGTA-resistant binding of SAM to the adsorbed serum layer and its content of negatively charged hyaluronic acid, a model was proposed (Culp, 1974) for Ca2+-mediatedcross-linking between negatively charged SAM and other cell surface macromolecules; further evidence to be described below indicates that such a model for cell-substrate adhesion is probably invalid. In a similar set of experiments, Weiss et al. (1975) evaluated the growth of normal and SV40-transformed mouse cells on substrates covered with material resistant to EDTA-mediated removal of cells (termed here EDTA-SAM to differentiate it from EGTA-SAM). The initiation of growth of either cell type was stimulated when cells were

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

339

inoculated at low densities onto EDTA-SAM-coated substrates; there was little growth stimulation or inhibition at high cell densities by these coatings. EDTA-SAM deposited by subconfluent cultures appeared to be more effective in stimulating growth than EDTA-SAM from confluent cultures. Moore (1976) reported the secretion of adhesion-promoting factors (APF) into serumless medium of confluent chick, mouse, or rat embryo fibroblasts. This APF activity increased the rute of attachment and the long-term stability of adherence of avian sarcoma virustransformed fibroblasts which had been trypsin-subcultured into dishes containing serumless medium. Preincubating cells with this factor was not effective in promoting adhesion, while preincubating the substrate was effective (not species-specific). Interestingly, normal fibroblasts secreted three to six times more of this activity during a 24-hour incubation than avian sarcoma vims-transformed cells. Mallucci and Wells (1976) have also reported the secretion of a factor into serumless medium which promoted the reflattening of cells. It will be interesting to determine if the active factors in these extracts are the LETS gl ycoprotein and/or GAG-containing proteoglycans. Another productive approach in identifying cell surface components which mediate cell-substrate adhesion involved the quantitative removal of cells from the substrate with the Ca2+-specificchelator EGTA, a removal process which minimizes damage to cell surface components and leaves a small portion of the cell surface protein and polysaccharide on the substrate for subsequent removal and analysis (Culp and Black, 1972a). This SAM (Terry and Culp, 1974) was tenaciously bound to the substrate, requiring alkali or SDS treatment for quantitative removal. Interestingly, BALB/c 3T3 and Con A-selected revertant variant cells left approximately 3 4 % of their cellular protein and 9-11% of their polysaccharide as SAM, while SV40-transformed 3T3 (SVT2)cells left 1-1.5% of their protein and 3 4 %of their polysaccharide (Culp and Black, 1972a). This may be an indication that the relative amounts of this material determine the more tenacious adhesive properties of the 3T3 and revertant cells. The minimal amounts of RNA and DNA found in this material confirmed that this material did not result from the persistence of whole cells or very large pieces of cytoplasmic material (see further evidence below on the origin of SAM). Its elevated content of polysaccharide also suggested its origin as cell surface material. Another possibility is that SAM is deposited on the substrate from medium-secreted proteins and polysaccharides. However, Culp et al. (1975) studied the accumulation of SAM as a function of cell growth

340

LLOYD A. CULP

and found that its accumulation ceased when cells had saturated the substrate surface, even though SVT2 cells continued to pile into dense layers and 3T3 cells continued to secrete large amounts of protein and polysaccharide into the culture medium. When cells were pulse-radiolabeled at different densities during culture growth, the deposition of SAM on the substrate was markedly diminished after normal or transformed cells had covered the substrate (Culp et al., 1975). When cells were grown as sparse colonies in medium containing radioactive precursors and then removed by EGTA treatment, the radioactive SAM (detected autoradiographically) was observed only at locations on the substrate of colonies and not in intercolony spacings (Culp, 1975).These experiments argue for direct contact between the cell and substrate as a prerequisite for SAM deposition and against secretion of macromolecules into the medium followed by their random binding to the substrate. High-resolution autoradiography was used by Culp (1975) to demonstrate that SAM protein and polysaccharide were not evenly spread on the substrate but were concentrated in pools. The density of these pools was similar for 3T3 and SVT2 cells and was similar to the density of the cellular footpads by which cells adhere to the substrate (Revel et aZ., 1974).When preradiolabeled cells were allowed to reattach to fresh substrate after EGTA-mediated suspension (Culp, 1975), the SAM deposited during the initial 24 hours of cell spreading and movement over the substrate exhibited two interesting properties: (1) The geometric pattern of the SAM mimicked the pattern of spreading of pseudopodial processes, and (2) the area of SAM deposition was approximately twice the area occupied by the cell at the time of its removal. The latter observation indicated that SAM was derived from two different sources: (1) cell-substrate adhesion sites directly participating in cellular adhesion which are disrupted during the EGTA treatment, and (2) cell-substrate adhesion sites which have been disrupted during the normal cell movement, leaving “footprints” on the substrate (Culp, 197613; Culp et al., 1978). Adhesion to the substrate has been shown to occur at focal regions of the undersurface of the cell called footpads (Revel et al., 1974). These adhesion sites remain firmly bound to the substrate during mitosis (Revel et d . ,1974; Abercrombie and Dunn, 1975).During gentle trypsinization, the cell body rounds up while remaining linked to the morphologically intact footpads via stressed retraction fibers which eventually break because of shearing forces generated in agitated cultures; these experiments indicate that the cell surface components mediating the footpad adhesion process are relatively inaccessible to trypsin (Revel et al., 1974).

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

34 1

The experiments of Revel et al. (1974) therefore established that gentle trypsinization of cells off the substrate is not necessarily mediated by damage to adhesive components in the footpad adhesion site but possibiy by damage to the subsurface cytoskeleton resulting in cell rounding and breakage of the elastic retraction fiber, releasing cells into suspension while footpads remain on the substrate. Rosen and Culp (1977),utilizing scanning electron microscopy (SEM), established a similar sequence of events for EGTA treatment, indicating that the chemical components present in SAM are not linked by simple divalent cationic mechanisms and that SAM components probably represent all the important constituents of the footpad adhesion site (Culp et al., 1978). SEM analysis (Rosen and Culp, 1977) established that footpads are generated by the branching of 0.l-pm-diameter filopodia which initially make contact with the substrate, followed by membrane extension across the substrate between these fingerlike processes to form mature footpads and thickening of the membranous extension between the cell body and the footpad. EGTA treatment of flattened cells, however, was shown to cause cell rounding and a pulling away from the persistent footpads, generating stressed retraction fibers between the cell body and the footpads which eventually broke because of culture agitation. The pools of SAM were shown by Rosen and Culp (1977) to be virtually intact footpads left on the substrate with a nub of membrane where the retraction fiber had presumably broken away after the removal of well-spread cells. When attaching cells were removed, however, SAM appeared to be partially disrupted footpad material; the material at the periphery of the footpad persisted, while internal material at the pinching-off site was missing, indicating more tenacious binding to the substrate of peripheral material during the formation of footpads. The existence of SAM as pinched-off footpad material is consistent with prior evidence (Culp and Black, 1972a) that SAM contained only 1-3% of the cell’s protein (perhaps even less when footprint material is subtracted from the total). 2. CHEMISTRY OF SUBSTRATE-ATTACHED MATERIAL:GAGs

The chemical composition of SAM polysaccharide was initially examined by Terry and Culp (1974). The majority of the glucosamineradiolabeled and alkali-extracted polysaccharide was identified as the GAG hyaluronic acid. Sulfate radiolabeling also indicated a pool of sulfated GAGs (Terry and Culp, 1974; Roblin et al., 1975a) which was not readily observed in glucosamine-radiolabeledpreparations. The major sulfated GAG’S in SDS-extracted SAM have recently been identified as heparan sulfate and chondroitin sulfate (B. Rollins and

342

LLOYD A. CULP

L. Culp, unpublished data). Some evidence indicated a different pattern of metabolism in the substrate-attached hyaluronate and the sulfated GAGs. Hyaluronate appears to accumulate as a function of cell growth without sizable turnover (its deposition is selectively inhibited in growth-inhibited 3T3 cultures), whereas the entire pool of sulfated GAGs in SAM turns over with a half-life of 2 4 hours with a reduced accumulation as a function of time (Roblin et al., 1975a; Culp, unpublished data; chase analysis of SAM-containing sulfated GAGs radiolabeled for short or long periods of growth resulted in a rapid and quantitative loss of sulfate radioactivity from SAM-a metabolically stable pool of sulfated GAGs would have accumulated during longer radiolabeling periods). These data suggest that both hyaluronic acid and sulfated GAGs may be important components during the formation of footpad adhesion sites but that reorganization of macromolecular components at the footpad adhesion site and/or loss of footpads as footprints during cell movement result in an extensive and specific loss of some sulfated GAGs (along with two other components to be described in Section II,B,3). Two principal questions to be asked are (1)whether these GAGs are organized into proteoglycan complexes similar to those observed in cartilage tissues (Hardingham and Muir, 1974; Heinegard and Hascall, 1974), and (2) whether the hyaluronate and the sulfated GAGs assume roles as positive or negative regulators of adhesive bond formation (or if their roles change during footpad maturation). The helical conformations observed for GAG polysaccharides (Dea et al., 1973; Atkins and Sheehan, 1973; Winter et al., 1975) may provide an important organizational matrix for other cell surface and serum macromolecules at the adhesion site. The sizable quantities of GAGs in SAM and the alteration of their synthesis upon virus transformation of normal cells may be indicators of their importance in the differing adhesiveness of normal and malignant cells (Defendi and Gasic, 1963; Hamerman et al., 1965; Kraemer, 1971). SV40 transformation of 3T3 cells has been shown to result in decreased rates of synthesis and cell-associated accumulation of sulfated GAGs (Goggins et al., 1972; Roblin et al., 1975a; Underhill and Keller, 1976; Cohn et al., 1976). Transformed cells treated with dibutyryl cAMP synthesized 2.5 times more sulfated GAGs than untreated control cells (Goggins et al., 1972); similarly, dibutyryl cAMP treatment of rat fibroblasts led to increased levels of hyaluronic acid (Koyama et al., 1976). Con-A-selected revertant cells have been shown to synthesize severalfold higher levels of both sulfated GAGs and hyaluronic acid when compared to their parental SVT2 cells (Roblin et al., 1975a). Carcinogen-transformed cells treated with high MW

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

343

dextran sulfates in serum-containing medium displayed reduced saturation densities, while their doubling time was unaffected (Goto et al., 1973). In general, the synthesis of hyaluronic acid appears to be more closely linked to cell growth than the production of sulfated GAGs. Density-inhibited fibroblasts were shown to turn off the production of hyaluronate selectively, while sulfated GAG synthesis persisted (Tomida et al., 1975; Moscatelli and Rubin, 1975; Cohn et al., 1976). The stimulation of growth-inhibited cells by fresh serum-containing medium (Tomida et al., 1975; Moscatelli and Rubin, 1975; Lembach, 1976) or mouse epidermal growth factor (Lembach, 1976) led to increased rates of synthesis of cell-associated hyaluronic acid. All these studies establish the sensitivity of GAG production to cellular growth (and movement?) and virus transformation, alterations which may affect cell-substrate adhesion because of the prominence of these components in SAM. However, consideration must also be given to the possibility that GAGs serve many cellular functions other than adhesion. Atherly et ul. (1977) selected variants of CHO cells which were resistant to trypsin-mediated detachment from substrates and examined their ability to synthesize GAGs. Although these variants synthesized sulfated GAGs at levels similar to those of the parent population and grew as substrate-adherent cells with increased resistance to trypsinor EDTA-mediated detachment, hyaluronic acid synthesis was greatly diminished. This suggests that hyaluronate per se is not stringently required for cell-substrate adhesion (unless serum-containing hyaluronate can substitute for cellular synthesis in these particular cells). Since trypsin sensitivity may reflect damage to cell surface components other than those directly involved at footpad adhesion sites (Revel et al., 1974), much more information on the molecular aspects of trypsin-mediated cell surface damage will be required to determine the true selective environment required to obtain these variants. In contrast to this trypsin-resistant variant with diminished hyaluronateproducing capacity, substrate-adherent variants from a population of suspension-adapted CHO cells synthesized a threefold higher level of hyaluronate (Kraemer, 1976). In any case, selection in either direction (for or against substrate adherence) identifies cells with preferentially altered hyaluronic acid production and altered adhesiveness. Spooner and Conrad (1975)reported that the inhibition of polysaccharide synthesis with 6-diazo-5-oxo-~-norleucine(DON, a glutamine analog which partially inhibits endogenous formation of glucosamine 6-phosphate) had minimal effects on cell movement, suggesting that

344

LLOYD A. CULP

GAG synthesis may not be required for the formation of new adhesion sites. However, these experiments are complicated by the persistence of low levels of polysaccharide synthesis in the presence of DON and sizable pools of cell surface GAGs, the synthesis of which would not immediately precede the exercising of their postulated functions in adhesion formation (Mapstone and Culp, 1976; Winer and Culp, unpublished data). Slab gel electrophoresis of SAM (Culp, 1976a) separated the glucosamine-radiolabeled GAGs into three size classes and the sulfateradiolabeled GAGs into two size classes. Leucine radiolabeling identified protein material that coelectrophoresed with the three classes of glucosamine-radiolabeled GAGs [called GAG-associated protein (GAP)]. Some evidence indicates that portions of the GAP are linked to the sulfated GAGs and perhaps the hyaluronic acid as proteoglycans (Culp, 1976b; Culp et al., 1978). The protein linked to these GAGs may provide the functional and organizational specificity for these proteoglycans. Unfortunately, little is known about this protein material or the nature of its association with the GAG polysaccharides (further analyses of the three GAG bands may reveal even greater complexity in these proteoglycan components). 3. CHEMISTRY OF SUBSTRATE-ATTACHED MATERIAL: GLYCOPROTEINS AND PROTEINS

In addition to the large amounts of GAGS observed in SAM, several additional cell “surface” components were enriched in SAM compared to surface membrane preparations (Culp, 1976a,b) and were identified subsequent to slab gel electrophoresis under reducing conditions and in the presence of SDS. These components are listed in Table I, and the nature of several of them has yet to be determined. The C, glycoprotein appears to be the LETS glycoprotein identified in hamster cells (Hynes, 1973,1976); a similar surface membrane component has been called the Z (Wickus and Robbins, 1973) or CSP protein (Yamada and Weston, 1974) in chick embryonic cells, and fibronectin (Vaheri and Ruoslahti, 1975; Keski-Oja et al., 1976) in human cells. This component is well-exposed on the outer cell surface for lactoperoxidase-catalyzed iodination and proteolytic cleavage (Hynes, 1976). It contains galactose-terminated (Hynes, 1976) and sialylated oligosaccharide chains (Critchley et al., 1976; Vessey and Culp, 1978) and is a major Con-A binding glycoprotein on the surface of the cell (Burridge, 1976). This surface membrane component is decreased in many, but not all, tumor cell lines (Pearlstein et al., 1976; Chen et al., 1976).

BIOCHEMICAL DETERMINANTS

345

OF CELL ADHESION TABLE I

CELLULAR PROTEINSFOUND IN SAM" Cell protein

Apparent MWb

Properties

GAP-1, -2, -3

cll

Large (bound to GAG) 220,000

C, CX

200,000 175,000

Not collagenous LETS glycoprotein (rapid turnover and transformation-sensitive) Myosinlike Glycoprotein found in glass-bound SAM Clycoprotein found in glass-bound SAM May not b e a-actininc ? Subunit protein of 100-hi-diameter filaments Protein found in blass-bound SAM Rapid turnover Actin ?

145,000 85,000 67,000 56,000 Cd

49,000 48,000 45,000 37,000 17,000 14,000 13,000 11,000

? Histone Histone Histone

These are the components found in SAM from a variety of normal and virus-transformed cell lines which have been characterized by Culp (1976a,b) and Culp et al. (1978). Major proteins are denoted with subscript numbers and minor proteins with subscript letters. Apparent MWs have been determined by slab SDS-PAGE analysis (Culp, 1976a). See Culp (1976b) and Culp et al. (1978).

Purified CSP glycoprotein has been shown to agglutinate formalintreated red cells (Yamadaet al., 1975),to reassociate in some unknown fashion with the surface of CSP-depleted cells (Yamada and Weston, 1975), and to catalyze greater spreading of transformed cells over the substrate with increased resistance to detachment (Yamada et al., 1976).These latter observations, coupled with decreased quantities of the LETS glycoprotein in the enriched surface membranes (Hynes, 1976) and SAM (Culp, 1976a,b) of many virus-transformed mammalian cells, and its prominence in the SAM of normal cells (Culp, 1976a,b), suggest a quantitative role for this component in the cellsiibstrate adhesion process.

346

LLOYD A. CULP

SAM also contains a protein (C,) which does not appear to be glycosylated and which shares many properties with nonmuscle myosin (Table I; see also Culp, 1976a); it has been identified as a cell surface component (Willingham et al., 1974; Painter et al., 1975; Olden et al., 1976), although its accessibility to lactoperoxidase-catalyzed iodination and trypsin-mediated cleavage is less than that of the LETS glycoprotein (Teng and Chen, 1976). Actin (C,) and protein C, (MW 56,000), which does not appear to be tubulin but which may be the subunit protein of the 100-A beta filaments present in these cells (McNutt et al., 1971), were present in SAM from normal, SV40-transformed, and revertant cell populations in similar proportions (Culp, 1976a). The prominence of actin in SAM is important in light of the ultrastructural association of actin-containing microfilaments at cellsubstrate adhesion sites (Goldman and Follett, 1969; Abercrombie et al., 1971; McNuttet al., 1971,1973; Perdue, 1973; Reaven and Axline, 1973; Bragina et al., 1976; Nicolson, 1976a,b) and the more highly organized arrays of these filaments in normal and revertant cells when compared to those in virus-transformed cells (McNutt et al., 1971, 1973; Pollack et al., 1975; Altenburg et al., 1976). Protein C, appears to be the subunit protein of the 100-A-diameter nonactin filaments, since it electrophoreses identically on one-dimensional SDSPAGE and two-dimensional gels (O’Farrell, 1975) with the 52K protein identified by Brown et al. (1976) in Triton cytoskeletons as the subunit protein of these filaments (Culp, unpublished data); this indicates, along with other data below, a close association of both classes of cellular microfilaments at the substrate adhesion site (but the mechanism of binding of one or both classes of microfilaments to the surface membrane has not been established, and further studies on the molecular organization of SAM should provide systems for delineating these mechanisms). The identities of several other proteins in SAM (Cb, C,, Cd, C,, and C,) remain to be determined (Culp, 1976a; see also Table I). Some evidence (Culp, 1976b) suggests that’Cbis not a-actinin, which has been implicated as the membrane-binding component for the actin-containing microfilaments at the tips of gut epithelial microvilli (Mooseker and Tilney, 1975). Histones (C, to CJ, which were particularly prominent in SVT2 SAM, appear to be artifactually bound to SAM during the treatment of cells with chelating agents and are therefore not intrinsic SAM components (Culp, 1976a). When substrate-bound cells were split in half by the method of Barland and Schroeder (1970), in such a manner that the lower half of the cell remained substrate-bound while the upper half of the cell was solubilized, Noonan et al (1976) found that the most of the cell’s actin,

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

347

the LETS gl ycoprotein, and myosinlike protein components were in the lower half of the cell, consistent with the prominence of these components in SAM. This asymmetric distribution of at least three of the chemical components implicated in substrate adhesion may also be a reflection of the nonadhesive nature of the upper cell surface (DiPasquale and Bell, 1974; DiPasquale, 1975; Vasiliev et al., 1975). That all the components of SAM, other than histones, are intrinsic and essential components of the cell-substrate adhesion site is indicated by several lines of evidence. (1)All these components were observed in SAM prepared from cells under different attachment, detachment, and growth conditions (Table 11; see also Culp, 1976a,b). (2)The morphological identification of SAM as partially disrupted cellular footpads reflects the persistence of membrane-associated microfilaments, even from a variety of transformed cells (Culp et al., 1978; Rosen and Culp, 1977). (3) The inability to extract selectively any of these components by treating SAM with nonionic detergents indicates that they are not extraneous membrane-contained or vesicle-trapped components (Cathcart and Culp, unpublished data). (4)The inaccessibility of these components, including the LETS glycoprotein, to lactoperoxidase-catalyzed iodination indicates that they are located within membrane-enclosed SAM footpads or between the cell-substrate interface where the enzyme cannot permeate (Chi and Culp, unpublished data). For these reasons, Culp (1976b) has suggested that these components act as a cell surface adhesion complex involving GAGTABLE I1 VARIOUS GROWTH, ATTACHMENT, AND DETACHMENT CONDITIONS USED TO DETERMINESAM COMPOSITION^ Long-term radiolabeling during exponential growth of cells on: Tissue culture plastic Glass SAM-coated substrates Attachment of preradiolabeled cells to: Serum-adsorbed substrates SAM-coated substrates Short-term radiolabeling during exponential growth of cells; also chase analysis Detachment of cells by treatment with EGTA, EDTA, or scraping with a rubber policeman; also serum-mediated detachment of SVT2 cells (Bradley and Culp, 1977). a The composition of SAM proteins and polysaccharides was measured under these different conditions by SDS-PAGE (Culp, 1976a,b).All the components listed in Table I for the appropriate substrate were found under all conditions, except for elevated concentrations of a few components under some conditions.

348

LLOYD A. CULP

containing proteoglycans, the LETS glycoprotein, and membrane-associated microfilaments (with their own associated proteins).

4. PROSPECTIVE MODELS FOR CELL-SUBSTRATE ADHESION AND CELLULAR MOVEMENT The highly associated nature of the SAM components described above suggests that the substrate adhesion site is organized through the interaction of two different cell surface compartments: (1)interaction of external cell surface components such as the LETS glycoprotein and/or GAG-containing proteoglycans with substrate-bound serum components and (2) linkage of actin-containing and/or nonactin-containing microfilaments to a transmembrane component which binds to the LETS glycoprotein and/or the proteoglycans (these microfilaments may be organized into arrays which strengthen the adhesion site by linkages to other membrane sites and to cytoplasmic microtubules) (Culp, 1976a,b). Two prospective models for cell-substrate adhesion are presented in Fig. 1, consistent with most of the evidence in the literature at this time for the surface orientation and association of these various components; however, definitive proof that any portion of these models is correct is still lacking. In model A, the adhesion is directly mediated by interaction of the LETS glycoprotein, which may exist as an integral transmembrane component (Hunt and Brown, 1975), with a serum protein or glycoprotein receptor called SA(CIG?). [However, evidence has been obtained that some of the cell surface LETS glycoprotein may not span the lipid bi1975);,Vesseyand Culp (1978)have obtained evilayer (Graham et d., dence for at least three topographically different pools of cell surface LETS glycoprotein which may reflect different functional roles at the cell surface.] The LETS glycoprotein in model A is also associated in the lipid bilayer with the myosinlike protein, which at this time appears to be the principal candidate for actin-binding to the membrane (the myosin-like protein has been shown to have ATPase activity (Shizuta et al., 1976) and to have less external orientation than the LETS glycoprotein). The recent demonstration of Con-A-induced coordinate rearrangement of myosin-containing filaments and a major Con-Abinding glycoprotein, perhaps the LETS glycoprotein (Burridge, 1976),in the membranes of rat kidney cells is also consistent with this association (Ash and Singer, 1976). Actin is polymerized into microfilaments, perhaps with lateral association with other sites in the membrane (Fig. 1A) as observed by Mooseker and Tilney (1975), via a second class of myosinlike components or some as yet unidentified

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

A

349

m c , EGTA

FIG.1. Proposed molecular models of mammalian cell-substrate adhesion. Two different models of cell-subshate adhesion are presented in (A) and (B); these are prospective, and not proven, models based on the limited amount of information available on the identity of SAM components and the orientation of these components in the membrane. Different molecules are represented by different geometric shapes. Sub, Plastic or glass substrate; L, the LETS glycoprotein; LB, lipid bilayer of the surface membrane; M, myosinlike protein; PG, proteoglycans containing hyaluronic acid and some sulfated GAGS; SA, serum receptor site for LETS in model A (CIG?); SB, serum receptor site for proteoglycans in model A; S, serum receptor site for proteoglycans in model B; mf, actin-containing microfilaments (and perhaps non-actincontaining filaments as well); mt, microtubules; a, a membrane linkage component for the microfilaments; b, microfilament cross-linking protein; c, protein linking microfilaments and microtubules; x and y, possible sites of action of EGTA. Specific molecular interactions are denoted with complementary geometric patterns. These models do not take into consideration the proteins in SAM which have not been identified.

macromolecule. Microfilaments would be organized into bundles by cross-linkage (11 in Fig. 1A) and cross-linked with microtubules (c in Fig. 1A). The large proteoglycans might then act as modulators of the adhesion site by binding to serum receptor sites and to the surface of the cell via the LETS glycoprotein (d in Fig. 1A) (perhaps as a negative effector by competing for binding to the LETS glycoprotein) or via some other protein moiety (e in Fig. 1A) (perhaps their own protein moieties). An alternative model is proposed in Fig. 1B in which the proteoglycans play a stringent bridging role in the adhesion process by cross-linking serum receptor sites with the LETS glycoprotein. In this latter model, LETS on the cell surface may bind to the heparan sulhte which is a prominent GAG in SAM (B. Rollins and L. Culp, unpublished data); heparan in turn may bind to CIG in the serum layer, forming a cross-bridge between cell and serum layers. There is insufficient evidence yet to determine if either of these

350

LLOYD A. CULP

models is correct, particularly since little is known about the molecular organization of the proteoglycans in SAM. The molecular complexity of these proteoglycans may provide the cell with a variety of mechanisms for effecting its adhesive processes. Neither of these models takes dynamic aspects of adhesion formation into account. Perhaps proteoglycans play an initial role in binding the cell surface to the substrate and then are rearranged from model B to model A, allowing direct LETS glycoprotein-serum receptor interaction. Culp (1976a,b) demonstrated by pulse-chase analysis that footpad adhesion sites contain higher concentrations of the LETS glycoprotein, the Cd protein (MW 48,000), and sulfated GAGs than footprints, all of which turn over with a half-life of 2-4 hours. This turnover might be the result of molecular reorganization of the footpad adhesion sites as they form and/or the evolution of footprints during natural cell movement across the substrate; these possibilities have not been differentiated with the evidence available. Probably the most poorly understood aspect of these models is the molecular organization and the cell surface orientation of the large proteoglycans. The turnover of sulfated GAGs in SAM has been shown to result from a loss of these components (Culp, 1976a, and unpublished data). But what is the metabolic fate of the LETS glycoprotein and the c d protein? Keski-Oja et al. (1976) recently showed that fibronectin (the human fibroblast equivalent of the LETS glycoprotein) could be covalently linked to unknown components, with resultant migration of the complexes at the top of the SDS-PAGE stacking and separating gels (perhaps by cross-linkage to the two major classes of high-MW proteoglycans and/or by cross-linkage to itself, generating complexes which coelectrophorese with the proteoglycans) by treatment of intact cells with physiological concentrations of activated factor XI11 (XIIIa, a plasma transglutaminase). If the recipient of the cross-linked fibronectin proves to be proteoglycans, this will be the first chemical evidence for an association between this glycoprotein and the proteoglycans at the cell surface. These experiments indicate that turnover of the LETS glycoprotein in SAM (as determined only by the loss of the 220,000 band in SDS-PAGE gels) could be due to a protease-mediated loss into the medium and/or perhaps transglutaminase-mediated cross-linkage to the proteoglycans or to itself. These two possibilities can be resolved experimentally. These prospective models of the adhesion site are consistent with several lines of experimental evidence. First, formation of adhesion sites is more sensitive to treatment with cytochalasin B than to treatment with colchicine (Weiss, 1972; Kolodny, 1972; Grinnell, 1974a;

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

35 1

Mapstone and Culp, 1976; Juliano and Gagalang, 1977),whereas cytoplasmic spread over the substrate is equally sensitive to colchicine (Ivanova et al., 1976); the same relationships are noted for SAM deposition (Mapstone and Culp, 1976; Murray and Culp, unpublished data). Second, trypsinization (Revel et al., 1974) or EGTA treatment (Rosen and Culp, 1977) of cells minimally affects the adhesive materials at the footpad site but appear to detach cells by the disorganization of surface membrane-associated cytoskeletal elements (although a pool of membrane-associated actin and nonactin microfilaments persists, presumably because trypsin acts at a-type sites in Fig. 1A and EGTA at sites x and/or y). [Approximately 8-10 moles of actin exist in SAM per mole of myosinlike protein. Trypsin or EGTA may also induce a breakdown in membrane impermeability with a loss of intracellular cyclic nucleotides which may be required to maintain the integrity of the cell’s cytoskeleton (Revel et al., 1974).1 Third, increasing the CAMP concentration within the cell increases the cell’s resistance to detachment b y presumably increasing the organization of membrane-associated cytoskeletal elements (Johnson and Pastan, 1972; Grinnell et al., 1973; Willingham et al., 1973; Shields and Pollock, 1974; Willingham and Pastan, 1975).And finally, cell detachment froin the substrate was achieved by cytochalasin-B treatment in addition to tangential centrifugal force (Helentjaris et al., 1976). Therefore cells can modulate their adhesion processes at two different levels of cell surface organization: (1)by affecting the integrity of externally oriented LETS glycoprotein and/or proteoglycans, or (2) by affecting the organization of membrane-associated microfilaments required to strengthen the adhesion site and to prevent pinching off of the adhesion footpad from the remainder of the cell. Pulse-chase analysis of radiolabeled SAM components has revealed some interesting aspects of natural cell movement across the substrate (Culp, 1976a,b; Culp et aZ., 1978). After pulsing a population of sparsely growing cells for 2 hours with radioactive leucine and chasing for an additional 24 hours, during which the cells moved more than one cell diameter away from their original locations during the pulsing period (Gail and Boone, 1972),analysis of SAM before or after the chase period revealed two metabolically different pools of components. The elevated levels of the LETS glycoprotein, the Cdprotein, and sulfated proteoglycans observed in footpads being formed during the 2-hour pulsing period were rapidly lost from SAM during the chase period during the evolution of footprints on the substrate, although a small portion of LETS persisted. However, chasing results in only a small loss of the myosinlike protein (CJ, protein C1, actin (CJ,

352

LLOYD A. CULP

and the GAG-associated protein (presumably that portion associated with hyaluronic acid which appears to be metabolically inert). This latter observation indicates that footprints left on the substrate by natural cell movement across the substrate result from footpad-containing material being left behind with membrane-associated microfilaments. A model of this is presented in Fig. 2 where localized disorganization of posterior cytoskeletal elements generates labile retraction fibers which break as a result of posterior stress created by generation of anterior adhesion sites, leaving membrane-enclosed footpad remnants on the substrate as footprints in a fashion similar to that of EGTA treatment (Rosen and Culp, 1977). It is not clear how the cytoskeletal elements in only the posterior footpad sites could become disorganized. Another important approach for determining structure-function relationships of cell surface components involved in the adhesion process is generation of prospective adhesion variants and analysis of their surface components. Two classes of these variants have already been discussed with regard to altered hyaluronate metabolism (Atherly et al., 1977; Kraemer, 1976). Pouyssegur and Pastan (1976) isolated two loosely adhering mutants from mutagenized populations of BALB/c 3T3 cells, which appeared to be defective in synthesizing certain carbohydrates (Pouyssegur and Pastan, 1977). Both retained a rounded cell morphology in sparse cultures but flattened in dense cultures or when treated with CAMP (both mutants possessed normal levels of CAMP).Lactoperoxidase-catalyzed iodination of one mutant revealed loss of proteins of MW 90,000 and 135,000 from the cell surface, while

FIG.2. Generation of footprints as cells move across the substrate. As cells move across the substrate (Sub), thin filopodia (F)containing microfilaments extend out from the leading lamella (LL) of the surface membrane and establish initial contacts with the substrate. These contacts eventually develop into mature footpads (Fp). (See Rosen and Culp, 1977.) At the posterior end of the cell, cytoskeletal organization may weaken in such a fashion that the cell body begins to separate from the footpads by thin, membranous retraction fibers (RF) which eventually break, leaving footprints (Ft) of footpad material with a small pool of membrane-associated microfilaments.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

353

the second mutant had lost a 230,000-MW component which may have been the LETS glycoprotein. Although the 135,000-MW protein has not been identified in BALB/c 3T3 SAM, the missing 90,000-MW component may be protein Cb which is commonly observed in SAM (Culp, 1976a,b). Further analysis of these mutants should reveal whether they put down footpads normally on the substrate but are defective in organizing membrane-associated cytoskeletal elements, or whether they are missing or contain modified external surface components which are critically important in the recognition of serum receptor sites. The latter possibility may be likely in light of the defectiveness of these mutants in making certain carbohydrates and the prominence of carbohydrate in SAM. In summary, substrate adhesion appears to b e mediated by a complex at the surface of the cell involving externally oriented glycoproteins and GAGs which form the anchorage site for actin-containing and nonactin-containing microfilaments. Such a complex permits the cell to modulate its adhesion processes at several levels of cell surface organization. The precise molecular topography of these components at the adhesion site and dynamic aspects of these topographical relationships have not been determined. Transformation of normal cells with oncogenic viruses may affect the adhesion process in at least three ways: (1) by diminishing the pool of surface membrane LETS glycoprotein; (2) by preventing extensive organization of membraneassociated microfilaments; and (3) by altering the synthesis and/or turnover of hyaluronic acid and sulfated GAGs. Analysis of the chemical organization of macromolecules in SAM, addition of prospective adhesion components to deficient cells to alter the adhesion process, and characterization of adhesion variants should permit a detailed map of the molecular topography (with a consideration of the dynamics) of the cell-substrate adhesion site in the not-too-distant future. C. Adhesion to Collagen

Since many cell types in animals interact in some unknown fashion with many of the collagen-containing matrixes in the tissues of animals (see, e.g., Trelstad et d.,1974; Meier and Hay, 1975),it has been obvious that analyses of this interaction in vitro with different cell types (Harris, 1973b; Letourneau, 1975)should indicate the molecular mechanisms involved. Several laboratories have initiated such studies. Analysis of SAM remaining after EGTA-mediated detachment of a

354

LLOYD A. CULP

mouse fibroblastic cell line (Swiss 3T3) or an endothelial cell line (BALB/c 3T3; Boone, 1975) revealed that adhesion to the tissue culture substrate might not be mediated by a cell-produced collagenous protein (Culp, 1976a).This does not eliminate the possibility that one of the substrate-bound serum proteins was in fact collagen, but this possibility seems remote, since none of the serum polypeptides extracted from the substrate by SDS treatment exhibited the appropriate size upon SDS-PAGE analysis of collagen chains or the purplish stainability with coomassie blue commonly observed with collagen proteins (Culp and Buniel, 1976; Culp, 1976a).Also, sera of animals may contain very little, if any, collagen. However, collagenous proteins may resist SDS extraction from the substrate or electrophorese as cross-linked components at the top of gels (and may therefore be obscured by coelectrophoresis with high MW proteoglycans). More evidence will be required to exclude rigorously the possibility of a substrate-bound collagen material being an important determinant in cell-substrate adhesion of some tissue-cultured cells. Klebe (1974) established an assay for measuring the kinetics of attachment of trypsinized SV3T3 cells to collagen-coated dishes. This adhesion process was strictly dependent upon the presence of serum in the medium, consistent with the prospective role of a collagenbound serum factor being required to mediate the proper adhesion process; cell atkachment to serum-conditioned collagen (Klebe, 1974) or tissue culture dishes (see Section I1,A) has been found to be dependent upon the concentration of serum, to need a 5-minute time lag before initiation of attachment, and to require Ca2+or Mg2+.Fractionation of serum by Klebe (1974) generated an attachment factor whose adhesion-promoting ability with collagenous substrates was resistant to pronase or trypsin digestion and chromatographed as high MW materials (GAGS?). More recently, Pearlstein (1976) utilized Klebe’s system for assaying a cell adhesion factor (CAF) which was urea-extracted from BHK cells. He suggested that the LETS glycoprotein was the active factor in this material, even though it was only 60% pure, and it appears quite likely that urea extracts high MW proteoglycans and many other cell proteins. PyBHK cells were used in this assay, since their surface membranes contain very low concentrations of the LETS glycoprotein, and serumless medium was used to differentiate CAF from Klebe’s (1974) factor. The CAF-mediated adhesion to collagen was sensitive to 0.1 M lysine (which perhaps inhibited an important function of the negatively charged GAG or the sialylated oligosaccharides of the LETS glycoprotein?). It would be interesting to determine in

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

355

this system the potential blocking capacity of intact immunoglobulin and particularly that of Fab fragments of an antiserum prepared to highly purified LETS glycoprotein. Linsenmayer et al. (1976) analyzed the effects of addition of serum or specific cartilaginous proteoglycans (Hardingham and Muir, 1974; Heinegard and Hascall, 1974) on the adhesion of EGTA-detached BALB/c 3T3 cells to collagen coatings. Serum-preincubated collagen substrates were as effective in promoting the adhesion of these cells as in initiating adhesion in serum-containing medium. Incubation of serum-preincubated collagen with purified hyaluronic acid did not affect the adhesion process. However, sulfated GAG-containing proteoglycans from cartilage inhibited serum-catalyzed adhesion, suggesting perhaps an antagonistic role for sulfated proteoglycans in the collagen-mediated adhesion process. The experiments described above indicate that cells do not adhere directly to collagen polypeptides but that serum components bind to collagen and catalyze the adhesion by a process very similar to that occurring on tissue culture substrates (Culp and Buniel, 1976). It will be interesting to determine if serum components (CIG in particular) extracted from preincubated collagenous substrates with SDS share properties with proteins extracted from plastic or glass tissue culture dishes (Culp and Buniel, 1976) or the SF of Grinnell(1976b). In light of this possibility, Klebe et al. (1977) recently reported the isolation of variant CHO cells selected for decreased adhesiveness to serumpreincubated collagen substrates; these variants also displayed decreased adhesiveness to serum-preincubated glass or plastic tissue culture substrates. In performing readdition experiments to test the role of a prospective serum factor, it will be imperative to purify these factors extensively and to test the blocking ability of Fab fragments of specific antibodies. Even these experiments may be complicated by the participation of more than one cell surface or serum macromolecule in the adhesive bond, which may mean that the absence of any one specific component or binding of Fab fragments of antibody directed toward it has a minimal effect on adhesion processes. The addition of purified surface membrane components to cells to measure their potential influence on adhesion must also take into consideration the mechanism of rebinding of these components to the cell surfacewhether binding and orientation are identical to the de novo pathway of cell surface organization or whether some artifactually induced orientation of cell surface components has been effected. With the possible exception of platelets (Jamieson et al., 1971), adhesion of fibroblasts (Klebe, 1974; Pearlstein, 1976) and endothelial cells (Lin-

356

LLOYD A. CULP

senmayer et al., 1976) does not appear to be mediated by the direct binding of a cell surface receptor to collagen polypeptides, but by serum factors which have been only partially characterized.

111.

CELL-CELL ADHESION

The study of cell-cell adhesion in most instances has been complicated (1) by the necessity for dissociating cells from tissue matrixes in which they have already been genetically programmed to adhere; (2) by the cell surface damage induced by dissociative conditions and by the need to repair this damage before normal and specific readhesion occurs; (3) by the heterogeneity of cell-cell adhesion sites determined by morphological criteria (McNutt and Weinstein, 1973; Revel, 1974; Overton, 1975); and (4) by the inability thus far to enrich for surface membrane regions which contain only the adhesion site for appropriate biochemical and topographical analyses. This last approach was utilized by Goodenough (1974) to isolate mouse hepatocyte gap junctions which contained two relatively small proteins. However, the proteolytic conditions used to isolate these gap junctions have been shown by Duguid and Revel (1975) to modify artifactually the proteins comprising these junctions and therefore to impose severe limitations on the interpretation of biochemical studies with these preparations. The basic approach of isolating and enriching for fragments of surface membranes participating in cell-cell adhesion sites deserves more attention in determination of the biochemical complexity of these sites. The most frequently utilized approach for studying cell-cell adhesion has been the addition of a prospective macromolecular cofactor to suspensions of dissociated cells in order to measure increased adhesion, the increase being measured in any one of a number of ways (Curtis, 1973). Obviously, the question arises as to whether the factor actively mediates the adhesion process or merely catalyzes the reorganization of other surface macromolecules that mediate formation of the adhesive bond. Only a limited amount of attention has been given to the possibility that some surface macromolecules may act as negative regulators of adhesion formation rather than as positive effectors. In any case, sufficient information has been obtained in several systems to indicate that specific cell surface glycoproteins and high MW proteoglycans may play a direct role in the formation of cell-cell adhesions.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

357

A. Simple Eukaryotic Systems

Although the adhesive processes of simple eukaryotic cells appear to be biochemically less complicated than those of avian and mammalian cells, there is still considerable complexity in utilizing these “simple” organisms as models for studying cell-cell adhesion. For example, a genetic analysis by Warren et al. (1976) of defective adhesion mutants of the slime mold Polysphondylium violaceum revealed that as many as 50 complementable gene products were involved in the simple adhesiveness of this organism. Although many of these gene products are probably indirectly involved in adhesion formation, it is likely that the macromolecules directly involved are many and complex.

1. YEAST AND Chlamydamonas Crandall and Brock (1968) initiated biochemical studies on the mechanism of agglutination of the yeast Hansenula wingei. Strain 5 (str 5) or Strain 21 (str 21): cells of this organism do not self-agglutinate but agglutinate in mixed suspensions. When str-5 cells were treated with a snail enzyme preparation or subtilisin to solubilize partially intact cell surface components, a factor was solubilized which had a MW greater than 500,000 and which agglutinated str-21 cells. The activity of 5-factor was destroyed by treatment with mercaptoethanol, trypsin, or pronase probably reflecting the importance of multivalent proteinbinding sites in mediating agglutinability. The only sugar found in 5factor was mannose. Trypsin digestion of str-21 cells (Crandall and Brock, 1968), however, liberated 21-factor which did not agglutinate str-5 cells but which blocked 5-factor-mediated agglutination of str-21 cells. Trypsin digestion had probably liberated a univalent fragment. Twenty-onefactor was adsorbable to str-5 cells but not to str-21 cells. This latter property was used in a purification sequence, including Sephadex G200 chromatography and starch block electrophoresis, to obtain a purified 21-factor which was 65% protein and 35%mannose. The specificities in adsorption of 21-factor to str-5 cells and in blocking agglutination mediated by 5-factor suggested that this molecule was indeed the receptor site for the complex 5-factor and was not a nonspecific inhibitor of agglutination. These two factors were later shown to form a complex in vitro (Crandall et a1 ., 1974)-alkaline treatment of the complex dissociated

358

LLOYD A. CULP

the two components into irreversibly damaged 21-factor and active 5factor which could be assayed by agglutination of str-21 cells. The complex displayed a MW of several million upon Sepharose-6B chromatography (reflecting extensive molecular aggregation) and considerable heterogeneity (probably not surprising in light of the proteoIytic treatments required to isolate these -kvo components). Five-factor was further shown to be heterogeneous, with mannose/protein ratios varying from 50 to 96%. Intact 5-factor had a sedimentation velocity of 16.7s that was reduced to 1.7s upon reduction of disulfide linkages. Whether 21-factor binds to specific mannose oligosaccharide arrays in 5-factor (these arrays presumably are not present in 21-factor) or whether these two factors bind via protein-protein interaction is not known. In the Chlamydomonas system, there are two complementary types of cells labeled (+) and (-) whose gametes adhere in mixed suspensions. Mixing of (+) and (-) gametes of Chlamydomonas moewusii leads to flagellar binding with eventual cross-bridge formation and fusion of the cell bodies. McLean and Bosmann (1975) demonstrated that the adherence of (+) and (-) flagella in the gametes, but not adherence of ( +) and (-) vegetative cells, led to increased ectoglycosyltransferase (cell surface-localized sugar transferease to be discussed in Section II1,C) activity with endogenous polysaccharide acceptors for galactose-, glucose-, N-acetylglucosamine-, sialic acid-, mannose-, and fucose-specific activities. Although these investigators argued that these data supported the model of Roseman (1970) in emphasizing the importance of cell surface sugar transferases as mediators of specific cell adhesion, it is not clear from these studies why all these enzyme activities should rise when gametes are mixed. First, such a concomitant increase raises some question about a lack of specificity in the elevation of these enzyme activities. Second, the specific activities of the so-called ectoglycosyltransferases were a major portion of the enzyme specific activity observed in whole-cell extracts, raising an additional question as to whether or not the cell treatment and enzyme assay conditions measured truly cell surface-localized enzyme activities (see Section II1,C for a discussion of these enzymes). In the sea urchin system, a factor which catalyzed the reaggregation of dispersed cells has been called ovacquenin (Kondo, 1974),although preparations of this material were also found to contain an inhibitor of ovacquenin activity. These studies are difficult to interpret, since purification and specificity have not been demonstrated in the case of either of these factors. McClay and Hausman (1975) recently developed a species-specific aggregation assay using two different species

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

359

of sea urchin (Lythechinus and Tripneustes) and provided evidence that specificity was derived from a paternal gene product. Such specificity should prove useful in the search for adhesion-promoting cell surface factors in this system.

2. SPONGE The chemical requirements for reaggregation of sponge cells have been fruitfully explored, because of the ease of dissociation of these cells and the retention of biological activity in solubilized factors (see Moscona, 1968; Lilien, 1969; Kuhns et al., 1974; Humphreys, 1975; Burger et al., 1975).The removal of divalent cations from seawater results in the dissociation of sponges into single cells and solubilization of an aggregation factor (AF) which catalyzes reaggregation of these cells upon the addition of Ca2+(Fig. 3). This AF-mediated reaggregation has been shown to be species-specific, utilizing the color differences between Microciona prolifera and Haliclona occulata (Humphreys, 1970). EDTA-dissociated cells were irreversibly damaged (but without loss of viability) and required cellular protein synthesis to reacquire aggregation ability; experiments also indicated that EDTA irreversibly damaged A F (Humphreys, 1970).Reaggregation of cells which had been dissociated by gentle trypsinization occurred in two stages: (1) a rapid and unstable aggregation requiring Ca2+and Mg2+which was insensitive to further trypsinization or to the addition of puroinycin (probably a nonspecific adherence) and (2) longer-term

FIG.3. Molecular processes in the disaggregation and reaggregation of sponge cells. Sponge cells adhere by binding of a large multivalent aggregation factor (AF) to cell surface receptor baseplates (BP), which may or may not be transmembrane components as depicted in these drawings. Conditions for dissociation and, in some cases, reassociation are shown.

360

LLOYD A. CULP

stable aggregation requiring addition of a solubilized A F and divalent cations (Muller and Zahn, 1973). Early attempts to purify A F indicated that it was a very large proteinaceous particle (Miiller and Zahn,

1973). Humphreys and his collaborators conducted a series of experiments to purify extensively and examine the chemical and physical properties of the A F from M . parthena (Henkart et aZ., 1973; Cauldwell et al., 1973). A combination of Sepharose-2B chromatography and centrifugation on glycerol and cesium chloride gradients was used to generate a relatively homogeneous population of high MW proteoglycans of MW 20 x lo6.These were highly negatively charged molecules, were 47% by weight protein and 49% carbohydrate, contained galactose, mannose, an unidentified uronic acid, glucosamine, and galactosamine, and appeared microscopically as a sunburst configuration (Fig. 4) (Henkart et aZ., 1973). The chemical and physical organization of this proteoglycan complex was stable to hyaluronidase digestion, but upon EDTA treatment it dissociated into a stable central core and 2 x lo5MW radial arm subunits (Fig. 4) (Cauldwell et al., 1973).Thus Ca2+was required to link five to eight subunits into one radial arm of the sunburst AF, and these arms no doubt contained the recognition sites for the cellular receptor molecules. The biological activity of A F was destroyed by protease treatment, perhaps as a result of loss of its multivalent nature and/or loss of protein-binding sites. These experiments suggest that Ca2+does not function as a simple bridging mechanism but is required to maintain the multivalency of the factor. The glycopeptide chains liberated by pronase digestion appeared to be very large (20,000-40,000 MW), with three to four polysaccharide chains per arm subunit. These glycopeptide chains were not effective in inhibiting aggregation activity of purified A F (Humphreys, 1975),

0

X l@MW

20 X IO'MW FIG.4. Organization of the AF of the sponge M . purthena. The sunburst configuration of this factor is shown with generally 11 to 15 arms per core, the arms being depicted with many polysaccharide side chains. The core is approximately 80 nm in diameter; the arms are 4.5 nm in diameter and 110 nm long. Upon EDTA or EGTA treatment, the arms separate from the core and dissociate into subunits irreversibly.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

361

raising a question as to the functional role of the polysaccharide in this proteogl ycan . However, Burger and his collaborators (Turner and Burger, 1973; Burger et al., 1975) have implicated the carbohydrate from similar high MW proteoglycans isolated from another species of sponge, M i crociona prolifera, in adhesion processes. This A F activity was sensitive to mild periodate oxidation, digestion with a glycosidase enzyme preparation from Helix pomatia, or addition of 0.1 M glucuronic acid. However, this factor was not as extensively purified or characterized as that from M . parthena (Humphreys, 1975), although both factors had many properties in common (including a hexose/uronic acid/protein ratio of 45: 5: 50). The factor obtained from M . prolifera did not appear to be strictly species-specific, since it also catalyzed the aggregation of Haliclona cells (Kuhns et al., 1974). This evidence, along with more recent evidence obtained using a factor from the sponge Geodia cydonium (Miiller et al., 1976),questions the species specificity of these factors and/or the criteria used to classify these different organisms originally. However, antibodies against intact Microciona or Haliclona cells were species-specific in blocking homospecific AFmediated aggregation, suggesting little cross- reactivity between the AFs of the two species. Some evidence has been presented (McClay, 1974) that species-specific aggregation may be affected by the presence or absence of cell surface factors which actively inhibit the action of AF. Burger and his colleagues (Weinbaum and Burger, 1973; Burger et al., 1975) also succeeded in isolating a prospective cell surface receptor site for M . prolifera A F b y treating dissociated cells with hypotonic salt solutions (Fig. 3). An activity [baseplate (BP)] was liberated which blocked AF-mediated reaggregation of dissociated cells (Fig. 3). This inhibitory activity was in turn blocked by the coaddition of glucuronic acid or by treatment of BP with periodate, further strengthening arguments for the importance of protein-pol ysaccharide interactions between BP and AF. Sephadex beads covalently linked to BP aggregated in the presence of, but not in the absence of, A F and Ca2+; these beads also adhered to hypotonically shocked cells but not to untreated cells. A model for BP- and AF-mediated agglutination of sponge cells has been presented by Burger et al. (1975). Similarly, AF-bound beads may provide an important ligand for specifically purifying the BP cell surface receptor molecules from several sponge species. The purification and utilization of highly purified AFs and cell surface receptor molecules should lead to a detailed analysis of the molecular specificity and topography of adhesion in many of these sponge species.

362

LLOYD A. CULP

3. SLIMEMOLD Adhesion studies with various species of slime mold overcome the requirement of physical and chemical separation of adherent cell populations, since deprivation of food supply in the medium of single cell suspensions of vegetative amebas initiates a genetic program of differentiation of these cells to form aggregates, with eventual conus and stalk formation for the generation of progeny spores (see, e.g., Gerisch, 1968; Loomis, 1972). Initial formation of aggregates was found to be somewhat insensitive to EDTA (Gerisch, 1968), eliminating any simple divalent-cation cross-bridging mechanism. A more recent study of aggregation in Dictyostelium discoideum suggests that there is initial “loose” adhesion of differentiating cells which is EDTA-sensitive and nonspecific (Alexander et al., 1975);the aggregation competence of these differentiating cells was destroyed by trypsinization and required protein synthesis for restoration. Aggregation is initiated by a chemotactic response to pulses of CAMP emitted by aggregation center cells (Bonner et al., 1969). Eventually, the large body of aggregated amebas is engulfed by a polysaccharide sheath containing N-acetylglucosamine, fucose, and mannose (Gerisch, 1968; Loomis, 1972). Several studies have indicated that increased polysaccharide synthesis is correlated with differentiation events. Newel1 et al. (1971) established that differentiation of vegetative cells to aggregation-competent cells was accompanied by the induction of UDP-glucose pyrophosphorylase. After the induction of enzyme formation, the addition of actinomycin D had little deleterious effect on enzyme synthesis or aggregation competence. However, cells in aggregates dissociated by mechanical means required new mRNA and enzyme production before aggregation could be reinitiated. Several experiments suggested that cell-cell contact was required for persistent synthesis of this developmentally regulated enzyme, even though stable mRNA for it was available. Acquisition of aggregation competence in slime mold amebas was accompanied by an increase in the concentration of Con A required to agglutinate cells half-maximally (Weeks, 1973). A SEM study of the availability of Con-A-binding sites established that many sites were available in vegetative cells and were sensitive to a-methylmannoside addition, whereas these sites were unavailable in competent cells, which appeared to be covered with a diffuse layer of material (Rossomando et al., 1974). The chemical nature of this material has not been determined.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

363

Beug et al. (1970) initiated studies on the effects of antibodies directed toward cell surface determinants of competent cells. Preadsorption of competent cells with Fab fragments of IgG prevented cell -cell contact formation. The Fab inhibitory activity was adsorbable with purified membrane preparations of competent cells only, and the active determinants on these membranes were found to be heat- and periodate-sensitive (glycoprotein?). Two classes of antigenic determinants were later identified in D. discoideum (Beug et al., 1973a,b)contact sites A, characteristic of aggregation-competent cells, and contact sites B, present on vegetative cells but not on competent cells. Fab fragments of antibody against A sites blocked end-to-end adhesion of cells, while Fab fragments against B sites blocked adhesion at cell surface locations other than the ends, suggesting topographical specificity in the formation and location of surface molecules mediating adhesion formation. A differentiation between two types of adhesion sites was also made by Sussman and Boschwitz (1975), who studied the properties of intact plasma membrane ghosts of D. discoideum. The two types of adhesive processes reflected different properties of vegetative or competent surface membranes. An EDTA-sensitive association of ghosts was observed for both types of membrane, while an EDTA-resistant aggregation was only observed using membranes from competent cells. Barondes and his collaborators initiated a series of studies (Rosen et al., 1973) on the biochemical nature of the surface components mediating the adhesion process in competent cells. Homogenization or sonication of EDTA-treated cells solubilized a factor which agglutinated formalin-treated sheep erythrocytes; the titer of this factor increased 400-fold during the development of competence. A lectinlike protein, called discoidin, was easily purified from extracts ofD. discoideum, because of its avidity for Sepharose4B and subsequent elution with D-galactose (Simpson et d.,1974). In the presence of galactose, discoidin had a MW of 100,000 (in its absence, it aggregated into large complexes). Analysis by SDS-PAGE identified subunits of MW 26,000 as the major portion of the protein and a minor component of MW 2A,OOO. These components were later called discoidin I and discoidin 11, respectively, since they were both galactose-specific lectins but varied in concentration during later phases of D. discoideum development and in their specificity toward red cells from different species (Frazier et al., 1975); they did not appear to associate with each other under nondenaturing conditions. No amino sugar or neutral hexoses were found in purified discoidin. Agglutination of formalin-treated red cells with discoidin could be reversed with D-galactose.

364

LLOYD A. CULP

A similar lectinlike protein was purified from extracts of another slime mold Polysphondylium pallidum (Rosen et al., 1974) and called pallidin (Simpson et al., 1975). Again, the titer of pallidin correlated with the development of aggregation competence; addition of this component to disaggregated competent cells stimulated aggregation; it was identified on the surface of competent€'. pallidum; and it agglutinated formalin-treated red cells. The last-mentioned reaction was used to purify pallidin by specific adsorption, followed by elution from red cells with D-galactose. This lectin displayed chemical properties different from those of discoidin (Simpson et al., 1975):(1) The native molecule displayed a MW of 250,000 with subunits of 25,000 MW (SDS-PAGE), and (2)it had an amino acid composition and an isoelectric point different from those of discoidin. Pallidin did not appear to be a glycoprotein. When vegetative or competent D.discoideum cells were glutaraldehyde-fixed and assayed for discoidin-mediated agglutination, vegetative cells were poorly agglutinated while competent cells were highly agglutinated, suggesting that development of aggi-egationcompeterce requires synthesis and integration into the surface membrane of both the multivalent lectinlike molecule and a receptor carbohydrate-containing moiety (Reitherman et al., 1975), although increased membrane fluidity with focal concentration of receptor molecules during the development of competence has not been ruled out by these experiments. The association constant of discoidin I or I1 for competent D.discoideum cells was estimated to be lo9per mole, whereas the association constant of discoidin for binding to fixed P . pallidum cells was lo8 per mole. Both erythrocyte-mediated rosette formation of only competent cells and surface-localized immunofluorescence using specific antibody made against purified lectin established the cell surface location of these components (Rosen et al., 1976). Fab fragments of antibody prepared against purified pallidin also inhibited the aggregation of differentiated P . pallidum. All the above evidence strongly suggests that adhesiveness of slime mold cells is mediated by synthesis and integration into the surface membrane of multivalent lectin-type molecules composed of many subunits. These events are also accompanied by the synthesis and integration into the surface membrane of a polysaccharide-containing component (glycoprotein or proteoglycan?) whose terminal sugar is likely to be D-galactose which is recognized by the lectin and which may be inaccessible to cis-oriented lectin (lectin present on the surface of the same cell). The molecular orientation of these two classes of components in the surface membrane has yet to be determined.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

365

The use of aggregation-defective mutants of slime molds (Gerisch, 1968; Brackenbury and Sussman, 1975) and more sophisticated molecular probes of surface membrane architecture should prove valuable in these analyses. A protein similar to, if not identical with, discoidin has also been shown to appear in plasma membrane preparations during the development of aggregation competence in D . discoideum (Siu et al., 1976), as determined by SDS-PAGE analysis. This carbohydratebinding protein (CBP) has a subunit MW of 26,000 (identical to that of discoidin) and was identifiable on the surface of cells by lactoperoxidase-catalyzed iodination. Two noncohesive mutants of D. discoideum lacked the CBP band identified on SDS-PAGE gels. Antibody to CBP immunoprecipitated both CBP and an associated protein of MW 56,000 from NP-40-solubilized membrane preparations. The nature of the binding of this associated protein to CBP has not been determined. In conclusion, the simplicity of the eukaryotic cells described above, the availability of highly purified factors which mediate adhesion in these various cell systems, and the utilization of adhesion-defective mutants should provide a detailed molecular map of the adhesion process (both static and dynamic aspects) in the near future. It is clear that in most systems protein-carbohydrate interactions may be the principal, and perhaps the only, means for providing specificity.

0. Avian and Mammalian Systems

'

Virtually the entire effort in studying cell-cell adhesion in higher systems has focused on tissue specificity in embryonic systems. Few attempts have been made to enrich for surface membrane regions participating in adhesion sites, because of a number of theoretical and technical problems. Most of the effort has been directed toward assaying for factors which catalyze the aggregation of dissociated cells. The major disadvantage of these latter systems is the necessity for dissociating tissues into single cells with a rather aggressive proteolytic treatment, the severity of this treatment no doubt reflecting the molecular complexity by which cells adhere to each other in the first place in embryonic tissues. The question has been, and continues to be, whether so-called adhesion factors truly mediate the reaggregation of cells by participating in adhesive bond formation or whether they merely catalyze the repair of surface membrane damage, allowing faster and more effective adhesions to form. Since trypsinized embryonic cells show some (but certainly not all-or-none) tissue specific-

366

LLOYD A. CULP

ity, these factors would have to interact more effectively with homotypic cells in catalyzing the repair of surface damage.

1. AMEIIGUITIES CONTRIBUTED BY TISSUE DISSOCIATION METHODS Unfortunately, little information is available on trypsin-mediated surface damage (see, e.g., Revel et al., 1974; Teng and Chen, 1976; Roblin et al., 1975b),a deficit which must be overcome if accurate molecular biological data on the mechanism of cell-cell adhesion are to be obtained. It is clear, however, that trypsinization destroys external cell surface components and also disrupts surface-associated cytoskeleta1 organization (see Section 11,B). (The discussion of cell surface damage in this section refers to damage to externally oriented cell surface components and surface-associated cytoskeletal elements, since little attempt has been made to differentiate the importance of the two types of damage on various aggregation processes.) The complexity of trypsin-mediated damage in interpreting cell aggregation assays was initially demonstrated by Gershman (1970), showing that aggregate size, rate of aggregate formation, and tissue-specific sorting-out behavior among chick embryonic cells probabIy reflect different properties of the cell surface (and perhaps different aspects of recovery from proteolytic damage). The tissue-specific aggregation of chick embryonic cells observed by Roth (1968) was later suggested to be a tissue-dependent effect of trypsinization (Curtis, 1970). Differences in aggregate size were also noted in chick neural retina cells before and after recovery from trypsin damage (Weiss and Maslow, 1972).Edwards et al. (1975) studied the effects of trypsinization on the aggregation of suspended BHK cells in the presence or absence of cycloheximide and divalent cations; aggregation after gentle trypsinization was insensitive to the inhibition of protein synthesis or to the absence of divalent cations, whereas aggregation of extensively trypsinized cells was sensitive. This latter process was also shown to be inhibited by colchicine or vinblastine (Waddell et al., 1974),suggesting that intact microtubules may be required to repair this surface damage and that surface-associated cytoskeletal components may be important comediators of the adhesion process. Many other studies have emphasized the effects of trypsin damage (Dunn et al., 1970; McClay and Baker, 1975; Cassiman and Bernfield, 1975; McGuire and Burdick, 1976) and the two-step nature of adhesion of trypsinized cells (Sheffield and Moscona, 1969; Umbreit and Roseman, 1975)-1oose initial adherence which is probably nonspecific, followed by stable adhesion where cell specificity plays a principal role.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

367

Detachment of cells cultured in vitro with EDTA or EGTA, rather than trypsin, yielded SV40-transformed 3T3 cell populations (1) which aggregated at a higher initial rate than 3T3 cells (Cassiman and Bemfield, 1975), (2) which were more mobile than 3T3 cells in penetrating three-dimensional aggregates of 3T3 cells (Gershman and Drumm, 1975), and (3) which sorted out from 3T3 cell populations (Gershman et al., 1976). One of these distinguishing properties (increased aggregation rate) was shown to be abolished by trypsinization (Cassiman and Bernfield, 1975), although Dorsey and Roth (1973) reported differences in the rates of adhesion of trypsinized 3T3 and SV40-transformed 3T3 cells to preformed aggregates. Unfortunately, these chelating agents are not effective dissociative reagents for adult or embryonic tissues. These studies suggest that viral transformation may lead to surface alterations which affect the nature and stability of cell -cell adhesions. More recently, Cassiman and Bernfield (1976) attempted to avoid trypsin-induced artifacts in tissue-specific adhesion of embryonic cells by (1)allowing aggregates to form from trypsinized tissue cells over a =-hour period during which repair of surface damage presumably occurred, and then (2) testing the adherence of these aggregates to a confluent population of cultured homotypic or heterotypic cells, which had also recovered from trypsin damage, in an assay analogous to that developed by Walther et al. (1973). The rate of binding of aggregates was higher for homotypic tissue interactions when compared to that for heterotypic binding; this difference was abolished by glutaraldehyde treatment, although some adhesion still occurred, suggesting that specificity in this system may be mediated by the rearrangement of mobile surface membrane components or that glutaraldehyde may destroy the active sites of adhesive molecules. In measuring the adhesion of suspended single cells (Walther et al., 1973; Winkelhake and Nicolson, 1976) or aggregates (Cassiman and Bernfield, 1976) to confluent layers of cultured cells, consideration must be given to evidence that the upper surface of these cells is essentially nonadhesive (DiPasquale and Bell, 1974; DiPasquale, 1975; Vasiliev et al., 1975),raising a question as to what these assays really measure -perhaps adhesion along the peripheral edges of cultured cells or to the tissue culture substrate. Perhaps these assays measure the ability of the attaching population to extend morphologically-and biochemically-mature microvilli into these regions. Walther et al. (1973) observed little adhesive specificity of trypsinized BHK, PyBHK, or 3T3 cells for homotypic or heterotypic layers; specificity was observed with embryonic cells of different tissues, indicating that some tissue-specific difference was being measured (morphological and/or

368

LLOYD A. CULP

biochemical?). A SEM evaluation of these relationships would prove useful in determining whether cell surface morphological criteria played a more significant role than external cell surface biochemical parameters. Since embryonic and adult tissues may be bathed in serumlike fluids in the animal, the role of serum components as positive or negative effectors of cell-cell. adhesion must be considered in in vitro assays. Unfortunately, little attempt has been made to evaluate systematically the influence of serum in various aggregation assays by directly affecting adhesive bond formation and/or by catalyzing the repair of surface damage, and many of these assays are performed in the absence of serum. However, Curtis and Greaves (1965)reported the isolation of an aggregation-inhibitory protein from horse serum which inhibited low-temperature aggregation, although the mechanism of action of this component has not been determined. Curtis and de Sousa (1975)recently reported the partial purification of a low MW mouse serum factor which acts as a positive or negative effector of mouse lymphocyte adhesion, depending on the cell combination being used. In view of the importance of specific serum components in mediating proper cell-substrate adhesion and accessibility of embryonic or adult tissues to circulating serum, further biochemical analysis of the prospective role of serum factors as effectors of cell-to-cell adhesion will be necessary.

2. ADHESION-PROMOTING FACTORS The initial success in obtaining and measuring adhesion-promoting factors (APFs) from simple eukaryotic systems, such as in the sponge studies described above, provided the impetus for searching for similar factors in higher systems. Lilien and Moscona (1967)and Lilien (1968) reported that tissue-cultured chick embryonic neural retina cells secreted material into serumless medium which catalyzed aggregation (increased aggregate size) of trypsinized suspensions of neural retina cells, but not of cells from other tissues. Lilien (1968) also provided some evidence for surface localization of this factor by demonstrating that neural retina cells, and not cells from other tissues, specifically adsorbed out this biological activity and that antiserum to this unpurified material (1)agglutinated suspended neural retina cells if they had been preadsorbed with this material and (2) specifically inhibited neural retina cell aggregation if added simultaneously with the factor. This factor was shown to bind to cells at 4"C, with no aggregation occurring at this temperature presumably because of a temperature-dependent

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

369

requirement to mobilize many surface molecules into a multivalent adhesion site. These preadsorbed cells aggregated when shifted to 37°C in medium lacking the factor (Lilien, 1969). This factor was effective in promoting aggregation only if added soon after the trypsin-mediated suspension of cells, which raises the question whether such a factor catalyzes the repair of proteolytic damage in some tissue-specific fashion or whether it acts as a multivalent ligand in adhesive bond formation. Later evidence identified both a retina-specific aggregation-promoting activity (RAPM) and a cerebral lobe-specific activity (CLAPM) (Balsamo and Lilien, 1974a,b). The supernatants were also found to contain a considerable amount of polysaccharide after cultivation of these cell types in medium containing radioactive glucosamine. Some of this radioactive polysaccharide-containing material was shown to bind to these cells, although the chemical heterogeneity of this material and the cellular location of binding were not examined. Interestingly, recovery from trypsin damage for a l-hour period reduced the amount of polysaccharide binding by 60%. Binding of a portion of this material was shown to be tissue-specific and displayed kinetic behavior consistent with cooperative binding (Balsamo and Lilien, 1974b). The complexity of this system was further demonstrated by Balsamo and Lilien (1974a) using glutaraldehyde-fixed cells. When trypsin-dissociated retina cells or trypsinized retina cells preadsorbed with RAPM were fixed with glutaraldehyde at room temperature, they failed to aggregate. However, RAPM-fixed cells catalyzed the aggregation (the aggregate size) of unfixed, freshly trypsinized retina cells, while unadsorbed, fixed cells had no effect upon the aggregation of freshly dissociated cells. When RAPM-fixed cells were added to freshly dissociated retina cells in the presence of cycloheximide, there was little aggregation. Balsamo and Lilien (1974a) interpreted these experiments as showing that a RAPM receptor on trypsinized cells, RAPM itself, and a third component which must be resynthesized by damaged cells mediate the adhesion process, although little additional evidence has been provided to support this model. A second interpretation, which is consistent with these experiments, is that RAPM does not mediate adhesion itself but catalyzes the repair of surface damage by any one of several mechanisms and that newly synthesized surface components mediate the formation of adhesive bonds. It should be emphasized that these supernatants are no doubt very complex mixtures of biochemical components from the cell. These experiments are important in demonstrating the complexity of cell surface functions required for factor-mediated adhesion.

370

LLOYD A. CULP

When RAPM was digested with pronase before binding and glutaraldehyde fixation of cells, there was no catalytic effect upon the aggregation of freshly dissociated cells (Balsamo and Lilien, 1974a). This indicates that polysaccharide chains per se were not effective in surface binding and/or catalyzing repair of damage (these possibilities were not differentiated). Balsamo and Lilien (1975) later showed that glycosidase treatment of these supernatants prevented binding of both protein and carbohydrate, suggesting that carbohydrate moieties may be important in binding to cells. RAPM binding was specifically sensitive to treatment with P-N-acetylhexosaminidaseor coaddition of N-acetylgalactosamine, whereas CLAPM binding was sensitive to amannosidase or coaddition of mannosamine. The complexity of this system was also demonstrated by adding RAPM-fixed cells to serumless rotation cultures of retina monolayers; these monolayers secreted a factor into their medium which catalyzed aggregation of the RAPMfixed cells but not unadsorbed, fixed cells (this effect was presumably not due to the detachment of live cells). All these effects were specific for retina or cerebral cells, and there may be many possibilities, both biochemical and morphological, for providing specificity of these various effects in these systems. Although these experiments indicate that so-called aggregation-promoting factors effect readhesion of cells by a complex sequence of events, fractionation of these extracts and analysis by this type of approach should prove valuable in determining the mechanism of readhesion of dissociated embryonic cells. The greater biochemical diversity of mammalian and avian cell surface components, in comparison to those of sponge cells, virtually guarantees a much more complicated mechanism of cell-cell adhesion than a simple bridging mechanism mediated by a receptor-ligand complex which may occur in the sponge cell system. Partially purified RAPM was shown to lack galactosyltransferase activity or acceptor activity (Garfield et al., 1974), a test of Roseman’s hypothesis (Roseman, 1970)that cell surface glycosyltransferases provide specificity in cell-cell adhesion (to be discussed in Section 111,C). Since there is limited evidence for any direct involvement of RAPM in adhesive bond formation, this evidence does not rigorously disprove this model. The supernatant of chick neural retina cultures incubated in serumless medium was fractionated by Sephadex G-200 and DEAE-cellulose chromatography to purify an RAPM factor (McClay and Moscona, 1974; Hausman and Moscona, 1975).An active factor which appeared to be a glycoprotein, to have an apparent MW of 50,000 (determined by SDS-PAGE), and to lack galactosyltransferase acceptor or donor ac-

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

371

tivity was obtained. Its carbohydrate contained N-acetylglucosamine, galactose, mannose, and sialic acid residues. This purified factor retained its tissue specificity, and its RAPM activity required integrity of its protein but not of its carbohydrate [in contrast to Balsam0 and Lilien’s (1975) data supporting the importance of the carbohydrate]. This purified RAPM retained its biological activity after the removal of SDS. It also persisted as a monomeric species upon sedimentation velocity analysis, raising the question as to whether this moiety can act as a multivalent ligand since multivalency in many other systems requires molecular self-association into large complexes. A factor with similar properties has recently been isolated from enriched preparations of chick neural retina surface membranes, consistent with the surface localization of this activity, by butanol extraction (Hausman and Moscona, 1976). This factor was obtained only from embryos less than 13 days of age and stimulated only the aggregation of pre-13-day neural retina cells. Factors isolated by the same experimental approaches from optic tectum or cerebrum failed to stimulate the aggregation of neural retina cells. The low doses of retina-specific factor required to obtain as much as a 1000-fold increase in aggregate volume suggest that this material is highly purified and that study of the effects of intact antibody or Fab fragments of antibody against this factor may prove valuable from a mechanistic viewpoint.

3. ADHESION-INHIBITORY ACTIVITIES

A different approach to an analysis of biochemical determinants important in embryologically specific chick cell adhesion was taken by Merrell and Glaser (1973). These investigators isolated enriched surface membrane preparations from chick neural retina or cerebellum tissues on discontinuous sucrose gradients after trypsin-mediated suspension of cells and after a 1 to 2-hour recovery period. (But have these cells in fact repaired their surface damage within this short time span?) These membrane preparations bound to homotypic but not to heterotypic cells or tissues and prevented homotypic aggregation. Binding was reversed by trypsinization. Although these workers suggested blockage of specific adhesion sites with these membrane fragments, it is surprising that these large membrane fragments and/or vesicles did not mediate an agglutination reaction, since no steps were taken to ensure the “uniualency” of these fragments. Plasma membranes prepared from whole tissues without trypsinization were then shown to display homotypic aggregation-blocking ac-

372

LLOYD A. CULP

tivity which paralleled specificity in the temporal development of optic tectum and neural retina tissues in the embryo (Gottlieb et al., 1974). In other words, membranes prepared from tissues of 7-, 8-, or 9-day embryos were more effective in blocking the aggregation of homotypic cells from embryos of the same age rather than cells from younger or older embryos. Trypsin treatment of the membranes destroyed inhibitory activity. These time-dependent changes in tissue specificity may be due to an alteration of the surfaces of many of the cells in these tissues or an alteration of the relative numbers of cell populations with different surface properties; these possibilities have not been differentiated. A geometric gradient of cell-cell adhesion within the tissue matrix of the developing chick neural retina has also been observed (Gottlieb et al., 1976)by measuring the adhesion of radioactive trypsin-suspended cells obtained from various regions of the tissue to monolayer cultures of various retinal cell populations. Extraction of these membrane preparations with lithium diiodosalicylate (LIS), which selectively extracts glvcoproteins (Marchesi and Andrews, 1971),generated solubilized material which displayed similar tissue- and time-specific inhibitory behavior (Merrell et al., 1975). The question arises as to how this glycoprotein fraction differs from the butanol-extracted membrane fraction obtained by Hausman and Moscona (1976) which stimulated aggregation-both glycoprotein fractions were obtained from chick neural retina. [Two very different assays for aggregation were used to measure these activities. Merrell et al. (1975) measured loss of single cells over a 30-minute time period, whereas Hausman and Moscona (1976) measured aggregate size after 24 hours. These assays no doubt detect very different parameters of recovery from trypsin-mediated damage and stability of adhesions.] A partial solution to this problem will require purification of the active glycoprotein from LIS extracts and biochemical comparison with the 50,000-MW glycoprotein of Hausman and Moscona (1976). A different approach to the biochemical mechanism of embryologically specific adhesion was recently undertaken by Rutishauser et al. (1976).These investigators fractionated secreted extracts, obtained by incubation of intact neural retina tissue (TCS) or trypsin-recovered monolayer cultures (MCS) of neural retina cells in serumless medium, on DEAE-cellulose and identified two uniquely interesting proteins (F1 of MW 140,000 from MCS and F2 of MW 110,000 from TCS); F2 was a dimer of two 55,000-MW subunits. Antibodies to F1 or F2 reacted with components on the surface of retinal cells; F1 and F2 had antigenic determinants in common. Moreover, partial degradation of F1 with trypsin liberated F2. Anti-F1 or anti-F2 immunoprecipitated

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

3 73

a major iodinatable cell surface protein from NP-40-solubilized retinal surface membranes which had a MW of 240,000-the size, iodinatability, and trypsin lability of this component indicate that it may be the chick cell surface CSP protein identified by Yamada and Weston (1974) which agglutinates formalin-treated red cells (Yamada et al., 1975). Anti-F2 also immunoprecipitated an NP-40-solubilized membrane component of MW 150,000 which also shared many antigenic determinants with the 240,000-MW surface component. Evidence was presented showing that F1, F2, and the 150,000-MW component appeared to be cleavage products of the 240,000-MW cell surface component. When Rutishauser et al. (1976) tested the effects of antibody to F1 or F2 on the adhesion of trypsin-recovered retinal cells to retinal cells immobilized on nylon fibers or on culture dishes, preincubation of suspended cells with anti-F2 inhibited adhesion by 75-90%, whereas anti-F1 displayed little inhibitory activity. The binding of retinal or brain cells, after recovery from trypsin damage, to nylon fibers coated with anti-F2 was dependent on the age of the embryo from which cells were harvested, reflecting the temporal specificity observed by Gottlieb et al. (1974). Rutishauser et al. (1976) constructed a model to explain their data, in which proteolysis of the precursor 240,000-MW [pro-cell adhesion molecule (pro-CAM)] membrane component liberates a fragment (MW 150,000) [cell adhesion molecule (CAM)] which mediates the adhesive process. Further breakdown of the membrane-bound 150,000-MW moiety solubilizes a component F2 (MW 110,000)which is a dimer of two 55,000-MW subunits which are disulfide cross-linked. There is no evidence yet to indicate how tissue and temporal specificities are conferred in this system. The data of Rutishauser et a2. (1976) may be related to observations from several other laboratories. Is the membrane precursor pro-CAM (MW 240,000) in retinal cells identical to the CSP glycoprotein found in chick embryo fibroblasts by Yamada and Weston (1974) [in fact, anti-F2 was shown by Rutishauser et al. (1976) to bind to chick embryonic fibroblasts rich in CSP]? CSP has been shown to agglutinate formalin-treated red cells (Yamada et al., 1975) and to catalyze stronger adhesion of cells to tissue culture substrates (Yamada et al., 1976). Similarly, the mammalian LETS glycoprotein, which is similar to CSP in many of its properties, has been shown to be a major constituent of SAM, which mediates cell-substrate adhesion; the LETS glycoprotein also displays a high turnover rate in SAM (Culp, 1976a,b). What aspect of pro-CAM synthesis, association with other membrane components, or proteolysis controls tissue and temporal specificity? Is

374

LLOYD A. CULP

the F2 monomer fragment of MW 55,000 similar to the 50,000-MW glycoprotein isolated by Hausman and Moscona (1976) which appears to catalyze increased aggregate size? Further chemical and immunological analysis of the membrane-bound pro-CAM and CAM moieties should determine their possible relationships with other factors identified by several other experimental approaches. It will also be interesting to determine if CAM molecules can self-associate, thereby mediating adhesive procksses on two different cell surfaces directly, or if CAM acts as a receptor site for a multivalent macromolecule which has not been identified and which bridges CAM sites on two different cells.

4. IMPLICATION OF COMPLEX POLYSACCHARIDES IN ADHESION Other indirect evidence for a role for surface polysaccharide in the reaggregation of trypsin-dissociated cells has also been obtained. Oppenheimer et al. (1969) observed an inhibition of the loss of single cells from suspension upon aggregation of mouse teratoma cells in the absence of L-glutamine, a deficit which was overcome by placing Dglucosamine and Dmannosamine in the aggregation medium. This was consistent with a glutamine requirement for the synthesis of amino sugars and the incorporation of amino sugars into complex polysaccharides which were lost as a result of trypsin-mediated damage. This glutamine-dependent aggregation was shown to occur in several suspension-cultured cell lines and to be prevented by two inhibitors of glutamine metabolism in cells, DON and azaserine (Oppenheimer,

1973). In an attempt to ptilize approaches which had proved successful in the sponge aggregation system, Oppenheimer and Humphreys (1971) measured the aggregation of trypsin-dissociated mouse teratoma embryoid bodies and observed a stimulatory effect upon the addition of mouse ascites fluid in which the embryoid bodies had been grown in situ. A factor was obtained from this fluid after fractionation whose activity was trypsin-sensitive and Ca2+-dependent.This factor was very large (MW > lo6)and highly negatively charged. Oppenheimer (1975) showed that this teratoma adhesion factor (TAF) was sensitive to P-galactosidase treatment and that its activity was partially inhibited by the coaddition of D-galactose. More recently, Meyer and Oppenheimer ( 1976) obtained evidence that this aggregation-promoting activity for teratoma cells required at least two components which were separable by DEAE-cellulose chromatography-a nonanionic frac-

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

3 75

tion rich in a high MW glycoprotein and a highly negatively charged fraction also rich in polysaccharide-containing components. Perhaps the active component in the latter fraction is a GAG-containing proteoglycan, although little is known about the polysaccharide moieties in these preparations. Other indirect evidence for the involvement of complex polysaccharides in adhesion processes is the following. Neuraminidase treatment affected the aggregation of some trypsin-dissociated cells (Kemp, 1970; Vicker and Edwards, 1972; Lloyd and Cook, 1974) but not others (McQuiddy and Lilien, 1971); these experiments may be complicated by contamination of these enzymes with protease activities, although Lloyd and Cook (1974) prevented this neuraminidasedependent effect by the coaddition of bovine submaxillary mucin (a highly sialylated glycoprotein). Inhibitory effects on aggregation have also been noted after the addition of hexosamines (Lloyd and Kemp, 1971; Kuroda, 1974a), fucose (Kuroda, 1974b), and dextran sulfates (Kuroda, 1974c) in experiments which may have been complicated by nonspecific effects at high concentrations of these reagents (particularly inhibition of polysaccharide synthesis b y monosaccharides or nonspecific blockage of adhesion sites by the polymers). Aggregationpromoting factors liberated into the medium of several cell lines were found to be sensitive to hyaluronidase treatment by Pessac and Defendi (1972), and the addition of hyaluronic acid stimulated aggregation. This latter study is interesting in light of (1) Oppenheimer and Humphreys’ (1971) discovery of a large, negatively charged molecule which stimulated mouse teratoma cell adhesion, (2)the importance of hyaluronic acid in cell-substrate adhesion (see previous sections), and (3) the prominence of lanthanum-stainable proteoglycans detected electron microscopically at cell-cell adhesion sites (Overton, 1969; Khan and Overton, 1969,1970). Examination of the possible role of Con-A-binding sites on the surface of cells in adhesion has generated conflicting data. Steinberg and Gepner (1973) tested the effects of a trypsinized Con-A preparation, which was supposedly monovalent, and found no effect upon the rate of aggregation or sorting behavior of chick embryonic tissue cells. Evans and Jones (1974), however, observed an inhibitory effect by trypsinized Con-A upon the rate of aggregation of similar chick embryonic cell populations. These studies were no doubt complicated by variable damage to surface-associated Con-A-binding polysaccharides induced b y proteolysis, both in terms of loss from the cell surface and altered mobility, and by the inability to determine accurately the valency of the trypsinized Con-A preparations.

376

LLOYD A. CULP

In any case, data are accumulating which suggest that cell-to-cell adhesion processes in avian and mammalian cells may be mediated by complex polysaccharides on the surfaces of cells, although investigation of the biochemical nature of these determinants and their functional roles has been slowed by the variability introduced by proteasemediated dissociation of cells. The inhibitory effects. of cytochalasin B upon (1)the early events in the reaggregation of chick embryonic cells (Maslow and Mayhew, 1972,1974; Appleton and Kemp, 1974) and (2) metabolic cooperation of contacting 3T3 cells (Stoker, 1975), as well as the partial inhibition by antimyosin antibody of the aggregation of chick muscle or liver cells (Kemp et al., 1971, 1973), suggest the involvement of surface membrane-associated microfilaments in adhesive processes. Cell-cell adhesion may be similar in biochemical complexity to cell-substrate adhesion; that is, it may be mediated by externally oriented glycoproteins and/or proteoglycans which may be linked in the membrane to anchorage components for the contractile microfilaments (see Section 11,B). Although studies of cell aggregation have been complicated in the past by a lack of awareness of cell surface damage caused by the dissociation of tissues into single-cell suspensions, attempts are now being made to avoid artifacts introduced by this damage and to utilize antibodies (particularly Fab fragments) against specific cell surface components to test their ability to block the formation of new adhesions. Investigators will have to differentiate between the formation of transient nonspecific adhesions and the formation of more stable, specific adhesions. It is quite likely that the exquisitely formed and dynamic cellular matrixes of embryonic tissues result from very complex molecular phenomena on the cell surface and that an awareness of this complexity will be important in future approaches.

C. Role of Cell Surface Glycosyltranrferarer(3)

Because of the importance of protein recognition of specific polysaccharides in several adhesion systems, Roseman (1970) proposed that this specificity was provided by a cell surface-localized glycosyltransferase (ectoglycosyltransferase) on one cell surface binding to polysaccharide on the surface of an adjacent cell whose terminal sugar residues were complementary to the enzyme’s active site. This enzyme-substrate complex would be stable in the absence of the specific nucleotide sugar required by the enzyme. Thus far, most effort has been focused on proving that ectoglycosyltransferases do exist, rather than establishing any potential role for them in adhesive

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

377

processes [for a inore complete review of this topic, see Shur and Roth (1975)l. However, Garfield et uZ. (1974) established that chick embryonic RAPM was not a galactosyltransferase, although the mechanism of action of this factor is not known. A receptor protein in the surface membrane of mammalian liver cells which binds galactoseterminated glycoproteins in the circulating serum of animals has been purified and also shown not to be a glycosyltransferase (Hudgin et aZ., 1974; Hudgin and Ashwell, 1974; Kawasaki and Ashwell, 1976). The first evidence that cell surface glycosyltransferases may exist was provided by Roth et al. (1971)who demonstrated that intact chick embryonic neural retina cells catalyzed the transfer of [14C]galactose from UDP galactose to endogenous or exogenous acceptors. Since trypsinization was used to suspend these retinal cells for assaying enzyme activity, the possible penetration of nucleotide sugar or a breakdown product of nucleotide sugar into the cell, followed by its utilization for polysaccharide synthesis via de novo pathways, must be a major consideration. In an attempt to eliminate these possibilities, Roth et al. (1971)demonstrated that the addition of galactose or galactose l-phosphate did not competitively inhibit the utilization of nucleotide sugar and that nucleotide sugar was not taken u p by cells. However, this latter point is difficult to prove, since (1)a nucleotide sugar with radioactivity located in the pyrimidine portion of the molecule rather than in the sugar portion was not used and (2) the concentration of nucleotide sugar was so low that uptake into the cell might result in the immediate transfer of sugar by intracellular sugar transferases to intracellular acceptors. However, the transfer of sugar to high MW exogenous acceptors which apparently remained in the extracellular fluid (but can these macromolecules shuttle in and out of surface-damaged cells?) suggests that some surface ectogalactosyltransferase activity may have existed (Roth et al., 1971),although the exogenous acceptor was not purified and characterized at the end of the reaction. Autoradiography experiments indicated that the endogenous reaction resulted in radiolabeling of greater than 90% of the cells, which indicated that a small population of heavily damaged cells was not solely responsible for the transferase activity. Evidence has been obtained that leakage of intracellular glycosyltransferases into the culture medium is not responsible for the activities being measured (LaMont et al., 1974; Bernacki, 1974). Roth and White (1972) reported evidence for ectogalactosyltransferases on the surfaces of normal and spontaneously transformed BALB/c. 3T3 cells. I n 3T3 cells the transfer of radioactivity from UDP galactose to endogenous acceptors was detected by autoradiographic means only in dense, contacting populations, whereas 3T12 cells gly-

3 78

LLOYD A. CULP

cosylated endogenous acceptors in both sparse and dense cultures. Roth and White (1972) suggested that potential acceptors on normal cells are unavailable to cis-located enzyme, requiring cell-cell contact to obtain trans-mediated acceptor-enzyme complex formation. However, Patt and Grimes (1974) were unable to repeat these results using normal and transformed Swiss 3T3 cells, raising the question whether this contact-mediated process is a generalized phenomenon. The surface localization of galactoglytransferase activity has been questioned by Deppert et al. (1974),who demonstrated that the incorporation of radioactive galactose from UDP galactose by BHK cells could be completely explained by the breakdown of nucleotide sugar to free sugar by presumably cell surface hydrolytic enzymes and the subsequent incorporation of free sugar by de no00 intracellular pathways. These investigators also showed that high concentrations of exogenously added galactose or galactose l-phosphate were required to inhibit effectively the incorporation of sugar donated by nucleotide sugar; perhaps the cell surface hydrolytic enzymes for nucleotide sugars preferentially shuttle their galactose product into the cell more effectively than the galactose transport system with its high K,. This latter observation questions the argument that decreased competition for transferase activity by exogenously added free sugars proves that the breakdown of nucleotide sugar and uptake of this sugar are not the principal mechanism of incorporation of radioactive sugar. Several laboratories have examined the potential existence of and differences in the ectoglycosyltransferases of normal and transformed cells (Bosmann, 1972; Webb and Roth, 1974; Patt and Grimes, 1974, 1975; Yogeeswaranet al., 1974). In general, types of controls similar to those described above were employed in attempts to exclude incorporation via intracellular pathways, and similar criticisms apply, particularly since surface-damaging conditions were used to suspend cells in order to measure enzyme activity. Webb and Roth (1974) noted greatly increased activities in trypsin-suspended mitotic 3T3 cells as compared to noncontacting interphase cells, activities which resembled those of interphase or mitotic 3T12 cells. Roth et al. (1974) demonstrated ectoglycosyltransferase activities for galactose, sialic acid, N-acetylglucosamine, and N-acetylgalactosamine in trypsinsuspended BALB/c 3T3, SV40-transformed 3T3, and revertant variant cells and noted no consistent patterns in the levels of these various activities in sparse and in dense cultures. Similarly, Patt and Grimes (1975) identified activities with EDTAsuspended normal or SV4O-transformed 3T3 cells for six different nucleotide sugars (5’-AMPwas also added to assays to prevent the breakdown of nucleotide sugars). Interestingly, very low activities toward

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

379

endogenous acceptors were noted in these studies with substratebound cells, whereas EDTA-suspended cells contained sizable levels of all activities, raising some question as to whether EDTA-induced leakiness of cells resulted in intracellular transfer from nucleotide sugar to endogenous acceptors. These experiments were no doubt complicated b y the lack of information on the precise nature of the cell surface damage caused by EDTA treatment and the rapidity of repair of this damage. Another alternative for the EDTA effect may be increased mobility of surface components, permitting more effective interaction of enzyme and substrate on the cell surface. Some transfer of galactose or sialic acid to appropriate exogenous acceptors by substrate-bound normal or transformed cells was observed by Patt and Grimes (1975), suggesting that ectoglycosyltransferase enzymes exist in monolayer cells but that they are inaccessible to cis-oriented endogenous acceptors and that viral transformation does not necessarily lead to accessibility of the substrate to an enzyme on the surface of the same cell. Several laboratories have reported sugar transferase activities in untreated blood-containing cells. Verbert et al. (1976) have reported that rat lymphocytes display ectogalactosyltransferase activity which cannot be inhibited by the addition of free sugar or phlorizin, an inhibitor of sugar transport; this activity was stimulated twofold by the addition of soybean agglutinin. Mammalian thymocytes may have ectogalactosyl (LaMont et al., 1974) and ectosialyl (Painter and White, 1976) transferases which are stimulated by the addition of Con A. Porter and Bernacki (1975) reported the transfer of radioactive sialic acid from CMP sialic acid to the plasma membrane of mouse leukemic cells as determined by electron microscope autoradiography (the vast majority of this radioactivity was removed by the treatment of intact cells with neuraminidase). However, Patt et al. (1976) reported very little ectoglycosyltransferase activity in mouse splenic lymphocytes which had not been treated with surface-disruptive reagents. Ectosialyltransferase activities have also been reported by Datta (1974). However, Hirschberg et al. (1976) demonstrated by use of CMP sialic acid containing radioactivity in its primidine or sugar portions that (1) NIL, BHK, and 3T3 cells efficiently broke this nucleotide sugar down to free sialic acid which was utilized by the cell for intracellular glycosylation, (2) considerable loss of cell integrity resulted from EDTA or trypsin treatment before the measurement of ectoglycosyltransferase activity, and (3)relatively high concentrations of exogenously added free sialic acid were required to inhibit nucleotide sugar-mediated reactions. Perhaps the most effective demonstration of ectoglycosyltransferase

3 80

LLOYD A. CULP

enzymes are the experiments of Yogeeswaran et al. (1974).The addition of glass beads linked covalently to specific glycosphingolipids to substrate-bound BHK,NIL, or their polyoma-transformed counterpart cells resulted in specific glycosylation using an appropriate glycosphingolipid acceptor and nucleotide sugar donor. Beads were washed with SDS after the reaction to remove glycolipids and glycoproteins adsorbed nonspecifically. The presence of nucleotide sugar was not required to obtain glycosylation of the bead-bound glycosphingolipid if the cells were being grown in an appropriate radioactive sugar precursor; this suggests that an activated form of the sugar (Lennarz, 1975) was present in the surface membranes of these cells with limited availability to endogenous acceptors. Ectoglycosyltransferases have also been reported in rat intestinal epithelial cells by Weiser (1973) and Podolsky et al. (1974). Crypt cells had considerably higher activities for N-acetylglucosamine, glucose, galactose, mannose, and fucose, whereas differentiating villus cells had higher sialic acid activity toward endogenous acceptors. The rather aggressive conditions required to dissociate these cells and the paucity of controls for measuring the leakage of nucleotide sugar or free sugar into the cell with resultant intracellular incorporation raise a question as to how much of the activities being measured in these experiments is actually surface-localized enzyme activity. The addition of Con A to these cell populations reduced the galactosyltransferase activity by 50%and had little effect on the other sugar transferases (Podolsky et al., 1974). Perhaps the only direct study of the possible role of an ectoglycosyltransferase activity in adhesion is that of Jamieson et al. (1971),which measured the adhesion of human platelets to collagenous substrates. Platelet cell surface glucosyltransferase was implicated in mediating the adhesion of these cells to galactosy-collagen sites by the following evidence. ( 1)Glucosyltransferase activity was enriched in surface membrane preparations; (2) p-chloromercuribenzoate, D-glUCoSamine, aspirin, and chlorpromazine inhibited both glucosyltransferase activity of platelets and the adhesion of platelets to collagen in a similar fashion; and (3) collagen glycopeptides with terminal galactose moieties inhibited the adhesion of platelets to collagen. Destruction of the carbohydrate moieties of collagen by periodate oxidation also resulted in reduced adherence by platelets, although more subtle damage to the protein cannot be ruled out in these studies (Brass and Bensusan, 1976). Unfortunately, platelet glucosyltransferase has not been purified to test the inhibitory potential of purified enzyme or antibody to it in this

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

38 1

system. And evidence, reported by Menashi et al. (1976),that native collagen was not a substrate for platelet glucosyltransferase and that platelet-bearing plasma contained appreciable levels of this enzyme which would presumably block platelet adherence to galactosyl-collagen, suggests that this sugar transferase may not mediate platelet adhesion. To summarize, evidence has been obtained for low levels of ectoglycosyltransferases on the surfaces of some cells, enzymes which apparently lack accessibility to cis-oriented acceptor molecules for some as yet undefined and potentially interesting reason. Establishing proof of their existence has been complicated by artifacts due to the breakdown of nucleotide sugars, with resultant intracellular incorporation of free sugar, and possibly to the intracellular utilization of intact nucleotide sugar by damaged cells. Unfortunately, different laboratories have utilized very different controls in their experiments and different means of suspending cells, complicating any comparative analysis of experiments. Little information is available on the possible role of ectoglycosyltransferases in mediating the formation of adhesive bonds between cells. It is quite likely that their functional role, if any, will be determined only when these enzymes have been purified and when specific antibodies have been prepared to them; both the pure enzymes and their antibodies should prove to be valuable ligands as potential inhibitors of adhesion formation. Information on the biochemical properties of these enzymes and their association with other macromolecules in the surface membrane of cells should also prove important in the future exploration of their function. IV.

CONCLUSION

Although investigation of the molecular mechanism of cellular adhesion processes has been a major undertaking in many systems during the past decade, there is still no description of the complete molecular topography of any adhesion site. However, specific glycoproteins and proteoglycans on the surface of cells have been implicated as mediating these processes in virtually all cell-cell and cell -substrate adhesion processes which have been examined. Specificity in some systems appears to be mediated by the binding of proteins which have complementary sites for polysaccharides with anomerically specific terminal residues-a specificity comparable to that observed in lectin-polysaccharide binding. In other systems, it is not yet clear whether adhesive bonds are mediated by protein-polysaccharide or protein-protein interactions. However, indirect evidence

382

LLOYD A. CULP

strongly suggests that carbohydrate-containing macromolecules on

the surface of the cell play an as yet undefined role in cell-substrate and cell-cell adhesions. The precise conformational properties of the molecules implicated in adhesion have not been determined in any system. There is accumulating evidence that divalent cations do not simply act as cross-bridging reagents between negatively charged macromolecules on adjacent cell surfaces, but that they are important in maintaining the multivalency of bridging proteoglycans in simpler cells. They may also be important in reorganization of the cell surfaceassociated cytoskeleton in higher cells, a cytoskeleton which appears to be linked to externally oriented glycoproteins and proteoglycans which directly form adhesive bonds. It is also possible that divalent cation binding to the hexuronic acid moieties in proteoglycans plays an important role in molecular organization, although little is known about the molecular composition and inter- and intramolecular association of these “supramolecules,” a shortcoming requiring extensive biochemical analysis of purified complexes. Most investigations thus far have pursued the identification of positive effectors of adhesive bond formation. Since cell surface molecules exist in a fluid matrix and can constantly change their association with cis-oriented macromolecules, consideration should also be given to the possibility that some cell surface macromolecules inhibit the function of a prospective adhesion factor glycoprotein and/or proteoglycan b y binding to this adhesion factor, thereby preventing normal self-association of many adhesion factor molecules into multivalent adhesion sites and/or binding the active site of adhesion factor in such a way that it is unavailable for trans-oriented binding to a component on another cell surface. Such interactions would indicate that the formation of cellular adhesions requires a sensitive balance of the biosynthesis and integration into the membrane of both positive and negative effectors. In any case, all the evidence to date suggests that cell adhesion is a very complex phenomenon mediated by the interaction of several classes of cell surface macromolecules with complementary binding sites, particularly glycoproteins and proteoglycans. The continuing purification and biochemical analyses of many different cell surface components should provide important reagents for testing the functionality of specific determinants in adhesion processes. Understanding the molecular complexity of adhesion processes should subsequently provide considerable insight into the mechanisms of embryonic development and of the malignant conversion process in cells of higher organisms.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

383

ACKNOWLEDGMENTS The author acknowledges funding for some of the studies reported here from the National Cancer Institute (research grant 5-ROl-CA-l3513), from the American Cancer Society (research grant BC-217 with partial support from the Ohio division), and from the National Institutes of Health (training grant 5-T01-GM-00171 to the Department of Microbiology). Appreciation is also expressed to Martha Cathcart, Barrett Rollins, Ben Murray, Robert Haas, and Howard Gershman, for critical analyses of this chapter. The author was a Harry H. Pinney Fellow in Cancer Research at Case Western Reserve University and continues as a Career Development Awardee of the National Cancer Institute (l-KO4-CA 70709). REFERENCES Abercrombie, M., and Dunn, G. A. (1975). Adhesions of fibroblasts to substratum during contact inhibition observed by interference reflection microscopy. Exp. Cell. Res. 92, 57-62. Abercrombie, M., Heaysman, J. E. M., and Pegmm, S. M. (1971).The locomotion of fibroblasts in culture. IV. Electron microscopy of the leading lamella. Exp. Cell Res. 67,359-367. Albrecht-Buehler, G. (1976). Filopodia of spreading 3T3 cells: Do they have a substrate-exploring function?J . Cell Biol. 69,275-286. Alexander, S., Brackenbury, R., and Sussman, M. (1975).Tryptic destruction of aggregative competence in Dictyostelium discoideum and subsequent recovery. Nature (London) 254,698-699. Altenburg, B. C., Somers, K., and Steiner, S. (1976). Altered microfilament structure in cells transformed with a temperature-sensitive transformation mutant of murine sarcoma virus. Cancer Res. 36,251-257. Appleton, J. C., and Kemp, R. B. (1974). Effects of cytochalasins on the initial aggregation in vitro of embryonic chick cel1s.J. Cell Sci. 14, 187-196. Ash, J. F., and Singer, S. J. (1976). Concanavalin A-induced transmembrane linkage of concanavalin A surface receptors to intracellular myosin-containing filaments. Proc. Natl. Acad. Sci. U.S.A. 73,4575-4579. Atherly, A. G., Barnhart, B. J., and Kraemer, P. M. (1977).Growth and biochemical characteristics of a detachment variant of CHO cel1s.J. Cell Physiol. 89, 375-385. Atkins, E. D. T., and Sheehan, J. K. (1973). Hyaluronates: Relation between molecular conformations. Science 179,562-564. Austen, F. K. (1974). Hageman-factor-dependent coagulation fibrinolysis and kinin generation. Transplant. Proc. 6 , 3 9 4 5 . Balsamo, J., and Lilien, J. (19744. Functional identification of three components which mediate tissue type-spceific embryonic cell adhesion. Nature (London)251, 522524. Balsamo, J., and Lilien, J . (197413). Embryonic cell aggregation: Kinetics and specificity of binding of enhancing factors. Proc. Natl. Acad. Sci. U.S.A. 71, 727-731. Balsamo, J., and Lilien, J. (1975). The binding of tissue-specific adhesive molecules to the cell surface: A molecular basis for specificity. Biochemistry 14, 167-171. Barland, P., and Schroeder, E. A. (1970). A new rapid method for the isolation of surface membranes from tissue culture cells. J . Cell Biol. 45, 662-668. Bases, R., Mendez, F., Mendez, L., and Anigstein, R. (1973). Stimulation of HeLa cellsurface attachment by histones. Exp. Cell Res. 76,441-444. Bemacki, R. J. (1974). Plasma membrane ectoglycosyltransferase activity of L1210 murine leukemic cells. J . Cell Physiol. 83,457-466.

384

LLOYD A. CULP

Beug, F., Gerish, G., Kempff, S., Riedel, V., and Cremer, G. (1970). Specific inhibition of cell contact formation in Dictyostelium by univalent antibodies. Exp. Cell Res. 63, 147-158. Beug, H., Katz, F. E., Stein, A., and Gerisch, G. (1973a). Quantitation of membrane sites in aggregating Dictyostelium cells by use of tritiated univalent antibody. Proc. Nutl. Acud. Sci. U S A . 70,3150-3154. Beug, H., Katz, F. E., and Gerisch, G. (1973b). Dynamics of antigenic membrane sites relating to cell aggregation in Dictyostelium discoideum. J. Cell Biol. 56,647-658. Bolund, L., Darzynkiewin, Z., and Ringertz, N. R. (1970). Cell concentration and the staining properties of nuclear deoxyribonucleoprotein. Exp. Cell Res. 62,76-89. Bonner, J., Barkley, D., Hall, E., Konijn, T., Mason, J., O’Keefe, G., and Wolfe, P. (1969). Acrasin, acrasinase and the sensitivity to acrasin in Dictyostelium discoideum. Deu. Biol. 20,72-87. Boone, C. W. (1975). Malignant hemangioendotheliomas produced by subcutaneous inoculation of BALB/3T3 cells attached to glass beads. Science 188,68-70. Bosmann, H. B. (1972). Cell surface glycosyltransferases and acceptors in normal and RNA- and DNA-virus transformed fibroblasts. Biochem. Biophys. Res. Commun. 48,523-529. Brackenbury, R., and Sussman, M. (1975). Mutant of Dictyostelium discoideum defective in cell contact regulation of enzyme expression. Cell 4,347-351. Bradley, W. E. C., and Culp, L. A. (1977). Contact-inhibited revertant cell lines isolated from simian virus 40-transformed cells. VII. Serum detachment-resistant revertant cells. J . Virol. 21, 1228-1231. Bragina, E. E., Vasiliev, J. M., and Gelfand, I. M. (1976). Formation of bundles of microfilaments during spreading of fibroblasts on the substrate. E x p . Cell Res. 97,241248. Brass, L. F., and Bensusan, H. B. (1976). On the role of the collagen carbohydrate residues in the platelet collagen interaction. Biochim. Biophys. Actu 444,43-52. Brown, S., Levinson, W., and Spudich, J. A. (1976). Cytoskeletal elemefits of chick embryo fibroblasts revealed by detergent extraction. J . Suprumol. Struct. 5, 119-130. Burger, M. M., Turner, R. S., Kuhns, W. J., and Weinbaum, G. (1975).A possible model for cell-cell recognition via surface macromolecules. Phil. Trans. R. Soc. Lond., Ser. B 271,379-393. Burridge, K. (1976). Changes in cellular proteins after transformation: Identification of specific glycoproteins and antigens in sodium dodecyl sulfate gels. Proc. Nutl. Acud. Sci. U.S.A. 73,4457-4461. Cassiman, J., and Bernfield, M. R. (1975). Transformation-induced alterations in fibroblast adhesion: Masking by trypsin treatment. Exp. Cell Res. 91,31-35. Cassiman, J . , and Bemfield, M. R. (1976). Use ofpreformed cell aggregates and layers to measure tissue-specific differences in intercellular adhesion. Deu. B i d . 52, 231245. Cauldwell, C. B., Henkart, P., and Humphreys, T. (1973). Physical properties of sponge aggregation factor: A unique proteoglycan complex. Biochemistry 12,3051-3055. Chen, L. B., Gallimore, P. H., and McDougall, J. K. (1976). Correlation between tumor induction and the large external transformation sensitive protein on the cell surface. Proc. Nutl. Acud. Sci. U.S.A. 73,3570-3574. Chipowsky, S., Lee, Y.C., and Roseman, S. (1973). Adhesion of cultured fibroblasts to insoluble analogues of cell-surface carbohydrates. Proc. Natl. Acud. Sci. U.S.A. 70, 2309-2312. Cohn, R. H., Cassiman, J., and Bernfield, M. R. (1976). Relationship of transformation, cell density, and growth control to the cellular distribution of newly synthesized glycosaminoglycan J . Cell Biol. 71,280-294.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

3 85

Crandall, M. A., and Brock, T. D. (1968).Molecular aspects of specific cell contact. Science 161,473-475. Crandall, M., Lawrence, L. M., and Saunders, R. M. (1974).Molecular complementarity of yeast glycoprotein mating factors. Proc. Natl. Acad. Sci. U.S.A. 71,26-29. Critchley, D. R., Wyke, J. A., and Hynes, R. 0. (1976). Cell surface and metabolic labelling of the proteins of normal and transformed chicken cells. Biochim. Biophys. Acta 436,335-352. Culp, L. A. (1974). Substrate-attached glycoproteins mediating adhesion of normal and virus-transformed mouse fibroblasts. J. Cell Biol. 63, 71-83. Culp, L. A. (1975). Topography of substrate-attached glycoproteins from normal and virus-transformed cells. E x p . Cell Res. 92,467-477. Culp, L. A. (19764. Electrophoretic analysis of substrate-attached proteins from normal and virus-transformed cells. Biochemistry 15,4094-4104. Culp, L. A. (1976b). Molecular composition and origin of substrate-attached material from normal and virus-transformed cells. J. Suprumol. Struct. 5,239-255. Culp, L. A., and Black, P. H. (1972a). Release of macromolecules from BALB/c mouse cell lines treated with chelating agents. Biochemistry 11,2161-2172. Culp, L. A., and Black, P. H. (197213). Contact-inhibited revertant cell lines isolated from simian virus 40-transformed cells. 111. Concanavalin A-selected revertant cel1s.J. Virol. 9, 611-620. Culp, L. A., and Buniel, J. F. (1976). Substrate-attached serum and cell proteins in adhesion of mouse fibrob1asts.J. Cell Physiol. 88,89-106. Culp, L. A,, Terry, A. H., and Buniel, J. F. (1975).Metabolic properties of substrate-attached glycoproteins from normal and virus-transformed cells. Biochemistry 14, 406-412. Culp, L. A., Buniel, J. F., and Rosen, J. J. (1978). Initial studies on the mechanism of substrate adhesion of normal and virus-transformed cells. Proc. Symp. Cell Surf. Curbohydr. Chem. 172nd Meet. Am. Chem. Soc., Sun Francisco Culg., 1976, in press. Curtis, A. S. G. (1970). On the occurrence of specific adhesion between cells. J. Embryol. E x p . Morphol. 23,253-272. Curtis, A. S. G. (1973). Cell adhesion. Prog. Biophys. Mol. Biol. 27,317-386. Curtis, A. S. G., and de Sousa, M. A. B. (1975). Lymphocyte interactions and positioning. I. Adhesive interactions. Cell. lmmunol. 19,282-297. Curtis, A. S. G., and Greaves, M. F. (1965). The inhibition ofcell aggregation by a pure serum protein. J. Embryol. E x p . Morphol. 13,309-326. Datta, P. (1974).Labeling of the external surface of hamster and mouse fibroblasts with ['4C]sialic acid. Biochemistry 13,3987-3991. Dea, I. C. M., Moorhouse, R., Rees, D. A., Arnott, S., Guss, J . M., and Balazs, E. A. (1973). Hyaluronic acid: A novel, double helical molecule. Science 179,560-562. Defendi, V., and Gasic, G. (1963). Surface mucopolysaccharides of polyoma virus transformed ce1ls.J. Cell. Comp. Physiol. 62, 23-31. Deppert, W., Werchau, H., and Walter, G. (1974). Differentiation between intracellular and cell surface glycosyl transferases: Galactosyl transferase activity in intact cells and in cell homogenates. Proc. Natl. Acad. Sci. U.S.A. 71, 3068-3072. DiPasquale, A. (1975).Locomotory activity of epithelial cells in culture. E x p . Cell Res. 94, 191-215. DiPasquale, A,, and Bell, P. B. (1974).The upper cell surface: Its inability to support active cell movement in culture. J. Cell Biol. 62, 198-214. Dorsey, J. K., and Roth, S. (1973). Adhesive specificity in normal and transformed mouse fibroblasts. Dev. Biol. 33,249-256. Duguid, J . R., and Revel, J. P. (1975).The protein components of the gap junction. Cold Spring Harbor Scymp. Quant. Biol. 10,45-47.

386

LLOYD A. CULP

Dunn, M. J., Owen, E., and Kemp, R. B. (1970). Studies on the mechanism underlying the inhibition by puromycin of cell aggregation in uitro. J . Cell Sci. 7,557-573. Edwards, J. G., Campbell, J. A., Robson, R. T., and Vicker, M. G. (1975). Trypsinized BHK-21 cells aggregate in the presence of metabolic inhibitors and in the absence of divalent cati0ns.J. Cell Sci. 19, 653-667. Erickson, C. A,, and Trinkaus, J. P. (1976). Microvilli and blebs as sources of reserve surface membrane during cell spreading. E x p . Cell Res. 99,375-384. Evans, P. M., and Jones, B. M. (1974).Studies on cellular adhesion-aggregation: Consideration of involvement of concanavalin A receptors. E x p . Cell Res. 8 8 , 5 6 4 2 . Follett, E. A. C., and Goldman, R. D. (1970). The occurrence of microvilli during spreading and growth of BHK21/C13 fibroblasts. E x p . Cell Res. 59, 124-136. Frazier, W. A., Rosen, S. D., Reitherman, R. W.,and Barondes, S. H. (1975). Purification and comparison of two developmentally regulated lectins from Dictyostelium discoideum. J . Biol. Chem. 250,7714-7721. Gail, M . H., and Boone, C. W. (1972).Cell-substrate adhesivity-A determinant of cell motility. E x p . Cell Res. 70,33-40. Garfield, S., Hausman, R. E., and Moscona, A. A. (1974). Embryonic cell aggregation: Absence of galactosyltransferase activity in retina-specific cell-aggregating Factor. Cell Differ. 3, 215-219. Gerisch, G. (1968). Cell aggregation and differentiation in Dictyostelium. Curr. Top. Deo. Biol. 3, 157-195. Gershman, H. (1970).On the measurement of cell adhesiveness.J. E x p . Zool. 174,391406. Gershman, H., and Drumm, J. (1975). Mobility of normal and virus-transformed cells in cellular aggregates. J . Cell Biol. 67,419-435. Gershman, H., Drumm, J,, and Culp, L. (1976).Sorting out of normal and virus-transformed cells in cellular aggregates. J . Cell Biol. 68,276-286. Goggins, J. F., Johnson, G. S., and Pastan, I. (1972).The effect of dibutryryl cyclic adenosine monophosphate on synthesis of sulfated acid mucopolysaccharides by transformed fibroblasts. J . Biol. Chem. 247, 5759-5764. Goldman, R. D., and Follett, E. A. C. (1969). The structure of the major cell processes of isolated BHK21 fibroblasts. E x p . Cell Res. 57,263-276. Goodenough, D. A. (1974). Bulk isolation of mouse hepatocyte gap junctions: Characterization of the principal protein, c0nnexin.J. Cell Biol. 61,557-563. Goto, M., Kataoka,*Y.,Kimura, T., Goto, K., and Sato, H. (1973). Decrease of saturation density of cells of hamster cell lines after treatment with dextran sulfate. E x p . Cell Res. 82,367-374. Gottlieb, D. I., Merrell, R., and Glaser, L. (1974).Temporal changes in embryonal cell surface recognition. Proc. Natl. Acad. Sci. U.S.A.71, 1800-1802. Gottlieb, D. I., Rock, K., and Glaser, L. (1976). A gradient of adhesive specificity in developing avian retina. Proc. Natl. Acad. Sci. U S A . 73,410-414. Graham, J. M., Hynes, R. O., Davidson, E. A., and Bainton, D. F. (1975).The location of proteins labeled by the '*51-lactoperoxidase system in Nil 8 hamster fibroblast. Cell 4,353-365. Grinnell, F. (1973).Concanavalin A increases the strength of baby hamster kidney cell attachment to substratum.J . Cell Biol. 58,602-607. Grinnell, F. (1974a). Studies on the mechanism of cell attachment to a substratum with serum in the medium: Further evidence supporting a requirement for two biochemically distinct processes. Arch. Biochem. Biophys. 165, 524-530. Grinnell, F. (197413).Studies on the mechanism of cell attachment to a substratum: Evidence for three biochemically distinct processes. Arch. Biochem. Biophys. 160, 304 -3 10.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

387

Grinnell, F. (1975). Cell attachment to a substratum and cell surface proteases. Arch. Biochem. Biophys. 169,474-482. Grinnell, F. (1976a). The serum dependence of baby hamster kidney cell attachment to a substratum. E x p . Cell Res. 97,265-274. Grinnell, F. (1976b). Cell spreading factor: Occurrence and specificity of action. E x p . Cell Res. 102,51-62. Grinnell, F., Milam, M., and Srere, P. A. (1973).Cyclic AMP does not affect the rate at which cells attach to a substratum. Nature (London),New Biol. 241,82-83. Grinnell, F., Tobleman, M. Q., and Hackenbrock, C. R. (1976). Initial attachment of baby hamster kidney cells to an epoxy substratum.J. Cell Biol. 70,707-713. Hamerman, D., Todaro, G. J., and Green, H. (1965). The production of hyaluronate by spontaneously established cell lines and viral transformed lines of fibroblastic origin. Biochim. Biophys. Actu 101,343-351. Hardingham, T. E., and Muir, H. (1974).Hyaluronic acid in cartilage and proteoglycan aggregation. Biochem. J . 139,565-581. Harris, A. (1973a). Location of cellular adhesions to solid substrata. Dew. Biol. 35,97-

114. Harris, A. (197311).Behavior of cultured cells on substrata of variable adhesiveness. E x p . Cell Res. 77, 285-297. Hausman, R. E., and Moscona, A. A. (1975).Purification and characterization of the retina-specific cell-aggregating factor. Proc. Natl. Acad. Sci. U.S.A. 72, 916-920. Hausman, R. E., and Moscona, A. A. (1976). Isolation of retina-specific c e h g g r e g a t i n g factor from membranes of embryonic neural retina tissue. Proc. Natl. Acad. Sci. U.S.A. 73,3594-3598. Heinegard, D., and Hascall, V. C. (1974). Aggregation of cartilage proteoglycans. 111. Characteristics of the proteins isolated from trypsin digests of aggregates. J . Biol. Chem. 249,4250-4256. Helentjaris, T. G . , Lombardi, P. S., and Glasgow, L. A. (1976). Effect of cytochalasin B on the adhesion of mouse peritoneal macr0phages.J. Cell Biol. 69,407-414. Henkart, P . , Humphreys, S., and Humphreys, T. (1973). Characterization of sponge aggregation factor: A unique proteoglycan complex. Biochemistry 12, 3045-

3050. Hirschberg, C. B., Goodman, S. R., and Green, C. (1976). Sialic acid uptake by fibroblasts. Biochemistry 15,3591-3599. Hudgin, R. L., and Ashwell, G. (1974).Studies on the role of glycosyltransferases in the hepatic binding of asialoglycoproteins. J . Biol. Chem. 249, 7369-7372. Hudgin, R. L., Pricer, W. E., Ashwell, G., Stockert, R. J., and Morell, A. G. (1974).The isolation and properties of a rabbit liver binding protein specific for asialoglycoproteins. J . Biol. Chem. 249,5536-5543. Humphreys, T. (1970).Species specific aggregation of dissociated sponge cells. Nature (London)228,685-686. Humphreys, T. (1975).Biochemical studies on cell association and its relevance to neoplasia. In “Cellular Membranes and Tumor Cell Behavior,” pp. 173-192. Williams (Ir Wilkins, Baltimore, Maryland. Hunt, R. C., and Brown, J. C. (1975). Identification of a high molecular weight transmembrane protein in mouse L ce1ls.J. Mol. Biol. 97,413-422. Hynes, R. 0. (1973).Alterations of cell surface proteins by viral transformation and by proteolysis. Proc. Natl. Acad. Sci. U.S.A. 70,3170-3174. Hynes, R. 0. (1976). Cell surface proteins and malignant transformation. Biochim. Biophys. Actu 458,73-107. Ivanova, 0. Y., Margolis, L. B., Vasiliev, J. M., and Gelfand, I. M. (1976). Effect of Colcemid on the spreading of fibroblasts in culture. E x p . Cell Res. 101,207-219.

388

LLOYD A. CULP

Jamieson, G. A., Urban, C. L., and Barber, A. J. (1971).Enzymatic basis for p1atelet:collagen adhesion as the primary step in haemostasis. Nature (London),New Biol. 234, 5-7. Johnson, C., and Pastan, I. (1972). Cyclic AMP increases the adhesion of fibroblasts to substratum. Nature (London),New Biol. 236,247-249. Juliano, R. L., and Cagalang, E. (1977).The adhesion of Chinese hamster cells. I. Effects of temperature, metabolic inhibitors and proteolytic dissection of cell surface macromolecules. J . Cell Physiol. 92,209-220. Kawasaki, T., and Ashwell, G. (1976). Chemical and physical properties of an hepatic membrane protein that specifically binds asialoglycoproteins. J . Biol. Chem. 251, 1296-1302. Kemp, R. B. (1970).The effect of neuraminidase on the aggregation of cells dissociated from embryonic chick muscle tissue. J . Cell Sci. 6,751-766. Kemp, R. B., Jones, B. M., and Groschel-Stewart, U. (1971). Aggregative behaviour of embryonic chick cells in the presence of antibodies directed against actomy0sins.J. Cell Sci. 9, 103-122. Kemp, R. B., Jones, B. M., and Groschel-Stewart, U. (1973). Abolition by myosin and heavy meromyosin ofthe inhibitory effect of smooth-muscle actomyosin antibodies on cell aggregation in oitro. J . Cell Sci. 12, 631-639. Keski-Oja, J., Mosher, D. F., and Vaheri, A. (1976). Cross-linking of a major fibroblast surface-association gl ycoprotein (fihronectin) catalyzed by blood coagulation factor XIII. Cell 9,29-35. Khan, T., and Overton, J. (1969). Staining of intercellular material in reaggregating chick liver and cartilage cells.]. Exp. Zool. 171, 161-174. Khan, T. A., and Overton, J. (1970). Lanthanum staining of developing chick cartilage and reaggregating cartilage cells. J. Cell B i d . 44,433-438. Klehe, R. J. (1974). Isolation of a collagen-dependent cell attachment factor. Nature (London)250,248-251. Klehe, R. J., Rosenherger, P. G., Naylor, S. L., Burns, R. L., Novak, R., and Kleinman, H. (1977). Cell attachment to collagen: Isolation of a cell attachment mutant. Exp. Cell Res. 104, 119-125. Kolodny, G. M. (1972).Effect of various inhibitors on readhesion of trypsinized cells in culture. Exp. Cell Res. 70, 196-202. Kondo, K. (1974). Demonstration of a reaggregation inhibitor in sea urchin embryos. E x p . Cell Res. 86, 178-181. Koyama, H., Tomida, M., and Ono, T. (1976).The enhanced production of byaluronic acid by cultured rat fibroblast cells treated with cyclic AMP and its dibutyryl derivative. J . Cell Physiol. 87, 189-198. Kraemer, P. M. (1971).Complex carbohydrates of animal cells: Biochemistry and physiology of the cell periphery. Zn “Biomembranes” (L. A. Manson, ed.) Vol. 1, pp. 67190. Plenum, New York. Kraemer, P. M. (1976). Association of cellular adherence with increased synthesis of hyaluronic acid. Abstr. Meet. Znt. Cong. Cell Biol., Boston, Mass. Kuhns, W. J., Weinbaum,’G., Turner, R., and Burger, M. M. (1974).Sponge aggregation: A model for studies on cell-cell interactions. Annals N . Y . Acad. Sci. 234,58-74. Kuroda, Y. (1974a).Effects of hexosamines and their acetyl derivatives on aggregation of rat hepatoma cells in rotation culture. Cancer Res. 34,403-409. Kuroda, Y. (1974b). Differential inhibition by fucoses of aggregation of rat hepatoma cells in rotation-mediated cell cultures. J. Natl. Cancer Znst. 52, 161-166. Kuroda, Y. ( 1 9 7 4 ~ )Inhibitory . effect of dextran sulfates on aggregation of embryonic quail liver cells in culture. Exp. Cell Res. 84,351-356.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

389

LaMont, J. T., Perrotto, J. L., Weiser, M. M., and Isselbacher, K. J. (1974). Cell surface galactosyltransferase and lectin agglutination of thymus and spleen lymphocytes. Proc. Na t l . Acad. Sci. U.S.A. 71,3726-3730. Lembach, K. J . (1976). Enhanced synthesis and extracellular accumulation of hyaluronic acid during stimulation of quiescent human fibroblasts by mouse epidermal growth fact0r.J. Cell Physiol. 89, 277-288. Lennarz, W. J. (1975).Lipid-linked sugars in glycoprotein synthesis. Science 188,986991. Letourneau, P. C. (1975).Cell-to-substratum adhesion and guidance of axonal elongation. Deu. B i d . 44,92-101. Lilien, J. E. (1968). Specific enhancement of cell aggregation in uitro. Deu. Biol. 17, 657-678. Lilien, J. E. (1969). Toward a molecular explanation for specific cell adhesion. Curr. Top. Deu. Biol. 4, 169-195. Lilien, J. E., and Moscona, A. A. (1967).Cell aggregation: Its enhancement by a supernatant from cultures of homologous cells. Science 157, 70-72. Linsenmayer, T. F., Toole, B. P., and Gross, J. (1976). Cellular adhesion to collagenous matrices. Abstr. Meet. Int. Congr. Cell Biol., Boston, Mass. Lloyd, C . W., and Cook, G. M. W. (1974).On the mechanism of the increased aggregation by netiraminidase of 16C malignant rat dermal fibroblasts in uitro. J. Cell Sci. 15,575-590. Lloyd, C. W., and Kemp, R. B. (1971). A biochemical investigation into the inhibitory effects of glucosamine and N-acetylglucosamine on the aggregation in uitro of embryonic chick muscle cells.]. Cell Sci. 9,85-101. Loomis, W. F. (1972).Role ofthe surface sheath in the control of morphogenesis in Dictyostelium discoideum. Nature (London), New Biol. 240,6-9. McClay, D. R. (1974).Cell aggregation: Properties of cell surface factors from five species of sp0nge.J. Exp. Zool. 188,89-102. McClay, D. R., and Baker, S. R. (1975).A kinetic study of embryonic cell adhesion. Deu. Biol. 43, 109-122. McClay, D. R., and Hausman, R. E. (1975).Specificity of cell adhesion: Differences between normal and hybrid sea urchin cells. Deu. Biol. 47,454-460. McClay, D. R., and Moscona, A. A. (1974).Purification of the specific cell-aggregating factor from embryonic neural retina cells. E x p . Cell Res. 87,438-443. McGuire, E . J., and Burdick, C. L. (1976).Intercellular adhesive selectivity. I. An improved assay for the measurement of embryonic chick intercellular adhesion. J . Cell B i d . 6 8 , 8 0 4 9 . McKeehan, W. L., and Ham, R. G. (1976).Stimulatip of clonal growth of normal fibroblasts with substrata coated with basic polymers. J. Cell Biol. 71, 727-734. McLean R. J., and Bosmann, H. B. (1975).Cell-cell interactions: Enhancement of glycosyl transferase ectoenzyme systems during Chlamydomonas gametic contact. Proc. N a t l . Acud. Sci. U.S.A. 72, 310-313. McNutt, N . S., and Weinstein, R. S. (1973).Membrane ultrastructure at mammalian intercellular junctions. Prog. Biophys. Mol. Biol. 26,45-101. McNutt, N. S., Culp, L. A., and Black, P. H. (1971). Contact-inhibited revertant cell lines isolated from SV40-transformed cells. 11. Ultrastructural stL1dy.J. Cell Biol. 50, 691-708. McNutt, N. S., Culp, L. A., and Black, P. H. (1973). Contact-inhibited revertant cell lines isolated from SV4O-transformed cells. IV. Microfilament distribution and cell shape in untransformed, transformed, and revertant BALB/c 3T3 ce1ls.J. Cell B i d . 56,412-428.

390

LLOYD A. CULP

McQuiddy, P., and Lilien, J. (1971).Sialic acid and cell aggregation.J . Cell Sci. 9,823833. Mallucci, L., and Wells, V. (1976).Determination of cell shape by a cell surface component. Nature (London)262,138-141. Mapstone, T. B., and Culp, L. A. (1976).Early events during substrate adhesion of normal and virus-transformed mouse fibroblasts. /. Cell Sci. 20,479-495. Marchesi, V. T., and Andrews, E. P. (1971). Glycoproteins: Isolation from cell membranes with lithium diiodosalicylate. Science 174, 1247-1248. Martin, G. R., and Rubin, H. (1974).Effects of cell adhesion to the substratum on the growth of chick embryo fibroblasts. E x p . Cell Res. 85,319-333. Martz, E., Phillips, H. M., and Steinberg, M. S. (1974).Contact inhibition of overlapping and differential cell adhesion: A suficient model for the control of certain cell culture morphologies. J . Cell Sci. 16,401-419. Maslow, D.E., and Mayhew, E. (1972).Cytochalasin B prevents specific sorting of reaggregating embryonic cells. Science 177,281-282. Maslow, D. E., and Mayhew, E. (1974). Histotypic cell aggregation in the presence of cytochalasin B. J . Cell Sci. 16,651-663. Meier, S., and Hay, E. D. (1975).Stimulation of corneal differentiation by interaction between c*ell surface and extracellular matrix. I. Morphometric analysis of transfilter “induction.” J . Cell Biol. 66,275-291. Menashi, S., Harwood, R., and Grant, M. E. (1976).Native collagen is not a substrate for the collagen glucosyltransferase of platelets. Nature (London) 264,670-672. Merrell, R., and Glaser, L. (1973). Specific recognition of plasma membranes by embryonic cells. Proc. Natl. Acad. Sci. U.S.A. 70,2794-2798. Merrell, R., Gottlieb, D. I., and Glaser, L. (1975).Embryonal cell surface recognition: Extraction of an active plasma membrane component. J . Biol. Chem. 250, 56555659. Meyer, J. T., and Oppenheimer, S. B. (1976).The multicomponent nature of teratoma cell adhesion factor. E x p . Cell Res. 102,359-364. Milam, M., Grinnell, F., and Srere, P. A. (1973).Effect of centrifugation on cell adhesion. Nature (London),New Biol. 244,83-84. Moore, E. G. (1976).Cell to substratum adhesion-promoting activity released by normal and virus-transformed cells in culture. J . Cell Biol. 70, 634-647. Mooseker, M. S., and Tilney, L. G. (1975).Organization of an actin filament-membrane complex: Filament polarity and membrane attachment in the microvilli of intestinal epithelial cells. J . Cell Biol. 67,725-743. Mori, Y., Akedo, H., and Tanigaki, Y. (1973). Changes induced by concanavalin A in morphological and adhesive properties of suspension cells. E x p . Cell Res. 78,360366. Moscatelli, D., and Rubin, H. (1975).Increased hyaluronic acid production on stimulation of DNA synthesis in chick embryo fibroblasts. Nature (London) 254, 65-66. Moscona, A. A. (1968). Cell aggregation: Properties of specific cell-ligands and their role in the formation of multicellular systems. Deu. Biol. 18,250-277. Miiller, W. E. G., and Zahn, R. K. (1973).Purification and characterization of a speciesspecific aggregation factor in sponges. Exp. Cell Res. 80, 95-104. Muller, W. E. G., Muller, I., Kurelec, B., and Zahn, R. K. (1976).Species-specific aggregation factor in sponges. IV. Inactivation of the aggregation factor by mucoid cells from another species. E x p . Cell Res. 98,31-40. Newell, P. C., Longlands, M., and Sussman, M. (1971).Control of enzyme synthesis by cellular interaction during development of the cellular slime mold Dictyostelium discoideum. J . Mol. Biol. 58, 541-554.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

39 1

Nicolson, G. L. (1974). The interactions of lectins with animal cell surfaces. Znt. Rev. Cytol. 39,89-190. Nicolson, G . L. (1976a). Transmembrane control of the receptors on normal and tumor cells. I. Cytoplasmic influence over cell surface components. Biochim. Biophys. Acta 457,57-108. Nicolson, G. L. (1976b).Transmembrane control of the receptors on normal and tumor cells. 11. Surface changes associated with transformation and malignancy. Biochim. Biophys. Acta. 458, 1-72. Noonan, K. D., and Burger, M. M. (1973).Binding of [3Hlconcanavalin A to normal and transformed cells. J. Biol. Chem. 248,4286-4292. Noonan, K. D., Lindberg, D. S., and McClure, J. A. (1976). Isolation and preliminary characterization of that part of the plasma membrane which is apposed to the substratum. In “Membranes and Neoplasia: New Approaches and Strategies” (V. I. Marchesi, ed.), Progress in Clinical and Biological Research, Vol. 9, pp. 245-260. Alan R. Liss, New York. O’Farrell, P. H. (1975). High resolution two-dimensional electrophoresis of proteins. J. Biol. Chem. 250,4007-4021. Olden, K., Willingham, M., and Pastan, I. (1976).Cell surface myosin in cultured fibroblasts. Cell 8,383-390. Oppenheimer, S. B. (1973). Utilization of L-glutamine in intercellular adhesion: Ascites tumor and embryonic cells. E x p . Cell Res. 77, 175-182. Oppenheimer, S. B. (1975). Functional involvement of specific carbohydrate in teratoma cell adhesion factor. Exp. Cell Res. 92, 122-126. Oppenheimer, S. B., and Humphreys, T. (1971). Isolation of specific macromolecules required for adhesion of mouse tumour cells. Nature (London)232, 125-127. Oppenheimer, S. B., Edidin, M., Orr, C. W., and Rosman, S. (1969). An L-glutamine requirement for intercellular adhesion. Proc. Natl. Acad. Sci. U.S.A. 63, 13951402. Oseroff, A. R., Robbins, P. W., and Burger, M. M. (1973). The cell surface membrane: Biochemical aspects and biophysical probes. Annu. Rev. Biochem. 42,647-682. Overton, J. (1969). A fibrillar intercellular material between reaggregating embryonic chick cells. /. Cell Biol. 40, 136-159. Overton, J. (1975). Experiments with junctions of the adhaerens type. Curr. Top. Dev. Biol. 10, 1-34. Painter, R. G., and White, A. (1976). Effect of concanavalin A on expression of cell surface sialyltransferase activity of mouse thymocytes. Proc. Natl. Acad. Sci. U.S.A.73, 837-844. Painter, R. G., Sheetz, M., and Singer, S. J. (1975). Detection and ultrastructural localization of human smooth muscle myosin-like molecules in human non-muscle cells by specific antibodies. Proc. Natl. Acad. Sci. U.S.A. 72, 1359-1363. Patt, L. M., and Grimes, W. J. (1974). Cell surface glycolipid and glycoprotein glycosyltransferases of normal and transformed cells. J . B i d . Chem. 249,4157-4165. Patt, L. M., and Grimes, W. J. (1975). Ectoglycosyltransferase activity in suspensions and monolayers of cultured fibroblasts. Biochem. Biophys. Res. Commun. 67,483490. Patt, L. M., Endres, R. O., Lucas, D. O., and Grimes, W. J. (1976). Ectogalactosyltransferase studies in fibroblasts and concanavalin A-stimulated lymphocytes. J. Cell Biol. 68,799-802. Pearlstein, E. (1976). Plasma membrane glycoprotein which mediates adhesion of fibroblasts to collagen. Nature (London) 262,497-500. Pearlstein, E., Hynes, R. O., Franks, L. M., and Hemmings, V. J. (1976). Surface pro-

392

LLOYD A. CULP

teins and fibrinolytic activity of cultured mammalian cells. Cancer Res. 36, 14751480. Pegrum, S. M., and Maroudas, N. G. (1975).Early events in fibroblast adhesion to glass. Exp. Cell Res. 96,416-422. Perdue, J. F. (1973).The distribution, ultrastructure, and chemistry of microfilaments in cultured chick embryo fibroblasts. J . Cell Biol. 58,265-283. Pessac, B., and Defendi, V. (1972). Cell aggregation: Role of‘ acid mucopolysaccharides. Science 175,898-900. Pinto da Silva, P., Martinez-Palomo, A., and Gonzalez-Roldes, A. (1975). Membrane structure and surface coat of‘Entamoehuhistolyticu. J . Cell Biol. 64,538-550. Podolsky, D. K., Weiser, M. M., LaMont, J. T., and Isselbacher, K. J. (1974). Galactosyltransferase and concanavalin A agglutination of cells. Proc. Natl. Acud. Sci. U.S.A. 71,904-908. Pollack, R., Osborn, M., and Weber, K. (1975).Patterns of organization of actin and myosin in normal and transformed cultured cells. Proc. Nutl. Acud. Sci. U.S.A.72,994998. Porter, C. W., and Bernacki, R. J. (1975).Ultrastructural evidence for ectoglycoslytransferase systems. Nuture (London) 256,648-650. Poste, G., Greenham, L. W., Mallucci, L., Reeve, P., and Alexander, D. J. (1973). The study of cellular “microexudates” by ellipsometry and their relationship to the cell coat. E x p . Cell Res. 78,303-313. Pouyssegur, J. M., and Pastan, I. (1976).Mutants of BALB/c 3T3 fibroblasts defective in adhesiveness to substratum: Evidence for alteration in cell surface proteins. Proc. Natl. Acud. Sci. U.S.A. 73,544-548. Pouyssegur, J. M., and Pastan, I. (1977). Mutants of mouse fibroblasts altered in the synthesis of cell surface glycoproteins.]. Biol. Chem. 252, 1639-1646. Reaven, E. P., and Axline, S. G. (1973). Subplasmalemmal microfilaments and microtubules in resting and phagocytizing cultivated macr0phages.J. Cell Biol. 59, 12-27. Reitherman, R. W., Rosen, S. D., Frazier, W. A., and Barondes, S. H. (1975).Cell surface species-specific high affinity receptors for discoidin: Developmental regulation in Dictyostelium discoideum. Proc. Natl. Acad. Sci. U.S.A.72,3541-3545. Revel, J. P. (1974).Contacts and junctions between cells. Symp. Soc. E x p . Biol. 28,447461. Revel, J. P., and Wolken, K. (1973). Electron microscope investigations of the underside of cells in culture. E x p . Cell Res. 78, 1-14. Revel, J. P., Hoch, P., and Ho, D. (1974). Adhesion of culture cells to their substratum. E x p . Cell Res. 84,207-218. Roblin, R., Albert, S. O., Gelb, N. A., and Black, P. H. (1975a). Cell surface changes correlated with density-dependent growth inhibition. Glycosaminoglycan metabolism in 3T3, SV3T3, and Con A selected revertant cells. Biochemistry 14,347-357. Roblin, R. O., Chou, I. N., and Black, P. H. (1975b). Proteolytic enzymes, cell surface changes, and viral transformation. Adu. Cancer Res. 22,203-260. Roseman, S . (1970). The synthesis of complex carbohydrates by multiglycosyltransferase systems and their potential function in intercellular adhesion. Chem. Phys. Lipids 5,270-297. Rosen, J. J., and Culp, L. A. (1977). Morphology and cellular origins of substrate-attached material from mouse fibroblasts. E x p . Cell Res. 107, 139-149. Rosen, S . D., Kafka, J. A., Simpson, D. L., and Barondes, S. (1973). Developmentally regulated, carbohydrate-binding protein in Dictyostelium discoideurn. Proc. Natl. Acad. Sci. U.S.A. 70,2554-2557.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

393

Rosen, S. D., Simpson, D. L., Rose, J. E., and Barondes, S. H. (1974). Carbohydratebinding protein from Polysphondylium pullidum implicated in intercellular adhesion. Nature (London)252, 128. Rosen, S. D., Haywood, P. L., and Barondes, S. H. (1976). Inhibition of intercellular adhesion in a cellular slime mold by univalent antibody against a cell-surface lectin. Nuture (Lottdon)263,425-427. Rosenberg, M. D. (1960).Microexudates from cells grown in tissue culture. Biophys. J. 1, 137-159. Rossomando, E. F., Steffek, A. J., Mujwid, D. K., and Alexander, S. (1974). Scanning electron microscopic observations on cell surface changes during aggregation of Dictyostelium discoideum. E x p . Cell Res. 85,73-78. Roth, S. (1968).Studies on intercellular adhesive selectivity. Deu. Biol. 18,602-631. Roth, S . , and White, D. (1972). Intercellular contact and cell surface galactosyltransferase activity. Proc. Nutl. Acad. Sci. U.S.A.69,485489. Roth, S., McGuire, E. J., and Roseman, S. (1971). Evidence for cell-surface glycosyltransferases: Their potential role in cellular recognition. J . Cell Biol. 51,536-547. Roth, S., Patteson, A., and White, D. (1974). Surface glycosyltransferases on cultured mouse fibroblasts./. Suprumol. Struct. 2, 1-6. Rutishauser, U., Thiery, J., Brackenhury, R., Sela, B., Edelman, G. M. (1976). Mechanisms of adhesion among cells from neural tissues of chick embryo. Proc. Nutl. Acad. Sci. U.S.A. 73,577-581. Sato, C., and Takasawa-Nishizawa, K. (1974).Change in cell-to-dish contact of cultured mammalian cells induced by concanavalin A. E x p . Cell Res. 89, 121-126. Schachter, H., Moscarello, M., and Sturgess, J. (1978). Structure and biosynthesis of membrane glycoproteins. C u n . Top. Membr. Transp. 11, 15-105. Sheffield, J. B., and Moscona, A. A. (1969). Early stages in the reaggregation of embryonic chick neural retina cells. E x p . Cell Res. 57,462-466. Shields, R., and Pollock, K. (1974). The adhesion of BHK and PyBHK cells to the substratum. Cell 3,31-38. Shizuta, Y., Davies, P. J. A., Olden, K., and Pastan, I. (1976). Diminished content of plasma membrane-associated myosin in transformed fibroblasts. Nature (London) 261,414-415. Shur, B. D., and Roth, S. (1975). Cell surface glycosyltransferases. Biochim. Biophys. Actu 415,473-512. Simpson, D. L., Rosen, S. D., and Barondes, S. H. (1974).Discoidin, a developmentally regulated carbohydrate-binding protein from Dictyostelium discoideum: Purifica\ tion and characterization. Biochemistry 13,3487-3493. Simpson, D. L., Rosen, S. D., and Barondes, S. H. (1975). Pallidin: Purification and characterization of a carbohydrate-binding protein from Polysphondylium pallidum implicated in intercellular adhesion. Biochim. Biophys. Acta 412, 109-119. Siu, C., Lerner, R. A., Ma, G., Firtel, R. A., and Loomis, W. F. (1976).Developmentally regulated proteins of the plasma membrane ofDictyosteZium discoideum: The carbohydrate-binding protein. J . Mol. B i d . 100, 157-178. Spooner, B. S., and Conrad, G. W. (1975). The role of extracellular materials i n cell movement. I. Inhibition of mucopolysaccharide synthesis does not stop ruffling membrane activity or cell movement.]. Cell Biol. 65,286-297. Steinberg, M. S., and Gepner, I. A. (1973).Are concanavalin A receptor sites mediators of cell-cell adhesion? Nature (London),New Biol. 241,249-251. Stoker, M. (1975).The effects of topoinhibition and cytochalasin B on metabolic cooperation. Cell 6,253-257.

394

LLOYD A. CULP

Sussman, M., and Boschwitz, C. (1975).Adhesive properties of cell ghosts derived from Dictyostelium discoideum. Dev. Biol. 44,362-368. Takeichi, M . (1971).Changes in the properties of cell-substrate adhesion during cultivation of chicken fibroblasts in vitro in a serum-free medium. E x p . Cell Res. 68, 88-96.

Takeichi, M., and Okada, T. S. (1972). Roles of magnesium and calcium ions in cell-tosubstrate adhesion. E x p . Cell Res. 74, 51-60. Taylor, A. C. (1961).Attachment and spreading of cells in culture. E x p . Cell Res., S u p p l . 8, 154-173. Teng, N. N. H., and Chen, L. B. (1976). Thrombin-sensitive surface protein of cultured chick embryo cells. Nature (London) 259,578-580. Terry, A. H., and Culp, L. A. (1974). Substrate-attached glycoproteins from normal and virus-transformed cells. Biochemistry 13,414425. Tomida, M., Koyama, H., and Ono, T. (1975). Induction of hyaluronic acid synthetase activity in rat fibroblasts by medium change of confluent cu1tures.J. Cell Physiol. 86, 121-130. Trelstad, R. L., Hayashi, K., and Toole, B. P. (1974).Epithelial collagens and glycosaminoglycans in the embryonic cornea: Marcrornolecular order and rnorphogenesis in the basement membrane.J. Cell Biol. 62,815-830. Trinkaus, J. B., Betchaku, T., and Krulikowski, L. S. (1971).Local inhibition of ruffling during contact inhibition of cell movement. E x p . Cell Res. 64,291-300. Turner, R. S., and Burger, M. M. (1973). Involvement of a carbohydrate group in the active site for surface guided reassociation of animal cells. Nature (London) 244, 509-5 10. Umbreit, J., and Roseman, S . (1975).A requirement for reversible binding between aggregating embryonic cells before stable adhesion J . Biol. Chem. 250, 9360-9368. Underhill, C. B., and Keller, J. M. (1976).Density-dependent changes in the amount of sulfated glycosaminoglycans associated with mouse 3T3 cells. J . Cell Physiol. 89, 53-64. Unhjem, O., and Prydz, H. (1973). Interference of serum and serum components with attachment of HeLa 71 cells to the surface of plastic culture vessels. E x p . Cell Res. 83,418-421. Vaheri, A., and Ruoslahti, E. (1975). Fibroblast surface antigen produced but not retained by virus-transformed human cells.]. Exp. Med. 142, 530-535. Vasiliev, J. M., Gelfand, I. M., Domnina, L. V., Zacharova, 0. S., and Ljubimov, A. V. (1975). Contact inhibition of phagocytosis in epithelial sheets: Alterations of cell surface properties induced by cell-cell contacts. Proc. Natl. Acad. Sci. U S A . 72, 719-722. Verbert, A,, Cacan, R., and Montreuil, J. (1976). Ectogalactosyltransferase: Presence of enzyme and acceptors on the rat lymphocyte cell surface. Eur.J. Biochem. 70,4953. Vessey, A. R.,and Culp, L. A. (1978).Contact-inhibited revertant cell lines isolated from simian virus 40-transformed cells VIII. Membrane and protein and glycoprotein composition. Virology, in press. Vicker, M. G . ,and Edwards, J. G . (1972).The effect of neuraminidase on the aggregation of BHK21 cells and BHK21 cells transformed by polyoma virus.]. Cell Sci. 10, 759-768. Waddell, S. W., Robson, R. T., and Edwards, J. G . (1974). Colchicine and vinblastine inhibit fibroblast aggregation. Nature (London) 248,239-241. Walther, B. T., Ohman, R., and Roseman, S. (1973).A quantitative assay for intercellular adhesion. Proc. Natl. Acad. Sci. U.S.A.70, 1569-1573.

BIOCHEMICAL DETERMINANTS OF CELL ADHESION

395

Warren, A. J., Warren, W. D., and Cox, E. C. (1976).Genetic and morphological study of aggregation in the cellular slime mold Polysphondylium violoceum. Genetics 83, 25-47. Webb, G . C., and Roth, S. (1974). Cell contact dependence of surface galactosyltransferase activity as a functjon of the cell cyc1e.J. Cell Biol. 63,796-805. Weeks, G . (1973). Agglutination of growing and differentiating cells of Dictyostelium discoicleum by concanavalin A. E x p . Cell Res. 76,467-470. Weinlxmm, G . , and Burger, M. M. (1973).Two component system for surface guided reassociation of animal cells. Nature (London)244,510-512. Weiser, M. M. (1973).Intestinal epithelial cell surface membrane glycoprotein synthesis. 11. Glycosyltransferases and endogenous acceptors of the undifferentiated cell surface membrane../. Biol. Chem. 248,2542-2548. Weiss, L. (1961). Studies on cellular adhesion in tissue culture. IV. The alteration of substrate by cell surfaces. E x p . Cell Res. 25,504-517. Weiss, L. (1970). Cell contact phenomena. In Vitro 5,48-78. Weiss, L. (1972). Studies on cellular adhesion in tissue culture. XII. Some effects of cytochalasins and colchicine. Exp. Cell Res. 74,21-26. Weiss, L., and Lachmann, P. J. (1964). The origin of an antigenic zone surrounding HeLa cells cultured on glass. Exp. Cell Res. 36,86-91. Weiss, L., and Maslow, D. E. (1972). Some effects of trypsin dissociation on the inhibition of reaggregation among embryonic chicken neural retina cells by cycloheximide. Dev. B i d . 29,482-485. Weiss, L., Poste, G., Mackearnin, A,, and Willett, K. (1975).Growth of mammalian cells on substrates coated with cellular microexudates. I. Effect on cell growth at low population densities. I . Cell B i d . 64, 135-145. Wickus, G. G., and Robbins, P. W. (1973). Plasma membrane proteins of normal and Rous sarcoma virus-transformed chick embryo fibroblasts. Nature (London),New Biol. 245, 65-67. Willingham, M. C., and Pastan, I. (1975).Cyclic AMP and cell morphology in cultured fibroblasts: Effects on cell shape, microfilament and microtubule distribution, and orientation to substratum. 1.Cell Biol. 67, 146-159. Willingham, M . C., Carchman, R. A., and Pastan, I. (1973).A mutant of 3T3 cells with cyclic AMP metabolism sensitive to temperature change. Proc. Natl. Acad. Sci. U.S.A. 70,2906-2910. Willingham, M. C., Ostlund, R. E., and Pastan, I. (1974).Myosin is a component of the cell surface of cultured cells. Proc. Natl. Acad. Sci. U.S.A. 71,4144-4148. Winkelhake, J. L., and Nicolson, G . L. (1976). Determination of adhesive properties of variant metastatic melanoma cells to BALB/3T3 cells and their virus-transformed derivatives by a monolayer attachment assay. J . Natl. Cancer Znst. 56,285-291. Winter, W. T., Smith, P. J. C., and Arnott, S. (1975).Hyaluronic acid: Structure of a fully extended 3-fold helical sodium salt and comparison with the less extended 4-fold helical forn1s.J. Mol. B i d . 99,219-235. Wisnieski, B. J., Parkes, J. C . , Huang, Y. O., and Fox, C. F. (1974).Physical and physiological evidence for two phase transitions in cytoplasmic membranes of animal cells. Proc. Nutl. Acad. Sci. U.S.A. 71,4381-4385. Witkowski, J . A,, and Brighton, W. D. (1971). Stages of spreading ofhuman diploid cells on glass surfaces. E x p . Cell Res. 68,372-380. Witkowski, J. A,, and Brighton, W. D. (1972). Influence of serum on attachment of tissue cells to glass surfaces. E x p . Cell Res. 7 0 , 4 1 4 8 . Yamada, K. M., and Weston, J. A. (1974).Isolation of a major cell surface glycoprotein from fibroblasts. Proc. Natl. Acad. Sci. U.S.A. 71,3492-3496.

396

LLOYD A. CULP

Yamada, K. M., and Weston, J. A. (1975). The synthesis, turnover, and artificial restoration of a major cell surface glycoprotein. Cell 5 , 7 5 4 1 . Yamada, K. M., Yamada, S. S., and Pastan, I. (1975).The major cell surface glycoprotein of chick emhryo fibroblasts is an agglutinin. Proc. Nutl. Acad. Sci. U S A . 72,3158-

3162. Yamada, K. M., Yamada, S. S., and Pastan, I. (1976).Cell surface protein partially restores morphology, adhesiveness, and contact inhibition of' movement to transformed fibroblasts. Proc. Natl. Acud. Sci. U.S.A. 73, 1217-1221. Yaoi, Y., and Kanaseki, T. (1972). Role of microexudate carpet in cell division. Nature

(London) 237,283-285. Yogeeswaran, G . , Laine, R. A., and Hakomori, S. (1974). Mechanism of cell contact dependent glycolipid synthesis: Further studies with glycolipid-glass complex. Bio-

chem. Biophys. Res. Commun. 59,591-599.

CURRENT TOPICS IN MEMBRANES AND TRANSPORT, VOLUME

11

Proteolytic Modification of Cell Surface Macromolecules: Mode of Action in Stimulating Cell Growth KENNETH D. NOONAN* Department of Biochemistry and Molecular Biology

JHM Health Center University of Florida Gainewille, Florida

I. Introduction . . . . . . . . . . . . . . . . . . 11. The Erythrocyte Membrane . . . . . . . . . . . . . 111. Stimulation of Cell Division in a Resting Lymphocyte Population . . A. Proteases , . . . . . . . . . , . . . . . . . B. Sodium Periodate. . . . . . . . . . . . . . . . C. Lectins. . . . . . . . . . . . . . . . . . IV. Protease Induction of Cell Division in Fibroblasts . . . . . . . A. Overgrowth Stimulating Factor . . . . . . . . . . . B. Chick Embryo Fibroblasts . , . . . . . . , . , . . C. Mouse Embryo Fibroblasts . . . . . . . . . . . . . D. Other Cell Lines . . . . . . . . . . . . , . . . E. Conclusions . . . . . . . . . . . . . . . . . V. Effects of Proteases on Fibroblast Surface Structure . . . . . . A. Change in Lectin Agglutinability . . . . . . . . . , . B. Protease-Mediated Modification of Membrane Peptides and , . . Glycopeptides . . . . . . . . . . . . . C. Proteolytic Modification of the Macroarchitecture of the Cell Surface . . , . . . . . . . . . . . . . . D. Contact Inhibition of Movement . . . . . . . . . . . VI. Protease-Induced Transmembrane Events . . . . . . . . . A. Protease Modification of Intracellular CAMP Concentrations . . . B. Protease-Mediated Modification of Inbacellular Actinlike Cables . VII. Limited Autolysis as a Mechanism for Inducing Cell Division . . A. Evidence for Surface-Localized Proteases . . . . . . . . B. Plasminogen Activator . . . . . . . . . . . . . VIII. Role Media Components Play in Protease-Stimulated Cell Division . . Role of Serum Factors in Protease-Stimulated Cell Growth . . . . IX. Summary , . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . Note Added in Proof . . . . . . . , . . . . . . .

.

.

.

.

.

.

. .

398 399 404 404 408 411 412 412 413 416 419 420 421 421 424 429 430 432 433 435 437 440 441 444 446 450 452 461

* Some ofthe work reported here was supported by grant DRG-1241 from the Damon Runyon Foundation. Copyright @ 1978 by Academic Press, Inc.

All rights of' reproduction in any form reserved. ISBN 0-12-151311-5

397

398

KENNETH D. NOONAN

1.

INTRODUCTION

Over the last decade proteases have proven to be very valuable tools which both the membrane biochemist and the cell biologist have used to investigate membrane-mediated phenomena. On the one hand, the membrane biochemist has used a variety of proteases to investigate the topological distribution of membrane proteins and glycoproteins within the intact plasma membrane. This approach, aimed at the dissection of membrane structure, has been most valuable in studying the relatively simple erythrocyte plasma membrane. The cell biologist, on the other hand, has been intrigued by the various reports which suggest that protease-mediated modification of the cell surface of thymocytes, lymphocytes, and fibroblasts can, under certain prescribed growth conditions, stimulate cell division in a previously quiescent cell population. This interest on the part of both cell biologists and biochemists in the effect(s) of proteases on membrane structure and cell division has gained new impetus from the many reports suggesting that cells (in particular transformed cells) may well contain and secrete specific proteases which play a role in determining the relative growth state of a particular cell population. (For an extensive review of the latter concept, see Clarkson and Baserga, 1974). This chapter attempts to synthesize the different approaches and questions posed by biochemists and cell biologists in order to resolve at least partially the dilemma of whether proteases, acting at the cell surface of various cell types, can stimulate a quiescent cell population to reenter the cell cycle and eventually complete at least one round of cell division. Equally important, this chapter attempts to analyze critically whether strong evidence exists to implicate one or more membrane peptides or glycopeptides as controlling element(s) in determining whether or not a cell population will be stimulated to divide following protease treatment of the cell surface. Furthermore, this chapter reviews the evidence for and against protease-induced transmembrane control of cell division. Evidence is presented which is at least supportive of the concept that proteolytic modification of a membrane peptide or glycopeptide can initiate a series of intracellular events which culminate in cell division. In this chapter we concentrate on the effects of proteolysis on the growth state of lymphocytes, thymocytes, and fibroblasts. In order to introduce the methodology and concepts underlying the use of proteases to modify membrane structure, we very briefly review the extensive work done in the area of protease modification of the erythrocyte membrane. Introduction of the erythrocyte membrane here is for

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

3 99

the sole purpose of pointing out some of the problems cell biologists must be aware of when proteases are used as stimulants of cell division. A much more complete analysis of erythrocyte membrane structure is presented by Tanner in this volume.

II. THE ERYTHROCYTE MEMBRANE

When one is faced with the question of whether or not proteases act to stimulate cell division via a modification of membrane structure, one is always forced to ask whether the protease in question acts only on the external surface of the target cell or whether, following its initial modification of the cell surface, it is taken up by the cell and subsequently acts intracellularly to modify the target cell further. Absolute exclusion of the latter possibility is frequently very difficult to provide, and therefore this crucial problem is frequently glossed over or totally ignored. However, in one system, that of the erythrocyte membrane, enough data have been accumulated to allow one to conclude safely that proteases, added to intact cells, do not penetrate the cell but act solely on surface proteins and glycoproteins. We very briefly analyze the work done with the erythrocyte membrane in order to point out the advantages and disadvantages encountered in using proteases to modify membrane structure. It must be noted, however, that the problems associated with an analysis of proteolytic modification of the erythrocyte cell surface must be considered the most minimal of problems which will be associated with a similar analysis of lymphocyte and fibroblast membranes. This is particularly true, since the erythrocyte is unique in that even prolonged protease digestion (e.g., a 60-minute incubation with 250-1000 pg/ml trypsin) rarely causes cell lysis in more than 5% of the cells. Furthermore, since the erythrocyte lacks the capacity for phagocytosis, it is unable to internalize extracellular proteases actively. As discussed in many of the later sections, phagocytosis of exogenous proteases is a major problem in determining the role of cell surface-localized proteins and glycoproteins in controlling the growth state of fibroblasts. Therefore, although much can be learned from the erythrocyte, the key point to be derived from this section is that one must take great care to avoid overextending interpretations of proteolytic modification of complex cell surfaces. Thus this section should not be used to vindicate those of us who have used proteases to study membrane structures and their role in stimulating cell division but rather should serve as a warning, pointing to the many pitfalls awaiting the unwary bio-

400

KENNETH D. NOONAN

chemist or cell biologist who indiscriminately applies proteases to quiescent cell populations and then attributes all the observed changes in the proliferative activity of the population to proteolytic activity at the cell surface. Figure 1 is a densitometric scan of a sodium dodecyl sulfate polyacrylamide gel electrophoretic (SDS-PAGE) separation of SDSsolubilized erythrocyte membrane components, displaying the major peptides and glycopeptides of the erythrocyte membrane. The numbering system used to identify particular coomassie blue or periodic acid-Schiff (PAS) staining peaks is the system originally described by Fairbanks and his collaborators (Fairbanks et al., 1971).There is now ample evidence in the literature, based predominantly on the use of nonpenetrating reagents to label surface-localized peptides and glyco-

B

PAS. 1

I 0.4 -

B

Q a 0.2-

01, 0

.

,

20

40

60

80

MIGRATION, mm

FIG.1. Erythrocyte membrane polypeptides and glycoproteins. (A) A densitometric scan of a gel stained for protein with coomassie blue after electrophoresis of 10 ~1 of packed ghosts (40 pg protein). (B) A similar gel stained for carbohydrate with the PAS reagent. (From Steck, 1974.)

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

401

peptides covalently [e.g., lZ5I(Phillips and Morrison, 1970; Hubbard and Cohn, 1972), B3H4 (Steck and Dawson, 1974), [35S]formylmethionyl sulfone (Bretscher, 1971),and [3Hldisulfonicstilbene (C3H]DIDS) (Cabantchik and Rothutein, 1974)l that the membrane peptides and glycopeptides of the erythrocyte are arranged asymmetrically; i.e., some peptides are confined to the external surface of the cell, others are confined to the internal surface ofthe cell, while still others (specifically PAS-1 and component 3) probably traverse the plasma membrane and are thus available to both the interior and exterior of the cell (reviewed in Steck, 1974). Among the peptides and glycopeptides identified as being present on the outer surface of the erythrocyte cell surface are the major glycoproteins (PAS l and component 3) plus other minor glycopeptides (see Fig. l),acetylcholinesterase, a NADase, the ouabain-binding site, and the DIDS-binding site. If proteases are to be used as effective probes of membrane structure, then it would be expected that only those components of the reythrocyte membrane which have been unequivocally assigned to the external surface of the cell would be affected by protease treatment. As a general statement this has proved true. Band 3, PAS-1, and acetylcholinesterase are clearly affected by protease treatment of the intact erythrocyte. PAS-2, which is probably an intreconvertible form of PAS-1 (Steck, 1974), is also cleaved by trypsinization. It must be noted in this regard, however, that even prolonged treatment of intact erythrocytes with pronase or chymotrypsin did not reduce choline transport, sodium ion efflux, or anion permeability, thereby demonstrating that at least the functional site of these (external) transport peptides or glycopeptides [identified by rHlDIDS binding (Cabantchik and Rothstein, 1974)l must be either resistant to or not accessible to these proteases. These data clearly imply that at least a portion of some membrane proteins which, because of their physiological function, must be exposed to the cell’s exterior is not functionally destroyed by proteolysis. That there are peptides or glycopeptides on the external surface of the erythrocyte membrane whose physiological functions are not destroyed by limited proteolysis suggests both an advantage and a disadvantage in using proteases as probes of membrane structure-function relationships. The advantage to be derived from this fact is that some proteases apparently do not indiscriminately reduce the physiological activity of all exteriorized membrane peptides, thus at least offering the possibility of identifying the specific membrane peptides responsible for maintaining growth control. The disadvantage pointed out by this work with the

402

KENNETH D. NOONAN

erythrocyte membrane is that a specific protease, because of its substrate specificity or the arrangement of the peptide within the membrane, may fail to attack the peptide responsible for maintaining the growth state of a cell population and may thus erroneously lead to the conclusion that membrane proteins play little or no role in growth control. Thus it is important to be aware of the fact that failure to observe protease stimulation of cell growth should not be accepted as ultimate evidence that there is no controlling element in the surface membrane which plays a role in maintaining growth control. A second cautionary note to be drawn from the exhaustive work on protease modification of the erythrocyte membrane is that not all proteases attack the same surface peptide. For example, component 3 (Fig. 1)is digested in intact cells by chymotrypsin and by pronase but not by trypsin (Steck, 1974). This finding should alert investigators to the fact that in more complex cell systems one must consider whether one has used a battery of proteases in investigating the role of surface modifications in controlling cell growth, or has depended on a single protease to modify the surface structure and subsequently evaluated its stimulatory activity with regard to cell division. Another cautionary note which should be drawn from the work on erythrocytes comes from Triplett and Carraway’s (1972) clear demonstration that resealed erythrocyte ghosts (which are frequently considered impermeant to large macromolecules such as proteases, Steck, 1974) show sequestration of labeled, inactivated proteases. This finding should alert the cell biologist to monitor continually the physiological state of the cells with which he or she is working, since even slightly “damaged” cells may be much more permeable to proteases than physiologically intact cells. Clearly, unless one is working with impermeant cells, one cannot ascribe the changes in growth properties frequently observed after protease treatment solely to activity of the protease on the cell surface. As pointed out above and as is evident from Fig. 1, the main tool used by the biochemist to separate membrane peptides and glycopeptides is SDS-PAGE. This tool is of course also used to demonstrate any proteolytic modification of membrane peptides. In this regard it must be noted that Fairbanks et al. (1971) reported a variety of SDSresistant peptidases which continued to cleave membrane peptides even after the membranes were solubilized in the anionic detergent. In fact, evidence exists that SDS-denaturation of membrane peptides may well expose protease-sensitive areas of particular membrane peptides, making them more accessible to cleavage by endogenous proteases released during membrane isolation (Fairbanks et d.,1971).

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

403

Thus care must be taken (usually by adding multifunctional protease inhibitors to the SDS solution) to be sure that any peptide modification detected on SDS-PAGE in fact resulted from protease treatment of the intact cell and is not a by-product of the activity of an endogenous protease. Finally, it is now clear from the work of Steck (1972), Wallach (1972), Morrison (Mueller et al., 1976), and their collaborators that the densitometric scan of the erythrocyte membrane peptides and glycopeptides shown in Fig. l is not representative of the true complexity of the erythrocyte membrane. For example, both Steck (1974) and Mueller et al. (1976) showed that PAS staining bands 1,2, and 3 are composed of multiple glycopeptides which can be identified as discrete components either by the variability observed in galactose oxidase-NaB3H, labeling (Steck and Dawson, 1974) or by separation of the membrane components in a gel system dependent on a discontinuous buffer (Laemmli, 1970) which has a greater resolving power (Mueller et al., 1976) than the gel technique introduced by Fairbanks et al. (1971). In this same context, Steck (1972),using chemical crosslinkers, demonstrated that many membrane peptides and glycopeptides exist in oligomeric arrays rather than as discrete components. With this in mind, one could envision a situation in which limited proteolysis frees components from an oligomeric complex without drastically changing the MW of any of the components of the complex. Such relatively minor changes in surface architecture might not be detected on SDS-PAGE. Thus the investigator might again be confronted with a change in membrane structure-function relationships which could not be detected because the tools used for detection were not adequate for the task. Thus, from a careful analysis of the elegant work done on the erythrocyte membrane, the cell biologist who wishes to approach the question of protease-induced modification of the cell surface and the possible role of such a modification in the stimulation of cell division can obtain an appreciation of the complexity of the question being asked and should heed all the warnings against overinterpretation of data which can be derived from the work performed on this relatively simple membrane. In particular it is our opinion that, when one considers the relative complexity of the fibroblast plasma membrane as compared to the erythrocyte plasma membrane (see Section V), one must be acutely aware that a one-dimensional separation of fibroblast membrane peptides following limited proteolysis may not be an adequate tool with which to demonstrate the complexity of the changes the membrane has undergone.

404

KENNETH D. NOONAN

111.

STIMULATION OF CELL DIVISION IN A RESTING LYMPHOCYTE POPULATION

A. Protearer

In what must be regarded as an astute observation by a group of clinicians, Mazzei et al. (1966)reported that the topical application of an antiinflammatory agent known to contain trypsin and chymotrypsin to an inflammed epidermal area produced 2-14% lymphocyte blast transformation in six different patients. In a brief note appearing in the British medical journal Lancet these workers further reported that the addition of either crystalline trypsin or chymotrypsin to cultured human lymphocytes produced a simliar degree of blast transformation within 96 hours. These observations suggested that protease treatment of the lymphocytes might in itself act as a mitogenic stimulus leading to blast transformation. Despite this interesting report, the basic observation that proteases could be mitogenic was not actively investigated until 8 years later when Vischer (1974) demonstrated that mouse B lymphocytes could be stimulated to enter the S phase of the cell cycle following a 3-day incubation in growth medium containing 2.5 pg/ml trypsin. Interestingly (and of significance to many of the later observations to be discussed), Vischer showed that protease stimulation of B-cell division occurred in the absence of serum factors, demonstrating that B cells could be induced to reinitiate cellular DNA synthesis without involvement of the various mitogenic factors known to be present in serum (Holley, 1974). Vischer’s (1974)findings were subsequently confirmed and refined by Kaplan and Bona (1974),who demonstrated that splenic (B) but not thymic (T) lymphocytes could be stimulated to reinitiate DNA synthesis following a 3- to 5-day incubation in media containing 0.1 pg/ml trypsin. Like Vischer (1974),Kaplan and Bona (1974)found that serum was not needed to elicit the mitogenic response. Furthermore, Kaplan and Bona (1974) demonstrated that trypsin-induced reinitiation of DNA synthesis could be prevented when the protease was preincubated with soybean trypsin inhibitor or heat-denatured. These latter results clearly suggested that protease modification of the cell and not an in vitro immune response to trypsin was responsible for the observed mitogenic stimulation. Kaplan and Bona (1974) also showed that pronase, a protease less specific than trypsin, could be used as a mitogen. However, quantitation of lymphocytic blast transformation in the presence of pronase was much more difficult than had been the case with trypsin.

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

405

In strong support of the hypothesis that trypsin preferentially stimulated B- and not T-cell proliferation, Kaplan and Bona (1974) demonstrated that peripheral lymphocytes, derived from athymic, nude (nulnu)mice, could be stimulated to reinitiate cell division by culturing them in the presence of trypsin. As a working hypothesis to explain the mitogenicity of trypsin, Kaplan and Bona (1974) suggested that both nonspecific mitogens [e.g., lectins or lipopolysaccharide (LPS)] and antigens act to initiate blast transformation via the release and/or activation of an endogenous lymphocytic protease which attacks the surface structure of its own or surrounding lymphocyte plasma membranes, thereby triggering the subsequent events of blast transformation. Evidence that lysosomal proteases might be released during the immune response and, as a result of their release, act on the lymphocyte cell surface had been previously discussed in a review by Weissmann et al. (1972). According to Kaplan and Bona’s hypothesis, exogenous protease (added in the form of trypsin or pronase) would act directly on the lymphocyte surface to initiate blast transformation, thereby bypassing the mitogen- or antigen-stimulated release of an endogenous protease. A similar hypothesis has recently received much attention with regard to protease stimulation of fibroblast cell division (Talmadge et al., 1974). This hypothesis is discussed in detail in Section VII. Recently, Chen et al. (197613) demonstrated that both trypsin and thrombin served as potent mitogens for mouse splenocytes. Both proteases were found to be mitogenic at concentrations of 2 pg/ml, and neither required any additional serum factors to induce the mitogenic response. It was also reported that both proteases had to be maintained in the growth medium for 24 hours or more in order to maximize the stimulation of DNA synthesis in the splenocyte culture. As in the work of Vischer (1974) and Kaplan and Bona (1974), Chen et al. (1976b) demonstrated that both trypsin and thrombin were approximately 50% as efficient as LPS or the lectin phytohemagglutinin (PHA) in stimulating cell division in the splenocyte culture. Figure 2A is a densitometric scan of an autoradiograph of [131111actoperoxidase-labeled splenocytes prior to thrombin treatment (Chen et al., 1976b). Figure 2B and C demonstrates that proteins with a MW of approximately 45,000 are the only peptides significantly reduced by trypsin or thrombin treatment of the splenocytes. This finding led Chen et al., (1976b) to suggest that proteolytic modification of surface peptides in this MW range might be sufficient in itself to induce the mitogenic response observed after trypsin or thrombin treatment. It must be noted, however, that the densitometric scan of an l3II autoradiograph presented in Fig. 2 gives a somewhat misleading im-

406

KENNETH D. NOONAN

FIG.2. Densitometric tracing of autoradiographs of SDS polyacrylamide gradient gel strips obtained by electrophoretic analysis of splenocyte surface proteins labeled by lactoperoxidase-catalyzediodination. (A) Untreated control; (B) treated with thrombin, 2 pg/ml; (C) treated with hypsin, 1 pg/ml. Abscissa: top of gel to bottom (left to right); ordinate: absorbance. (From Chen et al., 197613.)

pression of the number of iodinateable lymphocyte surface components. As can be seen in Fig. 3, autoradiographs of lectin or LPS-stimulated lymphocytes (Trowbridge et aZ., 1975) demonstrated that the “fine structure” of the autoradiograph is lost in the densitometric scan published by Chen et al. (1976b). Furthermore, Trowbridge et al. (1975) have reported that autoradiographs of isolated lymphocyte plasma membranes radiolabeled with amino acids are even more complex than the autoradiograph in Fig. 3. Taken together, the data of Trowbridge et aZ. (1975) leads one to suspect that Chen et al. (1976b)may have inadvertently overinterpreted their data in suggesting that protease-dependent cleavage of surface peptides (or glycopeptides) of approximately 45,000 MW are the only modifications in the cell surface brought about by limited trypsin- or thrombindirected proteolysis. Clearly, other peptides or glycopeptides may

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

407

FIG.3. Iodination of activated T and B cells. Autoradiographs of iodinated proteins from spleen cells stimulated with con A (a), LPS (b),allogeneic cells (c),and pea lectin (d).(From Trowbridge et al., 1975.)

have been modified by the protease treatments, and only studies employing more detailed autoradiographs will allow one to conclude with certainty how many surface components are modified by protease treatment of the lymphocyte. As pointed out earlier in this chapter, it is essential, if one is to assign the key role in the induction of a subsequent cell division to a surface alteration, that there be little or no opportunity for the protease to be taken up into the cell. All the work discussed above demonstrates that at least a %-hour incubation with the chosen protease is necessary in order to achieve maximal stimulation of cell division in the lymphocyte population. This finding, which all the investigators recognize as a weak point in assigning surface modifications the key role in the induction of lymphocyte blastogenesis, has been rationalized by Chen et al. (1976b) to result from a constant regeneration of the 45,000-MW surface peptides even in the presence of the protease. These workers therefore argue that trypsin or thrombin must be con-

408

KENNETH D. NOONAN

tinuously present in the growth medium in order to cleave the newly added 45,000-MW surface peptides, and that it is only after -24 hours of continuous absence of these peptides from the surface that the lymphocytes are committed to reinitiate cell division. Although this argument (Chen et al., 1976b) may be correct, it is essential that more controls be run to rule out properly the possibility that the surface modification detected by Chen et al. (1976b), although the result of proteolysis, is not the signal which initiates the in uitro mitogenic response. In a totally different approach to the use of proteases in probing the role of the lymphocyte surface architecture in the mitogenic response, Goodall et al. (1971) investigated the ability of limited proteolysis to modify lymphocyte blastogenesis stimulated by a lectin (in this case PHA). These investigators have reported that the treatment of lymphocytes from a variety of sources with 0.01% trypsin or 0.0005% pronase increased the lymphocyte response to PHA as measured by enhanced DNA synthesis. They suggest that various lymphocytes in the original population did not respond to PHA stimulation because lectin binding was sterically hindered, and that protease treatment might act to enhance the response of the lymphocytes to PHA by exposing more lectin receptors. However, to the best of our knowledge, this lead has not been followed up and the data provided in the original paper (Goodall et al., 1971) are insufficient to allow one to judge critically the value of the hypothesis. B. Sodium Periodate

Although this chapter deals primarily with the role of proteolytically modified surface structures in inducing cell division, it is important to point out that, in lymphocytes, modification of a simple carbohydrate moiety on the cell surface seems to be sufficient to induce blast transformation. Novogrodsky and Katchalski (1971) were the first to demonstrate that a 10-minute incubation of rat lymph node lymphocytes with 5x M sodium periodate induced a mitogenic response which was comparable, although not identical, to the mitogenic response produced by the lectins PHA and concanavalin A (Con A). It is of interest to note that 104- stimulation differed significantly from trypsin in that it was dependent on the continued presence of sera in the growth media. Novogrodsky and Katchalski (1971) clearly demonstrated that the percent of I04--mediated blastogenesis was directly

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

409

proportional to the amount (up to lo%,v/v) and type of sera in the growth medium. This early work of Novogrodsky and Katchalski was confirmed by Zatz et al. (1972)as well as by Parker et al. (1973),who demonstrated that an 8- to 10-minute incubation of human lymphocytes with approximately M 10,- produced a mitogenic response similar to that observed with PHA or pokeweed mitogen (PWM).Parkeret al. (1973)suggested that 104- acted via oxidation of the vicinal hydroxyl groups of cell surface carbohydrates. In support of this hypothesis, these workers demonstrated that the addition of excess exogenous carbohydrate to the growth medium, prior to the addition of periodate, prevented the 10,- stimulation of lymphocyte blast transformation. In a later study Parker et al. (1974)demonstrated that the 10,--mediated stimulus was most pronounced (and cell viability best maintained) under conditions of pH and temperature where the two terminal exocyclic carbon atoms of sialic acid could be cleaved without subsequent destruction of the pyranose ring or the glycosidic bond of the nonreducing surface carbohydrate. Better evidence that sialic acid is the site of 104-action on the cell surface comes from the work of Zatz et al. (1972), who first demonstrated that reduction of the cell surface with 0.5-1.0 x M NaBH4 immediately after 10,- oxidation prevented blast transformation. Since it had been previously demonstrated that 10,- oxidation of lymphocytes followed by NaB3H4reduction specifically labeled the surface sialic acid, the work of Zatz et al. (1972) suggested that oxidation of the sialic acid moiety was the principal reaction in 10,- treatment of the cell surface. These data were confirmed and strengthened by Novogrodsky and Katchalski (1972),who showed that IO,--treated cells subsequently reduced with borohydride, hydroxylamine, or semicarbizide all displayed a much reduced level of blast transformation. Subsequently, Novogrodsky and Katchalski (1972) demonstrated that the response of mouse spleen cells to 10,- was markedly reduced if, following 10,- treatment, they were enzymically modified with neuraminidase or papain. Taken together, the work of Parker et al. (1974),Zatz et al. (1972),and Novogrodsky and Katchalski (1972) suggested that the continued presence of oxidized sialic acid on the cell surface was necessary to induce mitogenesis in lymphocytes. Thus it was concluded that the membrane component(s) attacked by 10,- “includes a glycoprotein complex containing sialic acid” (Novogrodsky and Katchalski, 1972) and that the “aldehyde-moiety formed upon oxidation of sialic acid residues in the cell membrane is essential for

41 0

KENNETH D. NOONAN

[IO,-] induction [of cell division]” (Novogrodsky and Katchalski, 1972). In later work Novogrodsky and Katchalski (1973) reported that the sialic acid moieties oxidized on the cell surface were not the only sugar residues which, following oxidation, could induce blast transformation. It had been known from previous work (Morel1 et al., 1966) that neuraminidase treatment of the lymphocyte cell surface could expose galactose moieties which in turn could be oxidized at the C-6 position by galactose oxidase, yielding a 6-aldehydo analog of galactose. Using this earlier work as a stepping stone, Novogrodsky and Katchalski (1973) demonstrated that extensive blastogenesis could be induced in mouse spleen cells treated with neuraminidase and then galactose oxidase. Using the reducing agents borohydride and hydroxylamine, Novogrodsky and Katchalski ( 1973) again demonstrated that maintenance of the aldehyde moiety on the cell surface was essential for the induction of blastogenesis. This induction of blast transformation, like 10,- induction, was dependent on the continued maintenance of sera in the growth media. Two questions remain open with regard to 10,- stimulation of lymphocyte cell division. The first concerns the evidence (or lack of evidence) that 104-acts only at the cell surface, while the second concerns the molecular mechanism of a stimulus which depends solely on the oxidation of a sugar moiety to induce blast transformation. In response to the first question posed, it has been suggested that the site of action of 10,- must be the cell surface, “because the amount of glycoprotein [at the cell surface] makes it unlikely that any unreduced 10,- penetrates into the cell” (Parker et al., 1974). Better evidence for surface-mediated triggering of 10,--stimulated blastogenesis comes from the work with neuraminidase, which clearly indicates that oxidized sialic acid must be retained on the cell surface for blast transformation to occur. Thus it is likely that 10,- initiates a response eventuating in blast transformation by oxidizing the cell surface sialic acid, however, this does not exclude the possibility that intracellular oxidation of other components also occurs and that such oxidation plays a role in stimulating DNA synthesis. In answer to the question relating to the molecular mechanism(s) underlying 10,- stimulation of blastogenesis via oxidation of the cell surface carbohydrates, Novogrodsky and Katchalski (1973) have suggested that the I04--produced aldehyde groups may react with free amino groups on other proteins or glycoproteins on the cell surface, subsequently yielding a Schiff base between proteins and glycoproteins. These Schiff bases might behave as cross-linked surface struc-

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

41 1

tures, thereby producing the glycoprotein patches which appear to be essential for the induction of lymphocyte blastogenesis by lectins or antigens. C. Lectins

It is well known that lectins (plant agglutinins) can induce blast transformation in lymphocytes. This mitogenic activity is dependent on the binding of a specific lectin to a surface carbohydrate. The best evidence supporting the contention that lectins act at the cell surface rather than intracellularly comes from the work of Anderson et al. (1972), Betel and Van Den Berg (1972), and Greaves and Bauminger (1972),all of whom showed that lectins (specifically Con A and PHA) could induce lymphocyte blastogenesis even when derivatized to a solid matrix which prevented them from being internalized by the cell. In most of the studies reported to date it has been demonstrated that lectins must remain bound to the lymphocyte cell surface for relatively long periods of time (-20 hours) in order for blastogenesis to be induced. Furthermore, it has been demonstrated that lectin-mediated mitogenesis occurs only when the lymphocytes are maintained in media containing serum. These data suggest that lectin-mediated blastogenesis resembles 10,--induced mitogenesis, at least in its requirement for serum factors. Of interest to this chapter is the finding that there is some evidence that endogenous proteolytic activity may be involved in lectinmediated cell division. Specifically, Hirschorn et al. (1971) demonstrated that the protease inhibitor eaminocaproic acid (EACA),at concentrations of 0.05-0.01 M , inhibited 50% of the L3H]thymidine ([3H]Tdr)incorporation normally observed in PHA-stimulated lymphocytes. Protease inhibitors with different substrate specificities [such as p-toluene sulfonyl-L-arginine methyl ester (TAME), N-tosyl-L-lysine chloromethyl ketone (TLCK), and N-tosysl-phenylethyl chloromethyl ketone (TPCK)] also inhibited PHA-stimulated blastogenesis. These data led Hirschorn et al. (1971)to suggest that proteolysis may play a role in bringing about the early alterations in “macromolecular synthesis” accompanying lymphocyte stimulation, and that EACA may act to limit blastogenesis by inhibiting important endogenous proteases. (For a critical review of the use of such protease inhibitors in investigating cellular events, see Section VII.) Obviously, a review of lectin-stimulated blastogenesis is well beyond the scope of this chapter. For excellent, lengthy reviews of lectin-stimulated blast transformation, see Greaves and Janossy (1972), Edelman (1973),and Nicolson (1974).

41 2

KENNETH D. NOONAN

IV.

PROTEASE INDUCTION OF CELL DIVISION IN FIBROBLASTS

Nontransformed (normal) secondary or established cell lines frequently display a property referred to variously as contact inhibition of cell division (Abercrombie and Heaysman, 1954; Noonan and Burger, 1974), density-dependent inhibition (DDI) of cell division (Stoker and Rubin, 1967),or postconfluence inhibition of cell division (Martz and Steinberg, 1972). Phenomenologically, this property of the cell is defined as growth of a cell population into a cell sheet with a subsequent collection of greater than 90%of the cells in the G , (or Go) stage of the cell cycle. Contact-inhibited cells remain viable at the monolayer stage and can frequently be induced to reenter the cell cycle. Cells transformed by a variety of agents (including DNA and RNA tumor viruses, chemical carcinogens, and x rays) usually do not display the property of DDI of growth and therefore form multilayers under growth conditions where nontransformed cells display DDI of growth. Many attempts have been made to modify either the growth conditions or the environment under which nontransformed cells are grown, with the hope that such modifications will cause the normal cells to behave “as transformed cells” in that they will grow to a higher cell density. Holley and his collaborators (Holley and Kiernan, 1968),as well as Temin (1967) and his co-workers, have been successful in modifying fibroblast cell growth by modifying either the amount or type of serum components added to nontransformed cell cultures. A variety of attempts has been made to modify the cell surface of contact-inhibited cells in a manner which might disrupt the apparent “recognition” normal cells display for one another at the monolayer stage. It has been argued that, since contact inhibition of cell division is almost certainly related to various cell surface properties, a change in the surface architecture of the nontransformed cells might reinitiate cell division in a quiescent cell sheet. Proteases have proved to be the most popular enzymic probes in such work. A. Overgrowth Stimulating Factor

In 1970 Rubin reported that the addition of media taken from chick embryo fibroblast cultures infected with Rous sarcoma virus (RSV) to quiescent, noninfected chick embryo fibroblasts (CEFs) produced an increase in both rH]Tdr incorporation and the mitotic index of the density-inhibited culture which was complete within 65 hours after

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

41 3

the change in medium. These findings led Rubin to suggest that cells were infected with a RSV released material [overgrowth stimulating factor (OSF)] which stimulated continued multiplication among crowded but uninfected CEFs. In further work Rubin found that, although OSF was secreted into the culture medium only by RSV-infected chick cells, normal cells sequestered an activity similar to OSF which could be released by sonication and subsequent cell lysis. These data suggested that the release of OSF into the growth medium was related to both transformation and spec@ leakage of this component from the transformed cells (Bissell et al., 1971). In his early work on OSF Rubin stated, without presenting supporting data, that the growth-stimulating activity of OSF could be mimicked by the addition of 3 pg/ml of trypsin or pronase to the agar overlaying confluent, density-inhibited CEF cell cultures (Rubin, 1970). This finding lead Rubin to be the first to suggest that proteases, acting at the plasma membrane, might affect the structure-function relationships in the membrane, subsequently stimulating cell division. Furthermore, Rubin suggested that OSF itself might b e a protease which, upon release from the RSV-infected CEFs, could act extracellularly to initiate cell growth in a quiescent cell population. This was (and is) a particularly attractive hypothesis, since it is well known that tumors (which have clearly lost many features defined by the term “growth control”) produce and release peptidases into the surrounding extracellular environment (e.g., Sylven and Malmgren, 1957). Unfortunately, subsequent work by Burr and Rubin (1975), using a 100- to 200-fold purified OSF fraction isolated from RSV CEFs, conclusively demonstrated that OSF contained no proteolytic activity which could be detected by any of the assays currently employed in the detection of neutral proteases. Thus it is very unlikely that OSF itself is a protease. However, Rubin’s early work with OSF was of major importance to the field of protease-stimulated cell growth, since it prompted his laboratory, as well as other workers, to investigate the ability of exogenous proteases to stimulate cell growth in confluent, nontransformed cell monolayers. 6. Chick Embryo Fibroblasts

Sefton and Rubin (1970) were the first to demonstrate that extended trypsin treatment (> 24 hours) of quiescent CEFs produced a marked increase in [3H]Tdrincorporation and subsequent cell division among the nontransformed cells.

414

KENNETH D. NOONAN

This finding that secondary chick fibroblasts brought to quiescence in low serum (0.7%chick serum plus 1.5%calf serum) containing medium could be stimulated to reinitiate cell division following trypsin or pronase treatment was subsequently verified by Cunningham and Ho (1975), who showed that the addition of trypsin or pronase to the growth medium produced anywhere from a 30 to an 85% increase in cell number. Cunningham and Ho (1975) also demonstrated that the CEFs had to be treated with these proteases for an extended period of time in order for cell division to occur. As shown in Fig. 4, a 16- to 20-hour incubation with trypsin (followed by replacement of the protease-containing media with conditioned media) was necessary to produce an increase in CEF cell density. Thus Cunningham and Ho (1975) clearly demonstrated that plasma membrane modifications occurring after brief protease treatment [e.g., enhanced lectin agglutinability, release of the large, external transformation-sensitive (LETS) protein, and so on; see Section v] were not sufficient in themselves to stimulate cell division. Chen et al. (1975) also studied protease-induced cell division in CEFs. These workers grew CEFs to quiescence in media containing 0.5%calf serum. The quiescent cells were then washed and incubated in serum-free medium containing 2.5 pg/ml thrombin. Twelve hours later thrombin was removed from the culture and the cells were returned to medium lacking a serum supplement. As can be seen in Fig. 5, CEFs treated with thrombin showed a three-fold increase in cell

1.3k

4

i

I2

wuf6

16

2b

24

’

FIG.4. Duration of trypsin treatment required to stimulate proliferation of quiescent secondary CEFs. Chick fibroblasts were plated at 4.5 x 104 cells/cm2 in medium containing 0.7%chicken serum.and grown to quiescence. At time zero, trypsin was added to a final concentration of 2 pg/ml. At the indicated times this medium was removed, cultures were rinsed twice with conditioned medium, and conditioned medium was then added back to cultures. Cell number was monitored in all cultures at 24 hours. (From Cunningham and Ho, 1975.)

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

41 5

CiTWd&fl

FIG.5. Dose-response curve of mitogenic activity of bovine thrombin, calf serum, and bovine prothrombin. Resting CEFs were incubated for 12 hours with DME containing proteins to be tested. The cells in 1.5 ml of medium were then incubated for 1 hour with 2 p Ci/ml [3H]Tdr.Incorporation into insoluble material was then determined. (From Chen et nl., 1975.)

number over the next 4 days, despite the absence of serum from the growth medium. It must be noted that this report that thrombin can act as a mitogen toward CEFs in the absence of any exogenous serum is the only report, to our knowledge, of fibroblast cell division induced in the absence of at least a minimal serum concentration. This finding that serum is not required for thrombin-induced CEF cell division has been verified by at least one other group (Perdue, personal communication). The fact that thrombin-induced CEF cell division occurs in the absence of exogenous serum suggests that this protease-cell type combination might be the best system with which to study the role of proteolytic modifications of surface structure in the induction of fibroblast cell division. As with almost all the work described to date, it must be noted that extended incubation times were required for trypsin, pronase, or thrombin to induce CEF cell division. As we have emphasized before, this need for extended incubation times must make one cautious in assigning a site of action to the various proteases. Thus, although changes in surface structure clearly occur following limited proteolysis (see Section V), there is no evidence presently available which conclusively rules out an intracellular locus as the site of protease activity which might be responsible for the reinitiation of cell division in the quiescent CEFs. (See Note Added in Proof, p. 461.)

416

KENNETH D. NOONAN

C. Mouse Embryo Fibroblasts (313)

The cell line which has been the most extensively used in the study of DDI growth is the established mouse embryo fibroblast (MEF) cell line (3T3)first developed by Todaro and Green (1963).Under the culture conditions these workers employed [Dulbecco’s modified Eagle’s medium (DME).plus 10% calf serum] a specific 3T3 clone grew to approximately 3 x lo4 cells/cm2 (the “monolayer state”) and then displayed a virtually complete inhibition of DNA synthesis and cell division. For several years this cell line was the best example of a cell line in which cell-cell contact could apparently initiate a series of pleiotypic events which culminated in the inhibition of cell division. Burger (1970) was the first to demonstrate that a subclone (41) of 3T3 (which grew to -9 x lo4 cells/cm2 in DME plus 3% calf serum) could be stimulated to undergo one further round of cell division following a 45-minute incubation with serum-free DME containing 5 pg/ml ficin or 0.70 pg/ml trypsin. This induction of 3T3 cell division required that the cells be returned to the 2-day-old conditioned medium after the protease treatment. Bombik and Burger (1973), as well as Noonan and Burger (1973), subsequently verified this work using the same cell clone but growing it to quiescence (-4 x lo4cells/cm2) in medium containing 10% calf serum. Bombik and Noonan demonstrated that this particular 3T3 clone could be grown to confluence in DME plus 10% calf serum and then held in a quiescent state (Go)for 4 days. When these cells were subsequently treated with 10 p g pronase per milliliter phosphatebuffered saline (PBS) for 10 minutes and then returned to the conditioned medium which had been sterilely removed prior to protease treatment, they underwent one further round of cell division. Noonan and Burger (1973) demonstrated that this stimulatory cycle could be repeated twice, eventually producing a normal cell culture in which the final cell density approached that of a transformed derivative of the 3T3 cell line. Thus with this particular 3T3 cell clone it was apparent that a brief protease treatment could effectively induce cell division if the culture was returned to the conditioned medium following protease modification of the cell surface. Burger (1972) reported that insoluble, bead-bound trypsin, overlayered onto a confluent 3T3 cell population, could induce a new round of cell division even if the beads were washed off the monolayer within 5 minutes after addition. These data were taken to support the contention that protease induction of 3T3 cell division is a surface-mediated phenomenon. It is of importance to note that this particular 3T3 cell clone could

PROTEOLMIC MODIFICATION OF CELL SURFACE MACROMOLECULES

41 7

also be stimulated to resume cell division by an 18- to 24-hour incubation in conditioned medium supplemented with 0.055-0.110 U of insulin per milliliter (Bombik and Burger, 1973).This particular finding is of distinct interest, since insulin stimulation of 3T3 cell division was not found in many of the other 3T3 subclones tested (K. D. Noonan, unpublished observation). Despite the relatively large body of evidence generated by Burger’s group supporting protease-mediated stimulation of 3T3 cell division, controversy has arisen concerning the general applicability of these findings to the 3T3 cell type. Cunningham and his collaborators (Glynn et al., 1973) clearly showed that the 3T3 clone used in their studies (clone 42 of Todaro and Green) could not be stimulated to reinitiate cell division following treatment of a quiescent monolayer with trypsin or pronase. Using clone 42 (which grew to 3.0-3.7 x lo4 cells/cm2 in DME plus 10% calf serum), Glynn et al. (1973) demonstrated that treatment of these 3T3 cells with 5 pg/ml pronase for 10 minutes followed by a return to the conditioned medium used in growing the same cells to confluency produced no significant increase in the population density of the culture. Further work by Cunningham and Ho (1975) demonstrated that prolonged incubation (> 24 hours) of 3T3 cells with trypsin or pronase also did not produce an increase in cell density. A likely explanation for the apparent discrepancies between the data reported by Cunningham’s and Burger’s groups was recently published by Noonan (1976). In his study Noonan (1976) randomly selected 3T3 subclones from clones 41 and 42 of Todaro and Green. Each subclone selected was then tested to determine the percent serum required for growth to confluency (arbitrarily defined as 3.0 x lo4 cells/cm2). Eventually, four clones having different serum requirements for growth were chosen for the study (Table I). As can be seen, clone 3T3, (which was the clone used throughout all of Burger’s work) grew to the monolayer stage in DME plus 3% calf serum while, at the opposite end of the spectrum, clone 3T310 required DME plus 10%calf serum in order to reach a final cell density of -3.0 x lo4 cells/cm2. As can be seen in Fig. 6, only subclone 3T3, could be grown to confluency in DME plus 10%calf serum, treated with 10 pg/ml pronase for 5 minutes, and then returned to conditioned medium with a resultant increase in cell number. The remaining subclones required the addition of differing amounts of fresh serum to the media in order for a protease-induced increase in cell number to occur (Fig. 6).These data suggest that protease treatment of 3T3 cells may be sufficient to ini-

-

418

KENNETH D. NOONAN

TABLE I SERUM REQUIREMENTS OF INDIVIDUAL CELL LINESO Final cell density [(cells/cmz)(x I@’)]

Cell line

DME plus 3% calf serum

DME plus 5% calf serum

DME plus 7% calf serum

DME plus 10% calf serum

Conditioned medium

3T33 3T35 3T3, 37’310

2.2 1.5 0.35 0.30

2.5 2.0 1 .o 0.675

3.0 2.4 2.5 1.1

3.2 2.6 3.O 2.5

2.5 1.75 0.75 0.30

Saturation density was determined by plating cells at approximately 2.0 x 1V cells/cm* in DME plus the given concentration of calf serum and counting cells daily for the next 7 days. A complete monolayer is formed with these cells at approximately 2.0 x l(r cells/cm2. Cell size remained approximately the same for all cell lines tested.,Cell lines originally derived from an NIH Swiss mouse embryo fibroblast (3T3) cell line were used throughout. The original line was obtained as clone 41 from G. J. Todaro. Random stocks of this original clone had been stored in liquid nitrogen at different intervals for the past 5 years. Four of these cell stocks were used. All cell lines were maintained at 37.5”C in a moist incubator in which the carbon dioxide tension was held at 5%. DME plus 10% calf serum and 1% penicillin-streptomycin was used as growth medium for all stock cultures. The cells were demonstrated to b e free of contamination with pleuropneumonia-like organisms at the time of the experiments, by both autoradiography and mycoplasma broth media. No cell line was maintained in culture for more than 20 passages. Frozen stocks were removed from liquid nitrogen at intervals during the course of the experiments. Conditioned medium was produced by plating each cell line at 2.0 x 103 cells/cm* in DME plus 10%calf serum and allowing the cells to grow in this medium for 5 days at which time the medium was removed and filtered. In this regard it must be noted that the term “conditioned” refers to medium taken from specific cell lines. Thus the conditioned medium added to 3T3, cells is medium which has been depleted by 3T33 cells, while conditioned medium added to 3Tslocells has been depleted by 3T3,, cells. Since these cell lines have different serum requirements for growth, it might be expected that the conditioned media produced by the two cell types would also vary. (From Noonan, 1976.)

tiate a response that can culminate in cell division, but that in many 3T3 subclones serum components (which are depleted in conditioned medium) are required to maintain the progress of the cells through the cell cycle. Thus we believe that the apparent discrepancy between the work of Burger’s group and Cunningham’s group can be explained by the fact that each group used 3T3 cells with different serum requirements for

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

41 9

FIG.6. Protease stimulation of different 3T3 subclones in various serum concentrations. The various subclones were grown to confluency in DME plus 10% calf serum. After 2 days at the monolayer stage the cells were washed once with PBS and then incubated for 5 minutes with 10 k g of pronase. DME plus the desired concentration of serum was then added back to the cells. Open squares, DME plus 3%calf serum; open triangles, DME plus 5% calf serum; solid squares, DME plus 7% calf serum; solid triangles, DME plus 10%calf serum; solid circles, conditioned medium; open circles, untreated monolayer. (From Noonan, 1976.)

growth. In Burger’s work a clone with a low serum requirement was inadvertently chosen, thereby setting up a situation in which mild proteolysis could induce cell division in the presence of conditioned medium. Cunningham, however, worked with a more “traditional” 3T3 cell line which required more serum for efficient cell growth. Thus, in Cunningham’s work, simply returning the cells to conditioned medium after protease treatment was not sufficient to induce a new round of cell division, since serum factors required to maintain the cell cycle were apparently depleted from the conditioned medium. D. Other Cell lines

Cunningham and Ho (1975) studied the effects of mild protease treatment on the stimulation of cell division in human diploid foreskin

420

KENNETH D. NOONAN

cells and bovine embryonic brachial cells under conditions identical to those sufficient to support protease-induced cell division in CEFs. However, under the conditions chosen, neither of these cell lines could be induced to resume cell division. In light of Noonan’s (1976) data it could be argued that the serum requirements of these two cell lines were not properly evaluated and that their failure to divide was related to serum depletion. However, Noonan has also reported (without providing supporting data) that several quiescent cell lines (in particular an established rat embryo fibroblast (REF) cell line) did not resume cell division following protease treatment even when great care was taken to ensure that the cells’ serum requirements were met. These data from the laboratories of Cunningham and Noonan suggest that some cell lines are refractory to protease-stimulated cell division even when maintained under ideal growth conditions.

E. Conclusions

The phenomenological data set forth in this section suggest that cells which grow to the monolayer stage in media containing relatively low concentrations of sera (e.g., CEFs, 3T3J can be stimulated to resume cell division following mild protease treatment. In most cases the incubation with the protease must be an extended one. Apparently, only the 3T3 cells can, under ideal conditions, be stimulated to divide after brief protease treatment, and even these cells (with the probable exception of 3T3J require that fresh serum be added to the growth medium in order to ensure a subsequent cell division. It is possible that the addition of fresh serum to the 3T3 cells may itself act in a fashion analogous to the extended proteolysis required to induce cell division in CEFs. Alternatively, fresh serum may simply act as a source of new nutrients or serum factors necessary for maintenance of the cell cycle. As discussed in Section VIII, our present concept of protease-stimulated cell division agrees in essence with the theory put forth first by Pardee (1964) and later by Holley (1972) that modification ,of the normal cell surface by proteases or excess serum may influence the efficiency of uptake of nutrients or serum components and that it is these factors which are most directly responsible for the subsequent cell division. Under such an hypothesis, protease modification ofthe cell surface serves as one of many “signals” necessary to drive a quiescent cell population into and through a new cell division.

PROTEOLMIC MODIFICATION OF CELL SURFACE MACROMOLECULES

V.

42 1

EFFECTS OF PROTEASES ON FIBROBLAST SURFACE STRUCTURE

A. Change in Leetin Agglutinability

The surface property most frequently correlated with the transformed phenotype described by growth to high cell density is the surface architecture defined by enhanced agglutinability with plant lectins (Burger and Goldberg, 1967; Burger, 1969; Inbar and Sachs, 1969; Pollack and Burger, 1969). As can be seen in Fig. 7, a direct correlation has been reported between the relative ease of agglutinability of a cell line and the final density to which it grows. These data led to the hypothesis that the surface structure defined by enhanced lectin agglutinability might be responsible for the growth of a transformed cell line to a high cell density, Although the universality of this correlation has been questioned (most recently by Ukena et al., 1976), it is clear (as is demonstrated in Fig. 8) that the brief treatment of a nontransformed cell line, such as 3T3, with a variety of proteases significantly enhances the relative agglutinability of the nontransfonned cells. To the best of our knowledge there have been no reports to date of nonagglutinable, nontransformed cell lines which cannot be rendered agglutinable by mild protease digestion. However, although it is generally agreed that protease treatment of nonagglutinable cells renders such cells maximally agglutinable with a variety of lectins, it must be

SATURATION DENSITY [CELLS/cm?X 16'1 Correlation between loss of DDI of growth and agglutinability. (From FIG. 7. Burger, 1971.)

422

KENNETH D. NOONAN

0

0007

I

KTRYPSIN

0.027

IP FIG. 8. Dependence of agglutinability on trypsin concentration. For 3 minutes 4 x lo6 cells/ml were treated with the concentration of trypsin indicated. The reaction was stopped with a 2- to 10-fold excess of soybean trypsin inhibitor, and the cells tested for agglutinability. Open triangles, BHK cells; solid circles, PyBHK cells; open circles, 3T3 cells; open squares, Py3T3 cells. (From Burger, 1969.)

recognized that the molecular mechanism of this enhanced lectin agglutinability is still not understood. Although Nicolson and a variety of other workers (for a review of this work, see Nicolson, 1974) have reported that protease treatment of the nonagglutinable cell surface increases the mobility of the lectin receptors and that this increase in receptor mobility is responsible for increasing a cell’s agglutinability, several other workers have more recently reported that agglutinability, including protease-induced agglutinability, can occur in the absence of enhanced mobility of the lectin receptors (e.g., DePetris et al., 1973; Ukena et al., 1976; Ellison et al., 1977). For recent reviews directed toward the molecular basis of lectin-initiated cell agglutination, the reader is referred to Nicolson (1974), Nicolson and Poste (1976), and Noonan (1978). The fact that mild protease digestion of the nontransformed 3T3 cell rendered its surface architecture similar to that of its transformed counterpart (at least with regard to agglutinability with plant lectins) led Burger (1970)to attempt to stimulate cell growth by brief protease treatment of quiescent 3T3 cells. As discussed in Section IV,C, mild proteolysis induced a new round of cell division with a particular 3T3

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

423

cell clone. This lead Burger (1970) to suggest that modification of the 3T3 cell surface from the nonagglutinable to the agglutinable state might be sufficient in itself to induce a new round of cell division. This hypothesis was supported, but certainly not proved, by the work of Bombik and Burger (1973) and Noonan and Burger (1973), using 3T3 clone 41 (see Section IV,C). It is now clear that the change in the surface architecture described by enhanced lectin-initiated cell agglutination is not sufficient to induce a new round of cell division. Among the pieces of evidence which support this statement are the following:

1. Sefton and Rubin (1970), as well as Cunningham and Ho (1975), demonstrated (see Section IV,B) that prolonged incubation (>20 hours) with trypsin or pronase is necessary to induce a new round of cell division in a quiescent CEF cell culture. Since the maximal increase in C E F agglutinability with Con A occurs after a 10-minute incubation with 2 pg/ml trypsin (Cunningham and Ho, 1975),it is clear that a change in the relative agglutinability of this cell type is not sufficient to induce a new round of cell division. 2. Glynn et al. (1973), as well as Cunningham and Ho (1975), demonstrated that treatment of the 3T3 cell clone 42 with 5 pg/ml pronase for 10 minutes radically increased the cells’ agglutinability with Con A but failed to produce any significant increase in cell number within the quiescent cell population. 3. Noonan (1976) demonstrated that a majority of the 3T3 subclones he derived from 3T3 clone 41 or 42 cannot be stimulated to undergo a new round of cell division even after they are rendered maximally agglutinable with Con A by a 5-minute incubation with 10 pg/ml pronase. As mentioned in Section IV,C, it was necessary to add DME plus fresh serum following the pronase treatment in order to induce most of the 3T3 subclones to undergo a new round of cell division. 4. In an as yet unpublished study, Noonan and Noonan isolated a series of flat revertants (Pollack et al., 1968) of polyoma virus-transformed 3T3 (Py3T3) cells which, although displaying most of the properties associated with the normal phenotype (including growth to a low saturation density), remained maximally agglutinable with low concentrations of three lectins (Con A, PHA, and wheat germ agglutinin). Although treatment of these revertants with pronase does not increase their relative agglutinability with any of the lectins tested, incubation of a quiescent monolayer of these cells with 10 pg/ml pronase for 5 minutes, followed by the addition of DME plus 7% calf serum, induces these variants to go through a new round of cell division. These unpublished data suggest that the protease probably acts

424

KENNETH D. NOONAN

at a site other than that responsible for enhancing lectin agglutinability and further suggest that the change in surface architecture detected by enhanced agglutinability is probably not responsible for the protease-induced stimulation of cell division. Taken together, the data from the various laboratories clearly demonstrate that modification of the nontransformed surface architecture to the agglutinable state is not sufficient to induce a new round of cell division. The data reported by Noonan (1976) suggest that such a change in surface architecture might be sufficient to “pull” the quiescent cells out of Go into GI, but that various serum factors are necessary to drive the cells into S. However, the unpublished work of Noonan and Noonan argues more strongly that the modification of the surface structure responsible for pulling the stationary cells out of Go is not correlated with the surface structure defined by enhanced lectin agglutinability. Thus, given the evidence cited above, we suggest that the modification in surface architecture indicated by enhanced lectin agglutinability probably has little or no involvement in inducing a new round of cell division even under ideal culture conditions where proteolytic modification of the cell surface produces enhanced DNA synthesis and subsequent cell division.

B. Proteate-Mediated Modification of Membrane Peptidet and Glycopeptider

As should be obvious from the previous section, the subdiscipline of cell biology interested in protease-stimulated cell division most frequently relies on studies aimed at detecting differences in the membrane peptides of normal versus transformed cells to suggest particular membrane components which might respond to the type of protease treatment demonstrated to induce cell division. In light of this, one must be aware that over the last 3 years much energy has been expended in investigating the peptide and glycopeptide composition of normal and transformed plasma membranes. The most exhaustively studied membrane component reported to be sensitive to transformation by a wide variety of agents is the LETS glycoprotein (see Hynes et al., 1975, for a review of the work related to the LETS glycoprotein). This particular component, which has been reported to have a MW ranging from 205,000 to 250,000, is also known as the Z protein (Wickus et al., 1974), the a band (Gahmberg and Hakamori, 1973),the CSP (Yamada and Weston, 1974), and the SF antigen (Ruoslahti et al., 1973).

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

425

Hynes (1973) was the first to report that lactoperoxidase-directed iodination of nontransfopned NIL cells, followed by autoradiography of the iodinated components separated on SDS-PAGE, detected a prominent band at -250,000 MW which was absent from hamster sarcoma virus (HSV)-transformed NIL cells. This initial report was quickly confirmed by Hogg (1974), Pearlstein and Waterfield (1974), and Wickus et ul. (1974), using a variety of other normal-transformed cell combinations. Subsequent work demonstrated the LETS protein to be a surface glycoprotein which could be metabolically labeled with glucosamine or fucose (Hynes and Humphreys, 1974), as well as with NaB3H4following galactose oxidase-neuraminidase treatment of the cell surface (Critchley, 1974). Since these early observations confirming the preferential association of the LETS protein with the normal, nontransformed cell surface much work has gone into further characterizing the LETS protein. The findings to date are summarized in a variety of reports from different laboratories (see, e.g., Hynes and Humphreys, 1974; Hynes and Bye, 1974; Gahmberget al., 1974; Wartiovaara et al., 1974; Vaheri and Ruoslahti, 1974; Kuusela et al., 1975; Hynes et al., 1976). (For an exhaustive review of the biochemistry of the LETS protein as well as its putative role in cell-cell and cell-substratum adhesion, see the chapter by Culp in this volume.) From the point of view of this chapter the most important facet of the LETS protein is that it is extremely trypsin-labile (Hynes, 1973; Hynes and Humphreys, 1974). As can be seen in Fig. 9, treatment of nontransformed NIL cells for 10 minutes with 1 pg/ml of trypsin is sufficient to remove virtually all the LETS protein from the cell surface. It is this property of the LETS protein which has made it the focus of much attention among workers interested in protease stimulation of cell division (see Section V,B,l). Although most of the attention of investigators interested in differences between normal and transformed cell surfaces has been directed toward the LETS protein, a few other proteins have been reported to be preferentially associated with the normal as opposed to the transformed cell line and therefore have been considered potential candidates for a role in the proteolytic modification of the cell surface which produces enhanced cell division. Stone et al. (1974) reported the loss of a 39,000-MW peptide from RSV-infected CEF and NRK cells. However, since this component was also lost from cells infected with a nontransforming virus, it proved to be a very unlikely candidate for a role in proteasestimulated cell division in CEFs. Wickus and Robbins (1973), using

426

KENNETH D. NOONAN

FIG.9. Autoradiograph of SDS-PAGE gels of fibroblast proteins, labeled by iodinqtion, catalyzed by lactoperoxidase. (a and b) Comparison of normal hamster fibroblasts NIL 8 (a) with virus-transformed derivative NIL 8-HSV6 (b).(c and d) trypsin treatment after iodination of NIL 8: 3 wg/ml trypsin for 5 minutes (c) and 1 pg/ml trypsin for 10 minutes (d). (e and 9 NIL 8 cells iodinated and then overlaid for 24 hours with either NIL 8 (e)or NIL 8-HSV (0. NIL 8 cells at different growth stages: exponential ( g ) , confluent but still dividing (h), confluent arrested culture (i). (From Hynes, 1974.)

CEFs transformed by a temperature-sensitive mutant (ts-68) of RSV, first reported that the only detectable difference between cells maintained at the permissive versus the nonpermissive temperature was the absence, at the permissive temperature, of an [35S]methioninelabeled peptide of 45,000 MW. However, in subsequent work Wickus et al. (1974) demonstrated that this protein was not iodinateable and was lost relatively slowly with relation to the time course for the morphological transformation of the ts-68 RSV CEFs. Together these data suggested that decreased synthesis of this protein may occur secondarily to the earliest transformation events. Thus, except for the LETS protein, SDS-PAGE analysis of normal and transformed plasma membranes has yielded few obvious differ-

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

427

ences between the two membrane types. This has led many laboratories to investigate the possible correlation between the proteolytic susceptibility of the LETS protein and protease stimulation of cell growth.

1. ROLE OF THE LETS PROTEIN I N PROTEASE-INDUCED CELLDIVISION Following the demonstration that a major surface protein, found on normal but not on transformed cells, was extremely sensitive to proteolytic modification, there was a great rush to demonstrate that proteolytic cleavage of the LETS protein from the cell surface was responsible for protease-induced cell division. In a review emphasizing the importance of proteases and surface proteins in the process of cell transformation, Hynes, (1974) suggested “that loss or alteration of one or few [surface] proteins could have pleiotropic effects on the behavior of many surface proteins,” subsequently affecting cell growth properties. Since few differences had been detected in the normal versus transformed membrane other than the loss of LETS, Hynes (1974) proposed that the loss of this single glycopeptide might be responsible for protease induction of cell division in a population of quiescent fibroblasts. Support for the hypothesis that cleavage of the LETS protein might be of key importance in the proteolytic induction of cell division among quiescent CEFs came from Blumberg and Robbins (1975a) who demonstrated that treatment of these cells with trypsin or collagenase, which removed the LETS protein from the cell surface, produced a four- to six-fold increase in 2-deoxyglucose (2-dG) uptake, a four- to six-fold increase in C3H]Tdrincorporation into DNA and a 97% increase in cell number. Despite the appealing correlation between proteolytic removal of the LETS protein and enhanced cell division reported by Blumberg and Robbins (1975a), Buchanan, Chen, and Teng quickly presented data which made untenable the hypothesis that removal of the LETS protein from the cell surface was sufficient to induce cell division. First, Teng and Chen (1974) demonstrated that, although treatment of quiescent CEFs with 1pg/ml trypsin for 10 minutes removed most of the LETS protein from the cell surface, no increased incorporation of L3H]Tdrinto DNA was observed in these cells; i.e., the cells remained quiescent, despite removal of the LETS protein from the cell surface. Second, and probably more importantly, Teng and Chen (1974) further demonstrated that incubation of quiescent CEFs with l ,ug/ml

42%

KENNETH D. NOONAN

thrombin for 12 hours or 50 pg/ml thrombin for 30 minutes produced enhanced Tdr incorporation into DNA without removing any LETS protein from the cell surface. Finally, Teng and Chen (1974)reported that, although treatment of quiescent CEFs with insulin produced enhanced DNA synthesis, the amount of LETS protein on the cell was unaffected. The finding that thrombin could induce CEF cell division without changing the amount of LETS protein associated with the cell surface was subsequently verified by Blumberg and Robbins (197513). Taken together, these data clearly demonstrate that removal of the LETS protein from the cell surface is not necessary for proteolytic induction of cell division. The fact that cleavage of the LETS protein from the cell surface (e.g., 1 pg/ml trypsin, 10 minutes) does not in itself induce cell division strongly suggests that modification of the LETS protein is also not sufficient for the induction of cell division (Chen et aZ., 1975). 2. THROMBIN-SENSITIVE SURFACEPEPTIDE

Since thrombin did not cleave the LETS protein, Teng and Chen (1976) sought to establish which, if any, of the C E F surface proteins were cleaved by thrombin (a serine protease with a very limited substrate specificity). Using ['311]lactoper~xidaselabeling of the cell surface (which Teng and Chen demonstrated gave better autoradiographs than the conventional lZ5Ilabeling), it was demonstrated that a 30-minute incubation of CEFs with 50 pg/ml thrombin removed only one l3II-labeled peptide from the cell surface (Teng and Chen, 1976). This peptide had a MW of 205,000 and was also sensitive to trypsin treatment. Teng and Chen (1976)demonstrated that this peptide was a membrane peptide and not simply a contaminating serum component. It must be noted, however, that this thrombin-sensitive peptide, like the LETS protein, was not removed from the cell surface following insulin stimulation of quiescent CEFs (Teng and Chen, 1976). Thus it was obvious from the earliest work on this peptide that its removal from the cell surface was not necessary for the induction of cell division. Further work by Zetter et aZ. (1976) demonstrated that removal of the 205,000-MW surface peptide was not sufficient to induce a new round of cell division in quiescent CEFs. These workers demonstrated that, although the proteases a-protease, thermolysin, and papain all remove the 205,000-MW cell surface peptide, they do not induce a new round of cell division. On the basis of the morphological effects of serum and a variety of proteases on quiescent CEFs, Zetter et al. (1976) suggest that the 205,000-MW protein may play a role in

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

42 9

cellular elongation, whereas the LETS protein may play a role in cellcell adhesion (a conclusion also reached b y Yamada and Pastan; see Section V,D). Most recently, however, data have appeared from Buchanan’s laboratory (Zetter et d.,1977) which question whether thrombin exerts its biological activity via proteolytic attack on the cell surface or as a result of some action on cytoplasmic components. Specifically, Zetter et d.(1977), using electron microscope autoradiography of [1251]thrombin,clearly demonstrated that over a 6-hour incubation normal CEFs endocytose approximately 70% of the added thrombin. Since what activity, if any, the endocytosed thrombin expresses is unknown, it is no longer possible to assign thrombin induction of cell division strictly to its activity at the,cell surface. Martin and Quigley ( 1978) have similarly demonstrated that thrombin is differentially internalized by quiescent CEF’s. These authors have demonstrated that the internalized thrombin retains its catalytic activity and suggest that the thrombin may act at the level of the cell nucleus (specifically at the arginine-rich histones) in inducing a new round of cell division. Taken together, all the data accumulated with regard to protease stimulation of cell division and protease-mediated changes in surface peptides and glycopeptides suggest that, to date, protease-induced cell growth cannot be correlated with the removal of any cell surface protein or glycoprotein. C. Proteolytic Modification of the Macrwrchitecture of the Cell Surface

One of the features of the transformed membrane which apparently differentiates it from the nontransfonned membrane is the increased incidence of surface microvilli (projections 0.1 pm in diameter containing parallel 60-w filaments, Follett and Goldman, 1970) which can be observed with both transmission and scanning electron microscopy (Boyde et al., 1972; Porter et aZ., 1973a,b; Malik and Langenback, 1976; Borek and Fenoglio, 1976). It could be of importance to the topic discussed in this chapter that mild protease treatment (2.5 pg/ml trypsin for 5 minutes) of quiescent 3T3 cells radically increases the number of microvilli on the nontransformed cell surface (Willingham and Pastan, 1975), thereby making it possible that this protease-induced modification in surface structure plays a role in the induction of cell division. Despite the appeal that such data have in suggesting that a change

430

KENNETH D. NOONAN

in the number or distribution of surface microvilli might be involved in protease-stimulated cell division (possibly by increasing the amount of surface area available for serum factor or nutrient uptake; see Section VIII), it is clear that such a change in surface macrostructure is not sufficient to induce cell division in a quiescent 3T3 cell population. Specifically, Collard and Temmink (1976) showed that a 7- to 10-minute treatment of “microvilli-free,” quiescent 3T3 cells with M EDTA greatly increases the number of microvilli present on the cell surface. Since Noonan (unpublished observation) demonstrated that EDTA treatment of contact-inhibited 3T3 cells is not sufficient, even under ideal growth conditions (see Section IV), to induce cell division, a change in the distribution of surface microvilli cannot be sufficient to induce a new round of cell division. Thus, although it is possible that an increase in the number of surface microvilli may be necessary for protease-stimulated cell division, it is clear that such a change in the macrostructure of the cell surface is not sufficient to induce cell division. D. Contact Inhibition of Movement

One of the properties of normal cells which has often, but incorrectly, been associated with the phenomenon described as contact inhibition of cell division is contact inhibition of movement (Abercrombie and Heaysman, 1954). For a review of the latter property of nontransfonned cells, see Abercrombie (1970). In an interesting observation related to the reinitiation of movement in a contact-inhibited cell population, Weston and Roth (1969) demonstrated that treatment of a quiescent population of fibroblasts with low concentrations of urea increased the nuclear overlap ratio in the ureatreated culture relative to the untreated control, thereby suggesting that contact inhibition of movement had been overcome by the mild urea treatment. In an extension of this work, Weston and Hendricks (1972) demonstrated that incubation of a quiescent population of 3T3 cells for 24 hours with 200 mM urea in nutrient media containing 10% calf serum produced both an increase in nuclear overlap (Lea,a reduction in the degree of contact inhibition of movement displayed by the culture) and an increase in E3H1Tdr incorporation into DNA (i.e., a ureamediated stimulation of cell division). Weston and Hendricks (1972) noted that, when following urea treatment medium containing 10% calf serum was added to the treated cultures, the nuclear overlap ratios regressed and the cells slowly

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

43 1

reverted to a state resembling that of untreated cultures. Weston and Hendricks (1972) also noted that addition of cycloheximide to the nutrient medium to which the urea-treated cultures were returned prevented the reestablishment of contact inhibition of movement, suggesting that a protein(s) was removed from the cells by the urea treatment and that the cells could not reestablish contact inhibition of movement without resynthesizing the protein(s). On the basis of this evidence, Weston and Hendricks (1972) reasoned that the urea-extracted protein(s) should be present in the “urea extraction medium” (CMu) with which the cells had been treated. They therefore added dialyzed CMu to urea-stimulated cells and found that it rapidly (within 4 hours of addition) reestablished contact inhibition of movement in the treated cell population. This finding clearly suggested that the mild urea extraction removed component(s) from the cell which were essential to the maintenance of contact inhibition of movement and that, under appropriate conditions, these components could be restored to the cell, with a subsequent and rapid establishment of contact inhibition of movement. Using the urea extraction media as a source for cell component(s) purportedly responsible for maintaining both contact inhibition of movement and DDI of growth, Yamada and Weston (1974) subsequently showed that the major component removed from the cell by the urea treatment was an iodinateable surface peptide with a MW of 220,000. This component has since been shown to be the CSP (or the LETS protein, Yamada and Weston, 1975). In the initial fractionation of CMu (Yamada and Weston, 1974), it was demonstrated that components other than the CSP were extracted from the cell by mild urea treatment. Yamada and Weston (1974) suggested that the CSP alone could not restore both contact inhibition of movement and DDI of growth in urea-stimulated 3T3 cells. Rather they suggested that the minor cell components extracted by the urea treatment were essential to the reestablishment of contact inhibition of cell division, while the CSP might be sufficient to reestablish contact inhibition of movement in urea-treated cell cultures. Subsequently, Yamada and Weston (1975) demonstrated that metabolically labeled CSP could adsorb to the cell surface of ureaextracted cells and that the adsorbed glycopeptide could be removed from the cell surface by mild proteolysis. These data suggested that the effect(s) of the CSP on the growth phenomena investigated were related to CSP association with the cell surface. In later work, employing purified CSP (80-100% of the protein being the 220,000-MW surface glycopeptide), Yamada et d.(1975) demonstrated that the CSP

432

KENNETH D. NOONAN

could act as an agglutinin, agglutinating Formalin-treated erythrocytes, CEFs, and NRK cells. The latter work suggested to Yamada et al. (1975) that the CSP might exert its biological effect by modifying cell-cell or cell-substratum adhesion. In the most recent work with highly purified CSP, Yamada et al. (1976) demonstrated that the addition of 50 pg/ml CSP to cultures of SV4O-transformed 3T3 cells (SVT2) altered the morphology of the cells, causing them to flatten, elongate their cellular processes, and arrange themselves in parallel arrays, reminiscent of normal cell lines displaying the phenomenon of contact inhibition of cell movement. These data also suggest that addition of the CSP to the SVTS cells increases cell-substratum adhesion. In the work of Yamada et al. (1976) a fourfold reduction in SVTS nuclear overlap was observed after the addition of 50 pg/ml CSP to the SVTS cultures. However, using this highly purified CSP fraction, Yamada et al. (1976) found no effect of as much as 100 pg/ml CSP on the growth rate or on the final saturation density of the SVTS cells. These data suggest that the dialyzed CMumediated restoration of contact inhibition of cell division observed by Weston and Hendricks (1972) may have been produced by one of the minor contaminants of the extract and not by the major CSP peptide itself. Yamada et al. (1976) suggest that the effect of the CSP on contact inhibition of movement might be explained if the CSP could bind directly to both the cells and the substratum, thereby increasing the relative adhesiveness of the substratum. Several immunofluorescent studies have demonstrated that the CSP can bind to the substratum as well as to the cells (Wartiovaara et nl., 1974; Hynes et al., 1976). In support of the hypothesis of Yamada et al. (1976), Carter (1965) and Harris (1973) both showed that the relative adhesiveness of the substratum can modify the behavior of cells with regard to contract inhibition of movement. Thus the accumulated data with regard to the CSP (the LETS protein) suggests that it may be involved in cell-cell (Chen et d., 1976a) and cell-substratum adhesion and as such may play a role in determining contact inhibition of movement. However, as reported in Section V,B, there is no evidence that the CSP plays a role in determining contact inhibition of cell division.

VI.

PROTEASE-INDUCED TRANSMEMBRANE EVENTS

If proteases do act to stimulate cell division by preferentially modifying one or more cell surface peptides, it seems reasonable to as-

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

433

sume that this extracellular event might result in the modification of one or more intracellular components. In this regard various attempts have been made to identify a protease-induced intracellular event which might be directly involved in the initiation of cell division. In effect, several investigators have sought evidence for the proteaseinduced transduction of a signal across the membrane which might be similar to the intracellular events noted following peptide hormone binding to the cell surface. A. Proteare Modification of Intracellular CAMP Concentrations

It is generally considered to be proved that normal cells display higher intracellular concentrations of cAMP than their transformed counterparts. The reduced cAMP concentration of transformed cells has been reported to be correlated with (and possibly responsible for) their increased growth rate of growth and, in some cases, with their lack of DDI of growth (e.g., Sheppard, 1971, 1972; Otten et al., 1971, 1972a,b; Pastan et al., 1974; Sheppard and Bannai, 1974). This finding, together with several other apparent correlations between various properties ascribed to the transformed phenotype and lowered cAMP concentrations (Johnson et ul,, 1972, Johnson and Pastan, 1972a,b), led Pardee et al. (1974) to suggest that the protease treatment of nondividing cells which is sufficient to drive them from the quiescent to the proliferative state might be a “transitory, probablistic event involving a decrease of cAMP below a critical level” necessary for the maintenance of cellular quiescence. If the hypothesis of Pardee et d.(1974) is correct, one would expect that mild protease treatment of quiescent cells, as discussed in Section IV, would trigger a rapid drop in intracellular cAMP concentration which would eventuate in a new round of cell division. Sheppard (1972) was the first to demonstrate that treatment of contact-inhibited 3T3 cells with 0.05%trypsin for 5 minutes produced a 2.5-fold drop in intracellular cAMP concentration. This drop was demonstrated to be a transitory event, and the intracellular levels of cAMP returned to the pretrypsin levels 3 hours after treatment. The same transitory drop in intracellular cAMP concentration was reported to occur when the 3T3 cells were treated with insulin (80 mU/Ml) or serum (50%),agents which also induced anew round of cell division (Sheppard, 1972). Burger et ul. (1972), in support of Sheppard’s initial findings, observed a fivefold drop in intracellular cAMP concentration following a 2-minute treatment of quiescent 3T3 cells with 0.05% trypsin. Simi-

434

KENNETH D. NOONAN

larly, a 2.5- to 3-fold drop in cAMP concentration was observed after incubation of quiescent 3T3 cells with a host of other proteases (e.g., chymotrypsin, ficin, papain, or subtilsin). Otten et aZ. (1972b) reported that mild trypsinization of 3T3-42 cells produced a 10-fold drop in intracellular cAMP concentration within 10 minutes after treatment. Like Sheppard (1972) and Burger et aZ. (1972), Otten et al. (1972b) found that this change in intracellular cyclic nucleotide concentration was transitory, with &e intracellular concentration returning to its initial level 60 minutes after protease treatment. Taken together, these data suggested that agents which stimulate confluent normal cells to resume cell division might act by decreasing intracellular cAMP levels. It was assumed by many workers that this reduction in the intracellular cAMP concentration might be the signal directly responsible for initiating a new round of cell division. If a transitory decrease in cAMP concentration was responsible for protease induction of cell division, one would expect that the maintenance of high intracellular cAMP concentrations following trypsin treatment of the cells would prevent a new round of cell division. In support of this hypothesis, Bombik and Burger (1973) reported that, when a 4-day-old stationary culture of 3T3 cells was treated with 10 pg/ml pronase and immediately incubated in DME plus 10% calf serum plus 1 X M dibutyryl cAMP (dBcAMP), the induction of a new round of cell division was completely prevented. Interestingly, Burger et aZ. (1972), as well as Bombik and Burger (1973), found that the exogenous dBcAMP had to be added to the 3T3 cells within 3 minutes after protease treatment. These data suggested that, under the proper growth conditions, the drop in intracellular cAMP concentration was a trigger which initiated a new round of cell division, and that once the trigger was “pulled” the cells were committed to completing a new round of cell division. How protease treatment of the cell surface acts to reduce intracellular cAMP concentrations is still unknown. It has been suggested that the protease-induced modification of the cell surface acts t a inhibit the membrane-associated adenyl cyclase, to increase intracellular phosphodiesterase activity, or simply to increase the “leakiness” of the cell to small molecules such as CAMP. Despite the attractiveness of the cAMP trigger mechanism as an explanation for protease induction of cell division in a quiescent cell population, unpublished work from our laboratory shows that a reduction in intracellular cAMP is not sufficient in itself to induce a new round of cell division. Specifically, using a 3T3 cell clone requiring relatively high levels of serum in order to complete a round of

PROTEOLYTIC MODIFICATION

OF CELL SURFACE MACROMOLECULES

435

protease-stimulated cell division, we demonstrated that protease treatment of these cells reduced the intracellular cAMP concentration regardless of the serum concentration in the medium. Thus, for example, a 3T310clone (see Section IV) treated with pronase and then returned to conditioned medium, displays a 5- to 10-fold drop in intracellular cAMP concentration but does not resume DNA synthesis or complete a new round of cell division. Thus, under these culture conditions, a reduction in intracellular cAMP is not sufficient to induce a new round of cell division in this particular 3T310subclone. Despite evidence that a reduced intracellular cAMP concentration is not sufficient to induce a new round of cell division there have been no experiments, to our knowledge, which eliminate the possibility that a drop in intracellular cAMP Concentration is necessary for proteasemediated induction of cell division. B. Protease-Mediated Modification of Intracellular Actinlike Cables

Using fluorescent antiactin antibodies, Lazarides and his coworkers (Lazarides and Weber, 1974; Weber et al., 1974; Goldman et al., 1975) demonstrated that most, if not all, nontransformed cells have actinlike cables which course throughout their cytoplasm and apparently play a role in determining and maintaining cell shape and morph ol ogy. The presence of these well-defined, well-oriented actin cables in the cytoplasm appears to be a property of the nontransformed phenotype, since the transformed cell shows a diffuse, nonoriented distribution of fluorescent antiactin which can be easily distinguished from the antiactin distribution observed in the normal cell (e.g., Weber et al., 1974).Recently, it has been reported that the presence or absence of oriented actin cables in a cell’s cytoplasm can be correlated with the release of plasminogen activator (PA) (Section VIII) by a c e l l In turn, the relative amount of PA secreted by a cell may be associated with the ability of that cell to grow in suspension and produce tumors when injected into nude (nulnu)mice (Pollack et al., 1974; Shin et al., 1975). However, there now appears to be some disagreement regarding the latter point (Jones et al., 1975; Wolf and Goldberg, 1976). In an interesting article, Pollack and Rifkin (1975) demonstrated that secondary cultures of nontransformed REFS grown in plasminogen-free sera display fully extended, well-oriented actin cables which could be easily detected with fluorescent antiactin. Pollack and Riflcin

436

KENNETH D. NOONAN

(1975) showed that, when these cells were treated with purified dog plasminogen plus purified urokinase (thereby generating the active protease, plasmin), the percentage of cells displaying fully extended actin cables dropped within 3 hours of treatment from 90%to less than 20% of the cells in the population. Upon removal of the plasminogen from the growth medium, the cells gradually regenerated and reoriented their intracellular actin cables. These data suggested that the activity of a serine protease on the secondary REFs could modify the intracellular distribution of the actin cables. This finding of an effect of plasmin on the cytoskeleton was further strengthened by the fact that treatment of R E F s (grown in plasminogen-free sera) with 5 pg/ml trypsin for 15 minutes removed the detectable actin cables from -50% of the cells. Incubation of the cells with the same concentration of trypsin for 2 hours removed the cables from approximately 100%of the cells. Again dissolution of the actin cables was completely reversible, with 100% of the cells regaining their fully extended cables within 2 hours after the addition of fresh serum to the cells. Both chymotrypsin and thrombin were also capable of removing actin cables from the REFs, although both these proteases were less efficient in this regard than trypsin or plasmin. In order to demonstrate that the various proteases acted to dissociate the cables by modifying surface structures, rather than by entering the cell and actively hydrolyzing the cables, R E F s were incubated for 1 hour with insoluble trypsin. Pollack and Rifkin (1975)reported that this trypsin, which apparently was not phagocytosed by the REFs, also removed the actin cables from -90% of the cells. These data clearly suggest that proteolytic modification of the cell surface induces an intracellular event which is observable with fluorescent antiactin antibodies as dissociation of the actin cables. Whether this dissociation occurs because the actin attachment sites on the cell surface are perturbed or because the actin cables themselves [which Perdue (1973) demonstrated may penetrate the membrane] are modified by proteolysis is unknown. However, if the former hypothesis is correct, these data could be taken as further evidence for a transmembrane event induced by proteolytic modification of the cell surface. Whether or not dissociation of the intracellular actin cables plays a role in protease-induced cell division is simply unknown. Although it is conceivable that such an intracellular modification is necessary for the induction of cell division, it seems unlikely to us (see Section IV) that such an intracellular event is sufficient in itself for the induction of a new round of cell division.

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

VII.

437

LIMITED AUTOLYSIS AS A MECHANISM FOR INDUCING CELL DIVISION

One hypothesis which has gained some favor in regard to protease stimulation of cell division proposes that a transformed cell produces proteases which act on its own membrane or on that of a neighboring cell, consequently modifying the cell surface structure and thereby maintaining the transformed cells in a proliferative state. From this hypothesis comes a suggestion that the addition of exogenous protease to a nontransformed cell culture performs a function which transformed cells perform for themselves; i.e., the exogenous protease modifies the surface structure of the normal cell, consequently moving the cell from the quiescent to the proliferative state. The idea that exogenous proteases may mimic proteases produced by transformed cells has been discussed at length by Burger (Burger, 1971; Talmadge et d.,1974). Burger has suggested that membraneassociated proteases, forming a cascading series, may be activated by the exogenous protease and have to act in concert to produce a change in the normal cell surface structure, which is interpreted by the cell as a signal to reinitiate cell division (Talmadge et d.,1974). The fact that malignant cells produce more or different proteases than normal tissue has been known for many years. It has frequently been suggested that these proteases are responsible for or involved in neoplasia (see, e.g., Sims and Stillman, 1937; Fisher, 1946). However, it has only been within the last 5 years that several workers have suggested that transformed cells produce, or at least secrete, higher levels of neutral proteases than their normal counterparts. Burger (1971) first demonstrated that the malignant, ascitic L1210 cell line, when plated onto a quiescent population of 3T3 cells, produced a doubling in the number of 3T3 cells. Since this malignant cell-mediated response was inhibited when the L1210 cells were first incubated with a synthetic protease inhibitor (TAME or TPCK), Burger suggested that the L1210 cells secreted a protease(s) which acted on the 3T3 cells, thereby inducing a new round of cell division. Various investigators have presented data suggesting that transformed cells secrete more neutral proteases than normal cells. Schnebli (1972, 1974), using either 14C-labeledChZoreZZa hydrolysate or [3H]TAME as substrate, demonstrated that transformed 3T3 cells secreted more hydrolytic (proteolytic) enzymes than their nontransformed counterparts. Bosmann (1972),using 3H-labeled acetylated hemoglobin as substrate, demonstrated that both a trypsinlike and a

438

KENNETH D. NOONAN

cathepsinlike activity was elevated in Moloney sarcoma virus (MSV)and RSV-transformed 3T3 cells. Both Schnebli and Bosmann suggested that the elevated proteolytic activity of the transformed cells leads to sublethal self-autolysis which in turn might account for the differences in surface structure observed in normal and transformed cells (e.g., enhanced agglutinability of the transformed cells by plant lectins). If intracellular or membrane-bound proteases did play a role in maintaining the altered surface structure of transformed cells, it was argued that the incubation of transformed cells in medium containing synthetic protease inhibitors might alter both the surface architecture and the growth properties of a transformed cell line. Schnebli and Burger (1972) reported that the addition of TLCK, TPCK, TAME, or Trasylol (a pancreatic protease inhibitor) to cultures of SV40 or Py3T3 cells reduced the final cell density to which the transformed lines grew without affecting the growth of the parental, nontransformed 3T3 cell line. Significantly, Schnebli and Burger (1972) reported that “TLCK [50 pg/ml] treatment of (Swiss) Py3T3 cells renders these cells less agglutinable with WGA [8fold] and with Con A [Sfold], indicating that this [proteasel inhibitor prevents some of the surface modifications characteristic of transformed cells.” Goetz et al. (1972) subsequently produced data supporting Schnebli and Burger’s report that synthetic protease inhibitors could reduce the final cell density to which transformed BHK cells could grow. Goldberg (1974a) reported that, like Schnebli and Burger (1972), he had found that incubation of transformed cells in medium containing protease inhibitors reduced the agglutinability of the transformed cell lines under consideration. In a subsequent study Weber (1975) investigated the effect of TLCK (50 pg/ml) on three phenotypes associated with transformation in RSV CEFs, namely, increased rate of hexose transport, rounded cell morphology, and decreased adhesion of cells to the substrate. Weber (1975) reported that TLCK was very effective in changing the transformed CEFs to a more flattened shape, increasing cell-substratum adhesion and, over a 60-hour incubation, reducing the rate of 2-dG uptake to that seen in nontransformed cells (Section VIII). Despite the enthusiasm with which the initial data relating to protease inhibitors and their effect on transformed cell growth was received, they were quickly disputed. Chou et al. (1974a,b), as well as McIlhinney and Hogan (1974), reported that they could not find a TPCK-mediated effect on transformed cell growth which was not also observed in nontransformed cell lines. Thus, unlike Schnebli and Burger (1972), these two groups could find no evidence that the syn-

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

439

thetic protease inhibitor preferentially acted on the transformed cell type. More important to the interpretation of the data of Schnebli and Burger (1972) was the finding of Chou et al. (1974a,b) that incubation of SV3T3 cells with 100-200 pg/ml TPCK produced a 60-70% inhibition of protein synthesis. To support their hypothesis that the “growth plateau” observed by Schnebli and Burger (1972) resulted from the inhibition of protein synthesis rather than from the inhibition of intracellular proteolytic activity, Chou et al. (1974a,b) demonstrated that the TPCK-mediated inhibition of SV3T3 cell growth could be mimicked by the addition of low concentrations of cycloheximide to the growth medium. Subsequently, Pong et al. (1975),working with HeLa S, cells as well as with a transformed derivative of 3T3 and a myeloma cell line (MPC-ll), demonstrated that 20-30 pg/ml TPCK was a potent inhibitor of protein initiation. In later work Noonan and Noonan (1977) showed that TLCK, at the concentrations used by Schnebli and Burger to inhibit transformed cell growth, is a potent inhibitor of cellular RNA synthesis. These three groups (Chou et al., 1974a,b; Pong et al., 1975; Noonan and Noonan, 1977) all concluded that the apparent growth plateau observed by Schnebli and Burger (1972) was produced by a dynamic equilibrium between cell multiplication and cell death produced by an inhibition of protein or RNA synthesis. Support for the concept that a dynamic equilibrium between cell multiplication and cell death was responsible for the growth plateau observed by Schnebli and Burger came from the work of Schnebli and Hasmmerli (1974), who reported that SV3T3 cells, which were maintained as a monolayer by incubation in medium containing TLCK, continued to incorporate [3H]Tdr into their DNA despite the fact that they displayed a much reduced mitotic index. Cytophotometric data derived by these workers suggested that most of the “quiescent” TLCK-treated SV3T3 cells were in either the S or the G, phase of the cell cycle rather than the G , phase. In summary then it is our strong conviction that most, if not all, of the work relating to the use of soluble, synthetic protease inhibitors to modify cell growth and cell surface structure must be reconsidered. Clearly, this approach to elucidating the role of intracellular proteases in cell growth and cell surface architecture must, as Chou et al. (1974a) have pointed out, await the development of protease inhibitors which do not have adverse side effects on cellular metabolism. Until this stringent requirement is met, the use of soluble protease inhibitors to modify cell growth or cell surface structure will remain a bankrupt approach to the questions raised throughout this chapter.

440

KENNETH D. NOONAN

A. Evidence for Surface-Localized Proteases

As indicated in the previous section, the use of soluble, synthetic protease inhibitors to modify cell growth and surface architecture has proved to be a relatively fruitless approach in the attempt to evaluate the role of cell proteases in the mediation of these phenomena. One approach which offers some hope with regard to the use of protease inhibitors to modify the transformed phenotype is that taken by Talmadge et al. (1974). These workers bound ovomucoid (a relatively nonspecific protease inhibitor) to an insoluble polymer (Biogel P-10) prior to adding the inhibitor-polymer complex to cultures of Py3T3 cells. The advantage of this approach over that employing soluble protease inhibitors is that an effect of the protease inhibitor on cell growth can be at least tentatively ascribed to the ovomucoid-mediated inhibition of a surface-localized protease. In their study Talmadge et al. (1974) demonstrated that addition of the ovomucoid-polymer complex to Py3T3 cultures reduced the final cell density to which the cells grew by 60% without affecting the growth of the parental 3T3 cells. The difficulty which arises in interpreting these data, however, is that incubation of serum-containing medium with the ovomucoid beads prior to addition of the medium to the Py3T3 cells reduced the final cell density to which the Py3T3 cells grew by 20%, suggesting that only -40% of the inhibition of Py3T3 cell growth observed with the ovomucoid beads could even be tentatively attributed to surface proteases. Furthermore, it is distinctly possible that the ovomucoid beads modified Py3T3 cell growth by inhibiting a secreted rather than a membrane-bound protease. Thus, although this approach seems to be more likely to succeed than an approach employing soluble protease inhibitors, difficulties still exist with regard to interpretation of the data obtained. Better evidence that at least some proteases are surface-associated comes from the fluorescent antibody work of Sylven et al. (1974). These investigators demonstrated, using fluorescence microscopy, that an antibody prepared against a cathepsin-B, activity can be localized to the cell surface of a variety of transformed cell lines. More recently, Spataro et al. (1976) demonstrated that, in the absence of sera (see Section VII,B), quiescent RSV CEFs hydrolyze 3Hlabeled acetylated hemoglobin at a 6.3-fold higher rate than normal CEFs. These workers suggest that this hydrolytic activity is a neutral, surface-associated protease. Their evidence for a surface association is simply that no trypsinlike activity could be recovered from the medium which sustained the RSV CEFs. These investigators suggest, on

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

441

the basis of whole-cell electrophoretic data, that the RSV CEFassociated protease(s) is responsible for maintenance of a high negative charge on the transformed relative to the normal cell surface. However, the evidence supporting this latter contention is, in our opinion, weak. Hatcher et al. (1976, 1977), using [3H]casein as a substrate for cells maintained in a non-serum-containing medium, reported the presence of nonsecreted, surface-localized neutral protease(s) which do not appear to be related to plasmin or PA. These investigators noted that the proteolytic activity they investigated was three to four times more active on transformed as compared to normal cells. Furthermore, their data indicate a correlation between the doubling time of a cell line and the relative activity of the protease(s), the lower activity being associated with slower growth. Most recently (Hatcher et al., 1977), these workers have reported a significant reduction in proteolytic activity in density-inhibited normal fibroblasts relative to that of proliferating normal fibroblasts. These data have led them to propose that this neutral protease activity may be one of the elements controlling proliferation of both normal and transformed cells. Thus there is some evidence, albeit weak, to support the hypothesis that at least some proteases are surface-associated. Whether or not any of the putative protease activities described in this section play a role in the type of sublethal autolysis discussed in the previous section is open to question.

8. Plasrninogen Activator

The best characterized neutral serine protease both produced and secreted by cells growing in culture is PA which has been extensively studied by Reich and his collaborators (Unkeless et al., 1973; Ossowski et al., 1973b), Goldberg (1974b), and others (Christman and Acs, 1974). This serine protease hydrolyzes serum plasminogen to plasmin. The plasmin in turn apparently acts on the cells, producing a variety of effects, many of which have been previously defined as traits of the transformed phenotype. Since this protease is primarily secreted by transformed cells (Unkeless et al., 1973; Ossowski et al., 1973b), it has been considered a likely candidate for a protease which could carry out the type of limited self-autolysis described above. Although a review of the work on PA is clearly beyond the scope of this chapter, a summary of its most prominent features follows immediately. This brief review is derived, in part, from Quigley (1976).

442

KENNETH D. NOONAN

1. The levels of PA in the culture medium are substantially higher in media derived from transformed cell cultures as compared to those from normal cell cultures (Unkeless et al., 1973; Ossowski et al., 197313). 2. The appearance of PA in the culture medium is closely correlated with, and in fact may be, an early event in malignant transformation (Pollack et al., 1974; Unkeless et al., 1974). 3. The total PA activity is elevated 10-to 100-fold (relative to that in normal cells) in tumor virus-transformed primary cultures, in primary cultures of tumors, in chemically induced tumors, and in tumor cell lines (Lang et al., 1975; Riain et al., 1974; Yunis et al., 1974). 4. Under physiological conditions PA functions at neutral pH and, since it is actively secreted by transformed cells, should have catalytic access to the cell surface (Unkeless et al., 1973, 1974; Ossowski et al., 1973b). 5. PA activates serum plasminogen to plasmin, thereby generating additional proteolytic activity in serum-supplemented cultures (Ossowski et al., 1973a; Quigley et al., 1974; Unkeless et al., 1974). 6. Enhanced PA activity has been associated with the cell’s ability to grow in suspension and to produce tumors in nude (nulnu) mice (Pollack et al., 1974). [Recent evidence from Goldberg’s laboratory (Wolf and Goldberg, 1976), as well as from Jones et al. (1975), however, has questioned the apparent correlation between elevated PA activity and growth in suspension.]

With regard to the main topic of this chapter the primary question to be asked about PA activity is whether PA itself or the plasmin generated by PA in serum-containing cultures can modify the cell surface structure and/or induce a new round of cell division. In support of these points, Blumberg and Robbins (1975a) demonstrated that 1.0 Fg/ml human plasminogen, hydrolyzed to plasmin by streptokinase, removed the LETS protein from the cell surface of quiescent CEFs and produced a 1.8-fold increase in cell number. Furthermore, Blumberg and Robbins (1975a) have reported that the addition of a partially purified PA (the so-called crude harvest fraction) to serum-containing cultures of quiescent CEFs produced an increase in 2-dG uptake (see Section VIII), although it did not cleave the LETS protein from the cell surface. Thus the work of Blumberg and Robbins (1975a) suggests that PA-generated plasmin can modify surface structure(s) and induce cell division. This work also provides some weak evidence that PA itself might act on the cell to modify surface structure or induce cell division.

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

443

If PA itself is to act as an autolytic agent, then one might expect PA to be associated with the plasma membrane. Early reports (Unkeless et al., 1974; Christman et al., 1975) had suggested that PA was either bound to membranes or, at least, granules. Most recently, Quigley (1976) demonstrated that the PA activity of RSV CEFs copurified with a variety of plasma membrane markers (e.g., Na+,K+-ATPase, 5’nucleotidase, and [3H]fucose).In his study Quigley (1976) found that 60% of the cellular PA activity in RSV CEF cells copurified with a plasma membrane fraction which comprised < 8% of the total cellular protein. The remainder of the PA purified with other membranous components of the cells, in particular a putative Golgi apparatuscontaining fraction. Quigley’s (1976) work does not of course rule out the possibility that PA is a secreted protease tightly bound to the membrane rather than an integral membrane protein. Nevertheless, the observation that a significant portion of the cell’s PA copurifies with plasma membrane raises the possibility that this neutral serine protease could have pronounced effects on the cell surface properties of transformed cells. However, in light of the very limited substrate specificity of PA, it has been considered more likely the PA-activated plasminogen (plasmin) from the growth medium is the protease directly responsible for the surface changes detected in transformed cell cultures. Nevertheless, Quigley (1976) is quite correct in suggesting that, although “plasminogen is the only known substrate of PA, this does not preclude the possibility that other proteins within the microenvironment of the cell surface may also be substrates for . . . PA.” Chen and Buchanan’s (1975) demonstration that a proteolytic activity which is assayed with PA but which is not plasminogen-dependent is elevated in transformed as compared to normal cells increases the possibility that an endogenous proteolytic activity of PA might be responsible for at least some of the surface changes detected in transformed as compared to normal cells. One should be able to determine whether it is plasmin, PA, or both which are responsible for surface modifications associated with the transformed phenotype by growing transformed cells in plasminogendepleted serum. If PA itself is capable of producing surface modifications, one might expect these modifications to occur in the absence of plasminogen. In this regard it is interesting to note that Weber (1975) reported that the addition of inhibitors of fibrinolysis to the growth medium prevented transformed cells from rounding and the loss of cell-substratum adhesion but did not affect enhanced 2-dG transport. Thus it is at least possible that PA itself [or a proteolytic activity mea-

444

KENNETH D. NOONAN

sured with PA (Chen and Buchanan, 1975)l could be responsible for maintaining the elevated 2-dG transport observed in the transformed cell, VIII. ROLE MEDIA COMPONENTS PLAY IN PROTEASE-STIMULATED CELL DIVISION

In two papers published 8 years apart, Pardee (1964) and Holley (1972) both suggested that malignant cell growth, as well as in uitro transformed cell growth, could be influenced by, if not determined by, the availability to the cell of low-MW nutrients (e.g., sugars and amino acids). Holley (1972) has argued that “the growth of normal cells in uiuo (and in uitro) is controlled by hormones or growth factors [present in the growth media] that influence the uptake or availability, inside the cell, of the specific (low MW) nutrients that in turn regulate cell growth.” Holley (1972)and Pardee (1964)have gone on to suggest that “cancer (or transformation) results from changes in [the cell’s] uptake mechanisms caused by changes in the cell membrane.” Thus it has been suggested that transformed cells grow to a higher cell density than normal cells because they possess a surface architecture adapted to a more efficient uptake of nutrients. By following this line of reasoning it can be argued that transformed cells should be able to maintain cell growth in media in which the nutrient levels are limiting to normal cell growth. There is ample evidence in the literature suggesting that transformed cells are more efficient in amino acid and sugar uptake than normal cells. In one of the earliest studies on amino acid transport, Foster and Pardee (1969) showed that a-aminoisobutyric acid (a-AIB, a nonmetabolizable amino acid analog) was taken up at a three-fold higher rate by Py3T3 cells as compared to 3T3 cells. This enhanced a-AIB uptake by Py3T3 cells was demonstrated to be related to a change in the V,,, of the a-AIB transport protein. Subsequently, these findings were confirmed by Isselbacher (1972) using PyBHK, SV3T3, and their normal cell counterparts. More work has been directed at correlating changes in 2-dG uptake with transformation. In a relatively early report Hatanaka et al. (1970) demonstrated that MSV-transformed mouse embryo cells take up 2dG at a faster rate than nontransformed mouse embryo cells. In this work the enhanced uptake of 2-dG was related to a modification in both the V,,, and K m of the 2-dG transport protein. Martin et ul.

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

445

(1971), using CEFs infected with a ts mutant of RSV (T5), showed that the enhanced rate of 2-dG uptake was apparent at the permissive (transformed) but not at the nonpermissive temperature. Subsequent work with RSV t s mutant-infected CEFs by Hatanaka and Hanafusa (1972), as well as by Kletzien and Perdue (1974a,b,c), conclusively demonstrated that the enhanced rate of 2-dG uptake was observed only at the temperature permissive for transformation and was related to a change in the V,,, and not in the K, of the 2-dG transport protein. These data relating enhanced metabolite uptake to the transformed state add support to Pardee’s (1964) and Holley’s (1972) hypothesis that it may be the increased efficiency of metabolite uptake which is responsible for the transformed phenotype characterized by loss of DDI of growth. If one accepts this hypothesis, then it follows that proteolytic modification of the cell surface of quiescent cells which is capable of inducing a new round of cell division should modify the rate of nutrient uptake. In at least partial support of the latter statement, Sefton and Rubin (1971) clearly showed that, whereas a 13.4-fold drop in 2-dG transport occurred as CEFs reached confluency and subsequent quiescence, treatment of the quiescent cells with a 4 pg/ml trypsin for 5 minutes produced a 5- to 9-fold increase in 2-dG uptake, observable between 2 and 6 hours after trypsin treatment. Similar increases in 2-dG uptake by quiescent CEFs and 3T3 cells treated with trypsin were reported by Bradley and Culp (1974), as well as by Vaheri et al. (1972). In all instances the enhanced rate of 2-dG uptake was related to a 2.5- to 4fold increase in the V,,, for 2-dG uptake. Although much less work has been done on protease-stimulated uptake of small metabolites, Sefton and Rubin (1971) have reported that brief protease treatment of quiescent CEFs produced little or no change in the uptake of a-AIB, Tdr, or uridine, thus suggesting that the enhanced uptake of 2-dG was not related to nonspecific stimulation of the uptake of all low-MW metabolites. It is possible that enhanced 2-dG uptake is a necessary prerequisite for protease-stimulated cell division, however, it is not likely that enhanced sugar uptake is sufficient to induce a new round of cell division. This is clear from our unpublished data demonstrating that 3T310 cells treated with trypsin and then restored to nonideal culture conditions (see Section IV) displayed enhanced 2-dG uptake but never entered the S phase of the cell cycle. This point is further supported by Sefton and Rubin’s (1971) report that treatment of quiescent CEFs with trypsin for 5 minutes was sufficient to induce a three-fold increase in the rate of 2-dG uptake, even though such an abbreviated

446

KENNETH D. NOONAN

protease treatment was not sufficient (see Section IV) to induce a new round of CEF cell division. Role of Serum Factors in Protease-Stimulated Cell Growth

Besides the low-MW nutrients discussed in the previous section, it is clear that in vitro growth is modified, and to some extent deter-

mined, by the availability to the cell of the so-called growth factors in the serum. At least seven different serum factors (reviewed by Holley, 1974)have been identified which are apparently essential for efficient 3T3 cell growth. Other small peptide factors [including epidermal growth factor (EGF), fibroblast growth factor (FGF), insulin, and somatemedin] have also been implicated in the maintenance of nontransformed cell growth. Most recently, interest in various serum factors has centered almost exclusively on FGF, described in detail by Gospodarowicz (1975, 1976; Gospodarowicz and Moran, 1974).This factor, which is isolated from mammalian brain and pituitary, has been shown to stimulate cell division in a wide variety of cells including fibroblasts, chondrocytes, myoblasts, smooth muscle cells, and several adrenal cell types. However, cells of endodermal or ectodermal origin are apparently not sensitive to induction of cell division by FGF. As a result of much work developed over the last 5 years, Holley (1974)has proposed that “changes in the cell membrane are probably responsible for malignant [or transformed cell] growth.” Furthermore, Holley has suggested, quite aside from his earlier proposal (Holley, 1972), that changes in the cell membrane can alter the response of a cell “to all of the many different [serum] factors that are observed to control growth in culture” (Holley, 1974). Holley (1974) and others have thus suggested that the cell surface architecture of the transformed cell makes it more efficient in the uptake of serum factors from the growth medium, and that it is this uptake of serum factors which permits transformed cells to grow to a higher cell density than their normal counterparts. Given this hypothesis and the data which support it, the question immediately arises as to whether or not protease modification of the quiescent, nontransformed cell surface acts to increase the efficiency of serum factor uptake or utilization and thus to promote a new round of cell division. Circumstantial evidence supporting such an hypothesis has been discussed in Section IV (Noonan, 1976). The most interesting system presently available for testing this hy-

PROTEOLYTIC MODIFICATION

OF CELL SURFACE MACROMOLECULES

447

pothesis is protease stimulation of CEFs. This is particularly true, since Temin and his collaborators (e.g., Pierson and Temin, 1972; Dulak and Temin, 1973; Temin et al., 1974) have identified, isolated, and purified an insulinlike activity from serum which apparently plays a major role in maintaining cell division in CEFs. This factor [the multiplication stimulating factor (MSA)] has been characterized as a family of small polypeptides, all with a MW < 10,000 but different isoelectric points. The function of MSA [a nonsuppressible insulinlike activity (NSILA)] seems to be the initiation and maintenance of a cellular process which in CEFs begins in GI and eventually culminates in DNA synthesis. Thus MSA may act as a prod, driving CEFs out of GI and into S. The demonstration that MSA is an insulinlike activity is particularly interesting in light of the various reports that quiescent 3T3 cells (Bombik and Burger, 1973), BHK-21 cells (de Asu6 et al., 1973), human fibroblasts (Gavin et al., 1972),and chick embryo cells (Temin et al., 1974)can be induced to reinitiate cell division following the addition of insulin to the growth medium. That insulin acts to stimulate cell division by binding to specific receptors on the cell surface has been suggested by Gavin et al. (1972),as well as Hollenberg and Cuatrecasas (1975), who found that [12511insulinbinds to secondary cultures of human skin fibroblasts with a specificity similar to that observed in the binding of insulin to isolated liver membranes. With regard to the material presented in this chapter, the most interesting data published to date relating to insulin binding to quiescent cells, and a possible role for this binding in the stimulation of growth, is the recently reported work of Raizada and Perdue (1976). These workers have reported that substratum-attached CEFs display two classes of insulin-binding sites, one being a high-affinity, highcapacity site (K, = 2-6 x 108per mole) and the other a low-affinity, low-capacity site (K, = 0.8-3.0 x lo' per mole). Furthermore, Raizada and Perdue (1976) demonstrated that growing RSV-transformed CEFs bind 1.5 to 2 times more insulin than normal fibroblasts at the same cell density. This difference in insulin binding is accounted for primarily by differences in the V,,, of insulin binding to the low-affinity sites. These findings have been supported by the demonstration that CEFs infected with a t s mutant of RSV (T5) bind 3 times more insulin at 37°C (the permissive temperature) than they do at 41°C (the nonpermissive temperature). In as yet unpublished work, Raizada and Perdue (1978) showed that, whereas uninfected avian fibroblasts reduced the number of available insulin-binding sites fourfold as they approached quies-

448

KENNETH D. NOONAN

cence, RSV-transformed CEFs did not significantly reduce the number of insulin-binding sites at confluence. Thus RSV CEFs bound two to three times more insulin at confluence than uninfected CEFs. Taken together, these findings have led Raizada and Perdue (1976, 1978) to suggest “that the decrease in replication with an increase in the density of the culture under conditions of an adequate level of serum may be explained by a cell contact-dependent decrease in the number of exposed [insulin] receptors.” If these findings have any bearing on protease stimulation of cell division past the monolayer, one might expect that protease treatment of quiescent CEFs would expose more insulin-binding sites then are available on untreated, quiescent cells. Just such a result has been obtained by Raizada and Perdue (1978). These workers reported that treatment of quiescent CEFs with 1.0 pg/ml trypsin for 10 minutes produced a 1.5-fold increase in insulin binding, while treatment with 10 pg/ml trypsin for 10 minutes produced a 1.9-fold increase. These data clearly suggest that some of the insulin receptor sites on the cell surface are masked at confluence and can be exposed or modified by trypsin treatment. From their data, Raizada and Perdue (1978) suggest that normal CEFs control cell multiplication b y controlling the availability of insulin-binding sites on their cell surface. Since Perdue (personal communication) has now demonstrated that insulin can be competed off the cell surface by MSA (NSILA), one can tentatively hypothesize that, in vitro, the sites which have been titrated as insulin-binding sites are MSA-binding sites and thus at confluence normal cells become quiescent because MSA fails to provide the necessary stimulus to drive them through the cell cycle. However, transfonned cells, even at confluence, would bind sufficient quantities of MSA to ensure continued cell division. Normal cells, briefly treated with trypsin, would bind more MSA and would consequently be driven through a new round of cell division. It must be noted at this point that data which are essentially diametrically opposite those of Raizada and Perdue (1976) have been published by Thomopoulos et al. (1976). These workers, using normal and transformed derivatives of BALB/c 3T3 have reported that, while the growing normal cells bind little insulin, the amount of insulin bound increases three- to ninefold at confluence. However, Thomopoulos et al. (1976) have reported that, while the amount of insulin bound to growing transformed cells was low, the binding increased as these cells approached confluence. The apparent discrepancies in the data obtained by the two groups almost certainly results from the different approaches used in estimating the amount of insulin bound. Thus, whereas Raizada and Perdue

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

449

(1976, 1978) measured insulin binding to substratum-attached cells, Thompoulos et al. (1976) measured binding to 3T3 cells removed from the substrate by collagenase and hyaluronidase treatment (with as long as a 90-minute incubation for the removal of normal cells). Perdue (personal communication) demonstrated that removal of both CEF and 3T3 cells from the substrate by the method described by Thompoulos et al. (1976) significantly reduced the binding of insulin to normal, growing cells and increased insulin binding to quiescent normal cells. Thus it is likely that collagenase and hyaluronidase (or proteases contaminating the enzyme preparations) remove insulinbinding sites from normal, growing cells and unmask insulin-binding sites in confluent, quiescent normal cells. This suggestion is indirectly supported by Hale and Weber’s (1975) data which demonstrate that, whereas pretreatment of quiescent CEFs with low concentrations of trypsin can “sensitize” cells to the mitogenic properties of serum, higher concentrations of trypsin significantly reduce the mitogenic response of the cells to serum. Thus it seems very likely that cells may exercise control over their own replication by controlling the number of sites available for the binding of mitogens present in the serum. This hypothesis agrees with much of the work of Holley and Temin, who have frequently reported that transformed cells are more efficient in the uptake of serum factors and thus, under the same growth conditions, can grow to a higher cell density than nontransformed cells. Since in the first half of this section it was demonstrated that protease treatment of quiescent cells could increase the uptake of lowMW nutrients (in particular 2-dG), it is of interest to ask whether or not insulin treatment of such cells can itself stimulate the rate of uptake of low-MW nutrients. Griffiths (1972), using human fibroblasts, was one of the earliest to report that insulin treatment induced an increased uptake of leucine in these cells. Hollenberg and Cuatrecases (1975),again using human fibroblasts, demonstrated that insulin binding to human fibroblast monolayers stimulated a-AIB uptake 1.2- to 2-fold without affecting 3-0-methylglucose uptake. Raizada and Perdue (1976) demonstrated that insulin treatment of quiescent CEFs produced a 1.Sfold increase in 2-dG uptake. Most interestingly, they also showed that pretreatment of quiescent CEFs with trypsin [which should increase the number of available insulin-binding sites (Raizada and Perdue, 1976)]increased the insulin-stimulated uptake of 2dG by -40%. This suggests that increasing the number of available insulin-binding sites on the cell surface may also increase the rate of 2-dG uptake. From the data presented in this section we believe a valid hypoth-

450

KENNETH D. NOONAN

esis to explain trypsin stimulation of cell division is one in which trypsin acts on the quiescent fibroblast cell surface in some fashion to increase the number of available mitogen receptor sites. As a result of an increase in available mitogen-binding sites, serum mitogens (e.g., MSA, EGF, FGF, and so on) would bind to the cell surface in such a manner or to such a degree as to increase the uptake of both low-MW nutrients (e.g., glucose, leucine) and specific serum factors (see the review of Holley, 1974).This increased uptake of nutrients or serum factors would subsequently result in a reinitiation of cell division. In such a model proteolytic modification of the cell surface would not itself be a mitogenic stimulus but would “prepare” the cell surface to bind growth factors present in the serum. The binding of these growth factors by the cell would in turn increase the uptake of other components in the medium, subsequently inducing the cell to proceed through the cell cycle. IX.

SUMMARY

One of the major purposes of this chapter has been to gather together many of the beliefs which have come to surround the subdiscipline of protease-induced stimulation of cell division and to ask whether these beliefs stand up to harsh scrutiny. One of the earliest beliefs was that only brief treatment of quiescent fibroblasts with proteases was required to induce cell division. This very belief contained within it the dogma that proteolytic modification of the cell surface acted as a trigger committing quiescent fibroblasts to reenter the cell cycle. The belief was that, once the cells had reentered the cell cycle they would complete cell division. However, most of the evidence now available (see Fig. 4)suggests that only a relatively long incubation of the cells with a protease is sufficient to induce a new round of cell division. Thus it is now relatively well established that, although a brief protease treatment of quiescent fibroblasts can modify the cell surface, this modification is not sufficient to stimulate cell division. The second belief to which this field had become attached was that proteases induced cell division in a quiescent population of fibroblasts by specifically modifying a component of the cell surface, and that the change(s) in cell surface structure were identifiable. It had been suggested by Burger and his collaborators that the modification in the cell surface detected as enhanced agglutinability was either necessary or sufficient for protease induction of cell division. However, numerous workers have now demonstrated that this change in surface structure is neither necessary nor sufficient for the induction

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

45 1

of cell division. Over the last 3 years claims similar to those made for agglutinability have been proposed for the LETS protein and the thrombin-sensitive surface protein. However, it has now been shown that loss or modification of either or both of these surface peptides is not sufficient or necessary for the induction of cell division in a quiescent cell population. Thus at present there is no evidence that proteases act to stimulate cell division by specifically modifying surface peptide or glycopeptide. The most popular belief presently held is that proteases stimulate cell division by modifying the surface structure in a fashion which permits or produces enhanced cellular uptake of mitogens present in the serum. According to this belief it is the mitogen and not the protease which is most directly responsible for inducing a new round of cell division. The major argument against this belief lies in the work outlined in Fig. 5 from which it is clear that thrombin can induce cell division in the absence of exogenous serum factors. It can be (and has been) argued that thrombin itself has mitogenic activity and thus acts as a “promitogen,” preparing the surface to accept mitogen binding and then itself binding to the exposed mitogen receptors, subsequently inducing the intracellular events necessary for cell division. Only further work will demonstrate whether or not this popular belief is correct. Thus, at present, we suggest (in accordance with the last belief) that the most likely explanation for protease induction of cell division in a quiescent fibroblast population lies very near the arguments put forth by Pardee (1964) and Holley (1972, 1974). Specifically, we suggest that the protease acts on the cell surface in such a manner as to prepare the cell surface to take up serum factors more efficiently, and that the serum factors act on the cell to increase the utilization of nutrients present in the medium. Thus, according to this view, the protease is not itself the trigger which induces new cell division but rather acts to prepare the cell to bind the mitogens already present in the surrounding milieu. Despite the attractiveness of this hypothesis, however, it is immediately realized that this proposal still does not explain why extended proteolysis is necessary to induce cell division. Clearly, continued work is required to answer the various questions which remain open in this field and which we hope have been adequately highlighted in this chapter. ACKNOWLEDGMENTS

I thank Drs. James Perdue and Norine Noonan for making manuscripts available prior to publication. I also thank Mrs. Sharon Flakes for her expert help in the preparation of the manuscript.

45 2

KENNETH D. NOONAN

REFERENCES Abercrombie, M. (1970). Contact inhibition in tissue culture. In Vitro 6, 128-142. Abercrombie, M., and Heaysman, J. E. M. (1954).Observations on the social behavior of cells in tissue culture. 11. “Monolayering” of fibroblasts. E x p . Cell Res. 6,293-306. Anderson, J., Edelman, G. M., Moller, G., and Sjoberg, G. (1972). Activation of B lymphocytes by locally concentrated concanavalin A. Eur. J. Immunol. 2,233-235. Betel, I., and Van Den Berg, K. J. (1972). Interaction of concanavalin A with rat lymphocytes. Eur. J. Biochem. 30,571-578. Bissell, M. J., Rubin, H., and Hatie, C. (1971). Leakage as the source of OSF in RSV transformed cultures. E r p . Cell Res. 68,404-410. Blumberg, P. M., and Robbins, P. W. (1975a). Relation of protease action on the cell surface to growth control and adhesion. In “Proteases and Biological Control” (E. Reich, D. Rifkin, and E. Shaw, eds.), pp. %5-956. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Blumberg, P. M., and Robbins, P. W. (197513).Effects of proteases on activation ofresting chick embryo fibroblasts and on cell surface proteins. Cell 6, 137-147. Bombik, B., and Burger, M. M. (1973). Cyclic AMP and the cell cycle: Inhibition of growth stimulation. E x p . Cell Res. 80,88-94. Borek, C., and Fenoglio, M. (1976). Scanning electron microscopy of surface features of hamster embryo cells transformed in uitro by x-irradiation. Cancer Res. 36, 13251335. Bosmann, H. B. (1972). Elevated glycosidase and proteolytic enzymes in cells transformed by RNA tumor virus. Biochim. Biophys. Acta 264,339-343. Boyde, A., Weiss, R. A., and Vesely, P. (1972).Scanning electron microscopy of cells in culture. Exp. Cell Res. 71,313-324. Bradley, W. E., and Culp, L. A. (1974). Stimulation of 2-deoxyglucose uptake in growthinhibited BALB/c 3T3 and revertant SV-40 transformed 3T3 cells. E x p . Cell Res. 84,335-350. Bretscher, M. S. (1971).A major protein which spans the human erythrocyte membrane. J . Mol. B i d . 59,351-357. Burger, M. M. (1969). A difference in the architecture of the surface membrane of normal and virally transformed cells. Proc. Natl. Acad. Sci. U.S.A.62,994-1001. Burger, M. M. (1970). Proteolytic enzymes initiating cell division and escape from contact inhibition of growth. Nature (London)227, 170-171. Burger, M. M. (1971).The significance of surface structure changes for growth control under crowded conditions. Ciba Found. Symp. Growth Control Cell Cult. pp. 4569. Burger, M. M. (1972).Surface changes detected by lectins and implications for growth regulation in normal and in transformed cells. i n “Biomembranes” (L. A. Manson, ed.), pp. 247-270. Plenum, New York. Burger, M. M., and Goldberg, A. R. (1967). Identification of a tumor-specific determinant on neoplastic cell surfaces. Proc. Natl. Acad. Sci. U.S.A. 57,359-366. Burger, M. M., Bombik, B. M., Breckenridge, B., and Sheppard, J. R. (1972). Growth control and cyclic alterations of CAMP in the cell cycle. Nature (London),New Biol. 239,161-163. Burr, J. G., and Rubin, H. (1975). Partial purification and characterization of overgrowth stimulating factor. In “Proteases and Biological Control” (E. Reich, D. Rifkin, and E. Shaw, eds.), pp. 807-825. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Cabantchik, Z. I., and Rothstein, A. (1974).Membrane proteins related to anion permea-

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

453

bility of human erythrocytes. 11. Effects of proteolytic enzymes on disulfonic stilbene sites of surface proteins. J . Membr. Biol. 15,227-248. Carney, D. H., and Cunningham, D. D. (1977). Initiation ofchick cell division by trypsin action at the cell surface. Nature (London)268,602-606. Carter, S. B. (1965). Principles of cell motility: The direction of cell movement and cancer invasion. Nature (London) 208, 1183-1187. Chen, L. B., and Buchanan, J. M. (1975). Plasminogen-independent fibrinolysis by proteases produced by transformed chick embryo fibroblasts. Proc. Natl. Acad. Sci. U.S.A. 72, 1132-1136. Chen, L. B., Teng, N. N., and Buchanan, J. M. (1975).The mitogenic activity and related effects of thrombin on chick embryo fibroblasts. I n “Proteases and Biological Control’’ (E. Reich, D. Rifkin, and E. Shaw, eds.), pp. 957-965. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Chen, L. B., Callimore, P. H., and McDougall, J. K. (1976a). Correlation between tumor induction and the LETS protein on the cell surface.Proc. Natl. Acad. Sci. U.S.A.73, 3570-3574. Chen, L. B., Teng, N. H., and Buchanan, J. M. (1976b). Mitogenicity of thrombin and surface alterations on mouse splenocytes. E x p . Cell Res. 101,41-46. Chou, I., Black, P., and Roblin, R. (1974a). Effects of protease inhibitors on growth of 3T3 and SV3T3 cells. I n “Control of Proliferation in Animal Cells” (B. Clarkson and R. Baserga, eds.), pp. 339-350. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Chou, I., Black, P. H., and Roblin, R. 0. (1974b). Non-selective inhibition of transformed cell growth by a protease inhibitor. Proc. Nutl. Acad. Sci. U S A . 71, 17481752. Christman, J. K., and Acs, G. (1974). Purification and characterization of a cellular fibrinolytic factor associated with onocogenic transformation: T h e plasminogen activator from SV-40 transformed hamster cells. Biochim. Biophys. Acta 340,339-347. Christman, J. K., Acs, G., Silagi, A., and Silverstein, S. C. (1975). Plasminogen activator: Biochemical characterization and correlation with tumorigenicity. In “Proteases and Biological Control” (E. Reich, D. B. Rifkin, and E. Shaw, eds.), pp. 827-839. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Clarkson, B., and Baserga, R. (1974). “Control of Proliferation in Animal Cells.” Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Collard, J. B., and Temmink, J. H. M. (1976). Surface morphology and agglutinability with Con A in normal and transformed murine fibroblasts./. Cell B i d . 68,101-112. Critchley, D. R. (1974). Cell surface proteins of NIL 1 hamster fibroblasts labeled by galactose oxidase, tritiated borohydride method. Cell 3, 121-125. Cunningham, D. D., and Ho, T. S. (1975). Effects of added proteases on Con A-specific agglutinability and proliferation of quiescent fibroblasts. In “Proteases and Biological Control” (E. Reich, D. Riflcin, and E . Shaw, eds.), pp. 795-806. Cold Spring Harbor Lab., Cold Spring Harbor, New York. de Asua, L., Surian, E. S., Flavia, M. M., and Torres, H. M. (1973).Effects of insulin on the growth pattern and adenylate cyclase activity of BHK fibroblasts. Proc. Nutl. Acad. Sci. U.S.A. 70, 1388-1392. DePetris, S., Raff, M. C., and Mallucci, L. (1973).Ligand-induced redistribution of Con A receptors on normal, trypsinized and transformed fibroblasts. Nature (London), New Biol. 244,275-278. Dulak, N. C., and Temin, H. M. (1973). A partially purified polypeptide fraction from rat liver conditioned medium with multiplication stimulating activity for embryo fibroblasts. J . Cell Physiol. 81, 153-160.

454

KENNETH D. NOONAN

Edelman, G. M. (1973). Antibody structure and molecular immunology. Science 180, 830-840. Ellison, S., Mulholland, J., and Noonan, K. D. (1977). An analysis of lectin-initiated cell agglutination in two subclones which respond morphologically to growth in dibutyryl CAMP. 111. Relationship of lectin receptor mobility to agglutination. Erp. Cell Res. Submitted for publication. Fairbanks, G., Steck, T. L., and Wallach, D. F. H. (1971). Electrophoretic analysis of the major polypeptides of the human erythrocyte membrane. Biochemistnj 10,26062617. Fisher, A. (1946). Mechanisms of the proteolytic activity of malignant tissue cells. Nature (London) 157,442. Follett, E., and Goldman, R. D. (1970). The occurrence of microvilli during spreading and growth of BHK 21/C13 fibroblasts. Exp. Cell Res. 59, 124-136. Foster, D. O., and Pardee, A. B. (1969). Transport of amino acids by confluent and nonconfluent 3T3 and polyoma virus-transformed 3T3 cells growing on glass cover slips.]. Biol. Chem. 244,2675-2681. Gahmberg, C. G., and Hakamori, S . 4 . (1973). Altered growth behavior of malignant cells associated with changes in externally labelled glycoprotein and glycolipid. Proc. Natl. Acad. Sci. U S A . 70,3329-3333. Gahmberg, C. G., Keihn, D., and Hakamori, %-I. (1974). Changes in a surface-labelled galactoprotein and in glycolipid concentrations in cells transformed by a temperature sensitive polyoma mutant. Nature (London)248,413-415. Gavin, J. R., Roth, J., Jen, P., and Freychet, P. (1972). Insulin receptors in human circulating cells and fibroblasts. Proc. Natl. Acad. Sci. U.S.A. 69,747-751. Glynn, R. D., Thrash, C. R., and Cunningham, D. D. (1973).Maximal Con A specific agglutinability without loss of density-dependent growth control. Proc. Natl. Acad. Sci. U.S.A. 70,2676-2677. Goetz, I., Weinstein, C., and Roberts, E. (1972). Effects of protease inhibitors on growth of hamster tumor cells in culture. Cancer Res. 32,2469-3474. Goldberg, A. R. (1974a). Involvement of cell and serum proteases in exposing tumorspecific agglutinin sites. Ann. N.Y. Acad. Sci. 234,348-354. Goldberg, A. R. (1974b). Increased protease levels in transformed cells: A casein overlay assay for the detection of plasminogen activator production. Cell 2,95-102. Goldman, R. D., Lazarides, E., Pollack, R., and Weber, K. (1975).The distribution of actin in non-muscle cells: The use of actin antibody in the localization of actin within the microfilament bundles of mouse 3T3 cells. E x p . Cell Res. 90,333-344. Goodall, D. M., Pardoe, G. I., and Gregory, J. K. (1971). Clin. Erp. Immunol. 9,645655. Gospodarowicz, D. (1975). Purification of a fibroblast growth factor from bovine pituitary.]. Biol. Chem. 250,2515-2520. Gospodarowicz, D. (1976). Humoral control of cell proliferation: The role of fibroblast growth factor in regeneration, wound healing, and neoplastic growth. In “Membranes and Neoplasia: New Approaches and Strategies” (V. T. Marchesi, ed.), pp. 1-19. Alan R. Liss, New York. Gospodarowicz, D., and Moran, J. S. (1974). Stimulation of division in a sparse and confluent 3T3 cell population by a fibroblast growth factor, dexamethasone and insulin. Proc. Natl. Acad. Sci. U S A . 71,4584-4588. Greaves, M. F., and Bauminger, S. (1972). Activation of T and B cells by insoluble phytomitogens. Nature (London),New Biol. 235,67-70. Greaves, M.F., and Janossy, G. (1972). Stimulation of lymphocyte cell division. Transplant. Reu. 11,87-150.

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

455

Griffiths, D. B. (1972). Role of serum insulin and amino acid concentrations in contact inhibition of growth of human cells in culture. E x p . Cell Res. 72,47-56. Hale, A. H., and Weber, M. J. (1975). Hydrolase and serum treatment of normal chick embryo fibroblasts: Effects on hexose transport. Cell 5,245-252. Harris, A. (1973).Behavior of cultured cells on substrate of variable adhesiveness. E x p . Cell Res. 77,385-397. Hatanaka, M., and Hanafusa, H. (1972).Analysis of a functional change in membranes in the process of cell transformation by RSV: Alteration in the characteristics of sugar transport. Virology 41,647-652. Hatanaka, M., Augl, C., and Gilden, R. V. (1970).Evidence for a functional change in MSV-infected mouse embryo cells. J. Biol. Chem. 215,714-717. Hatcher, V. B., Wertheim, M. S., Rhee, R. Y., Tsien, G., and Burk, P. G. (1976). Relationship between cell surface protease activity and doubling time in various normal and transformed cells. Biochim. Biophys. Acta 451,499-510. Hatcher, V. B., Oberman, M. S., Wertheim, M. S., Rhee, C. Y., Tsien, G., and Burk, P. G. (1977). The Relationship between Surface Protease Activity and the Rate of Cell Proliferation in Normal and Transformed Cells. Biochem. Biophys. Res. Commun. 76,602-608. Hirschorn, R., Grossman, J., Troll, W., and Weissman, G. (1971).The effect of eaminocaproic acid and other inhibitors of proteolysis upon the response of human peripheral blood lymphocytes to PHA.J. Clin. Invest. 50, 1206-1217. Hogg, N. M. (1974). A comparison of membrane labeling of normal and transformed cells by lactoperoxidase labeling. Proc. Natl. Acad. Sci. U.S.A. 71,489-492. Hollenberg, M. D., and Cuatrecasas, P. (1975). Insulin and epidermal growth factor: Human fibroblast receptors related to DNA synthesis and amino acid uptake. J. Biol. Chem. 250,3845-3853. Holley, R. W. (1972). A unifying hypothesis concerning the nature of malignant growth. Proc. Natl. Acad. Sci. U.S.A. 69,2840-2841. Holley, R. W. (1974). Serum factors and growth control. I n “Control of Proliferation in Animal Cells” (B. Clarkson and R. Baserga, eds.), pp. 13-18. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Holley, R. W., and Kiernan, J. A. (1968). “Contact Inhibition” of Cell Division in 3T3 Cells. Proc. Natl. Acad. Sci, U S A . 60, 602-608. Hubbard, A. L., and Cohn, Z. A. (1972).The enzymatic iodination of the red blood cell membrane.J. Cell. B i d . 55,390-405. Hynes, R. 0. (1973). Alteration of cell surface proteins by viral transformation and by 70,3170-3174. proteolysis. Proc. Natl. Acad. Sci. U.S.A. Hynes, R. 0. (1974). Role of surface alterations in cell transformation: The importance of proteases and surface proteins. Cell 1, 147-156. Hynes, R. O., and Bye, J. M. (1974). Density and cell cycle dependence of cell surface proteins in hamster fibroblasts. Cell 3, 113-120. Hynes, R. O., and Humphreys, K. C. (1974). Characterization of the external proteins of hamster fibroblasts./. Cell B i d . 62,438-448. Hynes, R. O., Wyke, J. A., Bye, J. M., Humphreys, K. P., and Pearlstein, E. S. (1975).Are proteases involved in altering surface proteins during viral transformation?In “Proteases and Biological Control” (E. Reich, D. Rifkin, and E. Shaw, eds.), pp. 931944. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Hynes, R. O., Destree, A. T., and Mantner, V. (1976). Spatial organization at the cell surface. I n “Membranes and Neoplasia: New Approaches and Strategies” (V. Marchesi, ed.), pp. 189-202. Alan R. Liss, New York. Inbar, M., and Sachs, L. (1969).Interaction of the carbohydrate-binding protein concan-

456

KENNETH D. NOONAN

avalin A with normal and transformed cells. Proc. Natl. Acad. Sci. U S A . 63,14181425. Isselbacher, H. J. (1972). Increased uptake of amino acids and 2-deoxyglucose by virus transformed cells in culture. Proc. Natl. Acud. Sci. U.S.A. 69,585-589. Johnson, G. S., and Pastan, I. (1972a). Cyclic AMP increases the adhesion of fibroblasts to the substratum. Nature (London), New Biol. 236,247-249. Johnson, G. S., and Pastan, I. (1972b). Role of 3’,5‘-adenosine monophosphate in regulation of morphology and growth of transformed and normal fibroblasts. J . Natl. Cancer Inst. 48, 1377-1387. Johnson, G. S., Morgan, W. D., and Pastan, I. (1972). Regulation of cell motility by cyclic AMP. Nature (London) 235,54. Jones, P. A., Shim, J. S., Isaacs, H., and McAllister, R. M. (1975). The relationship between tumorigenicity, growth in agar and fibrinolytic activity in a line of human osteosarcoma cells. Znt. J . Cancer 16,616-621. Kaplan, J. G . , and Bona, C. (1974). Proteases as mitogens. The effects of trypsin and pronase on mouse and human lymphocytes. E x p . Cell Res. 88,388-394. Kletzien, R. F., and Perdue, J. F. (19744 Sugar transport in chick embryo fibroblasts. I. A functional change in the plasma membrane associated with the rate of cell growth. J. Biol. Chem. 249,3366-3374. Kletzein, R. F., and Perdue, J. F. (1974b). Sugar transport in chick embryo fibroblasts. 11. Alterations in transport following transformation by a temperature sensitive mutant of RSV. J. Biol. Chem. 249,3375-3382. Kletzien, R. F., and Perdue, J. F. (1974~).Sugar transport in chick embryo fibroblasts. 111. Evidence for post-transcriptional and post-translational regulation of transport following serum addition. J . Biol. Chem. 249,3383-3387. Kuusela, P., Ruoslahti, E., and Vaheri, A. (1975). Polypeptides of a glycoprotein antigen (SF) present in serum and on surface of normal but not transformed chick embryo fibroblasts. Biochim. Bioph!ls.Acat 379,295-303. Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature (London)227,680-685. Lang, W . E., Jones, P. A., and Benedict, W. F. (1975). Relationship between fibrinolysis of cultured cells and ma1ignancy.J. Natl. Cancer Inst. 54, 173-179. Lazarides, E., and Weber, K. (1974). Actin antibody. The specific visualization of actin filaments in non-muscle cells. Proc. Natl. Acad. Sci. U.S.A. 71,2268-2272. Mcllhenney, A., and Hogan, B. L. (1974). Effects of inhibitors of proteolytic enzymes on the growth of normal and polyoma transformed BHK cells. Biochem. Biophys. Res. Commun. 60,348-354. Malik, L. E., and Langenback, R. (1976). Scanning electron microscopy ofin uitro chemically transformed mouse embryo cells. J . Cell Biol. 68, 654-664. Martin, B. M. and Quigley, J. P. (1978). Binding and internalization of [Iz5I]thrombin in chick embryo fibroblasts: possible role in mitogenesis. (Submitted for publication). Martin, G. S., Venuta, S., Weber, M., and Rubin, H. (1971). Temperature dependent alterations in sugar transport in cells infected by a temperature sensitive mutant of RSV. Proc. Natl. Acad. Sci. U.S.A. 68,2739-2741. Martz, E., and Steinberg, M. S. (1972). The role of cell-cell contact in “contact inhibition” of cell division: A review and new evidence.J . Cell Physiol. 79,189-210. Mazzei, D., Novi, C., and Bazzi, C. (1966). Mitogenic action of trypsin and chymotrypsin. Lancet ii, 232. Morell, A. G., Van Den Hamer, C. J. A., Scheinberg, I. H., and Ashwell, G:(1966). Physical and chemical properties of ceruloplasmin. IV. Preparation of radioactive sialicacid free ceruloplasmin labeled with tritium on terminal D-gahCtOSe residues. J . Biol. Chem. 241,3745-3749.

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

45 7

Mueller, T. J., Dow, A. W., and Morrison, M. (1976). Heterogeneity of the sialoglycoproteins of the normal human erythrocyte membrane. Biochem. Biophys. Res. Commun. 72, 94-99. Nicolson, G . L. (1974). The interactions of lectins with animal cell surfaces. Int. Rev. Cytol. 39,89-190. Nicolson, G., and Poste, G. (1976). The cancer cell: Dynamic aspects and modification in cell-surface organization. New Engl. J. Med. 295, 195-203. Noonan, K. D. (1976). Role of serum in protease-induced stimulation of 3T3 cell division past the monolayer stage. Nature (London)259,573-576. Noonan, K. D. (1978). Lectins as probes of the transformed cell surface. In “Virus Transformed Cell Membranes” (C. Nicolau, ed.), (in press) Academic Press, New York. Noonan, K. D., and Burger, M. M. (1973). Induction of 3T3 cell division of the monolayer stage. Early changes in macromolecular process. Erp. Cell Res. 80,405-414. Noonan, K. D., and Burger, M. M. (1974). The role of the cell surface in contact inhibition of cell division. Prog. S u g . Membr. Sci. 8,245-284. Noonan, N. E., and Noonan, K. D. (1977). The effect of TPCK on transcription and its role in modifying cell growth. J . Cell Physiol. 92, 137-144. Novogrodsky, A., and Katchalski, E. (1971). Induction of lymphocyte transformation by periodate. FEBS Lett. 12,297-300. Novogrodsky, A., and Katchalski, E. (1972). Membrane site modified on induction of the transformation of lymphocytes by periodate. Proc. Natl. Acad. Sci. U.S.A.69,32073210. Novogrodsky, A., and Katchalski, E. (1973). Induction of Lymphocyte Transformation by Sequential Treatment with Neuraminidase and Galactose Oridase. Proc. Natl. Acad. Sci. U.S.A. 70, 1824-1827. Ossowski, L., Unkeless, J. C., Tobia, A., Quigley, J. P., Rifkin, D. B., and Reich, E. (1973a). An enzymatic function associated with transformation of fibroblasts by oncogenic viruses. 11. Mammalian fibroblast cultures transformed by DNA and RNA tumor viruses.J. Erp. Med. 137, 112-126. Ossowski, L., Quigley, J. P., Hellerman, G . M., and Reich, E. (197313).Fibrinolysis Associated with Oncogenic Transformation. Requirement of Plasminogen for Correlated Changes in Cellular Morphology, Colony Formation in Agar and Cell Migration. J. E x p . Med. 138,1056-1064. Ossowski, L., Quigley, J. P., and Reich, E. (1975). Plasminogen, a necessary factor for cell migration in uitro. In “Proteases and Biological Control” (E. Reich, D. B. Rifkin, and E. Shaw, eds.), pp. 901-913. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Otten, J., Johnson, G. S., and Pastan, I. (1971). Cyclic AMP levels in fibroblasts: Relationship to growth rate and contact inhibition of growth. Biochem. Biophys. Res. Commun. 44,1192-1198. Otten, j., Bader, J. P., Johnson, G. S., and Pastan, I. (1972a). A mutation in a Rous sarcoma virus gene that controls adenosine 3‘,5’-monophosphate levels and transformation.]. B i d . Chem. 247, 1632-1641. Otten, J., Johnson, G. S., and Pastan, I. (1972b). Regulation of cell growth by CAMP: Effect of cell density and agents which alter cell growth on CAMP levels in fibroblasts. J. Biol. Chem. 247, 7082-7087. Pardee, A. B. (1964). Cell division and a hypothesis of ancer. Natl. Cancer Inst., Monogr. 14, 7-20. Pardee, A. B., Jimenez d e Asua, L., and Rosengurt, E . (1974). Functional membrane changes and cell growth: Significanceand mechanism. In “Control of Proliferation in Animal Cells” (B. Clarkson and R. Baserga, eds.), pp. 547-551. Cold Spring Harbor Lab., Cold Spring Harbor, New York.

458

KENNETH D. NOONAN

Parker, J. W., O’Brien, R. L., Steiner, J., and Paolilli, P. (1973). Periodate-induced lymphocyte transformation. 11. Character of response and comparison with PHA and pokeweed mitogen stimulation. Exp. Cell Res. 78,279-286. Parker, J. W., O’Brien, R. L., Steiner, J. L., and Paolilli, P. (1974). Periodate-induced lymphocyte transformation. 111. Effects of temperature and pH. Exp. Cell Res. 83, 220-224. Pastan, I., Anderson, W. B., Carchman, R. A., Willingham, N. C., Russell, T. R., and Johnson, G. S. (1974). Cyclic AMP and malignant transformation. In “Control of Proliferation in Animal Cells” (B. Clarkson and R. Baserga, eds.), pp. 563-570. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Pearlstein, E., and Waterfield, M. D. (1974). Metabolic studies on [12511 labeled BHK cell plasma membranes. Biochim. Biophys. Acta 362, 1-12. Perdue, J. F. (1973).The distribution, ultrastructure, and chemistry of microfilaments in cultured chick embryo fibroblasts. ]. Cell Biol. 58,265-283. Phillips, D. R., and Morrison, M. (1970). The arrangement of proteins in the human erythrocyte membrane. Biochem. Biophys. Res. Commun. 40,284-289. Pierson, R. W., and Temin, H. M. (1972). The partial purification from calf serum of a fraction with multiplication stimulating activity for chick fibroblasts in cell culture and non-suppressible insulin like activity. ]. Cell Physiol. 79,319-330. Pollack, R., and Burger, M. M. (1969).Surface-specific characteristics of a contact inhibited cell line containing the SV40 viral genome. Proc. Natl. Acud. Sci. U S A . 62, 1074- 1076. Pollack, R., and Rifkin, D. (1975).Actin-containing cables within anchorage-dependent rat embryo cells are dissociated by plamin and trypsin. Cell 6,495-506. Pollack, R., Green, H., and Todaro, G. (1968).Growth control in cultured cells: Selection of sublines with increased sensitivity to contact inhibition and decreased tumor producing ability. Proc. Natl. Acad. Sci. U S A . 60, 126-133. Pollack, R., Risser, R., Conlon, S., and Rifkin, D. (1974).Plasminogen activator production accompanies loss of anchorage regulation in transformation of primary rat embryo cells by simian virus 40. Proc. Natl. Acad. Sci. U.S.A.71,4792-4796. Pong, S., Nuss, D. L., and Hoch, G. (1975). Inhibition of initiation of protein synthesis in tissue culture cells by TPCK.]. Biol. Chem. 250,240-245. Porter, K., Prescott, D., and Frye, J. (1973a).Changes in surface morphology and Chinese hamster ovary cells during the cell cycle. ]. Cell Biol. 57,815-836. Porter, K., Todaro, G. J., and Fonte, V. (1973b). A scanning electron microscope study of surface features of viral and spontaneous transformants of mouse BALB/3T3 cells. ]. Cell Biol. 59,633-642. Quigley, J. P. (1976).Association of a protease (plasminogen activator) with a specifib membrane fraction isolated from transformed cells. ]. Cell Biol. 71,472-486. Quigley, J. P., Ossowski, L., and Reich, E. (1974).Plasminogen, the serum proenzyme activated by factors from cells transformed by oncogenic viruses.]. Biol. Chem. 249, 4306-431 1. Raizada, M. K., and Perdue, J. F. (1976).Mitogen receptors in chick embryo fibroblasts: The kinetics, specificity, unmasking and synthesis of [‘251] insulin binding sites. ]. Biol. Chem. 251,6445-6455. Raizada, M. K., and Perdue, J. (1978).Alterations in the expression of mitogen receptors in virus-transformed chick embryo fibroblasts and 3T3s. Proc. Natl. Acad. Sci. U.S.A. (in press). Rifkin, D. B., Loeb, J. N., Moore, G., and Reich, E. (1974).Properties of plasminogen activators formed by neoplastic human cell cultures.]. E x p . Med. 139,1317-1328. Rubin, H. (1970).Overgrowth stimulating factor released from Rous sarcoma cells. Science 167,1271-1272.

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

459

Ruoslahti, E., Vaheri, A., Kuusela, P., and Linder, E. (1973). Fibroblast surface antigen: A new serum protein. Biochim. Biophys. Acta 322,352-358. Schnebli, H. P. (1972). A protease-like activity associated with malignant cells. Schweiz. Med. Wochenschr. 102,1194-1200. Schnebli, H. P. (1974).Growth inhibition of tumor cells by protease inhibitors: Consideration of the mechanisms involved. In “Control of Proliferation in Animal Cells” (B. Clarkson and R. Baserga, eds.), pp. 327-337. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Schnebli, H. P., and Burger, M. M. (1972).Selective inhibition of growth of transformed cells by protease inhibitors. Proc. Natl. Acad. Sci. U.S.A. 69,3825-3827. Schnebli, H. P., and Hammerli, G. (1974). Protease inhibitors d o not block transformed cells in the GI phase of the cell cycle. Nature (London)248, 150-151. Sefton, B. M., and Rubin, H. (1970). Release from density dependent growth inhibition by proteolytic enzymes. Nature (London) 227,843-845. Sefton, B. M., and Rubin, H. (1971). Stimulation of glucose transport in cultures of density inhibited chick embryo cells. Proc. Natl. Acad. Sci. U.S.A. 68,3154-3157. Sheppard, I., and Bannai, S. (1974). Cyclic AMP and cell proliferation. In “Control of Proliferation in Animal Cells” (P. Clarkson and R. Baserga, eds.), pp. 571-580. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Sheppard, J. R. (1971). Restoration of contact inhibition to transformed cells by dibutyryl cyclic AMP. Proc. Natl. Acad. Sci. U.S.A. 68, 1316-1320. Sheppard, J. R. (1972).Difference in CAMP levels in normal and transformed cells. Nature (London),New Biol. 236, 14-16. Shin, S., Freedman, V. H. Risser, R., and Pollack, R. (1975). Tumorigenicity of virus transformed cells in nude mice is correlated specifically with anchorage independent growth in oitro. Proc. Natl. Acad. Sci. U.S.A. 72,44354439. Sims, H. S., and Stillman, D. P. (1937).Substances affecting adult tissue in oitro. I. The stimulating action of hypsin on fresh adult tissue. J. Gen. Physiol. 20,603-620. Spataro, A. C., Morgan, H. R., and Bosmann, H. B. (1976). Neutral protease activity of Rous sarcoma (RSV) transformed chick embryo fibroblasts. J. Cell Sci. 21,407-413. Steck, T. L. (1972).Cross-linking of the major proteins of the isolated erythrocyte membrane. J. Mol. Biol. 66,295-305. Steck, T. L. (1974).The organization of proteins in the human red blood cell membrane: A review. J. Cell Biol. 62, 1-19. Steck, T. L., and Dawson, G . (1974). Topographical distribution of complex carbohydrates in the erythrocyte membrane. J. Biol. Chem. 249,2135-2142. Stoker, M. G. P., and Rubin, H. (1967). Density dependent inhibition of cell growth in culture. Nature (London)215, 171-172. Stone, K. R., Smith, R. E.,and Joklik, W. K. (1974).Changes in membrane polypeptides that occur when chick embryo fibroblasts and NRK cells are transformed with avian sarcoma virus. Virology 58,86-100. Sylven, B., and Malmgren, H. (1957). The histological distribution of proteinase and peptidase activity in solid tumor transplants: A histochemical study on the enzymic characteristics of the different tumor cell types. Acta Radiol., Suppl. 154. Sylven, B., Snellman, O., and Strauli, P. (1974). Immunofluorescent studies on the occurrence of cathepsin B, at tumor cell surface. Virchows Arch. B 17,97-112. Talmadge, K. W., Noonan, K. D., and Burger, M. M. (1974). T h e transformed cell surface: An analysis of the increased lectin agglutinability and the concept of growth control by surface proteases. In “Control of Proliferation in Animal Cells” (B. Clarkson and R. Baserga, eds.), pp. 313-325. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Temin, H. M. (1967).Studies on carcinogenesis by avian sarcoma viruses. VI. Differen-

460

KENNETH D. NOONAN

tial multiplication of uninfected and converted cells in response to insulin. J . Cell Physiol. 69,377-385. Temin, H. M., Smith, G. L., and Dulak, N. C. (1974).Control of multiplication of normal and RSV-transformed chick embryo fibroblasts by purified multiplication activity with non-suppressible insulin-like and sulfation factor activities. In “Control of Proliferation in Animal Cells” (B. Clarkson and R. Baserga, eds.), pp. 19-26. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Teng, N. N., and Chen, L. B. (1974). The role of surface proteins in cell proliferation as studied with thrombin and other proteases. Proc. Natl. Acad. Sci. U.S.A. 72,413417. Teng, N. N., and Chen, L. B. (1976). Thrombin-sensitive surface protein of cultured chick embryo fibroblasts. Nature (London)259,578-580. Thomopoulos, P., Roth, J., Lovelace, E., and Pastan, I. (1976).Insulin receptors in normal and transformed fibroblasts: Relationship to growth and transformation. Cell 8, 4 17-423. Todaro, G., and Green, G. (1963). Quantitative studies of the growth of mouse embryo cells in culture and their development into established 1ines.J. Cell B i d . 17,299313. Tripplett, R. B., and Carraway, K. L. (1972). Proteolytic digestion of erythrocytes, resealed ghosts and isolated membranes. Biochemistry 11, 2897-2902. Trowbridge, I. S., Ralph, P., and Beran, M. J. (1975).Differences in surface proteins of B and T cells. Proc. Natl. Acad. Sci. U.S.A. 72, 157-161. Ukena, T. E., Goldman, E., Benjamin, T. L., and Kamovsky, M. J. (1976). Lack of correlation between agglutinability, the surface distribution of Con A and post-confluence inhibition of cell division in ten cell lines. Cell 7,213-222. Unkeless, J. C., Tobia, C., Ossowski, L., Quigley, J. P., Rifkin, D. B., and Reich, E. (1973). An enzymatic function associated with transformation of fibroblasts by oncogenic viruses. I. Chick embryo fibroblast cultures transformed by avian RNA tumor viruses.]. Exp. Med. 137,85-111. Unkeless, J., Dano, K., Kellerman, G. M., and Reich, E. (1974). Fibrinolysis associated with oncogenic transformation: Partial purification and characterization of the cell factor, a plasminogen activator. J . Biol. Chem. 249,4295-4305. Vaheri, A,, and Ruoslahti, E. (1974). Fibroblast surface antigen molecules and their loss from virus-transformed cells: A major alteration in cell surface. In “Control of Proliferation in Animal Cells” (B. Clarkson and R. Baserga, eds.), pp. 305-312. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Vaheri, A,, Ruoslahti, E., and Nordling, S. (1972). Neuraminidase stimulates division and sugar uptake in density dependent cultures. Nature (London),New Biol. 238, 21 1-212. Vischer, T. L. (1974). Stimulation of mouse B lymphocytes by trypsin.J. Zmmunol. 113, 58-62. Wallach, D. F. H. (1972).The dispositions of proteins of the plasma membranes of animal cells: Analytical approaches using controlled peptidolysis and protein labels. Biochim. Biophys. Acta 2 6 5 , 6 1 4 3 . Wartiovaara, J., Linder, E., Ruoshlati, E., and Valevi, A. (1974). Distribution of Fibroblast Surface Antigens and Loss Upon Viral Transformation. J . E x p . Med. 140, 1522-1533. Weber, K., Lazarides, E., Goldman, R. D., Vogel, A., and Pollack, R. (1974).Localization and distribution of actin fibers in normal, transformed and revertant cells. Cold Spring Harbor Symp. Quart. Biol. 39,363-369. Weber, M. L. (1975). Inhibition of protease activity in cultures of RSV-transformed cells: Effects on the transformed phenotype. Cell 5,253-261.

PROTEOLYTIC MODIFICATION OF CELL SURFACE MACROMOLECULES

46 1

Weissmann, R. B., Zurier, R. B., and Hoffstein, S. (1972).Leucocytic proteases and the immunologic release of lysosomal enzymes. Am. J . Pathol. 68,539-564. Weston, J. A,, and Hendricks,H. L. (1972).Reversible transformation by urea of contactinhibited fibroblasts. Proc. Natl. Acad. Sci. U.S.A. 69,3727-3731. Weston, J. A., and Roth, S. (1969). The Cell Surface and Development. I n “Cellular Recognition” (R. T. Smith and R. A. Good, eds.), pp. 29-37. Appleton, New York. Wickus, G. G., and Robbins, P. W. (1973). Plasma membrane proteins of normal and RSV-transformed chick embryo fibroblasts. Nature (London),New Biol. 245,65-67. Wickus, G. G., Branton, D. E., and Robbins, P. W. (1974). RSV transformation of the chick cell surface. In “Control of Proliferation in Animal Cells” (B. Clarkson and R. Baserga, eds.), pp. 541-546. Cold Spring Harbor Lab., Cold Spring Harbor, New York. Willingham, M. C., and Pastan, I. (1975).CAMP modulates microvillus formation and agglutinability in transformed and normal mouse fibroblasts. Proc. Natl. Acad. Sci. U.S.A. 72, 1263-1267. Wolf, B. A., and Goldberg, A. R. (1976). RSV-transformed fibroblasts having low levels of plasminogen activator. Proc. Nut!. Acad. Sci. U.S.A.73,3613-3617. Yamada, K. M., and Weston, J. A. (1974). Isolation of a major cell surface glycoprotein from fibroblasts. Proc. Natl. Acad. Sci. U.S.A. 71,3492-3496. Yamada, K. M., and Weston, J. A. (1975).Synthesis, turnover and artificial restoration of a major cell surface glycoprotein. Cell 5,75-81. Yamada, K. M., Yamada, S. S., and Pastan, I. (1975).The major cell surface glycoprotein of chick embryo fibroblasts is an agglutinin. Proc. Natl. Acad. Sci. U.S.A. 72,31583162. Yamada, K. M., Yamada, S. S., and Pastan, I. (1976). CSP partially restores morphology, adhesiveness and contact inhibition of movement to transformed fibroblasts. Proc. Natl.Acad. Sci. U.S.A.73, 1217-1221. Yunis, A. A., Schultz, D. R., andSato, G. H. (1974). The secretion of fibrinolysin b y cultured rat ovarian tumor cells. Biochem. Biophys. Res. Cornmun. 52, 1003-1012. Zatz, M. M., Goldstein, A. L., Blumenfield, 0. O., and White, A. (1972). Regulation of normal and leukemic lymphocyte transformation and recirculation by sodium periodate oxidation and sodium borohydride reduction. Nature (London), New Biol. 240,252-255. Zetter, B. R., Chen, L. B., and Buchanan, J. M. (1976). Effects of protease treatment on growth, morphology, adhesion and cell surface proteins of secondary chick embryo fibroblasts. Cell 7,407-412. Zetter, B. R., Chen, L. B., and Buchanan, J. M. (1977).Binding and internalization of . U.S.A.74, thrombin by normal and transformed chick cells. Proc. Natl. A c Q ~ Sci. 596-600. NOTE ADDED I N PROOF Since the completion of this chapter Carney and Cunningham [Nature 268,602-606 (1977)l have demonstrated that quiescent CEFs can be stimulated to divide by trypsin immobilized on polystyrene beads. The stimulation could not be accounted for by release of trypsin from the beads into the medium or into the cells.

This Page Intentionally Left Blank

CURRENT TOPICS IN MEMBRANES

AND TRANSPORT. VOLUME 11

Glycoprotein Antigens of Murine lymphocytes MZCHELLE LETARTE' The Ontario Cancer Institute Toronto. Ontario. Canada

. . . . . . . . . . . . . . . . . . 464 1. Introduction I1. Methods of Analysis of Cell Surface Antigens . . . . . . . . . 467

111.

IV .

V.

v1. VII . VIII .

A . Production ofAntisera toCell Surface Antigens . . . .' . . B. Analysis of Specificities Defined by Antisera to Cell Surface Antigens . . . . . . . . . . . . . . . . . . . C . Microsequencing Techniques . . . . . . . . . . . . Isolation and Characterization of H-2 Antigens . . . . . . . . A . Tissue Distribution and Polymorphism of H-2 Antigens . . . . B . Isolation, MW. and Subunit Structure of H-2 Antigens . . . . C . Partial Amino Acid Sequence of H-2 Glycoproteins . . . . . D. CarbohydrateStructure of H-2 Glycoproteins . . . . . . . E . Nature of the Antigenic Determinants of H-2 Glycoproteins . . . Isolation and Characterization of Ia Antigens . . . . . . . . . A . Tissue Distribution and Polymorphism of Ia Antigens. . . . . B . Isolation. MW. and Subunit Structure of Ia Antigens . . . . . C . Partial Amino Acid Sequence and Preliminary Carbohydrate Structure of Ia Glycoproteins . . . . . . . . . . . . . . . D . Antigenic Determinants of Ia Molecules . . . . . . . . . Isolation and Characterization of Thy-1 Antigen . . . . . . . . A . Tissue Distribution and Number of Thy-1 Antigenic Sites per Cell . B . Isolation and MW Determination of Thy-1 Antigen . . . . . C . Amino Acid and Carbohydrate Composition of Thy-1 Glycoprotein . D . Antigenic Determinants of the Thy-1 Molecule . . . . . . . Preliminary Characterization of Tla Antigens . . . . . . . . . Preliminary Characterization of Ly-2.3 Antigens . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . .

467 469 476 477 477 478 480 483 485 487 487 488 490 492 493 493 493 495 499 500 502 503 505

Previous publications under the name of Letarte.Muirhead . The author is a Research Scholar of the National Cancer Institute of Canada and is supported by grants from the National Cancer Institute and the Medical Research Council of Canada. 463 Copyright 0 1978 by Academic Press. Inc . All rights of reproduction in any lorn reserved . ISBN 0-12-153311-5

464

MICHELLE LETARTE

1.

INTRODUCTION

Cell surface antigens can be described in very simple terms as plasma membrane components detectable b y antibodies. However, for immunogeneticists and immunologists, the term “cell surface antigens” refers primarily to alloantigens whose serological expression is controlled by allelic polymorphism. In the mouse genome, more than 60 loci determining cell membrane alloantigens have been identified so far, including erythrocyte alloantigen loci (which are not discussed in this chapter; see J. Klein, 1975),lymphoid tissue alloantigen loci, and histocompatibility loci. Murine lymphoid tissue alloantigen loci control the expression of antigens on lymphocytes (Ly loci), on thymocytes (Thy loci), on thymocytes and leukemic lymphocytes (Tla loci), and on plasmacytes (Pca loci) (J. Klein, 1975).These lymphoid tissue antigens are particularly useful as markers of classes and subclasses of cells; e.g., Thy-1 antigen is the T-lymphocyte marker in the mouse (Reif and Allen, 1964, 1966), whereas Ly-1 and Ly-2,3 antigens are now considered markers of functional subclasses of T cells (Cantor and Boyse, 1975a,b). (Characters in italics are used to designate loci, whereas roman characters refer to the antigens. For example, Thy-1 locus determines the expression of Thy-1 antigen. Alloantigens Thy-1.1 and Thy-1.2 are controlled by their respective alleles Thy-1a and Thy-1b. Antigens Ly-2 and Ly-3, although serologically different, are closely linked genetically and topologically expressed together. They have not yet been found to be expressed independently of one another and are often referred to as Ly-2,3 antigens.) Antigens expressed in varying amounts on lymphoid cells but not present on all other tissues are called differentiation antigens (Boyse and Old, 1969), and they can be valuable tools in studying the differentiation of progenitor cells. Bone marrow precursors of T cells, for example, acquire in the thymus the differentiation antigens Thy-1, Tla, Ly-1, and Ly-2,3; upon differentiation into mature T cells, thymocytes completely lose Tla antigens, express much less Thy-1 antigen, and upon further functional commitment express either Ly-1 or Ly-2,3 antigens (see review in Cantor and Weissman, 1976). Histocompatibility loci control the expression of antigens detectable by tissue (primarily skin and tumor) transplantation. More than 30 histocompatibility loci have been defined in the murine genome (J. Klein, 1975). The major histocompatibility locus in the mouse is the H-2 complex which is in fact the only strong histocompatibility locus, i.e., “a locus such that a difference between donor and host at this

465

GLYCOPROTEIN ANTIGENS

locus will prevent the progressive growth of nearly all tumor homotransplants and cause the rapid rejection of skin homografts” (Counce et al., 1956). The H-2 gene complex is located on chromosome 17 and is an extremely complex region consisting possibly of several hundred closely linked loci. It is presently divided into five regions plus subregions (the map of the subregions is not definitive) as illustrated in Table I. Only loci representing cell surface antigens are listed; H-2K and H-2D alloantigens coded for by the H-2K and H - 2 D loci are the major histocompatibility antigens and are often referred to as transplantation antigens; they are present on most tissues. The l a loci code for Ia antigens (Ia stands for I-region-associated) which are expressed mostly on B lymphocytes but also on other cells and can also be classified as differentiation antigens. The H - 2 gene complex plays a fundamental role in such diverse immunological functions as graft rejection, lymphocyte recognition and collaboration, control of antibody response to several antigens, and recognition of virus-infected cells (J. Klein, 1975). Murine alloantigens can be genetically defined and serologically identified, and their role in immunological reactions can often be tested because of the stage of development of mouse immunology. However, the biochemical characterization of these lymphocyte membrane components has been hampered by the small amount of material available for purification, and very often by the lack of sensitive microtechniques for the isolation and quantitation of membrane components. Cell surface antigens are mostly membrane glycoproteins and are partially embedded in the hydrophobic core of the

TABLE I THEH-2 GENECOMPLEX: PRESENT NOMENCLATURE“

K region

S region

G region

D region

I-A, I-B, 1-J, I-E, I-C

-

-

-

Ia-1, la-2, Ia-4, la-S,la-3

-

I region

~

Tentative subregions Loci controlling cell membrane antigens a

-

H-2K

H-2G H-2D (erythrocyte antigen)

For references, see J. Klein (1975); Klein et al. (1975); and Shreffler (1978).

466

MICHELLE LETARTE

plasma membrane. Their solubilization and purification in an intact form requires the use of detergents (Robinson and Tanford, 1975). The biochemical data accumulated so far on murine cell surface antigens is rudimentary and stresses the necessity for using more quantitative techniques and particularly micromethods in order to isolate and purify intact cell surface antigens in sufficient quantity to characterize their structure and function. We discuss the techniques presently utilized to identify and quantitate cell surface antigens before summarizing the biochemical data available on H-2 and Ia, the antigens of the major histocompatibility gene complex, and on Thy-1, Tla, and Ly-2,3, T-cell differentiation antigens. In order to remain brief, this chapter is restricted to murine antigens which have been partially characterized biochemically. [Discussion of Glx antigen, a murine leukemia virus (MuLV)-associated antigen is not included; GIx is one of the antigenic determinants expressed on gp70, the major glycoprotein component of the MuLV envelope (Obataet al., 1975; Tunget al., 1975a,b).]The nature ofthe antigens coded for by the major histocompatibility complex has also been investigated in other species, particularly in humans and guinea pigs; from preliminary biochemical data presently available, it seems that both transplantation antigens and “Ia-like” antigens show interspecies homology. (A whole section of Cold Spring Harbor Symposdum, Vol. 41, 1976, is concerned with a discussion of the molecular products of the major histocompatibility complex, as well as their structure, genetics, and immunological functions.) Little information is available concerning T-cell differentiation in species other than murine, but it is likely that such antigens will be described soon and that their biochemical properties will be found to be similar to those of murine glycoprotein antigens. Immunological phenomena often appear complex and, in our view, this is a reflection of a lack of understanding of the molecular mechanisms underlying these phenomena. A direct approach to this problem is to purify, in a functionally intact form, molecules believed important in immune reactions (such as Ia antigens, for example), and to test their ability to replace a given cell type or factor in these reactions. Not all glycoprotein antigens may be involved directly in immunological mechanisms, but the expression of a given antigen at particular stage(s) of differentiation of cells (e.g., Thy-1 or Ly-antigens) could be correlated with the particular functions of these cells. To biochemists, the glycoprotein antigens of lymphocytes represent another class of membrane molecules to be investigated, and the questions of interest are thus: structure of the molecules, role of the

GLYCOPROTEIN ANTIGENS

467

carbohydrate side chains, localization of antigenic determinants, integration in the membranes, possible enzymic function, and so on. This chapter stresses the fact that biochemical characterization of glycoprotein antigens is still in its infancy and is very much dependent on micromethodology.

II. METHODS OF ANALYSIS OF CELL SURFACE ANTIGENS

A. Production of Antisem to Cell Surface Antigens

Antisera to cell surface antigens can be produced between strains of a given species (referred to as alloantisera) and recognize alloantigens determined by allelic polymorphism within the species. Alloantisera to lymphocyte antigens are produced by multiple injections of lymphocytes; antibodies against H-2 histocompatibility antigens can be produced by skin graft, by lymphocyte injection, or by a combination of both. An alloantiserum to a cell surface protein may recognize minor biochemical differences, such as substitution of a single amino acid (Boyse et nl., 1971b), and can thus be a useful tool in studying genetic variations in membrane proteins of different strains. The same antiserum can also b e used to study the tissue distribution of the protein bearing the antigenic determinant defined by the serum. A great advantage of antisera to murine lymphocyte alloantigens is that they can be genetically defined; congenic strains of mice can be produced by back-crossing, such that ideally they differ only at the locus coding for the antigen selected for (Snell, 1958). The identification of cellular antigens with alloantisera is limited to genetic polymorphisms resulting in detectable changes in antigenicity and to the number of available inbred lines of mice; however, it is also possible to define cellular antigens using xenoantisera (also called heteroantisera) (Williams, 1977).Heteroantisera to cellular antigens are produced between species by immunization with either cells or membranes, the rationale being that most proteins have diverged sufficiently in the course of evolution to be antigenic in another species. Xenoantisera usually contain antibodies against a variety of antigens and must consequently be analyzed with great care and absorbed extensively before they can define single antigens. Very often, antibodies raised against purified subclasses of lymphocytes recognize mostly lymphocyte-specific antigens rather than antigens associated with the functional subclasses. For example, a recent study

468

MICHELLE LETARTE

(Fabre and Williams, 1977) showed that rabbit antisera to rat thoracic duct lymphocytes or to purified T cells from the same source contained antibodies directed mostly at an antigen shared by bone marrow cells, thymocytes, macrophages, and peripheral lymphocytes. Much smaller amounts of antibody were specific for thoracic duct lymphocytes and were directed at an antigen shared by B and T cells, and only very small amounts of antibodies were even potentially specific for T cells (Fabre and Williams, 1977).The approach of raising an antiserum against whole cells or membranes bearing a given cell surface protein (identified in a mutant, for example), in order to obtain an antiserum specific for that particular component by absorption with cells known not to express this protein, is not often rewarding. In order to obtain xenoantisera specific for membrane proteins, it is thus advisable, whenever possible, to immunize with purified or at least partially purified material. Once a cellular antigen has been obtained in a purified form, following its purification with a genetically defined alloantiserum or with a well-characterized and adsorbed xenoantiserum shown to have restricted specificity, a xenoantiserum specific for the purified antigen can then be produced (usually with no or minimal absorption) and used routinely to identify it. Both alloantisera and xenoantisera can thus be used to identify cellular antigens and follow their purification. In some instances, as in the case of Thy-1 antigen, it has been possible to show that a xenoantiserum and an alloantiserum can recognize different antigenic determinants located on the same molecule (Morris et

al., 1975). A cautionary remark should be added at this point concerning the presence of antiviral antibodies in alloantisera. Mice from a wide variety of inbred strains produce antibodies against endogenous RNA type-C virus (Nowinski and Kaehler, 1974; Aaronson and Stephenson, 1974; Lee and Ihle, 1975). Certain alloantisera prepared in mice against either H-2 or Ia antigens were shown to be very cytotoxic to murine leukemia and sarcoma cells in culture. This unexpected reactivity of the antisera was demonstrated to be due to the presence of antibodies against murine leukemia virus proteins p15 and gp70; sera from aged, unimmunized mice of the same strains as the immunized mice also contained similar antibodies, indicating their natural occurrence in mice (P.A. Klein, 1975;Nowinski and Klein, 1975).In studying the expression of alloantigens on cultured tumor cells, one should be aware that neither direct cytotoxicity nor binding of alloantisera to tumor cells implies the presence of the corresponding alloantigens on these cells and that such effects could easily be due to viral antigens; it is necessary to look at the inhibition by tumor cells of the binding of

GLYCOPROTEIN ANTIGENS

469

alloantisera to normal tissues carrying the alloantigens. However, normal mouse tissues also carry murine leukemia virus; thymocytes of several strains of mice express gp70 and/or p30 viral proteins (Strand et al., 1974). This means that the reactivity of alloantisera with nontumor cells could also be due, at least in part, to the presence of viral antigens. We have demonstrated the presence of antiviral antibodies in Thy-1.2 sera (AKR anti-C3H) and shown that the reactivity of the antisera for DBA/2 thymocytes could be inhibited as much as 60% by absorption with disrupted gross passage A murine leukemia virus, whereas only 30%of the reactivity of the serum for thymocytes could be inhibited by purified Thy-1 antigen (MacDonald, M. and Letarte, M., unpublished results). 6. Analysis of Specificities Defined by Antisera to Cell Surface Antigens

1. CYTOTOXICITY ASSAY The cytotoxicity assay is the most popular method for the detection of cell surface antigens. It relies upon the killing of cells by antibodies in the presence of complement. The dead cells are scored for their inability to exclude from the cytoplasm a vital dye such as trypan blue or eosin Y (Gorer and O’Gorman, 1956), or for their ability to release into the medium incorporated W r (Sanderson, 1964). Several modifications of the cytotoxic test have been described ( J . Klein, 1975).Cytotoxic assays are quick, simple, and require very few cells. However, they are not quantitative, and variations in the complement activity of the rabbit or guinea pig sera remain a major problem. The reactivity of antisera with various target cells can be assessed directly by measuring the percentage of cell lysis at various antiserum concentrations. To establish if the reactivity of a given serum with various cell types is due to cross-reacting antigens and/or to measure the relative amount of a given antigen on different tissues, inhibition assays can be performed. The antiserum at a constant dilution known to give 50-75% killing of the target cells is incubated, without complement, with increasing dilutions of the test cells, and the residual cytotoxic activity is then measured against a fixed number of target cells in the presence of complement ( J . Klein, 1975).

2. FLUORESCENT ANTIBODY TEST The specific fluorescence emitted by fluorescent antibodies offers a rapid way of visualizing antigens on cells in suspension or on sections

470

MICHELLE LETARTE

of frozen or fixed tissue (Coons et at., 1955).The indirect method most widely used consists of incubating cells or tissue sections with the antiserum directed at the cellular antigen being studied, washing, and testing for any antibody remaining specifically bound by using a fluorochrome-conjugated antiimmunoglobulin serum of appropriate species specificity. Thus a single preparation of labeled anti-IgG can be used for the detection of any specific antibody raised in a given species. A practical difficulty of the immunofluorescence technique is the occurrence of “nonspecific” fluorescence, i.e., not attributable to the antibody-antigen reaction under study, which must be prevented by suitable preparation of samples and reagents. IgG fractions of antiimmunoglobulin sera must be purified before they can be coupled to fluorochromes. Fluorescein isothiocyanate, because of its high fluorescence emission intensity and ease of coupling to proteins, is the principal fluorochrome used for immunological tracing; rhodamine T is the usual additional fluorochrome in double-labeling studies. Methods involved in the coupling of fluorochrome to immunoglobulins and in purification of the conjugated products are discussed in the following articles: Rinderknecht (1962), Cebra and Goldstein (1965), Wood et al. (1965), The and Feltkamp (1970a,b), and Johnson and Holborow (1973). A modification of the indirect fluorescent labeling method, called the “hapten-sandwich” procedure, has also been described (Wofsy et al., 1974). Haptens are azocoupled to immunoglobulins specific for a cell surface antigen, and the hapten-modified cell-bound antibodies are visualized by adding fluorescent antihapten antibody. As many different purified anti-azophenyl-hapten antibodies are available, it makes it feasible to distinguish more than one membrane antigen in a single labeling experiment (Wofsy et al., 1974). The azocoupling procedure allows the addition of amplifying hapten groups to immunoglobulin fractions from specific antisera; it is particularly useful for the labeling of alloantisera where amplification with antiimmunoglobulin reagents is complicated by the presence of surface immunoglobulins in the cell suspension or tissue being studied. Another fluorescent sandwich technique now available involves the fluorescence-labeled protein A from Staphylococcus aureus which binds to the Fc portion of IgG molecules and can thus detect surface-bound antibodies (Ghetie et al., 1974). Fluorescent antigen-antibody reactions can be observed by microscopy under conditions optimizing maximal specific fluorescence (Johnson and Holborow, 1973). It is also possible to analyze and sort cells specifically labeled with fluorescent antibodies with a laser ana-

GLYCOPROTEIN ANTIGENS

47 1

lyzer-a fluorescence-activated cell sorter (Bonner et al., 1972; Hulett et al., 1973). This system allows the rapid separation of distinct groups of viable cells on the basis of their fluorescence intensity, their lightscattering characteristics, or different combinations of these two variables. Cells expressing various amounts of a given surface antigen can thus be sorted, recovered, and tested in functional assays. The cell sorter is a very sensitive analysis system; cells bearing only a few thousand fluorescent molecules can be detected and sorted. Thus extreme care must be taken in the preparation of reagents to avoid sorting the cells on the basis of fluorescence not due to the antigen-antibody reaction being studied.

3. CELLULAR RADIOIMMUNOASSAY Cytotoxicity assays and fluorescent antibody tests are very useful in the study of cellular antigens, but they do not permit quantitative measurement of the binding of antibodies to cellular antigens. Radioimmunoassays, however, are very effective in studying antigenantibody reactions quantitatively and constitute specific, sensitive, and practical tests, particularly in endocrinology, for the microdetermination of specific proteins in unfractionated mixtures (Hunter, 1073). Radioimmunoassays make use of radioactively labeled purified antigens; however, the cellular antigens in which we are interested are not usually available in a purified form. The problem can be overcome by the use of a cellular radioimmunoassay in which the labeled entity is a purified antiimmunoglobulin reagent. The cells bearing the antigen are first incubated with a specific antiserum, and the antigenantibody reaction is quantitated by a second incubation with lz5I-labeled purified antiimmunoglobulin antibody. Thus the specific antibody bound to a cellular antigen on a target cell behaves in turn as an antigen for the labeled antiimmunoglobulin antibody; one is thus dealing with two antibody-antigen reactions. The technicalities of the cellular radioimmunoassay, also referred to as the indirect radioactive binding assay have been described by Acton and colleagues (1974), Morris and Williams (1975), and Williams (1977). The antiimmunoglobulin reagents are antibodies purified by affinity and treated with pepsin; F(ab’)2fragments are always used as labeled reagents to avoid aggregation and anomalous binding via the Fc portion of the IgG molecule (Jensenius and Williams, 1974). Iodination of molecules is performed by the chloramine-T method, ensuring that the chloramine T/protein ratio is minimum to preserve biological activity (Sonoda and Schlamowitz, 1970). Both

472

MICHELLE LETARTE

xenoantisera and alloantisera can be analyzed by the cellular radioimmunoassay. However, with alloantisera, it is necessary to minimize the binding of the labeled antiimmunoglobulin reagent to surface Ig of the target cells, particularly when using lymph node or spleen lymphocytes as targets. It is possible to do this by using an antiimmunoglobulin reagent prepared against the purified Fc portion of the IgG molecule; it has been demonstrated that the binding of anti-Fc antibodies to cell surface immunoglobulin is much less than the binding of anti-IgG or anti-Fab antibodies (Williams, 1975). Rabbit antimouse IgG-Fc can be used successfully in the assay of Ia and H-2 alloantigens on spleen or lymph node lymphocyte targets (Letarte and Meghji, manuscript in preparation). The cellular radioimmunoassay is thus of general application for alloantisera and xenoantisera of different species, providing suitable purified antiimmunoglobulin reagents are available. The murine antigens Thy-1, H-2, and Ia (see above), as well as the Tla antigen (Esmon and Little, 1976), have been studied using this assay. Complex xenoantisera such as rabbit anti-rat brain serum (Morris and Williams, 1975; Morris et al., 1975), rabbit anti-rat thoracic duct lymphocytes (Fabre and Williams, 1977), rabbit anti-rat thymocyte membranes (Morris and Williams, 1978), and rabbit anti-erythrocyte ghosts (MacDonald et al., 1978)have been analyzed by the cellular radioimmunoassay and shown to contain several antigenic specificities which can be removed independently by selective absorptions. As mentioned in Section II,B,2, protein A from S . aweus can bind specifically to the Fc portion of IgG molecules (Forsgren and Sjoquist, 1966). An indirect radioactive binding assay has thus been set up using '251-labeledprotein A instead of '251-labeledantiimmunoglobulin as the labeled reagent (Dorval et al., 1975; Welsh et al., 1975), and histocompatibility antigens of the human, mouse, and rat have been shown to be detected specifically by this system. This modification of the indirect binding assay, if proven to be generally applicable, would simplify greatly the preparation of antiimmunoglobulin reagents, since protein A can bind IgG molecules of most mammalian species Kronvall et al., 1970) and can be prepared easily (Hjelm et al., 1972; Kronvall, 1973). However, the use of protein A in the indirect binding assay has to be investigated further and quantitated before it can be said to be as sensitive and specific as the double-antibody assay. The cellular radioimmunoassay can be used to estimate the number of specific antibody molecules bound per cell, providing that all the antigenic sites are saturated b y the specific antisera and that in turn all the antibodies bound are saturated by the antiimmunoglobulin re-

GLYCOPROTEIN ANTIGENS

473

agent. The ratio between the two antibodies can be estimated as follows. After incubation with specific antisera, an aliquot of the washed cells is solubilized, and the amount of immunoglobulin bound is measured b y inhibition of a radioimmunoassay for the given immunoglobulin; another aliquot of the washed cells is incubated with lZ5I-labeled antiimmunoglobulin, and the counts per minute bound to the cells estimated (Morris and Williams, 1975). Ratios have been calculated for several antiimmunoglobulin reagents, and it appears that, for anti-Fab or anti-Fc reagents, the ratio is between 1: 1 and 2 : 1, whereas for antiwhole IgG, the ratio is 4 : 1 (Williams, 1977).The number of antigenic sites per cell can be deduced from the number of antibody molecules bound per cell if the ratio between antigen and antibody is known. For Thy-1 antigen, it was possible, by comparing the number of antibody molecules bound under saturating conditions with the amount of Thy-1 estimated per cell by calibration of the cellular radioimmunoassay with purified Thy-1, to estimate that this ratio was 1 (Williams et

d,1976). It is also possible, using the cellular radioimmunoassay to estimate the concentration of antibodies in the antiserum in the presence of an excess of target cells. In order to do this, it is necessary to demonstrate, at the concentration of labeled antiimmunoglobulin reagent used in the assay, that the reaction between the first and second antibodies has come to an equilibrium so that all the molecules bound in the first incubation are detected (Hunter, 1973). Figure 1 illustrates the titration of a rabbit anti-murine erythrocyte ghost serum with an excess of target erythrocytes. T h e percentage of the total counts per minute of '251-labeled anti-rabbit IgG added to the assay which bound to the antibody-coated targets was measured at two concentrations of labeled reagent (1.0 and 0.03 pug per assay). The rabbit antibody bound is detected proportionally in the linear range of both curves (25% binding is obtained at antiserum dilutions of 65 and 2000 using 1.O and 0.03 p g of antiimmunoglobulin, respectively). However, the lower concentration of antiimmunoglobulin reagent becomes limiting at dilutions of first antibody smaller than 250 and is thus only quantitative at large dilutions of the latter. The concentrations of erythrocytedirected antibodies present in the antisera can be calculated from the concentration of antiimmunoglobulin bound, using a ratio of four molecules of horse anti-rabbit whole IgG bound per molecule of rabbit IgG (Morris and Williams, 1975). A value of 500 pg/ml of antiserum can be approximated from both curves at low dilutions of antisera. It should be stressed that the antisera being analyzed very often contain a mixture of antibodies at different concentrations and with different

474

MICHELLE LETARTE

Reciprocal antiserum dilution

l 7 . f

FIG.1. Titration of rabbit anti-mouse erythrocyte ghost serum on erythrocytes at two concentrations of antiimmunoglobulin reagent. The binding of rabbit anti-erythrocyte ghost serum (circles) or normal rabbit serum (squares) to an excess of 3 x 106 erythrocytes was determined using either 1 pg (solid symbols) or 0.03 p g (open symbols) of '2sI-labeled F(ab'), horse IgC anti-rabbit IgC (HAR) per sample. Target cells (25 p l ) were incubated for 1 hour at 4°C with the normal serum or antiserum; after two washings, the antibody-coated cells were incubated with the labeled reagent for 1 hour, washed twice, and counted. The results are expressed as the percentage of the total counts per minute added which are bound to the cells during the reaction (MacDonald et al., 1978).

affinities, and that the concentration of antibodies measured should be taken only as an estimate of an antiserum potency. The sensitivity of the quantitative assay, i.e., the extent to which the labeled reagent can be diluted and still titrate linearly the amount of specific antibody bound to the target cells in excess, is dependent on the affinity and concentration of antibodies in the serum being tested. These two parameters affect the equilibration time of the reaction; for practical purposes, incubation times of 1-2 hours are chosen for cellular radioimmunoassays and it must be demonstrated that the reaction has come to equilibrium in this short period of time before antibody concentration can be estimated. We observed that, for certain sera, the reaction did not reach equilibrium by 2 hours, even at high concentrations of antiimmunoglobulin reagent, but did so by 16 hours. Thus, when comparing the potency of different antisera, it is necessary to show that all the reactions have come to equilibrium or, in other words, that no more binding can be detected by increasing the time of reaction. The cellular radi,oimmunoassays can also be used under conditions where the amount of '"I-labeled antiimmunoglobulin bound is not in the proportionality range with respect to the specific antiserum bound as described in Fig. 1. It is, for example, desirable to perform in-

GLYCOPROTEIN ANTIGENS

475

hibition studies at low concentrations of antiserum, as less material is required for absorption. Inhibition studies can be done in a relatively quantitative manner by selecting a concentration of antiimmunoglobulin reagent (in our system, of the order of 0.03 p g per assay is suitable), titrating the specific antisera at that concentration of labeled reagent, and choosing for the inhibition assays a concentration of antisera on the linear portion of the curve. Such an assay can be used to analyze antibody specificities (see Fig. 4), to study relative tissue distribution of antigens, and to follow the purification of antigens. As target cells can be fixed with glutaraldehyde without affecting their antigenicity (Williams, 1973), inhibition studies can be preformed on detergent extracts. As most cell surface antigens are hydrophobic membrane proteins, they can be isolated only in an intact form with the help of detergents (Robinson and Tanford, 1975; Letarte-Muirhead et al., 1974), and their activity during purification can be measured by inhibition of the binding of antisera to glutaraldehyde-fixed target cells. The cellular radioimmunoassay, although it cannot be equated to the classic radioimmunoassay as a quantitative way of measuring antibody-antigen reactions, is a very useful tool in studying cellular antigens. It is in our view the best way of analyzing complex antisera and of following the purification of cellular antigens.

4. INDIRECTIMMUNOPRECIPITATION METHOD A technique which is routinely used for the identification of murine alloantigens is indirect immunoprecipitation which was first described by Schwartz and Nathenson (1971). Lymphocytes or tumor cells bearing the antigens are first labeled with radioactive amino acids or sugars in vitro for 4-6 hours; surface-labeling techniques can also be used. (Glycoprotein-labeling methods are described by Juliano in this volume). Cell membranes are solubilized with nonionic detergents, and the specific alloantiserum is reacted with the radiolabeled solubilized fraction. Soluble antibody-antigen complexes are then precipitated by the addition of an antiimmunoglobulin reagent. The immunoprecipitate, after washing, is solubilized in sodium dodecyl sulfate (SDS) and analyzed by SDS polyacrylamide gel electrophoresis. The gels are cut into thin slices, and the radioactivity in each gel slice determined (Schwartz and Nathenson, 1971). The SDS polyacrylamide gel pattern of the immunoprecipitate under reducing and nonreducing conditions can reveal how many peaks are present, ideally how many antigens are precipitated, their approximate MW and,

476

MICHELLE LETARTE

if a single antigen is detected, the number of associated subunits or polypeptide chains observed. The technique has been used to identify H-2 antigens (Schwartz and Nathenson, 1971),Ia antigens (Cullen et al., 1974), Tla antigens (Anundi et al., 1975), Thy-1 antigen (Trowbridge et al., 1975), and Ly-3 antigen (Durda and Gottlieb, 1976). A modification of the indirect immunoprecipitation technique has been described recently (Cullen and Schwartz, 1976), again making use of the fact that brotein A from S. aweus binds to the Fc portion of the IgG molecule (Forsgren and Sjoquist, 1966; Kronvall and Frommel, 1970). After incubation with alloantisera, the soluble immune complexes are bound to the protein-A sites on the S. aureus, the latter having been heat-inactivated and fixed prior to the assay. The immune complexes can be easily eluted from the staphylococci for electrophoretic analysis. Precipitation with staphylococci is rapid, more efficient, and less artifact-prone than precipitation with antiimmunoglobulin reagents (Cullen and Schwartz, 1976). Although the immunoprecipitation technique is very useful, it is not quantitative in the sense that the number of antigenic sites on cells, as well as the concentration of antibodies present in the antiserum, cannot be estimated. Unless solubilization conditions are optimized, certain antigens may not be detected; e.g., Thy-1 antigen is solubilized very poorly by Nonidet P40 (NP-40), a detergent used by many workers (Letarte-Muirhead et al., 1974). Nonspecific precipitation of material can occur, as well as dissociation of antibody-antigen complexes with low-affinity serum during the washing of the precipitates. Immunoprecipitation is not a technique that can be easily used to analyze complex antisera or to follow antigen activity during purification (Williams, 1977). However, it is a good method for identifying antigens with well-defined antisera, especially when small amounts of material are available. C. Microsequencing Techniques

One of the major problems involved in the isolation of cellular antigens from mouse tissues is the small amount of material available, and thus microtechniques for isolation and structural analysis are more than welcome. Microsequencing techniques have been used recently for investigating preliminary sequences of H-2 antigens (Silver and Hood, 1975; Vitetta et al., 1976a; Ewenstein et al., 1976; Henning et al., 1976), Ia antigens (Cecka et al., 1978), and HLA antigens (Bridgenet al., 1976; Ballou et al., 1976). Radiolabeled antigens isolated by immunoprecipitation and electrophoretic analysis (see Section II,B,3)

477

GLYCOPROTEIN ANTIGENS

can be eluted from the gels, concentrated, and sequenced in the presence of a carrier. The radioactive phenylthiohydantoin derivatives obtained by sequential protein degradation can be analyzed by thinlayer chromatography and counted (Silver and Hood, 1975,1976a;Vitetta et al., 1976a). Several variations of the microsequencing techniques have been reported including the coupling of high specific [35Slphenylisothiocyanate to a protein band eluted from SDS gels (Bridgen et al., 1976) rather than the coupling of “cold” phenylisothiocyanate to radiolabeled protein bands. It is also possible to sequence proteins in which all amino acids have been labeled by incubating cells with [14C]aminoacids and [14C]pyruvate.One introduces a quantitative analytical step to identify the labeled amino acid derivatives (Ballou et al., 1976). So far, microsequencing techniques have given only partial sequences ofthe N-terminal20 to 30 residues of cellular antigens. However, the art is in its infancy, and it should be possible to obtain data on internal peptides once the overlap and background problems have diminished. Additional discussion of these techniques and of their problems is available in Bridgen (1976), Ballou et al. (1976), and Capra et ul. (1976).

111.

ISOLATION AND CHARACTERIZATION

OF

H-2 ANTIGENS

A. Tissue Distribution and Polymorphism of H-2 Antigens

H-2 antigens are present in most if not all tissues tested, but considerable variation in the amount of H-2 antigen is observed in different tissues. Spleen has the highest amount of H-2 antigen (relative amount l.O), followed by lymph node (0.7),liver ( O S ) , thymus, lung, adrenal, gut (0.3), kidney, salivary glands (0.l), erythrocytes, heart, brain, testis, and skeletal muscle (0.01)as measured by inhibition of the cytotoxicity assay (J. Klein, 1975). The H-2 system is very polymorphic. In early studies, H-2 antigens were assigned specificity numbers, and a linear genetic map of H-2 specificities was constructed (a specificity being defined here as the portion of the antigen molecule which reacts with the combining site of the antibody molecules used to define it). However, further genetic studies revealed that at least two genes coded for H-2 antigens, which were subsequently named H-2K and H-2D genes (Shreffler et al., 1971; Snell et al., 1971). Specificities which are uniquely determined by either the H-2K or the H-2D gene are called private specificities; each H-2

478

MICHELLE LETARTE

haplotype (chromosome)has two private specificities, one determined by the K gene and one by the D gene. Specificities which map either in K or D or both and which can be shared by nonidentical haplotypes are called public specificities. For example, haplotype H-2 contains H-2Kband H-2Db regions; private specificities 33 and 2 are defined by the Kb and the Db regions, respectively. The public specificities assigned to the Kb region are 5,27,28,29,35,36,39, and 46; and, to the Dbregion, 6,27,28, and 29 (J. Klein, 1975). Inbred strains bearing the same H-2 haplotype (68 haplotypes have been defined so far) share the same H-2 specificities (J. Klein, 1975). The large number of H - 2 haplotypes is an attestation of the unusually high degree of polymorphism of H-2 antigens, which may be important in their recognition function at the cell surface. As discussed in Section III,E, the antigenicity of H-2 appears to be associated with the polypeptide chain, and thus specificities must reflect variations in amino acid sequence. B. Isolation, MW, and Subunit Structure of H-2 Antigens

H-2 antigens are membrane glycoproteins, and thus their isolation requires solubilization from the hydrophobic core of the plasma membrane. Attempts to isolate intact H-2 antigens in high yield in a biologically active form have been hampered by the small amount of starting material available, the inability to measure antigenic activity in detergent extracts using cytotoxicity assays, and the complexity of H-2 polymorphism. The approaches yielding the most information have included: 1. Limited proteolysis of membranes or whole cells with papain which releases H-2 antigens in a water-soluble form (Shimada and Nathenson, 1967, 1969) 2. Solubilization of cell membranes with the nonionic detergent NP-40 followed by immunoprecipitation of the detergent-soluble antigens with H-2 antisera and identification by polyacrylamide gel electrophoresis (Schwartz et al., 1973). The proteolytically released H-2 antigens, although fragments of intact H-2 molecules, carried at least some of the antigenic specificities [some antigenic specificities were recovered in low yield after papain digestion (Shimada and Nathenson, 1969)]and were partially purified chemically and allowed establishment of the glycoprotein nature of the molecules bearing H-2 alloantigenic sites. The fragments were purified by several steps including ammonium sulfate precipitation,

GLYCOPROTEIN ANTIGENS

4 79

gel and ion-exchange chromatography, and disk gel electrophoresis (Shimada and Nathenson, 1969). Purified fractions obtained were homogeneous, although the K and D gene products were copurified, showing an increase of 300- to 700-fold in specific activity, but the overall yield was less than 0.1% (Shimada and Nathenson, 1969). The major papain fragments had a MW of 37,000 and contained 90% protein and 10% carbohydrate; fragments differing in their H-2 haplotype (H-2b and H-2d) were similar in overall chemical analysis (Shimada and Nathenson, 1969). The glycoprotein nature of the intact H-2 antigen obtained by immunoprecipitation of NP-40-solubilized spleen cells was established by a double-labeling experiment (Schwartz et al., 1973). Immunoprecipitates obtained from cells labeled with both ['4Clleucine and [3H]fucose were shown to coelectrophorese in a single band, and the protein and sugar label could be dissociated only after exhaustive pronase digestion. The approximate MW of the NP40-H-2 antigen complex was 380,000 by gel filtration. This value probably includes large amounts of NP-40 bound to the H-2 molecules, thus increasing their apparent size and affecting elution behavior and MW determinations (Schwartz et al., 1973; Tanford et al., 1974). When the NP-40-H-2 complex was solubilized in SDS, a MW estimate of 45,000 was obtained for the H-2 glycoprotein (Schwartz et al., 1973). Papain cleavage presumably leaves an 8000-MW fragment (45,000 for the intact H-2 glycoprotein and 37,000 for the papain fragment) in the membrane. Although this fragment has not been recovered, it was shown by partial sequence analysis that the 45,000-MW and the 37,000-MW papain fragments had the same N-terminal sequence (Ewenstein et al., 1976).Thus the papain fragment carries the carbohydrate moiety and the private as well as some of the public antigenic specificities of H-2 molecules (Shimada and Nathenson, 1969). Two separable H-2 glycoproteins, determined by the H-2K and H 2 0 genes, respectively, were shown to be present in cell extracts; the allocation of specificities defined serologically to isolated H-2 glycoproteins was studied by the immunoprecipitation of radiolabeled cells. For example, specificities 33 and 2 were assigned, respectively, to the K and the D gene products of H - 2 haplotype, as they could not be coprecipitated; public specificity 5 could be precipitated by either anti-33 or anti-5 sera but not by anti-2 sera, suggesting that it was present on the H-2Kb glycoprotein (Nathenson and Cullen, 1974). Similar experiments were done with other specificities and haplotypes, and it was concluded that, in homozygous cells, at least two glycoproteins carried the antigenic specificities of a given haplotype, the private specificities being associated with either the H-2K or the H-2D gly-

480

MICHELLE LETARTE

coprotein, whereas the public specificities could be assigned to either or to both H-2 glycoproteins (Nathenson and Cullen, 1974).In H-2 heterozygous cells, four separable gene products can be immunoprecipitated independently: one for each of the H - 2 K genes of the parental haplotypes and another for each of the H - 2 D genes of the parental haplotypes (Cullen et al., 1972). It has been shown recently that H-2 antigens are noncovalently associated with a small polypeptide chain of approximately 12,000 MW which has been identified as a homolog of p,-microglobulin on the basis of antigenic cross-reaction with and its ability to bind with antihuman &-microglobulin (Rask et al., 1974)and anti-rat &-microglobulin (Natori et al., 1974; Vitetta et al., 1976b) and from preliminary sequence data (Silver and Hood, 1976a; Law and Appella, 1976). The H-2 glycoproteins obtained either by NP-40 solubilization or papain treatment of radiolabeled spleen cells can be precipitated by antisera to &-microglobulin or to H-2 allospecificities, giving two peaks of radioactivity on SDS gels, at 45,000 and 12,000 MW for the intact glycoprotein, and 37,000 and 12,000 MW for the papain fragment, respectively (Rask et nl., 1974; Silver and Hood, 1974). It has been shown that the human &-microglobulin gene is on a different chromosome than the H L A gene complex (major histocompatibility complex in human) (Goodfellowet al., 1975),suggesting that the association of& microglobulin with histocompatibility antigens (seen in all species studied so far) is not due to their respective genes being linked (see review on /3,-microglobulin, Poulik and Reisfeld, 1975). C. Partial Amino Acid Sequence of H-2 Glycoproteins

Amino acid analyses have been performed on the H-2 alloantigens fragments obtained by papain treatment of H-2b and H-2d spleen cells whose purification was reported earlier (Section III,B), and only small differences were observed (Shimada and Nathenson, 1969). Although the data suggested a similiarity in the overall amino acid composition of the H-2 glycoprotein fragments obtained from two different haplotypes, it must be kept in mind that the preparations from haplotypes b and d, which were used for analysis, contained both the H-2K and H2D antigens and that they appeared to be present in different amounts in the two preparations. Amino acid compositions of purified H-2K and H-2D glycoproteins from different haplotypes could thus show more variation. The peptide compositions of H-2 glycoproteins from different strains have also been compared. In early studies, the peptides pro-

GLYCOPROTEIN ANTIGENS

481

duced by cyanogen bromide cleavage and tryptic digestion of H-2b and H-2d papain (37,000-MW) fragments were compared by a cellulose thin-layer peptide mapping microtechnique (Shimada et al., 1970).It was estimated that approximately 70% of the theoretical maximum number of peptides was visualized, i.e., 38 peptides were detected and shown to be identical for the H-2 papain fragments obtained from different strains of the same haplotype. However, 3 peptides of the H-2b and 4 peptides of the H-2d papain fragments appeared unique for their respective haplotype (Shimada et al., 1970). The preparations tested each contained both the H-2K and H-2D gene products; their peptide maps (as well as the amino acid composition mentioned above) reflect the summation of at least 2 molecular species, and it was suggested that the degree of difference observed was a minimum estimate because of the insensitivity of the techniques used (Brown et al., 1974). More recently, the peptide composition of the individual H-2K and H-2D gene products of haplotypes b and d has been investigated by using more sensitive techniques and has in fact revealed notable diversity (Brown et al., 1974). The individual H-2 glycoproteins were obtained from radiolabeled spleen cells by immunoprecipitation with monospecific antisera and were purified by molecular sieve chromatography in SDS. The purified 45,000-MW peaks obtained were digested with trypsin, and the peptides separated by gel chromatography. Double-label comparison of peptides of products of K or D genes showed dissimilarity: Kb and Db showed 11 similar peptide peaks out of a total of 21 and 24 peaks obtained, respectively; Kd and Ddproducts showed only 7 similar peptides out of 20 and 25, respectively. When comparing products of the same gene from different haplotypes, such as Kb versus Kd and Db versus Dd, only 8 to 9 of 20 to 26 peptide peaks showed similar chromatographic behavior (Brown et d.,1974). This method of analysis tends to overemphasize existing structural differences, but the results nevertheless suggest that H-2 glycoproteins can show quite an extensive degree of variability in peptide composition which would be consistent with their genetic polymorphism. T h e definitive way to assess variations in the primary structure of H-2 glycoproteins is to compare their amino acid sequences. With the recently introduced microsequencing techniques mentioned in Section II,C, preliminary data are now available from several laboratories and are summarized in Table 11. AlthGugh limited information is available, the data show several positions of identity in the four proteins reported, i.e., Kk,Kb, Db,and Dd.In addition, K and D molecules of the same haplotype share residues at several other positions. It is, however, unwise to extrapolate to the degree of homology found in

GLYCOPROTEIN ANTIGENS

483

the overall sequence of H-2 glycoproteins. Complete sequence data will be required; let us hope that it will be feasible to apply microtechniques to such a purpose, as the amount of purified material available for structural studies remains a problem in the study of murine antigens. A discussion of the genetic implications and evolutionary significance of the variations in the amino acid sequence of histocompatibility antigens has been presented by Silver and Hood (1976b).

D. Carbohydrate Structure of H-2 Glycoproteins

Elucidation of the structure of the carbohydrate portion of H-2 antigens has not been very fashionable lately, and the earlier work of Nathenson and co-workers remains the only information on the subject (Nathenson and Muramatsu, 1971; Nathenson and Cullen, 1974). Purified papain-solubilized H-2 fragments, as well as H-2 glycoproteins obtained by immunoprecipitation with anti H-2 serum from cells labeled with radioactive sugars, contain galactose, glucosamine, mannose, fucose, and sialic acid (Nathenson and Muramatsu, 1971). Glycopeptides were generated from detergent-solubilized or papainsolubilized partially purified H-2 antigens by exhaustive pronase digestion of the polypeptide backbone in order to investigate the properties of the carbohydrate chains and their sugar sequence. The glycopeptides generated from normal spleen cells and from tumor cells of haplotypes b and d could not be distinguished on the basis of their chromatographic profiles. A single peak was observed on Sephadex G-50 corresponding to an estimated MW of 3300 ? 500, whereas 2 major glycopeptides were identified on DEAE-Sephadex. The H-2 glycopeptides, which were estimated to carry 1 or 2 amino acid residues and 12 to 15 monosaccharide residues, appeared to be unique in size. As the carbohydrate content of H-2 was caIculated as 8-12% of the weight of the papain fragment (MW 37,000), it can be deduced that probably one, but certainly no more than two carbohydrate side chains are present per molecule of H-2 (Nathenson and Muramatsu, 1971). It must be reemphasized that both the H-2K and H-2D gene products were present in the preparations submitted to gel and ionexchange chromatography and that their respective glycopeptides (K versus D) could be charged differently, which would account for their differential migration on DEAE-Sephadex. Observations following the enzymic digestion and limited acid and alkali treatments allowed the deduction of a hypothetical model for

484

MICHELLE LETARTE

6-N-acetylglucosaminidase

[(MA"OSE)z Endo-glycosidase

(GLUCOSAMINE)l or

Or

____)

T

FUCOSE

[H1' Sensitivity

4

( G L U C S M N E ) l oI

b

[OH]- Stability

the H-2 glycopeptide which is illustrated in Fig. 2 (Nathenson and Muramatsu, 1971). The model applies for both H-2b and H-2d glycopeptides and is based on the following findings:

1. The sialic acid residues could be removed easily with Vibrio cholerae neuraminidase. 2. The galactose residues could be released (70%)as free galactose b y a P-galactosidase fraction from Diplococcus pneumoniae without release of mannose or glucosamine. 3. Half of the glucosamine and 80% of the galactose could be removed by incubation of the glycopeptides with a mixture of P-Nacetylglucosaminidase and P-galactosidase (both from D.pneumoniae) and neuraminidase. 4. At least 80% of the mannose and 25-30% of the glucosamine of the glycopeptide were released as an oligosaccharide (while nearly all the fucose stayed attached to the peptide portion) by an endoglycosidase from D. pneumoniae, in the presence of the exoglycosidases mentioned above. The enzyme appears to be of the endo-P-Nacetylglucosaminidase type and was shown to release the oligosaccharide (Man),-Glu-NH, from the glycopeptide of a myeloma IgG (Muramatsu, 1971), leaving intact the Fuc-Glu-NH, bound to the peptide. The enzyme has been purified further and shown to have strict specificity with respect to the oligomannosyl core of glycopeptides;

GLYCOPROTEIN ANTIGENS

485

in the presence of exoglycosidases, i.e., neuraminidase, P-galactosidase, and P-N-acetylglucosaminidase,the enzyme can act on intact glycoproteins (Koide and Muramatsu, 1974). 5. Eighty to ninety percent of the fucose was eluted as a monosaccharide after acid treatment and gel chromatography of the oligopeptide, suggesting that it was in a terminal position near the carbohydrate-peptide bond. 6. Treatment of glucosamine-labeled H-2 antigen with alkali or alkaline sodium borohydride followed by gel chromatography showed that all the label was still associated with the protein, suggesting that the carbohydrate protein linkage was of the glucosamine-asparagine type (Nathenson and Muramatsu, 1971). If the core were made up of three mannose residues and two glucosamine residues (for a total of five glucosamine residues for the glycopeptide), the structure would be very similar to that of the triple-branched glycopeptide B of porcine thyroglobulin (Toyoshima et al., 1972).

E. Nature of the Antigenic Determinants of H-2 Glycoproteins

Comparison of the amino acid and carbohydrate sequences of K and D gene products from different haplotypes will reveal the polymorphism of H-2 antigens and will allow the determination of invariable as well as variable and hypervariable residues. The assignment of antigenic specificities to amino acid or carbohydrate residues, or to a peptide or oligosaccharide, can be confirmed by isolating putative antigenic, structures, and testing for specific inhibition of antibody binding to the intact H-2 molecule. Indirect evidence can be obtained by showing that the removal of sugars by enzymic digestion or proteolytic cleavage of the polypeptide portion, or by chemical modification of a given residue, is associated with a loss of antigenic reactivity. As only preliminary N-terminal amino acid sequence data are available on a few haplotypes (Table 11) and, as the carbohydrate structure is still in a hypothetical form (Fig. 2), the direct approach is not as yet convincing with respect to the localization of antigenic specificities. However, as mentioned in Section III,C, comparative peptide mapping of different H-2 glycoproteins from several halotypes has shown variations in the peptide profiles of K and D gene products of the same haplotype and of products of alleles of the same gene (Brown et al., 1974). Preliminary amino acid sequence data (see Table 11) also suggest invariable and variable amino acid residues (Silver and Hood, 1976b), as well as interspecies homology with human and guinea pig histocompatibility antigens (Silver, 1976).

486

MICHELLE LETARTE

Indirect evidence accumulated so far, which is now summarized briefly, favors the expression of H-2 polymorphism within the amino acid sequence rather than the sugar structure of the glycoproteins. Isolated glycopeptides of H-2 antigens were unable to inhibit anti-H-2 activity, and radiolabeled glycopeptides could not be precipitated by antibodies directed at different H-2 specificities (Nathenson and Muramatsu, 1971). Enzymic removal of 100%of the sialic acid, 70%of the galactose, and 25% of the glucosamine did not affect the antigenicity of the H-2 molecule; treatment with the endoglycosidase from D . pneumoniae (mentioned in Section III,D), which removed 90%of the carbohydrate of H-2, did not impair its antigenicity (Nathenson and Cullen, 1974). Sensitivity to reagents which affect protein conformation has been reported for H-2 antigens; the treatment of tumor extracts with 6 M urea, or extremes of pH or proteolytic digestion, resulted in the loss of biological activity of H-2 (measured by time of graft survival) (Kandutsch and Reinert-Wenck, 1957). Reagents selectively affecting only certain amino acid residues can result in the loss of some or all antigenic specificities of H-2; the modification of tyrosine residues with either N-acetylimidazole or tetranitromethane resulted in a doserelated decrease in antigenic specificities 4 and 31 from H-2d molecules and 5 and 33 from H-2Kb molecules, respectively, but no given antigenic determinants were selectively lost by this treatment (Pancake and Nathenson, 1973). These results suggested that the tyrosine residues may not be part of the antigenic site of any particular specificity but that their modification affected the protein conformation necessary for intact antigenicity. Chemical modification of lysine by reductive methylation resulted in inactivation of specificity 4 with retention of 80% of specificity 31 from H-2d glycoprotein; it also destroyed specificity 33 with little effect on specificities 2 and 5 from H-2b glycoprotein. As specificities 33 and 5 are on the same molecule (H-2Kb),it can be concluded that inactivation of specificity 33 is not due to an overall conformational change and that it is likely that lysine is part of antigenic determinant 33, or even if located some distance from the antigenic site is essential in the conformation of the region bearing the determinant (Pancake and Nathenson, 1973). The polymorphism of H-2 is manifested not only in the expression of serologically detectable antigenic specificities, but also in graft rejection, mixed lymphocyte reactivity (MLR), cell-mediated lymphocytotoxicity (CML), and graft-versus-host response (GVH) (J. Klein, 1975). However, it is not known if the antigenic sites detected serologically are in fact the same as the sites involved in any of the phenom-

GLYCOPROTEIN ANTlG ENS

487

ena just mentioned, and the question of the assignment of structural elements to these phenomena is still open. Some H-2 mutants (H-2 haplotypes arising from parental types supposedly through mutations) have been described recently which give positive MLR, CML, GVH, and graft rejection between themselves and their parents but do not give rise to serologically detectable antibodies by reciprocal immunization (Nathenson et al., 1976). The H-2K glycoproteins from these mutants, B6.C-H-2ba(Hzl) and B6.H-2bd(M505), have been isolated arid compared to each other and to the parent B6.H-2b glycoprotein by tryptic peptide map analysis (Brown and Nathenson, 1977). The peptide profiles revealed small but significant differences among H-2Kba, H-2Kbd, and H-2Kb glycoproteins (Brown and Nathenson, 1977). These findings suggest that alterations in the H-2Kb products in the mutants are sufficient to give positive MLR and CML without affecting serologically detected antigenic sites on the molecule, and favor separation of the sites responsible for these different determinants on H-2 molecules (see Nathenson et al., 1976, for more detail). Present evidence favors the view that the polymorphism of H-2 is found within the polypeptide part of H-2 glycoproteins, although no specificity has been assigned to particular amino acid residue(s) as yet. However, one cannot rule out the possible involvement of the carbohydrate portion in determining part of the polymorphism (some functions of H-2 could be mediated by the carbohydrate portion and could express a given polymorphism). The elucidation of the carbohydrate structure (including linkage determination) of H-2 glycopeptides from different haplotypes should shed some light on this matter.

IV.

ISOLATION AND CHARACTERIZATION OF la ANTIGENS

A. Tissue Distribution and Polymorphism of la Antigens

The history of Ia antigens is only 5 years long, but what an eventful existence they have had so far! They have been implicated in so many immunological phenomena that a vast literature has already accumulated. A recent issue of Transplantation Reviews (Volume 30, 1976), to which we will often refer for brevity, is dedicated to the Ia antigens and should be consulted for a more elaborate discussion. As mentioned in Section I, Ia antigens are coded by genes mapping in the Z region of the H-2 complex. Reciprocal immunization with lymphocytes of two intra-H-2 recombinant lines of mice, namely,

488

MICHELLE LETARTE

A.TH and A.TL, which are identical at the K and D ends but different in the central regions of the H - 2 complex, has led to identification of the Ia antigens (David et al., 1973). Anti-Ia antibodies were also shown to be present in many anti-H-2 sera (Sachs and Cone, 1973; Hauptfeld et aZ., 1973; Gotze et d., 1973). Several anti-Ia sera have now been produced; initial studies suggested the presence of multiple antibodies in the sera, and their analysis revealed the existence of several Ia specificities and thus of another polymorphism within the H-2 complex. Twenty-one specificities have so far been defined, certain of which are restricted to a single haplotype and behave as private specificities, whereas others are shared by several haplotypes and behave as public specificities (David, 1976). Although five Z subregions and l a loci have tentatively been defined (see Table I), strong evidence exists only for 2 Z region-associated molecules, namely, those associated with the 1-A and Z-C subregions (Klein and Hauptfeld, 1976). Most of the Ia specificities defined so far can be mapped in either the Z-A or Z-Csubregion; some specificities still have not yet been clearly assigned to a given subregion (David, 1976). For simplicity of discussion, we thus talk about two Ia molecules which bear the Ia antigenic specificities and are the gene products of the Z-A and Z-C subregions, respectively. Ia antigens are predominantly expressed on B lymphocytes; T lymphocytes do not have or have much smaller amounts of Ia antigens than B lymphocytes. Different functional subpopulations of T cells express different amounts of la antigens; e.g., thymocytes contain very small amounts of Ia, whereas concanavalin A (Con A)-induced T-cell blasts contain larger amounts. It has been suggested that the activation of some T-lymphocyte subpopulations may lead to an increase in the amount of Ia antigen on their surface; the Ia marker could become useful in differentiating certain functionally defined subsets of T cells. Ia antigens can also be detected on macrophages and epidermal cells, and possibly on spermatozoa and certain teratoma cells, but are absent from erythrocytes, brain, kidney, and liver (Hammerling, 1976; Niederhuber and Frelinger, 1976). B. Isolation, MW, and Subunit Structure of la Antigens

Techniques described for the isolation of H-2 antigens (see Section II1,B) have also been applied to the study of Ia antigens. Fragments of Ia antigens have been released from spleen cells by papain hydrolysis, followed by purification with ion-exchange and gel chromatog-

GLYCOPROTEIN ANTIGENS

489

raphy (Hess, 1976a).A very low yield of Ia antigens was obtained (less than 0.1% relative to the activity of the cells before papain hydrolysis); the purified material migrated as a homogeneous fraction on SDS acrylamide gel electrophoresis. It could be stained with coomassie blue or Schiff’s reagent, suggesting its glycoprotein nature; the MW was estimated at 28,000 on 10% acrylamide gels under nonreducing conditions and at 26,000 on 14% acrylamide gels after reduction (Hess, 1976b). Although the purified Ia papain fragments appeared homogeneous on SDS gels, they were not chemically pure, as the fraction eluted from the gels could specifically inhibit the cytotoxic activity of anti-Ia sera directed at specificities expressed either on Z-A or Z-Csubregion gene products. The products from at least two different Z subregions were thus purified concominantly, stressing the need for more effective separation techniques. Papain treatment of spleen cells thus releases, in very low yield, soluble fragments bearing Ia antigenic specificities; the la molecules are not released intact from the membranes by papain and probably have lost their biological activity, rendering this method of isolation of limited value. supernatant from radioIa antigens can be released in a 6 x 106 gmin labeled lymphocyte membranes using the nonionic detergent NP-40; the labeled material, -which retains antigenic activity, can be isolated by immunoprecipitation with anti-Ia sera and characterized by polyacrylamide gel electrophoresis (Cullen et d.,1974). It was observed that partially purifying the NP-40 extract by affinity chromatography to lentil lectin greatly improved the efficiency of the immunoprecipitation reaction (Cullen and Schwartz, 1976).Upon gel electrophoresis (10% acrylamide, after boiling the samples in 2% SDS and 2% 2mercaptoethanol), Ia antigens appear as two peaks of approximately 35,000 and 25,000 MW; in the absence of a reducing agent, an additional peak of about 58,000 MW is seen. It can in turn be reduced by treatment with 2-mercaptoethanol into two peaks of 35,000 and 25,000 MW, respectively. However, the 58,000-MW peak is not always observed under nonreducing conditions and, when observed, is always accompanied by the 35,000- and 25,000-MW peaks (Cullen et al., 1976b). The Ia molecules isolated may thus represent a mixture of complex Ia molecules, some having two disulfide-linked subunits and others having noncovalently linked subunits. However, the 58,000MW peak apparently not dissociated by SDS could represent an artifact arising during the isolation procedure. The two polypeptide chains of Ia molecules are thus associated at least through hydrophobic bonds, if not through disulfide bonds, and Ia antigenic determinants could be present on only one of the chains. The antigenic poly-

490

MICHELLE LETARTE

peptide chain would be determined by an I-region gene, and the nonantigenic chain could be determined by a gene outside the H - 2 complex (Cullen et al., 1976b). The MW pattern just described is observed for Ia molecules isolated by immunoprecipitation from different haplotypes and bearing antigenic specificities associated with the gene products of either the I-A or I-C subregion. Specificities mapping in different I subregions were shown to be separable by sequential immunoprecipitation (Cullen et al., 1974). Ia specificities mapping in the same subregion in a given haplotype are coprecipitated by monospecific anti-Ia sera (Cullen et al., 1976a),suggesting that the several specificities mapping in a given subregion may be present on a single Ia molecule in a situation analogous to that in H-2 antigens. Ia-like molecules associated with the major histocompatibility complex have also been isolated from humans (Humphreys et al., 1976; Barnstable et al., 1976) and guinea pigs (Geczy et aZ., 1975). C. Partial Amino Acid Sequence and Preliminary Carbohydrate Structure of la Glycoproteins

The two polypeptide chains of Ia molecules have been called a (MW 35,000) and /3 (MW 25,000) and are designated Aab, Apb,Cab,and Cobfor the a and p chains of the I-A and I-C subregion gene products, respectively (here of haplotype b). Microsequence analysis has been applied to Ia molecules isolated by immunoprecipitation of radiolabeled cells (see Section II,C),and preliminary data on the a-polypeptide chain are shown in Table 111. Although the data are few, it can be seen that so far no residues are shared between the A, and the C, chains. The A, chains from b and k haplotypes show five out of six positions identical where residues have been assigned to both haplotypes. Peptide map comparisons of a chains from r and k haplotypes suggested that these a polypeptides were similar but that several peptides could distinguish them (Cullen et al., 197613).Mouse A, chains do not show similarity with the 34,000-MW (p34) chain of human Ia-like molecules (Springer et aZ., 1978; Barnstable et al., 1976), whereas mouse C, chain shows five out of six residues identical to those of human p34 chains. It is thus possible that certain Ia polypeptides of mice and humans are homologous and could have retained, throughout evolution, a relationship with the major histocompatibility complex. The data available on the mouse &polypeptide chain is too limited (four residues) to afford any meaningful comparison (Cecka et al., 1978). As more data become available, it should be interesting to compare the

492

MICHELLE LETARTE

various Ia molecules and to assign variable and iqvariable positions in corresponding molecules of different haplotypes. Ia antigens have been shown to be glycoproteins, as they incorporate radioactive mannose, galactose, glucosamine, and fucose; the pattern of incorporation is similar to that observed for H-2 glycoproteins and suggests a homology between the carbohydrate portions of Ia and H-2 antigens. [3H]fucose-labeled Ia molecules isolated by immunoprecipitation with anti-Ia.7 sera revealed two different glycoprotein bands corresponding to C, and Cp chains. However, glycopeptides obtained b y pronase digestion of [3H]fucose-labeled Ia.7 molecules showed only a single sharp peak of MW 3100 by gel chromatography. Thus both the C, and Cp chains of Ia.7 molecules are labeled with [3H]fucose,and the glycopeptides derived by digestion of both chains with pronase appear to be homogeneous in size, suggesting that both polypeptide chains carry a carbohydrate moiety of similar MW (Cullen et al., 1976b). In fact, glycopeptides derived from either H-2K or H-2D antigens or from the &, AD, C,, or Cp chains of Ia antigens all have a similar MW (3100 to 3300). This suggests that Ia antigens have a carbohydrate structure similar to that described for H-2 antigens (Section 111,C).Cleavage of Ia glycopeptides with D. pneumoniae endoglycosidase released a fucose-labeled glycopeptide of MW 800, and it has been postulated that the protein-carbohydrate linkage was of the asparagine -glucosamine type (Freed and Nathenson, 1978). D. Antigenic Determinants of la Molecules

Very little direct evidence is available on the localization of Ia antigenic specificities to either the carbohydrate or the polypeptide portion of the molecule. Ia.7 isolated from H-2d strains was found to be sensitive to brief extremes of pH, and pronase digestion removed the ability of Ia.7 to bind anti-Ia.7 sera. Digestion with a mixture of endoglycosidases from D. pneumoniae did not result in a loss of antigenic reactivity relative to that in control digests (Freed and Nathenson, 1978). Preliminary amino acid sequence as well as limited peptide map comparisons available so far between Ia molecules of different haplotypes favor localization of antigenicity in the polypeptide portion of the molecules. The presence in normal mouse sera of low-MW Ia antigens bearing carbohydrate structures able to block anti-Ia sera has been reported (Parish et al., 1976a,b). The Ia antigenicity of the serum was found to be in a glycolipid fraction and was susceptible to periodate oxidation and neuraminidase treatment. Individual sugars have been claimed to block anti-Ia sera specifically. The results have

493

GLYCOPROTEIN ANTIGENS

not been reproduced in other laboratories, and it is felt that the antigens detected b y Parish and co-workers are not related to the alloantigens carried b y the Ia glycoproteins but may be part of an independent antigenic system of the H - 2 complex (Freed and Nathenson, 1978). V.

ISOLATION AND CHARACTERIZATION OF Thy-1 ANTIGEN

A. Tissue Distribution and Number of Thy-1 Antigenic Sites per Cell

Thy-1 antigen (formerly called theta antigen) is expressed in large amounts on brain and thymus of mouse and rat, and in smaller amounts on other lymphoid tissues and on epidermal cells and fibroblasts (Reif and Allen, 1964, 1966; Douglas, 1972; Scheid et al., 1972; Stern, 1973; Acton et al., 1974). In mouse studies, the Thy-1 antigen is used as the T-cell marker par excellence; it is present on 95% of thymocytes, 50-60% of lymph node cells, 30% of spleen cells, and only 1% of bone marrow cells (Williams et al., 1976). However, in the rat, Thy-1 cannot be considered a T-cell marker, although it may turn out to be marker of a subpopulation of T-cell precursors; it is present on over 90% of thymocytes, 5% of lymph node cells, 15% of spleen cells, and 40% of bone marrow cells (Williams, 1976). The number of anti-Thy-1 molecules bound per rat thymocyte was estimated at 600,000 under saturating conditions, as described by Acton and co-workers (1974), and b y Morris and Williams (1975) (see Section II7B,3). Hammerling and Eggers (1970) had estimated previously that more than 400,000 molecules of anti-Thy-1.2 serum were bound per mouse thymocyte. It was also possible, using purified Thy1 antigen, to estimate the number of molecules per thymocyte under nonsaturating conditions of anti-Thy-1 sera. It was calculated from the amount of Thy-1 and the corresponding number of thymocytes giving the same percentage inhibition of the binding to thymocytes of the anti-Thy-1 sera that 600,000 molecules of Thy-1 were present per thymocyte. An average rat thymus thus contains on the order of 40 p g of Thy-1, whereas an average rat brain contains approximately 270 p g (Williams et al., 1976).

B. Isolation and MW Determination

of Thy-1 Antigen

Thy-1 antigen has been purified from rat thymus (Letarte-Muirhead et al., 1975), rat brain (Barclay et al., 1975), and mouse Thy-1.2 brain

494

MICHELLE LETARTE

Letarte and Meghji, 1978).The solubilization of Thy-1 antigenic activity from rat whole thymocytes or membranes by several nonionic detergents and by deoxycholate was investigated first. It was established that, with the exception of NP-40, detergents did not affect the antigenicity of Thy-1, but only Lubrol-PX and deoxycholate gave effective solubilization as measured by the antigenic activity present in the 6 x 106 gminsupernatant (Letarte-Muirhead et aZ., 1974). However, when the soluble extracts were examined by gel filtration and sucrose gradient centrifugation, the Thy-1 activity of the Lubrol-PX extract was contained in a high-MW, low-density complex, presumably because of large amounts of detergent bound to the antigen. The Thy-1 activity in the deoxycholate extract behaved as a single component with the following properties: s ~ , , , = ~ 2.4, V = 0.70, Strokes radius = 3.1 nm, from which a MW of 28,000 was calculated for the Thyl-deoxycholate complex (Letarte-Muirhead et al., 1975). Similar values have also been obtained for the Thy-l-deoxycholate complex solubilized from rat brain (Barclay et al., 1975). Purification of the antigen from either brain or thymus involved the preparation of crude membranes and solubilization in deoxycholate, followed by gel filtration and affinity chromatography on antibody or lectin columns (Letarte-Muirhead et al., 1975; Barclay et d.,1976). With thymocytes, it is possible to prepare membrane fragments quickly using the nonionic detergent Tween 40; these fragments are very similar in composition, after sucrose gradient fractionation, to those obtained by disrupting cells by shearing forces (Standring and Williams, 1978).Tween 40 released Thy-1 antigen from thymocytes in a 3000 g supernatant, but subsequent centrifugation at 100,000 g pelleted the Thy-1 activity of the supernatant in a crude membrane fraction; 50-60% of the initial Thy-1 activity of the cells was recovered in the latter fraction and could be solubilized in deoxycholate (Morris et al., 1975; Letarte-Muirhead et al., 1975). The soluble Thy-1-deoxycholate complex, when chromatographed on Sephadex G-200 in 0.5%deoxycholate, eluted in the same position as ovalbumin, i.e., giving a Stokes radius of 3.1 nm as mentioned above. By chromatography on lentil lectin columns, the Thy-1 activity from rat thymus was always recovered in two forms (50%of each), one binding to lentil lectin (Thy-1L+)and the other one not (Thy-lL-). Both forms had the same detectable antigens and were of similar size, although Thy-1L-, which was further purified by affinity to anti-Thy-1 sera, appeared somewhat larger and more heterogeneous upon electrophoresis (Letarte-Muirhead et al., 1975). Thy-1 activity from rat or mouse brain did not show this behavior on lentil lectin columns; more

GLYCOPROTEIN ANTIGENS

495

than 80% of the activity was specifically bound and eluted (Barclay et al., 1975; Letarte and Maghji, 1978). Thy-lL+ activity from rat thymus was purified 260-fold (relative to whole cells) with a yield of 10% and appeared homogeneous on polyacrylamide gel electrophoresis in SDS (Fig. 3 ) .Thy-1 activity from rat brain was purified approximately 600fold with a yield of 15% and also appeared homogeneous on electrophoresis (Barclay et al., 1975). It was established that reduction had no effect on the mobility of Thy-1 antigens, suggesting that the molecule contained a single polypeptide chain. The apparent MW of the purified antigens was determined, under reducing and alkylating conditions, using different concentrations of acrylamide, since it is known that many glycoproteins behave anomalously and appear larger than predicted when run on gels of low-percentage acrylamide (Glossmann and Neville, 1971; Segrest et al., 1971). All fractions showed a decrease in apparent MW with increasing acrylamide concentration (Table IV). The values obtained in 12.5% gels represent a more accurate determination of the apparent MW of glycoproteins. Here Thy-lL+ has an apparent MW of 27,200 and is more heterogeneous; brain Thy-1 ran as a symmetrical peak of about 24,100 MW (see Barclay et al., 1976, for comparative gels). These values are in close agreement with those obtained by gel filtration and sucrose density gradient analysis of Thy-l-deoxycholate extracts, where a MW of 28,000 was calculated for thymus Thy-1 and a MW of 27,000 for brain Thy-1. In more recent experiments, the MW of brain and thymus Thy-1 glycoproteins has been estimated by sedimentation equilibrium to be 17,500 and 18,700 respectively (Kuchel et al., 1978).

C. Amino Acid and Carbohydrate Composition of Thy-1 Glycoprotein

Purified preparations of Thy-1 have been analyzed for their amino acid composition. Table V shows that the analyses were very similar for rat brain Thy-1 and rat thymus Thy-lL+ and Thy-1L-. The only differences which appear significant after statistical analysis are those between the leucine and glycine content of brain Thy-1 versus those of thymus Thy-1. Any difference in the protein structure will have to be revealed by sequence analysis. The amino acids are grouped on the basis of their hydrophobicity (in Table V), and it can be seen that Thy-1, which appeared fairly hydrophobic from its behavior in detergent (Letarte-Muirhead et al., 1974), does not show a high content of hydrophobic residues but is rather notable for its large proportion of

496

MICHELLE LETARTE

FIG.3. Assay of Thy-1.1 and Thy-I xenoantigenic activities after SDS polyacrylamide gel electrophoresis. Thy-IL+ (2.5 pg) purified from rat thymus by lentil lectin affinity was electrophoresed on a 5.6% acrylarnide gel containing 0.1% SDS with lZ5I-labeled whale myoglobin (Myo) and '25-labeled hen ovalbumin (Ov) as markers. After electrophoresis, the gel was sliced and assayed for Thy-1.1 activity (squares) and Thy-I xenoantigen activity (circles) by inhibition of indirect binding assays. Also shown is a photograph of a gel on which 7pg of Thy-I was electrophoresed under the same conditions and stained with coomassie blue. Reproduced from Letarte-Muirhead et uZ. (1975), with permission from Blochem. J .

TABLE IV APPARENT MW OF PURIFIEDTHY-1ANTIGEN ON SDS POLYACRYLAMIDE GEL ELECTROPHORESIS"

Acrylamide concentration Antigen

5.6%

12.5%

Brain Thy-1 Thymus Thy-lL+ Thymus Thy-1L-

31,300 f 1,000 (7) 31,600 k 400 (5) 34,800 f 380 (7)

* 150 (4)

24,100 25,300 27,200

2 2

240 (4) 390 (5)

Thy-IL+ represents the fraction of Thy-1 from thymus which binds to a lentil lectin column, whereas Thy-1L- represents the fraction which does not bind. The number of experiments done is shown in parenthesis. Data from Letarte-Muirhead et al. (1975) and Barclay et a l . (1975).

TABLE V

AMINO ACID COMPOSITION OF THY-1 GLYCOPROTEINS" Number of amino acids per 100 residues Brain Thy-1 As x Glx His Lys A% Thr Ser Pro Ala cys GlY TY r Val Ile Leu Phe Met

12.7 9.1 4.1 6.9 7.5 7.6 7.4 3.6 3.3 3.1 6.0 2.0 7.3 3.9 10.4 4.0 1.1

Thymocyte Thy-1L' 12.6 9.1 3.8 7.2 7.5 8.6 7.1 3.8 2.9 3.2 4.9 2.0 7.4 4.1 11.2 3.7 0.9

Thy-1L13.0 9.1 4.0 7.0 7.2 8.3 7.9 3.O 3.2 3.2 5.0 2.0 7.4 4.0 11.1 3.8 0.8

" Amino acid analyses are expressed as the mean number of each amino acid residue per 100 residues and are the result of at least four analyses in each case. Data reproduced from Barclay et al. (1976), with permission from Nature.

498

MICHELLE LETARTE

hydrophilic residues. It is likely that, as in the case of glycophorin (Segrest et al., 1972), a hydrophobic sequence is involved in holding the protein in the membrane and that the hydrophilic sequences are present on the outside of the membrane. Purified Thy-1 fractions from brain or thymus gave positive staining with the periodic acid-Schiff reagent. Thy-1L-, although it did not bind to lentil lectin, is also a glycoprotein, suggesting heterogeneity in the carbohydrate portion of thymus Thy-1. The carbohydrate composition of Thy-1 glycoproteins is illustrated in Table VI. The differences between Thy-1L+ and Thy-1L- are as follows. Thy-1L- seems to have extra mannose and glucosamine residues and possibly one extra sialic residue, while lacking galactose residues. These differences may or may not be sufficient to account for the different behavior of these two glycoproteins on lentil lectin; it is also possible that some variations exist in the linkages joining the core sugars, which could give rise to steric hindrance at certain lentil lectin-binding sites. The differences between brain Thy-1 and thymus Thy-1 are more numerous (see Table VI) and may reflect substantial differences in the TABLE VI CARBOHYDRATE COMPOSITlON OF THY-1 GLYCOPROTEINS" Carbohydrate residues per 100 amino acid residues ~~

~~

Brain

Fucose Mannose Galactose Glucose Glucosamine Galactosamine Sialic acid Amino sugars by ion-exchange chromatography Glucosamine Galactosamine Sialic acid by fluorescent method Sialic acid Percentage by weight of carbohydrate

~~

~~

~

Thymocyte

Thy-1

Thy-lL+

Thy-1L-

1.8 11.9 1.8 0.6 6.3 1.o 0.2

1.0 10.6 5.5 1.3 7.0 0.0 1.8

0.9 9.4 6.9 1.1 8.3 0.0 2.2

8.3 1.0

9.4 0.0

11.7 0.0

0.3 2.9%

2.1 32%

2.9 35%

Results shown are the means of four analyses for brain Thy-1 and thymocyte Thy1L+ and two analyses for Thy-1L- by gas-liquid chromatography. Data reproduced from Barclay et al. (1976), with permission.

GLYCOPROTEIN ANTIGENS

499

carbohydrate structure of these two glycoproteins which have homologous amino acid compositions and antigenicity. The percentage by weight of carbohydrate in Thy-1 glycoproteins is on the order of 30% and at least two carbohydrate side chains are present (Kuchel et al., 1978). D. Antigenic Determinants of the Thy-1 Molecule

Reif and Allen (1964)demonstrated that Thy-1 antigen (then called 8) existed in the mouse in two allelic forms: Thy-1.1 (formerly 8-AKR) alloantigen and Thy-1.2 (formerly 8-C3H) alloantigen. Rat tissues carry only Thy-1.1 alloantigen (Douglas, 1972). It was shown by Golub (1971) that rabbit anti-mouse brain serum reacted with mouse lymphocytes as if it were a potent anti-Thy-1 serum. Analysis of rabbit anti-rat brain serum and of its reactivity toward rat and mouse lymphocytes showed that, after complete liver absorption, the antiserum behaved as an anti-Thy-1 serum (Morris and Williams, 1975).It was then demonstrated that Thy-1.1 alloantigen and the xenoantigenic determinants recognized by rabbit anti-rat brain serum were present on the same molecule (Morris et al., 1975). Figure 3 demonstrates that purified Thy-lL+ antigen from rat thymus carries both the xenoantigenic determinants and Thy-1.1 alloantigen (Letarte-Muirhead et al., 1975). All purified fractions of brain and thymus Thy-1 carry xenoantigenic as well as alloantigenic determinants. A rabbit antiserum raised against purified rat brain Thy-1 contained 70% antibody specific for rat Thy-1 xenoantigen, 20% specific for rat-mouse cross-reacting xenoantigen, and only 4% possibly recognizing Thy-1.1 alloantigen (Barclay et d . ,1975). In Fig. 4,the cross-reaction between rat and mouse Thy-1 antigens is demonstrated using a rabbit antiserum raised against Thy-1 purified from mouse (Thy-1.2)brain. There is complete cross-reactivity between mouse brain Thy-1 and mouse thymocyte Thy-1 of either Thy-1.1 or Thy-1.2 [the 5% difference seen between Thy-1.1 and Thy-1.2 thymus is not significant (Fig. 4)].There is 60% cross-reaction between rat thymus and mouse brain Thy-1. These findings confirm the fact that the proteins bearing Thy-1.1 or Thy-1.2 alloantigen from either rat or mouse brain or thymus are closely related (Letarte and Meghji, 1978). Localization of the antigenic determinants on Thy-1 glycoproteins has not yet been clearly elucidated. Preliminary experiments showed that the antigenic activity of purified Thy-1 was destroyed after a 10minute exposure to temperatures of 70.40"C. The effects of proteolytic digestion were evaluated by measuring the loss of antigenic activity and the disappearance of the Thy-1 band on SDS polyacryla-

500

MICHELLE LETARTE

Log of cell concentration during absorption

an,

FIG.4. Absorption of the binding of rabbit anti-Thy-1 [purified from mouse (Thy-1.2) brain] to Thy-1.2 thymocytes by AKR thymocytes (triangles)bearing the Thy-1.1 alloantigen, by Bl0.BR thymocytes (circles) bearing the Thy-1.2 alloantigen, and by rat thymocytes (squares). The antiserum concentration during absorption was 1 :500, and the log of cell concentration during absorption is shown on the abscissa. The target cells were lo8 glutaraldehyde-fixed thymocytes and the concentration of 1251-labeledF(ab'), horse anti-rabbit IgC in the second incubation was 0.03 p g per assay.

mide gel electrophoresis. After pronase digestion for 24 hours at 37"C, 80-95% of the antigenic activity of all Thy-1 determinants was destroyed. Only partial loss of activity was obtained with papain, trypsin, or chymotrypsin, and in all cases it was correlated with the amount of material left on the gels (quantitated by densitometer tracing). Following digestion, no proteolytic fragments could be identified on the gels, except for traces in the case of papain treatment (Barclay et al., 1976; Williams et al., 1976). Although these results and particularly those of the pronase digestion suggest that the antigenic determinants may be present in the polypeptide portion, no definitive statement should be made at this point and one should await further structural analysis.

VI.

PRELIMINARY CHARACTERIZATION OF Tla ANTIGENS

Thymus leukemia alloantigens (Tla) are expressed in high concentration on the surface of 80-90% of mouse thymocytes. Neither B cells nor mature T cells react with anti-Tla antibodies; their distribution is restricted to thymocytes and leukemia cells (Boyse and Old, 1969). The testing of anti-Tla sera with normal thymocytes divides inbred strains into three groups, as shown in Table VII. The determinants

50 1

GLYCOPROTEIN ANTIGENS

coding for Tlu phenotypes and genotypes behave as alleles at the same locus, the Tla locus, which is closely linked to the H-2 complex. All strains of mice, irrespective of whether they express Tla antigens in their thymuses, can give rise to Tla-positive leukemias which can be classified in four groups (Table VII). The Tla phenotypes of thymus and leukemias are the same in only one group; in other groups leukemia cells express Tla antigens not found on thymocytes of the same strain (Boyse and Old, 1969). One possible explanation for the differential expression of Tla antigens on thymocytes and leukemia cells could be the existence of structural and regulator genes for Tla antigens. All strains carry structural genes for Tla-1 and Tla-2, since they are present in all leukemias, whereas only certain strains carry structural genes for Tla-3 or Tla-4; a regulator gene could act by repressing the structural genes in normal thymocytes and derepressing them in leukemia cells. Since Tla-1 and Tla-2 structural genes are identical in all strains, it must be the regulator rather than the structural gene which is linked to the H-2 complex (Boyse and Old, 1969; J. Klein, 1975). Tla antigens have been isolated by immunoprecipitation from radiolabeled thymocytes or leukemic cells solubilized in NP-40 (Vitetta et al., 1972; Ostberg et al., 1975), or by partial purification and/or immunoprecipitation following the treatment of cells with papain (Muramatsu et aZ., 1973; Anundi et al., 1975; Stanton et al., 1976).Immune complexes formed with anti-Tla 1,3 sera and thymocytes from strain A solubilized in NP-40 contained two polypeptide chains of MW 50,000 and 12,000 (by SDS acrylamide gel electrophoresis under reducing conditions); the large subunit carried the antigenic activity, and the small subunit appeared to be &-microglobulin, as in the case of H-2 antigens (Ostberg et al., 1975;Vitetta et al., 1976b).Tla antigens soluTABLE VII

Tla

Prototype strain A C57BL/6 BALB/c DBA/2

PHENOTYPES AND GENOTYPES"

Thymocyte Tla phenotype 1,2,3, -

----, 2, -, 2, -, -I

Leukemia cell phenotype and presumed Tla genotype of mouse 1 , 2 , 3, 1 , 2 , -, 4 1,2, 1, 2, -, 4 -1

Data reproduced from Boyse and Old (1969),with permission from Annual Review

of Genetics.

502

MICHELLE LETARTE

bilized by papain treatment of leukemic cells radiolabeled with amino acids and fucose were isolated by immunoprecipitation as a doubly labeled glycoprotein fragment of MW 37,000; exhaustive pronase digestion gave a glycopeptide of approximately 4500 MW (Muramatsu et al., 1973). Tla antigens were also isolated from crude membrane fractions prepared by freezing and thawing thymocytes; antigenic activities were then released in a so-called intact form, with a very low yield, by incubating the cells for 12 hours in EDTA-containing buffer (Anundi et al., 1975). The Tla antigens were then purified by gel chromatography and affinity to an anti-µglobulin column. Although a poor recovery of Tla antigens was obtained from the affinity column (15%), the antigens appeared relatively pure by SDS gel electrophoresis. The MW of the antigens solubilized by EDTA-containing buffer was calculated as 120,000 from gel filtration and sucrose gradient centrifugation analyses. When immunoprecipitates prepared from Tla antigens solubilized under these conditions were run on SDS gels in the absence of reducing agents, three peaks were seen: 90,000-100,000, 50,000, and 12,000 MW; when the immunoprecipitates were analyzed under reducing conditions, only the 50,000- and 12,000-MW peaks were visible (Anundi et al., 1975). Although it was suggested from these data that intact Tla antigens were composed of two disulfidelinked 50,000-MW polypeptide chains and two p,-microglobulin molecules, the evidence was not conclusive and many artifacts could have arisen during the isolation procedure (i.e., a 12-hour incubation of crude membranes prepared by freezing and thawing). Individual Tla antigens have not been isolated, and the biochemical characterization of Tla antigens is at the moment rudimentary.

VII.

PRELIMINARY CHARACTERIZATION OF Ly-2,3 ANTIGENS

Several systems of murine alloantigens called Ly antigens have now been defined. We limit this brief discussion to the Ly-1 and Ly-2,3 systems whose expression is restricted to thymocytes and peripheral T cells (Boyse et al., 1968, 1971a). Each Ly system comprises a genetic locus; the Ly-l locus is on chromosome 19, and the Ly-2 and Ly3 loci are closely linked on chromosome 6. Each locus has two alternative alleles defining two alloantigens; all inbred mice express either alloantigen Ly-1.1 or Ly-1.2 and, similarly, Ly-2.1 or Ly-2.2 and Ly3.1 or Ly-3.2. Recent interest in these Ly antigens has been generated

503

GLYCOPROTEIN ANTIGENS

by the observation that they appear to be markers of functional subclasses of T cells..Peripheral T cells which exert a helper function bear Ly-1 antigen, whereas cells which exert killer or suppressor functions bear both Ly-2 and Ly-3 antigens (Cantor and Boyse, 1975a,b). However, the majority of peripheral T cells and thymocytes express Ly-1, Ly-2, and Ly-3 antigens and may be the precursors of functional subclasses bearing restricted expression of Ly antigens (Huber et al., 1976).The only biochemical information available so far on Ly antigens has been derived from immunoprecipitation data on anti-Ly3.1 sera of NP-40 extracts from radiolabeled thymocytes of the Ly-2.1,3.1 type. When the precipitates were solubilized in SDS, reduced, and analyzed by polyacrylamide gel electrophoresis (10% gels), a major band of MW estimated at 35,000 was detected (Durda and Gottlieb, 1976). The antigenic subunit appeared to be of glycoprotein nature, since it could be specifically precipitated whether the thymocytes were labeled with ['251]lactoper~xida~e or with sodium borohydride and galactose oxidase. Specific precipitates obtained using Ly-2.1 antisera gave a similar profile on SDS gels. Preliminary results showed that, following immunoprecipitation with anti-Ly-3.1 sera, Ly-2.1 antigenic activity could no longer be precipitated, suggesting that both antigenic activities are closely associated in the NP-40 extract (Durda and Gottlieb, 1976).It has not been possible so far to find cells expressing Ly-2 antigen independently of Ly-3 antigen, and vice versa, and it has been postulated that Ly-2 and Ly-3 may comprise a complex locus Ly-2,3 with respective antigenic determinants expressed on the same molecule (Itakura et al., 1972). Further biochemical characterization of Ly-2 and Ly-3 antigens should test the latter hypothesis.

VIII.

CONCLUSION

When membrane components can be defined b y antibodies, it becomes feasible to utilize immunochemical techniques for their study. Our intention, in this chapter, was to discuss the various methods available for the detection, quantitation, and isolation of cell surface antigens. We stress that immunochemical approaches have been applied to the study of several membrane glycoproteins, thus providing extra tools for their identification and isolation. Immunofluorescent antibodies, for example, can serve to visualize cell surface antigens on various tissues and/or to separate cells bearing different amounts of the antigens. Cellular radioimmunoassays permit one to dissect complex antisera with respect to their multiple antigenic specificities,

504

MICHELLE LETARTE

to measure the number of antigenic sites per cell and the concentration of antibodies in a given antiserum, and to follow antigenic activity during the purification of cellular antigens. Immunoprecipitation techniques are very useful in the identification and preliminary biochemical characterization of antigenic specificities recognized b y either complex or monospecific antisera. The isolation of murine alloantigens has been carried out mostly by partially purifying fragments produced by papain treatment of cells, or b y immunoprecipitation of these antigens from solubilized radiolabeled cells. Although both methods have yielded useful information concerning the biochemical nature of murine alloantigens, neither method allows the isolation of large amounts of material in a biologically active and biochemically intact form. In order to perform structural analyses of murine glycoprotein antigens and to carry out functional assays with purified antigens, methods which permit the isolation of these molecules in an intact form must be used. T h e amount of material available in murine studies is limited, and microtechniques must be established for structural as well as functional characterization of membrane glycoproteins. The biochemical characteristics of murine glycoprotein antigens discussed in the review are summarized in Table VIII. It can be seen that these molecules have relatively low MWs and contain probably TABLE VIII SUMMARY OF THE BIOCHEMICAL CHARACTERISTICS OF MURINEGLYCOPROTEIN ANTIGENS" Antigen

Subunit structure and MW

Carbohydrate structure

H-2

Heavy chain 45,000, light chain 12,000 (f?z-rnicroglohuIin)

One or two glycopeptides of MW 3300 per heavy chain

Ia

a Chain 35,000, p chain 25,000

Thy-1

One chain 24,000

Tla

Heavy chain 50,000, light chain 12,000 (&-microglohulin)

Both chains carry a glycopeptide of MW 3200 Carbohydrate content of 30%; two glycopeptides per molecule One glycopeptide of MW 4500 on the heavy chain

Ly-2,3

One chain 35,000

Carbohydrate present

See text for referenbes.

505

GLYCOPROTEIN ANTIGENS

either one or two glycopeptides per molecule. Both H-2 and Tla antigens are associated with a small subunit closely related to @,-microglobulin. Evidence accumulated so far for H-2, Thy-1, and Ia favors the localization of antigenic determinants in the polypeptide portion of these molecules. The evidence is, however, not conclusive and does not exclude the possibility that the carbohydrate portion of these molecules is important in certain immunological mechanisms in which the antigens participate. Further biochemical and functional characterization of murine glycoprotein antigens should clarify the relationship between the structure of these membrane proteins and their immunological significance. REFERENCES Aaronson, S. A., and Stephenson, J. R. (1974).Widespread natural occurrence of high titers of neutralizing antibodies to a specific class of endogenous mouse type-C virus. Proc. Natl. Acad. Sci. U.S.A.71, 1957-1961. Acton, R. T., Morns, R. J., and Williams, A. F. (1974). Estimation of the amount and tissue distribution of rat Thy-1.1 antigen. Eur. J . Zmmunol. 4, 598-602. Anundi, H., Rask, L., Ostherg, L., and Peterson, P. A. (1975). The subunit structure of thymus leukemia antigens. Biochemistry 14,5046-5054. Ballou, B., McKean, D. J., Freedlender, E. F., and Smithies, 0.(1976).HLA membrane antigens: Sequencing by intrinsic radioactivity. Proc. Natl. Acad. Sci. U . S . A . 73, 4487-4491. Barclay, A. N., Letarte-Muirhead, M., and Williams, A. F. (1975). Purification of the Thy-1 molecule from rat brain. Biochem. J . 151, 699-706. Barclay, A. N., Letarte-Muirhead, M., Williams, A. F., and Faulkes, R. A. (1976).Chemical characterization of the Thy-1 glycoproteins from the membranes of rat thymocytes and brain Nature (London)263,563-567. Barnstable, C. J., Jones, E . A,, Bodmer, W. F., Bodmer, J. C., Arce-Gomez, B., Snary, D., and Crumpton, M. J. (1976). Genetics and serology of HLA-linked human Ia antigens Cold Spring Harbor S!/mp.Quant. Biol. 41,443-456. Bonner, W. A., Hulett, H. R., Sweet, R. C., and Herzenberg, L. A. (1972). Fluorescence activated cell sorting. Rev. Sci. Znstrum. 43,404-409. Boyse, E. A., and Old, L. J. (1969). Some aspects of normal and abnormal cell surface genetics. Annu. Rev. Genet. 3, 269-290. Boyse, E. A., Miyazawa, M., Aoki, T., and Old, L. J. (1968).Ly-A and Ly-B: Two systems of lymphocyte isoantigens in the mouse. Proc. R. Soc., Ser. B 170,175-193. Boyse, E. A., Itnkura, K., Stockert, E., Iritani, C. A., and Miura, M. (1971a).Ly-C: A third locus specifying alloantigens expressed only on thymocytes and lymphocytes. Transplantation 11,351-353. Boyse, E. A., Old, L. J., and Scheid, M. (1971h). Selective gene action in the specification of cell surface structure. Am. J. Pathol. 65,439-450. Bridgen, J. (1976).High sensitivity amino acid sequence determination: Application to proteins eluted from polyacrylamide gels. Biochernistr!l 15, 3600-3604. Bridgen, J., Snary, D., Crumpton, M. J., Barnstable, C., Goodfellow, P., and Bodmer, W. F. (1976). Isolation and N-terminal amino acid sequence of membrane-bound human HLA-A and HLA-B antigens. Nature (London)261,200-205.

506

MICHELLE LETARTE

Brown, J. L., and Nathenson, S. G. (1977). Structural differences between parent and mutant H-2K glycoproteins from two H-2K gene mutants: B6.C-H-2ba( H z l )and B6H-2bd(M505).J . Zmmunol. 118,98-102. Brown, J . L., Kato, K., Silver, J., and Nathenson, S. G. (1974). Notable diversity in peptide composition of murine H-2K and H-2D alloantigens. Biochemistry 13, 31743178. Cantor, H., and Boyse, E. A. (1975a). Functional subclasses of T lymphocytes bearing different Ly antigens. I. The generation of functionally distinct T-cell subclasses is a differentiative process independent of antigen. ]. E x p . Med. 141, 1376-1389. Cantor, H., and Boyse, E. .4. (1975h). Functional subclasses of T lymphocytes hearing different Ly antigens. 11. Cooperation between subclasses of Ly+ cells in the generation of killer activity.]. E x p . Med. 141, 1390-1399. Cantor, H., and Weissman, I. (1976). Development and function of subpopulations of thymocytes and T lymphocytes. Prog. Allergy 20, 1-64. Capra, D., Vitetta, E. S., Klapper, D. G., Uhr, J. W., and Klein, J. (1976). Structural studies on protein products of murine chromosome 17: Partial amino acid sequence of an H-2Kb molecule. Proc. Natl. Acad. Sci. U.S.A. 73,3661-3665. Cebra, J. J., and Goldstein, G. (1965). Chromatographic purification of tetramethylrhodamine-immune globulin conjugates and their use in the cellular localization of rabbit y-globulin polypeptide chains. J. Zmmunol. 95, 230-245. Cecka, M., McMillan, M., Murphy, D., Silver, J., McDevitt, H., and Hood, L. (1978). Partial amino acid sequence analyses of Ia molecules. In “Ir Genes and Ia Antigens” (H. 0. McDevitt, ed.), pp. 275-286. Academic Press, New York. Coons, A. H., Leduc, E. H., and Connolly, J . M. (1955).Studies on antibody production. I. A method for the histochemical demonstration of specific antibody and its application to a study of the hyperimmune rabbit.J. E x p . Med. 102,49-72. Counce, S., Smith, P., Barth, R., and Snell, G. D. (1956). Strong and weak histocompatibility gene differences in mice and their role in the rejection of homografts of tumors and skin. Ann. Surg. 144, 198-204. Cullen, S. E., and Schwartz, B. D. (1976). An improved method for isolation of H-2 and Ia alloantigens with immunoprecipitation induced by protein A-hearing staphylococci.]. Zmmunol. 117, 136-142. Cullen, S. E., Schwartz, B. D., Nathenson, S. G., and Cherry, M. (1972). The molecular basis of codominant expression of the histocompatibility-2 genetic region. Proc. Natl. Acad. Sci. U.S.A. 69, 1394-1397. Cullen, S. E., David, C. S., Shreffler, D. C., and Nathenson, S. G. (1974). Membrane molecules determined by the H-2 associated immune response region: Isolation and some properties. Proc. Natl. Acad. Sci. U.S.A. 71,648-652. Cullen, S. E., David, C. S., Cone, J. L., and Sachs, D. H. (1976a). Evidence for mqre than one Ia antigenic specificity on molecules determined by the I-A subregion of the mouse major histocompatibility complex. J. Zmrnunol. 116, 549-553. Cullen, S. E., Freed, J . H., and Nathenson, S. G. (1976b). Structural and serological properties of murine Ia alloantigens. Transplant. Rev. 30,236-270. David, C . S. (1976). Serological and genetic aspects of murine Ia antigens. Transplant. Rev. 30,299-322. David, C. S., Shrefffer, D. C., and Frelinger, J. A. (1973). New lymphocyte antigen system (Lna) controlled by the Zr region of the mouse H-2 complex. Proc. Natl. Acad. Sci. U.S.A. 70,2509-2514. Dorval, G . ,Welsh, K. I., and Wigzell, H. (1975). A radioimmunoassay of cellular surface antigens on living cells using iodinated soluble protein A from Staphylococcus aureus. 1.Immunol. Methods. 7,237-250.

GLYCOPROTEIN ANTIGENS

507

Douglas, T. C. (1972).Occurrence of a theta-like antigen i n rats.J. E x p . Med. 136,1054-

1062. Durda, P. J., and Gottlieb, P. D. (1976).The Ly-3 antigens on mouse thymocytes: Immune precipitation and molecular weight characterization. J. Exp. Med. 144,476-

493. Esmon, N. L., and Little, J. R. (1976).An indirect radioimmunoassay for thymus leukemia (TL) antigens.J. Immunol. 117,911-918. Ewenstein, B. M., Freed, J. H., Mole, L. E., and Nathenson, S. G. (1976). Localization of the papain cleavage site of H-2 glycoproteins. Proc. Natl. Acad. Sci. U.S.A. 73,915-

918. Fabre, J. W., and Williams, A. F. (1977). Quantitative serological analysis of a rabbit anti-rat lymphocyte serum and preliminary biochemical characterization of the major antigen recognized. Transplantation 23,349-359. Forsgren, A., and Sjoquist, J. (1966).“Protein A” from S. aureus. I. Pseudoimmune reaction with human y-globulin. J. Immunol. 97, 822-827. Freed, J. F., and Nathenson, S. G. (1978).Antigenic structure ofthe Ia glycoprotein molecules. In “Ir Genes and Ia Antigens” (H. D. McDevitt, ed.),pp. 263-273. Academic Press, New York. Geczy, A. F., D e Weck, A. L., Schwartz, B. D., and Shevach, E. M. (1975).The major histocompatibility complex of the guinea pig. I. Serologic and genetic studies. J.

Zmmunol. 115, 1704-1710. Ghetie, V., Nilsson, K., and Sjoquist, J. (1974). Identification of cell surface immunoglobulin markers by protein A-containing fluorescent staphylococci. Scand. J. Immunol. 3,397-403. Glossmann, H., and Neville, D. M. (1971).Glycoproteins ofcell surfaces: A comparative study of three different cell surfaces of the rat. J . Biol. Chem. 246,6339-6346. Gotze, D., Reisfeld, R. A., and Klein, J. (1973). Serological evidence for antigens controlled by the Ir region in mice.j. Exp. Med. 138, 1003-1008. Golub, E. S . (1971). Brain-associated 0 antigen: Reactivity of rabbit anti-mouse brain with mouse lymphoid cells. Cell. Zmmunol. 2,353-361. Goodfellow, P. N., Jones, E. A., van Heyningen, V., Solomon, E., and Bobrow, M. (1975). The µglobulin gene is on chromosome 15 and not in the HL-A region. Nature (London) 254, 267-269. Corer, P. A,, and O’Gorman, P. (1956).The cytotoxic activity of isoantibodies in mice. Transpkant. Bull. 3, 142-143. Hammerling, G. J. (1976).Tissue distribution of Ia antigens and their expression on lymphocyte subpopulations. Transplant. Reu. 30,64-82. Hammerling, U . , and Eggers, H. J. (1970). Quantitative measurement of uptake of alloantibody on mouse lymphocytes. Eur. J. Biochem. 17,95-99. Hauptfeld, V., Klein, D., and Klein, J. (1973). Serological identification of an Ir-region product. Science 181, 167-169. Henning, R., Milner, R. J., Reske, K., Cunningham, B. A,, and Edelman, G. M. (1976). Subunit structure, cell surface orientation and partial amino acid sequences of murine histocompatibility antigens. Proc. Natl. Acud. Sci. U.S.A. 73, 118-122. Hess, M . (19764. Immune response region-associated antigens of the mouse: Biochemical properties of detergent and papain solubilized molecules. Eur. J. Immunol. 6,

188- 193. Hess, M. (1976b).Ia antigens: Isolation, chemical modification and structural characterization. Trurisplant. Reu. 30,40-63. Hjelm, H., Hjelm, K., and Sjoquist, J. (1972).Protein A from Staphylococcus aureus: Its isolation by affinity chromatography and its use as an immunosorbent for isolation of immunoglobulins. FEBS Lett. 28, 73-76.

508

MICHELLE LETARTE

Huber, B., Cantor, H., Shen, F. W., and Boyse, E. A. (1976). Independent differentiative pathways of Lyl and Ly23 subclasses of T cells.]. E x p . Med. 144, 1128-1133. Hulett, H. R., Bonner, W. A., Sweet, R. G., and Herzenberg, L. A. (1973). Development and application of a rapid cell sorter. Clin. Chem. 19,813-816. Humphreys, R. E., McCune, J. M., Chess, L., Herrman, H. C., Malenka, D. J., Mann, D. L., Parham, P., Schlossman, S. F., and Strominger, J. L. (1976). Isolation and immunologic characterization of a human, B-lymphocyte-specific, cell surface antigen. J . E x p . Med. 144,98-112. Hunter, W. M. (1973). Radioimmunoassay. In “Handbook of Experimental Immunology” (D. M. Weir, ed.), pp. 17.1-17.36. Blackwell, Oxford. Itakura, K., Hutton, J. J., Boyse, E. A., and Old, L. J. (1972). Genetic linkage relationships of loci specifying differentiation alloantigens in the mouse. Transplantation 13,239-243. Jensenius, J. C., and Williams, A. F. (1974). The binding of anti-immunoglobulin antibodies to rat thymocytes and thoracic dust lymphocytes. Eur. J. Immunol. 4, 91-97. Johnson, G. D., and Horborow, E. J. (1973). Immunofluorescence. In “Handbook of Experimental Immunology” (D. M. Weir, ed.), pp. 18.1-18.20. Blackwell, Oxford. Kandutsch, A. A., and Reinert-Wenck, U. (1957). Studies on a substance that promotes tumor homograft survival (the “enhancing substance”).J. E x p . Med. 105, 125-139. Klein, J. (1975). “Biology of the Mouse Histocompatibility-2 Complex.” Springer-Verlag, Berlin and New York. Klein, J., and Hauptfeld, V. (1976). Ia antigens: Their serology, molecular relationships, and their role in allograft reactions, Transplant. Reo. 30,83-100. Klein, J., Hauptfeld, V., and Hauptfeld, M. (1975).Evidence for a fifth (G) region in the H-2 complex of the mouse. lmmunogenetics 2, 141-150. Klein, P. A. (1975). Anomalous reactions of mouse alloantisera with cultured tumor cells. I. Demonstration of widespread occurrence using reference typing sera. J . Immunol. 115,1254-1260. Koide, N., and Muramatsu, T. (1974). Endo-P-N-acetylglucosaminidaseacting on carbohydrate moieties of glycoproteins. J . Biol. Chem. 249,4897-4904. Kronvall, G . (1973). Purification of staphylococcal protein A using immunosorbents. Scand. J . Immunol. 2,31-36. Kronvall, G . , and Frommel, D. (1970). Definition of staphylococcal protein A reactivity for human immunoglobulin G fragments. Immunochemistry 7 , 124-127. Kronvall, G., Seal, U. S., Finstad, J., and Williams, R. C. (1970). Phylogenetic insight into evolution of mammalian Fc fragment of y G globulin using straphylococcal protein A. J . Zmmunol. 104, 140-147. Kuchel, P. W., Campbell, D. G., Barclay, A. N., and Williams, A. F. (1978). Molecular weights of the Thy-1 glycoproteins from rat thymus and brain in the presence and absence of dioxycholate. Biochem. J . 169,411 4 1 7 . Law, L. W., and Appella, E. (1976). Biological and biochemical properties of solubilized histocompatibility-2 (H-2) alloantigens. Contemp. Top. Mol. Immunol. 5,69-85. Lee, J. C., and Ihle, J. N. (1975). Autogenous immunity to endogenous RNA tumor virus: Reactivity of natural immune sera to antigenic determinants of several biologically distinct murine leukemia viruses. J . Natl. Cancer Inst. 55,831-838. Letarte, M., and Meghji, G. (1978). Purification and characterization of Thy-1 glycoprotein from mouse brain. J . Immunol. (submitted). Letarte-Muirhead, M., Acton, R. T., and Williams, A. F. (1974). Preliminary characterization of Thy-1.1 and Ag-B antigens from rat tissues solubilized in detergents. Biochem. J . 143,51-61.

GLYCOPROTEIN ANTIGENS

509

Letarte-Muirhead, M., Barclay, A. N., and Williams, A. F. (1975). Purification of the Thy-1 molecule, a major cell-surface glycoprotein of rat thymocytes. Biochem. J. 151,685-697. MacDonald, M. E., Letarte, M., and Bernstein, A. (1978). Genetic studies on the control of erythrocyte membrane antigen expression during Friend cell differentiation. J. Cell. Physiol. in press. Morris, R. J., and Williams, A. F. (1975). Antigens on mouse and rat lymphocytes recognized by rabbit antiserum against rat brain: The quantitative analysis of a xenogeneic antiserum. Eur. J . lmniunol. 5 , 274-281. Morris, R. J., and Williams, A. F. (1978). Serological and preliminary biochemical characteristics of a T lymphocyte differentiation antigen detected by rabbit antiserum to rat thymocyte membranes. Eur. J . Immunol. 7,360-365. Morris, R. J., Letarte-Muirhead, M., and Williams, A. F. (1975). Analysis in deoxycholate of three antigenic specificities associated with the rat Thy-1 molecule. Eur. J . Immunol. 5,282-285. Muramatsu, T. (1971). Demonstration of an endo-glycosidase acting on a glycoprotein. J . Biol. Cheni. 246,5535-5537. Muramatsu, T., Nathenson, S. G., Boyse, E. A., and Old, L. J. (1973). Some biochemical properties of thymus leukemia antigens solubilized from cell membranes by papain digestion. J. E x p . Med. 137, 1256-1262. Nathenson, S. G., and Cullen, S. E. (1974). Biochemical properties and immunochemical-genetic relationships of mouse H-2 alloantigens. Biochim. Biophys. Actu 344, 1-25. Nathenson, S. G., and Muramatsu, T. (1971). Properties of the carbohydrate portion of mouse H-2 alloantigen glycoproteins. I n “Glycoproteins of Blood Cells and Plasma” (G. A. Jamieson and T. J. Greenwalt, eds.), pp. 245-266. Lippincott, Philadelphia, Pennsylvania. Nathenson, S. G., Brown, J. L., Ewenstein, B. M., Nisizawa, T., Sears, D. W., and Freed, J. H. (1976). Structural differences between parent and variant H-2K glycoproteins from mouse strains carrying H-2 gene mutations. Cold Spring Harbor Symp. Quant. Biol. 41,343-350. Natori, T., Katagiri, M., Tanigaki, N., and Pressman, D. (1974). The 11,000 dalton component of mouse H-2. Transplantation 18,550-555. Niederhuber, J. E., and Frelinger, J. A. (1976). Expression of Ia antigens o f T and B cells and their relationship to immune-response functions. Transplant. Rev. 30, 101121. Nowinski, R. C., and Kaehler, S. L. (1974). Antibody to leukemia virus: Widespread occurrence in inbred mice. Science 185, 869-870. Nowinski, R. C., and Klein, P. A. (1975). Anomalous reactions of mouse alloantisera with cultured tumor cells. 11. Cytotoxicity is caused by antibodies to leukemia viruses. J. lmmunol. 115, 1261-1268. Obata, Y., Ikeda, H., Stockert, E., and Boyse, E. A. (1975). Relation of GIx antigen of thymocytes to envelope glycoprotein of murine leukemia virus. J . E x p . Med. 141, 188-197. Ostberg, L., Rask, L., Wigzell, H., and Peterson, P. A. (1975). Thymus leukaemia antigen contains p2-microglobulin. Nature (London)253,735-737. Pancake, S . J., and Nathenson, S. G. (1973). Selective loss of H-2 antigenic reactivity after chemical modification. J. lmmunol. 111, 1086-1092. Parish, C. R., Chilcott, A. B., and McKenzie, I. F. C. (1976a). Low molecular weight Ia antigens in normal mouse serum. I. Detection and raising of a xenogeneic antiserum. Inmunogenetics 3, 113-128.

51 0

MICHELLE LETARTE

Parish, C. R., Jackson, D. C., and McKenzie, I. F. C. (1976b). Low molecular weight Ia antigens in .normal mouse serum. 111. Isolation and partial chemical characterization. Immunogenetics 3,455-463. Poulik, M. D., and Reisfeld, R. A. (1975).p,-microglobulins. Contemp. Top. Mol. Immunol. 4, 157-204. Rask, L., Lindblom, J. B., and Peterson, P. E. (1974). Subunit structure of H-2 alloantigens. Nature (London)249,833-834. Reif, A. E., and Allen, J. M. V. (1964). The AKR thymic antigen and its distribution in leukemias and nervous tissues.]. E x p . Med. 120,413-433. Reif, A. E., and Allen, J. M. V. (1966). Mouse nervous tissue iso-antigens. Nature (London) 209,523, Rinderknecht, H. (1962). Ultra-rapid fluorescent labelling of proteins. Nature (London) 193, 167-168. Robinson, N. C., and Tanford, C. (1975). The binding of deoxycholate, Triton X-100, sodium dodecyl sulfate and phosphatidylcholine vesicles to cytochrome b,. Biochemistry 14,369-378. Sachs, D. H., and Cone, J. L. (1973).A mouse B-cell alloantigen determined by gene(s) linked to the major histocompatibility complex. J . E x p . Med. 138, 1289-1304. Sanderson, A. R. (1964). Cytotoxic reactions of mouse iso-antisera: Preliminary considerations. Br. J . E x p . Pathol. 45,398-408. Scheid, M., Boyse, E. A., Carswell, E. A., and Old, L. J. (1972). Serologically demonstrable alloantigens of mouse epidermal ce1ls.J. E x p . Med. 135,938-955. Schwartz, B. D., and Nathenson, S. G. (1971).Isolation of H-2 alloantigens solubilized by the detergent NP40.J. Immunol. 107, 1363-1367. Schwartz, B. D., Kato, K., Cullen, S. E., and Nathenson, S. G. (1973).H-2 histocompatibility alloantigens: Some biochemical properties of the molecules solubilized by NP-IO detergent. Biochemistrfi 12,2157-2164. Segrest, J. P., Jackson, R. L., Andrews, E. P., and Marchesi, V. T. (1971). Human erythrocyte membrane glycoprotein: A re-evaluation of the molecular weight as determined by SDS polyacrylamide gel electrophoresis. Biochem. Biophys. Res. Commun. 44,390-395. Segrest, J. P., Jackson, R. L., and Marchesi, V. T. (1972). Red cell membrane glycoproteins: Amino acid sequence of an intramembranous region. Biochem. Biophys. Res. Commun. 49,964-969. Shihada, A., and Nathenson, S. G. (1967).Solubilization of membrane H-2 isoantigens: Chromatographic separation of specificities determined by a single H-2 genotype. Biochem. Biophys. Res. Commun. 29,828-833. Shimada, A., and Nathenson, S. G. (1969). Murine histocompatibility-2 (H-2) alloantigens: Purification and some chemical properties of soluble products from H-2b and H-2d genotypes released by papain digestion of membrane fractions. Biochemistry 8,4048-4062. Shimada, A., Yamane, K.,and Nathenson, S. G. (1970). Comparison of the peptide composition of two histocompatibility-2 alloantigens. Proc. Natl. Acad. Sci. U.S.A. 65, 691-696. Shreffler, D. C. (1978). Serology, Genetics and Mapping of Ia Antigens. In “Ir Genes and Ia Antigens” (H. 0. McDevitt, ed.) pp. 3-7. Academic Press, New York. Shreffler, D. C., David, C. S., Passmore, H. C., and Klein, J. (1971). Genetic organization and evolution of the mouse H-2 region: A duplication model. Transplant. Proc. 3, 176-179. Silver, J . (1976). Partial amino sequences of MHC products. Cold Spring Harbor Syrnp. Quant. Biol. 41,405-406.

GLYCOPROTEIN ANTIGENS

51 1

Silver, J., and Hood, L. (1974). Detergent-solubilised H-2 alloantigen is associated with a small molecular weight polypeptide. Nature (London) 249,764-765. Silver, J., and Hood, L. (1975). An approach to amino acid sequence analysis of transplantation antigens. Nature (London) 2 5 6 , 6 3 4 4 . Silver, J., and Hood, L. (1976a). Structure and evolution of transplantation antigens: Partial amino acid sequences of H-2K and H-2D alloantigens Proc. Natl. Acad. Sci. U.S.A. 73, 599-603. Silver, J., and Hood, L. (1976b). Preliminary amino acid sequences of transplantation antigens: Genetic and evolutionary implications. Contemp. Top. Mol. Immunol. 5, 35-68. Snell, G. D. (1958).Histocompatibility genes of the mouse. 11. Production and analysis of isogenic resistant lines.]. Natl. Cancer Inst. 21,843-877. Snell, G . D., Cherry, M., and Demant, P. (1971). Evidence that H-2 private specificities can be arranged in two mutually exclusive systems possibly homologous with two subsystems of HL-A. Transplant. Proc. 3, 183-186. Sonoda, S., and Schlamowitz, M. (1970). Studies of Iz5I-tracelabeling of immunoglobulin G by chloramine-T. lmmunochemistry 7,885-898. Springer, T. A., Kaufman, J., Terhorst, C., and Strorninger, J. L. (1978).Nature (London) 268,213-218. Standring, R., and Williams, A. F. (1978). Glycoproteins and antigens of membranes prepared from rat thymocytes after lysis by shearing or with the detergent Tween-40. Biochim. Biophys. Acta. 508,85-96. Stanton, T. H., Bennett, J. C., and Wolcott, M. (1976). Quantitative isolation of mouse thymus leukemia antigens, TL. Tissue Antigens 7, 158-164. Stern, P. L. (1973). Theta alloantigen on mouse and rat fibroblasts. Nature New Biol. London 246,76-78. Strand, M., Lilly, F., and August, J. T. (1974). Host control of endogenous murine leukemia virus gene expression: concentrations of viral proteins in high and low leukemia mouse strains. Proc. Nat. Acad. Sci. USA 71,3682-3686. Tanford, C., Nozaki, Y., Reynolds, J. A., and Makino, S. (1974). Molecular characterization of proteins in detergent solutions. Biochemistry 13,2369-2376. The, T. H., and Feltkamp, T. E. W. (1970a). Conjugation of fluorescein isothiocyanate to antibodies. I. Experiments on the conditions of conjugation. Immunology 18,865873. The, T. H., and Feltkamp, T. E. W. (1970b). Conjugation of fluorescein isothiocyanate to antibodies. 11. A reproducible method. Zmmunology 18,875-881. Toyoshima, S., Fukada, M., and Osawa, T. (1972). Chemical nature of the receptor site for various phytomitogens. Biochemistql 11,4000-4005. Trowbridge, I. S., Weissman, I. L., and Bevan, M. J. (1975). Mouse T-cell surface glycoprotein recognized by heterologous anti-thymocyte sera and its relationship to THY-1 antigen. Nature (London)256,652-654. Tung, J. S., Fleissner, E., Vitetta, E. S., and Boyse, E. A. (1975a). Expression of murine leukemia virus envelope glycoprotein gp69/71 on mouse thymocytes-Evidence for two structural variants distinguished by presence vs absence of GMantigen. ]. E x p . Med. 142,518-523. Tung, J. S., Vitetta, E. S., Fleissner, E., and Boyse, E. A. (1975b). Biochemical evidence linking the Glx thymocyte surface antigen to the gp69/71 envelope glycoprotein of murine leukemia virus. J . Exp. Med. 141, 198-205. Vitetta, E., Uhr, J . W., and Boyse, E. A. (1972). Isolation and characterization of H-2 and T L alloantigens from the surface of mouse lymphocytes. Cell. Immunol. 4, 187191.

512

MICHELLE LETARTE

Vitetta, E. S., Capra, J. D., Klapper, D. G., Klein, J., and Uhr, J. W. (1976a). The partial amino acid sequence of an H-2K molecule. Proc. Natl. Acad. Sci. U.S.A. 73, 905909. Vitetta, E. S., Poulik, M. D., Klein, J., and Uhr, J. W. (1976b). Beta 2-microglobulin is selectively associated with H-2 and TL alloantigens on murine lymphoid cells. J . E x p . Med. 144, 179-192. Welsh, K. I., Dorval, G., and Wigzell, H. (1975). Rapid quantitation of membrane antigens. Nature (London)254,67-69. Williams, A. F. (1973).Assays for cellular antigens in the presence of detergents. Eur. J . Immunol. 3,628-632. Williams, A. F. (1975). IgG, and other immunoglobulin classes on the cell surface of rat lymphoid cells. Eur. J. Immunol. 5, 883-885. Williams, A. F. (1976). Many cells in rat bone marrow have cell-surface Thy-1 antigen. Eur. J . Immunol. 6,526-528. Williams, A. F. (1977). Differentiation antigens of the lymphocyte cell surface. Contemp. Top. M o l . Immunol. 6,83-116. Williams, A. F., Barclay, A. M., Letarte-Muirhead, M., and Morris, R. J. (1976). Rat Thy-1 antigens from thymus and brain: Their tissue distribution, purification and chemical composition. Cold Spring Harbor Symp. Quant. Biol. 4 1 , 5 1 4 2 . Wofsy, L., Baker, P. C., Thompson, K., Goodman, J., Kimura, J., and Henry, C. (1974). Hapten-sandwich labelling. I. A general procedure for simultaneous labelling of multiple cell surface antigens for fluorescence and electron microscopy. J . E x p . Med. 140,523-537. / Wood, B. T., Thompson, S. H., and Goldstein, G. (1965). Fluorescent antibody staining. 111. Preparation of fluorescein-isothiocyanate-labelledantibodies. J . Immunol. 95, 225-229.

Subject Index A

C

Acetylcholinesterase composition of, 164 properties of, 148-149,164-166 N-Acetylglucosaminyltransferases biosynthesis of, 53-55 &D-N-Acetylhexosaminidases, properties and functions of, 185-186 N-Acetyllactosamine oligosaccharides, elongation of, 53-58 Actinlike cables, intracellular, proteaseinduced modification of, 435-436 Active transport, concept of, 5 Adhesion, of cells, see Cell adhesion Alkaline phosphatases composition of, 164 polarities of, 152 properties of, 166-169 Anomeric configuration, in glycosylation,

CAMP, intracellular, protease modification of, 433435 Carbohydrates of enveloped viruses, 240 Carrier concept of, 5 CMP sialic acid biosynthesis of, 40-42 Cell, surface antigens of, analysis, 467-

50-51 Antigens, glycoprotein type, see Glycoprotein antigens viral glycoproteins as, 254-255 Aminopeptidases in membrane transport, 205-206 polarity of, 152 properties of, 149 Arylsulfatase A composition of, 164 properties of, 150,187-188 Ashwell’s mammalian lectin structure of,

27-28

477 Cell adhesion biochemical determinants of, 327-396 cell-cell type, 356-381 cell substrate type, 328-356 cellular movement and, 348-353 GAGS in, 341-344 serum macromolecules in, 329-337 surface adhesion complex, 337-353 to collagen, 353-356 complex polysaccharide role in, 374-

376 factors inhibiting, 371-374 factors promoting, 368-371 glycosyltransferases in, 376-381 Cell division erythrocyte membrane and, 399-403 limited autolysis as inducer of, 437444 protease stimulated, media component role in, 444450 proteolytic modification of surface in,

397-461

Asn-GlcNAc oligosaccharides, biosynthesis of, 48-49 Autolysis, cell division induction by, 437-

444 B Bilayer membrane concept, 4 Birds, cell-cell adhesion in, 365-376 Brush border hydrolases, properties and functions of, 170-183

stimulation of, in lymphocytes, 404411 Cell fusion, viral glycoprotein role in, 252 Cell membranes asymmetry of, 8, 16-19 biogenesis of, 67-85 carbohydrate asymmetry across, 16-19 enzymes of, 145-231 association, 188-194 of erythrocytes, glycoproteins of, 281-

316 historical review of, 1-13 513

514

SUBJECT INDEX

isolation of, 115-116 model systems of, 34-37 protease modification of, 424429 macroarchitectural changes, 429430 research papers on, 2 of viruses, see Enveloped viruses Cellular radioimmunoassay, of glycoprotein antigens, 471-475 Chick embryo fibroblasts, division of,

413415 Chlamydamonas cell-cell adhesion in,

357-359 Collagen cell adhesion to, 353-356 Contact inhibition of movement, in cell division, 430432 Cytochrome b, structure of, 28-29 Cytochrome b, reductase polarity of, 152 properties of, 148, 164 structure of, 29-30 Cytoskeleton, role in plasma membrane biogenesis, 79 Cytotoxicity assay, of cell surface antigens, 469 Division, of cells, see Cell division

D Denaturing solvents, glycoprotein fractionation by, 119-120 Detergents, glycoprotein fractionation by,

117-119 Disaccharidases, in membrane transport,

204 -205 Dolichol a-D-mannopyranosyl phosphate, biosynthesis of, 44-46 Dolichol phosphate monosaccharides, biosynthesis of, 4 2 4 4 Dolichol pyrophosphate N,N’-diacetylchitobiose, biosynthesis of, 44 Dolichol pyrophosphate oligosaccharides, biosynthesis of, 46-53

composition of, 164 Enveloped viruses arrangement of envelope components of, 240-242 assembly of, 260-260 event sequence, 263-266 macromolecular interactions, 266-

268 glycoproteins of, 233-277 carbohydrate function in, 255-258 structure and function, 242-260 effect on lipid bilayer, 258-260 list of, 235 membrane glycoproteins of, 233-277 structure of, 28-29 Enzymes of glycoprotein membranes, 145-231 biosynthetic and developmental aspects, 206-210 amino acids, 194 carbohydrate role, 195-201 composition, 164, 195 functional interrelationships, 201-206 list and properties, 148-150 monosaccharide composition, 167 membrane association, 188-194 polarities, 159 Epiglycanin, structure of, 26 Erythrocytes glycoproteins of, 279-325 membrane of, surface modification of, 399-403 origin and turnover of, 280-281

F F enzyme, properties of, 149 Fibroblasts, protease induction in cell division of, 412-420 Fluorescent antibody test, for glycoprotein antigens, 469471 Fucosyltransferase, biosynthesis of, 55

E

G

Ectoenzymes, as glycoproteins, 210 Endoplasmic reticulum, glycoproteins of,

GAGS, chemistry of, 341-344 Galactose oxidase, as surface-labeling agent, 111-112 Galactosyltransferases, biosynthesis of,

83-84 Enterokinase polarity of, 152 properties of, 150, 176-177

55-57 GDP fucose, biosynthesis of, 39

SUBJECT INDEX

GDP mannase, biosynthesis of, 38-39 Gel electrophoresis, of glycoproteins, 122 Gel filtration chromatography, of glycoproteins, 121 Genetic analysis, of membrane glycoproteins, 133-134 y-Glucosyltranspeptidases, polarities of, 152 p-Glucuronidase composition of, 164 properties of, 150, 183-184 a-Glutamyl-pnaphthylamidase, properties of, 150 YGlutarnyltranspeptidase

composition of, 149 in membrane transport, 206 properties of, 149, 180-182 Glycoprotein(s) (membrane) biosynthesis of diagram, 20 dolichol pathway, 51-53 carbohydrate structure of, 131-132 chemical analysis of, 129-133 of endoplasmic reticulum, 83-84 of enveloped viruses, 233-277 enzymes of, 145-231 of erythrocytes, 279-325 carbohydrate-bound, 311-3 16 functions, 304-316 organization, 285-288 periodate-stainable, 282-284 structure, 288-304 surface changes, 309-311 variants, 307-309 fractionation of, 117-129 by denaturing solvents, 119-120 by detergents, 117-119 by polyacrylarnide gels, 120-122 function of, 11 gel electrophoresis of, 122 gel filtration chromatography of, 121 genetic analysis of, 133-134 historical review of, 1-13 identification of, 108-115 ion-exchange chromatography of, 121122 isoelectric focusing of, 122 lectin affinity studies of, 122-129 list of, 17 metabolic labeling of, 116-117

51 5 molecular weights of, 132-133 polypeptides of, 130-131 structure of, 19-33 sugar analysis of, 131 surface labeling of, 108-109 of viruses migration through host cell, 79-83 Glycopeptides of cell membranes protease modiication of, 4 2 4 4 2 9 Glycophorin, structure of, 24 Glycophorin A, see Sialoglycoprotein Glycoprotein antigens, 463-512 antisera production from, 4 6 7 4 6 9 cellular radioimmunoassay of, 471-475 cytotoxicity assay of, 469 fluorescent antibody test for, 4 6 9 4 7 1 H-2 antigens, 477-487 Ia antigens, 4 8 7 4 9 3 immunoprecipitation identification of, 475-476 Ly-2,3 antigens, 502-503 microsequencing techniques for, 476477 Thy-1 antigen, 493-500 Tla antigens, 500-502 Glycoproteins, of enveloped viruses, 233277 Clycosidases, use in studies of glycoprotein enzymes 200-201 Glycosylation reaction N-acetyllatosamine oligosaccharide formation, 53-58 dolichol pyrophosphate oligosaccharide formation, 46-53 mechanisms of, 37-61 nucleotide sugar formation, 37-42 polyprenol phosphate sugar formation by, 42-46 ser (thr)-GalNAc oligosaccharide biosynthesis by, 58-61 subcellular sites of, 61 -67 autoradiography, 61-64 Glycosyltransferases, 147, 151-157 properties of, 148, 151-153 subcellular localization of, 64 surface location of, 153-157 roll in cell adhesion, 376-381 GM,-ganglioside-pgalactosidase, properties of, 150, 184-185

516

SUBJECT INDEX

Golgi complex, role in plasma membrane biogenesis, 70-78

H H-2 antigens, 477-487 amino acid sequence of (partial), 480-

483 antigenic determinants of, 485-487 carbohydrate structure of, 483-485 isolation and properties of, 4 7 8 4 8 0 tissue distribution and polymorphism of, 477-478 H-2 gene complex, nomenclature for, 465 Hen oviduct system, glycosylation of, 49-

50 Hexosaminidases, properties of, 150

I

Lectin affinity glycoprotein studies by, 122-129 enzyme studies, 196-200 LETS glycoprotein, structure of, 28 LETS protein, role in protease-induced cell division, 4 2 7 4 2 8 Lipid bilayer of enveloped viruses, 238-239 glycoprotein effects on, 258-260 Lipid mosaic membrane theory,'4 Lipids, viruses containing, membranes of,

233-236 L y 4 3 antigens, preliminary characterization of, 502-503 Lymphocytes cell division in, 404-411 glycoprotein antigens of, 463-512 Lysosomal acid hydrolases, properties and functions of, 183-187

Ia antigens, 487-493 amino acid structure (partial) of, 490-

492 antigenic determinants of, 492-493 carbohydrate structure of, 490-492 isolation and properties of, 488-490 tissue distribution and polymorphism

of, 487-488 Immunoprecipitation, of glycoprotein antigens, 475-476 Influenza virus, glycoproteins of, 242-246 Intracellular membranes, biogenesis of,

83 -85 Ion-exchange chromatography, of glycoproteins, 121-122 Isoelectric focusing, of glycoproteins, 122

1 Lactase-phlorizin hydrolase composition of, 164 properties of, 149, 175-176 Lactoperoxidase, as surface-labeling agent, 109-11 1 Lectin(s) Ashwell's mammalian, structure of, 27-

28 effect on lymphocyte cell division,

411 protease effect on agglutinability of,

421 -424

M Maltase-gl ycosamylase composition of, 164 polarity of, 152 properties of, 149, 173-175 Membrane transport, glycoprotein enzymes in, 203-206 Membranes, of cells, see Cell membranes Metalloendopeptidase, properties of, 150,

182-183 Microsequencing techniques, for glycoprotein antigens, 476-477 Mitochondria, glycosylation in, 84 Mobile carrier, concept of, 5 Mosaic membrane concept, 4 Murine histocompatibity antigens, structure of, 26-27 Myelin glycoproteins of, 30-32 proteolipid protein of structure, 32-33

N NA activity, viral glycoprotein role in, 252

-254

+

(Na ,K + )Mg2-ATPase composition of, 152 properties of, 148, 161-163 Nuclei, glycosylation in, 84-85 5'-Nucleotidase

51 7

SUBJECT INDEX

composition of, 152 properties of, 148, 158-161 Nucleotide pyrophosphatase composition of, 164 polarity, 152 properties of, 148, 157-158 Nucleotide sugars, formation of, in glycosylation reaction, 37-42

0 Oligoaminopeptidases polarities of, 152 properties of, 149, 177-180 composition of, 164 Overgrowth-stimulating factor, in fibroblast division, 412-420

P Paramyxoviruses, glycoproteins of, 249-

250 PAS-2 (glycoprotein), isolation and structure of, 295-296 PAS-3 (glycoprotein), isolation and structure of, 296-297 Periodate, effect on lymphocyte cell division, 408-411 Plasma membrane biogenesis of, 70-83 glycosylation in, 49 Plasminogen, activator of, in cell division,

44 1-444 Polypeptide 3 (glycoprotein) function of, 304-306 isolation and structure of, 297-304 Proteases effect on fibroblast surface, 421-432 effect on lymphocyte cell division, 404-

408 induction of, in fibroblast division, 412-

420 transmembrane events induced by, 432

Rhodopsin, structure of, 25-26 Ribosome, role in glycosylation, 64-67

5 S enzyme, properties of, 149 Secretory vesicles, glycosylation in, 85 Ser(thr)-galNAc oligosaccharides, assembly of, 58-61 Sialoglycoprotein antigens and receptors present on, 292295 isolation and structure of, 288-292 Sialyltransferases, biosynthesis of, 57-58 Slime mold, cell-cell adhesion in, 362365 Sponge, cell-cell adhesion in, 359-361 Sucrase-isomaltase composition of, 164 polarity of, 152 properties of, 149, 170-172 Surface-attached material (SAM) cellular proteins of, 345 chemistry of, 341-348 composition of, 347

T 3T3 fibroblasts, division of, 416-419 Thrombin, cell-surface peptide sensitive to, 428-429 Thy-1 antigen, 493-500 amino acid and carbohydrate composition of, 495-499 antigenic determinants of, 499-500 isolation and MW of, 493-495 tissue distribution and number of, 493 Tla antigens, preliminary characterization of, 500-502 Togaviruses, glycoproteins of, 247-249 Transglutaminases, as surface-labeling agents, 112-113 Transmembrane proteins, surface labeling of, 113

-436

U

R Receptors, viral glycoprotein adsorption to, 250-252 Rhabdovirus G , glycoproteins of, 246-247

UDP N-acetylgalactosamine, biosynthesis of, 3 9 4 0 UDP galactase, biosynthesis of, 38 UDP N-acetylglucosamine, biosynthesis

of, 39-40

51 8

SUBJECT INDEX

V Viruses enveloped, structure of, 28-29 glycoproteins of, migration to host cell, 79-83

membranes of, components of, 236-240 Voltage clamp procedure, 5

Y Yeast, cell-cell adhesion in, 357-359