Advances in Molecular and Cell Biology Volume 15b BIochemical Technology

ADVANCES IN MOLECULAR A N D CELL BIOLOGY

Volume 75B

1996

BIOCHEMICAL TECHNOLOGY

This Page Intentionally Left Blank

ADVANCES IN MOLECULAR AND CELL BIOLOGY BIOCHEMICAL TECHNOLOGY Series Editor: E. EDWARD BITTAR Department of Physiology University of Wisconsin Madison, Wisconsin

Guest Editors: BENGT DANIELSSON LElF BULOW Department of Pure and Applied Biochemistry Chemical Center Lund University Lund, Sweden

VOLUME 15B

1996

@ JAl PRESS INC. Greenwich, Connecticut

London, England

Copyright 0 1996 by ]A/ PRESS INC. 55 Old Post Road No. 2 Greenwich, Connecticut 06836 JAl PRESS LTD. The Courtyard 28 High Street Hampton Hill, Middlesex TW12 1PD England All rights reserved. No part of this publication may be reproduced, stored on a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, filming, recording, or otherwise, without prior permission in writing from the publisher. ISBN: 0-7623-01 14-7 Manufactured in the United States of America

CONTENTS (Volume 15B) ...

LIST OF CONTRIBUTORS

Xlll

PREFACE Klaus Mosbach

xxvii

INTRODUCTION

xxxiii IV. BIOSENSORS

BIOSENSORS: AN INTRODUCTION Bengt Danielsson

349

OVERVIEW OF BIOSENSOR TECHNOLOGY Frieder W. Scheller, Ulla Wollenberger, Dorothy Pfeiffer, and Florian Schubert

353

CURRENT STATE OF BIOSENSORS George G. Guilbault

365

BIOSENSORS AND MICROMACHINING lsao Karube, Kenji Yokoyama, Yuji Murakami, and Masayuki Suda

375

RECYCLING SENSORS BASED ON KINASES Ulla Wollenberger, Florian Schubert, Dorothea Pfeiffer, and Frieder W. Scheller

381

ELECTRON CONDUCTING ADDUCTS OF WATER-SOLUBLE REDOX POLYELECTROLYTES AND ENZYMES loanis Katakis, Mark Vreeke, Ling Ye, Atsushi Aoki, and Adam Heller

391

CONTENTS (Volume 156)

vi

SCREENING AND DESIGN OF IMMOBILIZED BIOCATALYSTS BY MEANS OF KINETIC CHARACTERIZATION ON ENZYME THERMISTOWHERMAL ASSAY PROBE Peter Germeiner, Vladimir Stefuca, and Bengt Danielsson

41 1

AMPEROMETRIC BIOSENSORS BASED ON CARBON PASTE ELECTRODES CHEMICALLY MODIFIED WITH REDOX ENZYMES L. Gorton, G. Marko-Varga, 6. Persson, Z. Huan, H. Lid& E. Burestedt, S. Ghobadi, M. Smolander, S. Sahni, and T. Skotheim

42 1

ENZYME BASED DIFFUSION BADGE FOR THE DETECTION OF FORMALDEHYDE R. Feldbrugge, K.P. Rindt, and A. Borchert

45 1

BlOSENSlNG OF HEAVY METAL IONS BASED ON SPECIFIC INTERACTIONS WITH APOENZYMES lkuo Satoh

461

DESIGN OF HIGH-ANNEALING TEMPERATURE PCR PRIMERS AND THEIR USE IN THE DEVELOPMENT OF A VERSATILE LOW-COPY NUMBER AMPLIFICATION PROTOCOL Michael W. Mecklenburg

473

ON-LINE MONITORING OF INDUSTRIAL FERMENTATIONSUSING A SPLIT-FLOW MODIFIED THERMAL BIOSENSOR M. Rank and B. Danielsson

491

MINIATURlZED THERMAL BIOSENSORS U. Hedberg, B. Xie, and B. Danielsson

499

V. AFFINITY TECHNIQUES FOR SEPARATION A N D BlORECOCNlTlON INTRODUCTORY REMARKS Per-Olof Larsson

509

Contents (Volume 158)

vii

AFFINITY CHROMATOGRAPHY A N D RELATED TECHNIQUES, PERSPECTIVES, A N D TRENDS Christopher R. Lowe

51 3

ONE-STEP AFFINITY PURIFICATION OF A RECOMBINANT CYCLODEXTRIN GLYCOSYL TRANSFERASE BY (Cu[IIl, Zn[IIl TANDEM COLUMN) IMMOBILIZED METAL ION AFFINITY CHROMATOGRAPHY F! Berna, F.F. Moraes, J.N. Barbotin, D. Thomas, and M.A. Vijayalakshmi

523

AFFINITY PURIFICATION OF ENZYMES USING TEMPERAT URE-I NDUC ED PHASE S EPARATION Folke Tjerneld

539

POSSIBLE BINDING SITES ON PROTEINS INVOLVED IN THlOPHlLlC ADSORPTION Alexander Schwarz and Meir Wilchek

547

NEW OPPORTUNITIES FOR USING IMMOBILIZED LIGANDS TO CHARACTERIZE MACROMOLECULAR RECOGNITION A N D DESIGN RECOGNITION MOLECULES Irwin Chaiken, David Myszyka, and Thomas Morton

553

PEPTIDES: MULTIPLE PURPOSE TOOLS Jean-Luc Fauch&re

569

SUPERPOROUS AGAROSE--A NEW MATERIAL FOR CHROMATOGRAPHY Per-Olof Larsson

585

HYDROPHILIC A N D AMPHIPHATIC MONOMERS AND USE OF THEIR GELS AS SEPARATION MEDIA Branko Kozulic and Urs Heimgartner

593

AN INTEGRATED APPROACH IN THE ANALYTICAL DESCRIPTION OF AFFINITY CHROMATOGRAPHY, BIOSENSORS, IMMOBILIZED BIOCATALYSTS, A N D SIMILAR SYSTEMS Volker Kasche

605

CONTENTS (Volume 15B)

viii

VI. MOLECULAR RECOGNITION MOLECULAR RECOGNITION: AN INTRODUCTION Ian A. Nicholls

62 1

THE ROLE OF GEOMETRIC FIT BETWEEN PROTEIN MOLECULES AND THEIR LIGANDS IN DETERMINING BIOLOGICAL SPEC1FlClTY Ephraim Katchalski-Katzir, lssac Shariv, Miriam Eisenstein, Asher A. Friesem, Claude Aflalo, and llya A. Vakser

623

MODELS OF THE BINDING SITES OF ENZYMES: TEMPLATE INDUCED PREPARATION OF SPECIFIC BINDING SITES IN CROSSLINKED POLYMERS Gunter Wulff

639

MOLECULAR IMPRINTING: THE CURRENT STATUS AND FUTURE DEVELOPMENT OF POLYMER-BASED RECOGNITION SYSTEMS Lars I. Andersson, Ian A. Nicholls, and Klaus Mosbach

65 1

AN APPROACH TOWARD THE SEMIQUANTITATION OF MOLECULAR RECOGNITION PHENOMENA IN NON-COVALENT MOLECULARLY IMPRINTED POLYMER SYSTEMS: CONSEQUENCES FOR MOLECULARLY IMPRINTED POLYMER DESIGN Ian A. Nicholls

669

CONTENTS (Volume 15A) ...

LIST OF CONTRIBUTORS

Xlll

PREFACE Klaus Mosbach

xxvii

INTRODUTION

xxxiii

1. ENZYME TECHNOLOGY ENZYMOLOGY-AN INTRODUCTION Mats-Olle Mdnsson

3

TAILORING THE MICROENVIRONMENT OF ENZYMES IN WATER-POOR MEDIA Bo Mattiasson, Patrick Adlercreutz, Ernst Wehtje, and Marina Otamiri

5

MODIFICATION OF ENZYMES AND PROTEINS WITH BlOlMPRlNTlNG PROCEDURES Mats-Olle Mdnsson

15

A NEW KIND OF ABZYMES: ANTI-IDIOTYPIC ANTIBODIES EXHIBITING CATALYTIC ACTIVITIES Alain Friboulet, Catherine Bedel-Cloutour, and Daniel Thomas

23

MODULATION OF THE CATALYTIC PATHWAY OF CARBOXYPEPTIDASE A BY CONJUGATION WITH POLYVINYLALCOHOLS Beka Solomon and Leon Goldstein

33

NEURAL NETWORKS IN ENZYMOLOGY Yi-HongZhu, Susan Linko, and Pekka Linko

47

ix

CONTENTS (Volume 15A)

X

MODERN ENZYMOLOGY OF PLANT PEROXIDASES I.G. Gazaryan and A.M. Egorov

59

CAPACITY OF THE YEAST TRIGONOPSIS VARIABILIS (DSM 70714) FOR THE ENANTIOSELECTIVE REDUCTION OF ORGANOSILICON COMPOUNDS Lutz Fischer

69

TWO-STEP SULFATE-ENHANCED REFOLDING: RECOMBI NANT Pneumocystis carinii DIHYDROFOLATE REDUCTASE Michel Goldberg, Lawrence A. Greenstein, Avigdor Levanon, and Moshe M. Werber

79

MARINE BIOTECHNOLOGY BASED ON MARINE MICROORGANISMS Koji Sode, 1. Grant Burgess, and Tadashi Matsunaga

95

ENZYME ENGINEERING-THEN AND N O W PV: Sundaram

103

REFLECTIONS ON THE HISTORY A N D SCIENTIFIC CHARACTER OF BIOCHEMICAL ENGINEERING Klaus Buchholz

117

PART II. IMMOBILIZED CATALYSTS

IMMOBILIZED BIOCATALYSTS: A N INTRODUCTION Staffan Birnbaum

137

B I OT RA NS FORMAT10NS USING IMM0B ILIZED BIOCATALYSTS-PAST, PRESENT, AND FUTURE Malcolm D. Lilly

141

INDUSTRIAL APPLICATIONS OF IMMOBILIZED BIOCATALYSTS AND BIOMATERIALS lchiro Chibata

151

PREPARATION OF IMMOBILIZED PROTEINS COVALENTLY COUPLED THROUGH SILANE COUPLING AGENTS TO INORGANIC SUPPORTS Howard H. Weetall

161

Contents (Volume 75B)

xi

LONG -T ERM STA BI LI TY 0F C0NTI NU0USLY PERFUSED ANIMAL CELLS IMMOBILIZED ON NOVEL MACROPOROUS MICROCA RR I ERS H.Katinger, A. Assadian, G. Bliiml, N. Borth, A. Buchacher, 0. Doblhoff, T. Gaida, M. Reiter, C. Schmatz, K. Strutzenberger, W. Steinfellner, F. Unterluggauer, and N. Zach

193

EFFECTS OF IMMOBILIZATION ON THE CATALYTIC PROPERTIES A N D STABILITIES OF ENZYMES. A SURVEY B. Szajdni, L. Boross, M. Abrahdm, and L.M. Simon

209

CHARACTERISTICS OF LACCASE IMMOBILIZED ON DIFFERENT SUPPORTS FOR WINE-MAKING TECHNOLOGY A. Lante, A. Crapisi, G. Pasini, A. Zamorani, and P Spettoli

229

LlGNlN PEROXIDASE PRODUCTION WITH A N IMMOBILIZED FUNGUS BIOREACTOR Susan Linko and Reetta Haapala

237

CHELATE MEDl ATE D I MMOBI L lZATl0N 0F PROTEINS Dieter Kirstein

247

111. ENZYMATIC ASPECTS OF CELLULAR METABOLISM

ENZYMATIC ASPECTS OF CELLULAR METABOLISM: AN INTRODUCTION Leif Bulow

259

ARE SUBSTRATES CHANNELED IN THE KREBS CITRIC ACID CYCLE? Paul A. Srere, Richard D. Brodeur, Craig R. Malloy, A. Dean Sherry, and Balazs Sumegi

263

xii

CONTENTS (Volume 15A)

INTERACTION BETWEEN CH LOROPLAST PHOSPHOGLYCERATE KINASE A N D GLYCERALDEHYDE-3-PHOSPHATE DEHYDROC ENASE Louise E. Anderson, Xiao-yi Tang, Gote Johansson, Xingwu Wang, lvano A. Marques, and jerzy Macioszek

2 73

METABOLIC CONTROL ANALYSIS IN SITU: PROBLEMS AND PARADIGMS G. Rickey Welch

281

METABOLIC ENGl NEERlN G James E. Bailey

289

"TOGETHERNESS" BETWEEN PROTEINS GENERATED BY GENE FUSION Leif Bulow, Helen Carlsson, Peter Ljungcrantz, Mats Persson, and Christer Lindbladh

297

THE ESCHERlCHlA COLl CHAPERONE DNAK: PROPERTIES AND POTENTIALS Sven-Olof Enfors, Kristina Gustavsson, Shaojun Yang, and Andres Veide

31 1

ELICITATION OF CULTIVATED PLANT CELLS AS A TOOL IN BIOTECHNOLOGY AND BASIC BIOCHEMISTRY Peter E. Brodelius

319

INHIBITION OF PLANT GROWTH BY THE TETRAPEPTIDE DES-ARC ENTEROSTATIN (VPDP) Charlotte Erlanson-Albertssonand Per-Ake Albertsson

341

LIST OF CONTRIBUTORS

M. Abraha’m

Department of Biochemistry Attila J6zsef University, Szeged, Hungary

Patrick Adlecreutz

Department of Biotechnology Chemical Center, Lund University Lund, Sweden

Claude A flalo

Department of Biochemistry The Weizmann Institute of Science Rehovot, Israel

Per-Ake Albertsson

Department of Biochemistry Chemical Center, Lund University Lund, Sweden

Louise E. Anderson

Department of Biological Sciences University of Illinois at Chicago Chicago, I IIinois

Lars 1. Anderson

Department of Pure and Applied Biochemistry Chemical Center, Lund University, Lund, Sweden

Atsushi Aoki

Department of Chemical Engineering University of Texas Austin, Texas

A. Assadian

Institute of Applied Microbiology University of Agricu I tu re Vienna, Austria

lames E. Bailey

Institute of Biotechnology ETH-Honggerberg Zurich, Switzerland

...

Xlll

LIST OF CONTRIBUTORS

xiv

J. N. Barbotin

Laboratoire de Technologie Enzymatique Universite de Technologie de Compiegne Centre de Recherche de Royallieu G6nie Biologique Compiegne, France

P. Bema

Laboratoire Interaction M o l h l a i r e et de Technologie de Separation Universite de Technologie de Compiegne Centre de Recherche de Royallieu Genie Biologique Compiegne, France

Staffan Birnbaum

Department of Pure and Applied Biochemistry Chemical Center, Lund University Lund, Sweden

G.Bluml

Institute of Applied Microbiology University of Agriculture Vienna, Austria

A. Borchert

FH-Ostfriesland Emden, Germany

L. Bososs

Department of Chemistry and Biochemistry University for Horticulture and Food Industry Budapest, Hungary

N. Borth

Institute of Applied Microbiology University of Agriculture Vienna, Austria

Peter E. Brodelius

Department of Plant Biochemistry Lund University Lund, Sweden

Richard D. Brodeur

Department of Veterans Affairs Medical Center Dallas, Texas

A. Buchacher

Institute of Applied Microbiology University of Agriculture Vienna, Austria

xv

List of Contributors Klaus Buchholz

lnstitut fur Landwirdschaftlische Technologie und Zuckerindustrie Technical University Braunschweig Braunschweig, Germany

). Grant Burgess

Department of Biotechnology Faculty of Technology Tokyo University of Agriculture and Technology Tokyo, Japan

Leif Bulow

Department of Pure and Applied Biochemistry Chemical Center, Lund University Lund, Sweden

E. Burestedt

Department of Analytical Chemistry Chemical Center, Lund University Lund, Sweden

He/& Carlsson

Department of Pure and Applied Biochemistry Chemical Center, Lund University Lund, Sweden

Irwin Chaiken

SmithKline Beecham Pharmaceuticals King of Prussia, Pennsylvania

lchiro Chibata

Tanabe Seiyako Co. Ltd. Osaka, Japan

A. Crapisi

Dipartimento di Biotecnologie Agrarie Universiti di Padova Padova, Italy

Bengt Danielsson

Department of Pure and Applied Biochemistry Chemical Center, Lund University Lund, Sweden

0. Doblhoff

Institute of Applied Microbiology University of Agriculture Vienna, Austria

xv i

LIST OF CONTRIBUTORS

A. M . Egorov

Chemical Department Moscow State University Moscow, Russia

Miriam Eisenstein

Department of Structural Biology The Weizmann Institute of Science Rehovot, Israel

Sven-Olof Enfors

Department of Biochemistry and Biotechnology Royal Institute of Technology Stockholm, Sweden

Charlotte Erlanson-Albertsson

Department of Medical and Physiological Chemistry Lund University Lund, Sweden

Jean-Luc FauchPre

lnstitut de Recherches Servier Paris, France

R. Feldbrijgge

lnstitut fur Chemo- and Biosensorik Munster Roxel, Germany

Lutz Fischer

Institute of Biochemistry and 6 iotechnology Technical University of Braunschweig Braunschweig, Germany

Alain Friboulet

Laboratoire de Technologie Enzymatique Universitt5 de Technologie de Compiegne Compiegne, Cedex, France

Asher A. Friesem

Department of Electronics The Weizmann Institute of Science Rehovot, Israel

7;

Gaida

1. G. Gazaryan

Institute of Applied Microbiology University of Agriculture Vienna, Austria Chemical Department Moscow State University Moscow, Russia

xvii

List of Contributors Peter Gernejner

Institute of Chemistry Slovak Academy of Sciences B ratislava, Slovakia

S. Ghobadi

Department of Analytical Chemistry Chemical Center, Lund University Lund, Sweden

Michel Goldberg

Bio-Technology General Kiryat Weitzman Rehovot, Israel

Leon Goldstein

Department of Molecular Microbiology and Biotechnology Tel Aviv University Ramat Aviv, Israel

Lo Gorton

Department of Analytical Chemistry Chemical Center Lund University Lund, Sweden

Lawrence A. Greenstein

Bio-Technology General Kiryat Weitzman Rehovot, Israel

George Guilbault

Department of Chemistry University of New Orleans New Orleans, Louisiana

Kristina Gustavsson

Department of Biochemistry and B iotechnology Royal Institute of Technology Stockholm, Sweden

Reetta Haapala

Laboratory of Biotechnology and Food Engineering Helsinki University Espoo, Finland

Ulrika Harborn

Department of Pure and Applied Biochemistry Chemical Center Lund. Sweden

xviii

LIST OF CONTRIBUTOR!

Urs Heimgartner

Elchrom AG Horgen, Switzerland

Adam Heller

Department of Chemical Engineering University of Texas Austin, Texas

Z. Huan

Department of Analytical Chemistry Chemical Center, Lund University Lund, Sweden

Gote Johansson

Department of Biochemistry University of Lund Lund. Sweden

lsao Karube

Research Center for Advanced Science and Technology University of Tokyo Tokyo, japan

Volker Kasche

AB Biotechnologie I I Hamburg, Germany

loanis Katakis

Department of Chemical Engineering University of Texas Austin, Texas

Ephraim KatchaIski- Katzir

Department of Membrane Research and Biophysics The Weizmann Institute of Science Rehovot, Israel

H. Katinger

Institute of Applied Microbiology University of Agriculture Vienna, Austria

Dieter Kirstein

Max Delbruck Center for Molecular Medicine University of Potsdam lnstiutte of Biochemistry and Molecular Physiology Berlin, Germany

Branko Kozulic

Elchrom AG Horgen, Switzerland

xix

List of Contributors

A. Lante

Dipartimento d i B iotecnologie Agra r ie Universith di Padova Padova, Italy

Per-Olof Larsson

Department of Pure and Applied Biochemistry Chemical Center, Lund University Lund, Sweden

A. Levanon

Bio-technology General Kiryat Weizman Rehovot, Israel

G. Lilius

Department of Pure and Applied Biochemistry Chemical Center, Lund University Lund, Sweden

Malcolm D. Lilly

The Advanced Centre for Biochemical Engineering Department of Chemical and Biochemical Engineering University College London London, England

Christer Lindbladh

Department of Pure and Applied Biochemistry Chemical Center, Lund University Lund, Sweden

H. Lind6n

Department of Analytical Chemistry Chemical Center, Lund University Lund, Sweden

Pekka Linko

Laboratory of Biotechnology and Food Engineering Helsinki University of Technology Espoo, Finland

Susan Linko

Laboratory of Biotechnology and Food Engineering Helsinki University of Technology, ESPOO, Finland

xx

LIST OF CONTRIBUTORS

Peter Ljungcrantz

Department of Pure and Applied Biochemistry Chemical Center,-Lund University Lund, Sweden

Christopher R. Lowe

Institute of Biotechnology University of Cambridge Cambridge, England

Jerzy Maciozek

Department of Biological Sciences University of Illinois at Chicago Chicago, IIIinois

Craig R. Malloy

The University of Texas Southwestern Medical Center at Dallas Rogers Magnetic Resonance Center Dallas, Texas

Mats-Olfe Mdnsson

Department of Applied Biochemistry Chemical Center, Lund University Lund, Sweden

G. Marko-Varga

Department of Analytical Chemistry Chemical Center, Lund University Lund, Sweden

fvano A. Marques

Department of Biological Sciences University of Illinois at Chicago Chicago, I I Iinois

Tadashi Matsunaga

Department of Biotechnology Faculty of Technology Tokyo University of Agriculture and Technology Tokyo, Japan

Bo Mattiasson

Department of Biotechnology Chemical Center, Lund University Lund, Sweden

Michael Mecklenburg

Department of Pure and Applied Biochemistry Chemical Center, Lund University Lund, Sweden

xxi

List of Contributors

F. F. Moraes

Departamento Engenharia Quimica Universidad Estadual de Maringa Maringa, Parana, Brasil

Klaus Mosbach

Department of Pure and Applied Biochemistry Chemical Center, Lund University Lund, Sweden

Thomas Morton

SmithKline Beecham Pharmaceuticals King of Prussia, Pennsylvania

Yuji Murakami

Research Center for Advanced Science and Technology University of Tokyo 441 Komaba, Tokyo 153, Japan

David Myszka

SmithKline Beecham Pharmaceuticals King of Prussia, Pennsylvania

Ian A. Nicholls

Department of Pure and Applied Biochemistry Chemical Center, Lund University Lund, Sweden

Marina Otamiri

Department of Biotechnology Chemical Center, Lund University Lund, Sweden

G. Pasini

Dipartimento di Biotecnologie Agrarie UniversitA di Padova Padova, Italy

Mats Persson

Department of Pure and Applied Biochemistry Chemical Center, Lund University L und, Sweden

B. Persson

Department of Analytical Chemistry Chemical Center, Lund University Lund, Sweden

Dorothea Pfeiffer

Max Delbruck Center for Molecular Medicine Berlin, Germany

LIST OF CONTRIBUTORS

xxii

M. Rank

Department of Pure and Applied Biochemistry Chemical Center, Lund University Lund, Sweden

M. Reiter

Institute of Applied Microbiology University of Agriculture Vienna, Austria

Klaus P Rindt

Dragenwerk AG Lubeck, Germany

S. Sahni

Moltech Corporation Stony Brook, New York

lkuo Satoh

Department of Chemical Technology Kanagawa Institute of Technology Kanagawa-ken, Japan

Frieder W. Scheller

Max Delbruck Center for Molecular Medicine Berlin, Germany

C. Schmatz

Institute of Applied Microbiology University of Agriculture Vienna, Austria

Florian Schubert

Physikal isch-Technische Bundesanstalt Berlin, Germany

Alexander Schwarz

Department of Membrane Research and Biophysics Weizmann Institute of Science Rehovot, Israel

Isaac Shariv

Department of Electronics The Weizmann Institute of Science Rehovot, Israel

A. Dean Sherry

Department of Chemistry The University of Texas at Dallas Richardson, Texas

L. M. Simon

Department of Biochemistry Attila Jdzsef University Szeged, Hungary

xxiii

List of Contributors

T. Skotheim

Moltech Corporation Stony Brook, New York

M. Smolander

Biotechnical Laboratory Technical Research Centre ESPOO, Finland

Koji Sode

Department of Biotechnology Faculty of Technology Tokyo University of Agriculture and Technology Tokyo, Japan

Beka Solomon

Department of Molecular Microbiology and Biotechnology Tel Aviv University Ramat Aviv, Israel

P. Spettoli

Oipartimento di Biotecnologie Agrarie Universita di Padova Padova, Italy

Paul A. Srere

Department of Veterans Affairs Medical Center Dallas, Texas

Vladimir Stefuca

Institute of Chemistry Slovak Academy of Sciences Bratislava, Slovakia

W. Steinfellner

Institute of Applied Microbiology University of Agriculture Vienna, Austria

K. Strutzenberger

Institute of Applied Microbiology University of Agricu Iture Vienna, Austria

M. Suda

Seiko Instruments Inc. Chiba, Japan

Balasz Sumegi

Department of Biochemistry University Medical School Pecs, Hungary

xxiv

LIST OF CONTRIBUTORS

P.V. Sundaram

Centre for Protein Engineering and Biomedical Research The Voluntary Health Services Madras, India

8.Szajdni

Reanal Factory of Laboratory Chemicals Budapest, Hungary

Xiao-yi Tang

Shanghai Institute of Plant Physiology Academia Sinica Shanghai, China

0. Thomas

Laboratoire de Technologie Enzymatique Universite de Technologie de Compiegne Centre de Recherche de Royallieu, Genie Biologique Compiegne, Cedex, France

Folke oerneld

Department of Biochemistry Chemical Center, Lund University Lund, Sweden

F. Unterluggauer

Institute of Applied Microbiology University of Agriculture Vienna, Austria

llya A. Vakser

Department of Membrane Research and Biophysics The Weizmann Institute of Science Rehovot, Israel

Andres Veide

Department of Biochemistry and Biotechnology Royal Institute of Technology Stockholm, Sweden

M. A. VQayalakshmi

Laboratoire Interaction Moleculaire et de Technologie de Sbparation Universite de Technologie de Compikgne Centre de Recherche de Royallieu Genie Biologique Compiegne, Cedex, France

Mark Vreeke

Department of Chemical Engineering University of Texas Austin, Texas

xxv

List of Contributors

Xingwu Wang

Department of Biological Sciences University of Illinois at Chicago Chicago, Illinois

Howard H. Weetall

National Institute of Standards and Technology Biotechnology Division Gaithersburg, Maryland

Ernst Wehtje

Department of Biotechnology Chemical Center, Lund University Lund, Sweden

G. Rickey Welch

Department of Biological Sciences University of New Orleans New Orleans. Louisiana

Moshe M. Werber

Bio-Technology General (Israel), Ltd. Kiryat Weizmann Rehovot, Israel

Meir Wilchek

Department of Membrane Research and Biophysics Weizmann Institute of Science Rehovot, Israel

Ulla Wollenberger

Frauenhofer Institute for Microstructure Technique Berlin, Germany

Gunter Wulff

Institute of Organic Chemistry and Macromolecular Chemistry Heinrich-Heine University Dusseldorf, Germany

9.Xie

Department of Pure and Applied Biochemistry Royal Institute of Technology Stockholm, Sweden

Shaojun Yang

Department of Biochemistry and Biotechnology Royal Institute of Technology Stockholm, Sweden

xxvi

LIST OF CONTRIBUTORS

Ling Ye

Department of Chemical Engineering University of Texas Austin, Texas

Zhu Yi-Hong

Laboratory of Biotechnology and Food Engineering Helsinki University of Technology ESPOO, Finland

Kenji Yokoyama

Research Center for Advanced Science and Technology University of Tokyo Tokyo, Japan

N. Zach

Institute of Applied Microbiology University of Agriculture Vienna, Austria

A. Zamorani

Dipartimento di Biotecnologie Agrarie Universiti di Padova Padova, Italy

PREFACE I understand it to be quite unusual for a man to whom a Festschrift has been dedicated to contribute himself. However, as I listened during the symposium to the many superb lectures in various disciplines, areas in which we have worked over the years, a number ofreflections came to my mind which I would like to share with the reader. Maybe some lessons can be learned from “our past” which may be of guidance for others in their planning for the future, both positive and negative experiences. First, though, I would like to take the opportunity to thank all the contributors and guests, from Sweden and abroad, and to express my sincerest gratitude to the organizers of the symposium. These were very intense days for me. During conferences or symposia I normally slip away from some lectures simply to have a break. On this occasion this would have seemed very impolite of me (and to make things worse I was seated in the front row). However, during this symposium there was not even a thought in this direction, all the lecturers influenced me greatly and left me with a deep sense of indebtedness to them for taking the time from their many other pressing tasks to participate in this Symposium. Well, a year ahead of the symposium, I was puzzled to notice that my senior co-workers, Professors Bulow, Danielsson, Larsson and Mhsson, were holding frequent meetings in our small library. Each time I entered the room they turned completely quiet. At one time I was even afraid they were plotting against me, maybe because they felt neglected due to my heavy travelling commitments and wanted another head of the department. I then found out, however, that they were xxvii

xxviii

PREFACE

planning for this symposium, which they as organizers chose to call “Symposium on Biochemical Technology.” I feel they chose a worthwhile title and I will return to this aspect later. How did it all start? I spent most of my initial research life working with secondary natural products, in particular on their biosynthesis in fungi and lichens, the latter representing fascinating symbiotic organisms comprisingalga and fungus, of which more than 20,000 species are known today. Prior to our chancy approach using 14C-labeledC02 and other precursors, they were not considered amenable to scientific investigationsbecause of their extremely slow growth. This work resulted in an assistant professorship at the University of Lund in 1964. On the side, I spent many evenings in my father’s company in Malmo doing “moonshineresearch.” My father’s work on the use of various acrylates as polymer components in paints and in transparent coatings, made me think of trying to entrap lichen cells or enzymes in such gels. Would such entrapped biomolecules still be active? There was not much thought on how this could be used initially-it was simply fun and “play” research. Anyhow, it resulted eventually in the first publication on the use of immobilized, and more specifically entrapped, enzymes and cells for biochemical production (Mosbach and Mosbach, 1966). Slowly but surely interest grew in such immobilized preparations, and new applications were discovered. Along with these “trial and error” experiments, arising from a rather playful approach (perhaps a consequence of my music teacher’s desire to make me a concert pianist) and my later concentration on more specific applications,I have always been concerned by fundamental scientific questions. A particular interest of mine has been to try to understand the many aspects of metabolic events and cycles taking place in a living cell; this is an interest the foundation for which was laid down by my late teacher, professor Gosta Ehrensviird. Later on, when founding our institute I deliberately chose the name Pure and Applied Biochemistry,reflecting both of these aspects, a combination of mutual benefit scientifically, and useful not least from a funding point of view. The variety of topics the organizers have included in this symposium, enzyme technology, immobilized biocatalysts, enzymatic aspects of cellular metabolism, biosensors, separation and molecular recognition seems heterogeneous at first sight. However, there is a common “thread” or denominator for all the different aspects as they practically all concern binding or immobilization in some manner. Thus, having come to grips with immobilization technology per se, we started applying this know-how in many areas. 1will mention only briefly here some early efforts carried out mainly with the organizers of this meeting or former students. For instance, once we understood how to immobilize enzymes (immobilized biocatalysts), an obvious extension was to immobilize two or more acting in sequence in proximity to one another, thereby mimicking metabolic multi-step systems (Mosbach and Mattiasson, 1970). Later, as an extension of these studies,

Preface

xxix

and as a reflection of our interest in the topic of enzymatic aspects of cellular metabolism, the preparation of enzyme complexes by gene fusion was undertaken (Bulow et al., 1985). Living microorganisms were applied for steroid transformations followed by many other studies in the area of enzyme technology (Mosbach and Larsson, 1970). Subsequently plant cells were immobilized (Brodelius et al., 1979) as well as animal cells (Nilsson et al., 1983). Influenced by the strong tradition at Lund University in microcalorimetry pioneered by Professor I. Wadso, and having the techniqueof immobilization at hand, it was only obvious to combine these two technologies. This led to the development of the enzyme thermistor, a new biosensor, capable of measuring heat evolved from enzymic or metabolic reactions (Mosbach and Danielsson, 1974). During the most hectic period in which we “played” with immobilization of all sorts of molecules, the potential of such preparations for purification purposes became obvious. I must say, though, that these studies were initially met with skepticism. I particularly remember one Nobel laureate, with whom I worked for some time, telling me when I asked whether he believed this concept could be of any use, “Why complicate things by binding an inhibitor to a bead to purify an enzyme? Look, we get our enzyme in pure crystalline form simply by ammonium sulfate precipitation.” Perhaps in those days I had too much respect for authority, so I waited for a while, during which the area, later to become known as affinity chromatography, gained popularity and recognition. Thus, I decided to focus on more general aspects of this technique; for instance, the affinity ligand. With new enzymes which have to be isolated it is uncommon to have an inhibitor initially, rather, only the substrate is known (Brodelius and Mosbach, 1973).Another aspect studied was the use of general ligands such as coenzymes, as they can be applied potentially for many enzymes (Mosbach et al., 1971). Thus, our interest inseparation developed in the department. In parallel with these investigations we looked into various ways of achieving the ultimate goal in separation technology, that is being able to design tailor-made specific adsorbents.We tested and pursued several approaches over many years and with varying intensity. One has been to make specific imprints in plastics of the molecule later to be “fished out,” an area only recently to have achieved practical utility. In order to be able to design more powerful affinity ligands and better understand these interactions, molecular recognition became of growing importance to us. After this brief account of “our history,” I would like to comment on the title. The organizers have deliberately chosen as a general title for this symposium “Biochemical Technology.” I consider this an appropriate one, as it should not be as broad a term as biotechnology. For instance, it precludes engineeringaspects and large-scale operations, and correctly reflects the ongoing work at this department. An alternativetitle would have been “applied biochemistry” but this does not sound as “catchy.” In this context, I mention a meeting to which I was invited and which

xxx

PREFACE

was organized as early as 1981 in Japan on “towards the establishment of biomolecular technology.” This title, “biomolecular technology,” has also its merits (although it may appear less multidisciplinary than biochemical technology. For example, the area of biosensors would not fit into it as well). (Incidentally, the meeting dealt with two aspects, genetic engineering with invited lectures by Professors C. Weissmann, S . Numa, J. Schell and D.V. Goeddel and enzyme technology with Professor Fukui as chairman and apart from myself Dr. I. Chibata and Professor E. Katchalski-Katzir,both of whom have taken the time to attend this meeting.) Together with the organizers of this symposium I feel that we should continue having meetings on “biochemical technology,” which should be multidisciplinary in nature and concern the search for new concepts in this field. In reflecting on our studies, we should pose the question: Have any really useful results come from them over the years? The answer is definitely yes. We have received many patents, Swedish and international companies have adopted and exploited techniques and procedures in various areas. Our approach has been to find interested partners as soon as possible as we do not (and are not supposed to) have experience of optimization or scale-up technology. At times the search for partners has been very diffkult and time-consuming. In a few cases the formation of start-up companies has been chosen as an alternative. This too, however, has not always been easy. It is a source of constant frustration that Europe does not have a tradition of capital investment like the United States. In closing, let me express my thoughts on what direction(s) I would like the department to pursue in the future. I feel strongly that aspects of biomolecular recognition will increase in importance both for the understandingof biomolecules and for their increased application. Increasingly, we will make use of ways to manipulate biomolecules, for instance enzymes and polynucleotides, to change their properties. We may do this indirectly by means of genetic engineeringbut also directly, physically, using for instance atomic force microscopy or similar techniques. With all due respect for Mother Nature it should be possible to change even lifeforms. This may be done for example by creating new and more efficient biochemical pathways or multienzyme-coenzymecomplexes using fused enzymes. Not only the “normal” biochemical building blocks, amino acids, sugars, nucleotides etc. should be used; these may in part be replaced by “unnatural” molecules obtained from polymer or organic chemistry. Utilizing molecular diversity, new useful entities, for instance, drugs, may be created. We should try to overcome our conventional thinking, and not be guilty of a modern phlogiston complex. We should begin to judge compounds by looking at their functionality and shape and forget their biological or non-biological origin.

Preface

xxxi

Let me close by thanking again all of the participants, all my good friends who came, as well as my present and former co-workers and students alike for what they have done for me over the years. Klaus Mosbach University of Lund Lund, Sweden

REFERENCES Brodelius, P., & Mosbach, K. (1973). The utilization of immobilized substratdproduct in affinity chromatography. A model study using a-chymotrypsin. Acta Chem. Scand. 27.263&2638. Brodelius, P., Deus, B., Mosbach, K., & Zenk, M.H. (1979). Immobilized plant cells for the production and transformation of natural products. FEBS Lett. 103,93-97. Biilow, L., Ljungcrantz, P., & Mosbach, K. (1985). Preparation of a soluble bifunctional enzyme by gene fusion. Bio/Technology 3,821-823. Mosbach, K. & Mosbach, R. (1966). Entrapment of enzymes and microorganisms in synthetic crosslinked polymers and their application in column techniques. Acta Chem. Scand. 20.2807-2810. Mosbach, K. & Larsson, P.-0. (1970). Preparation and application of polymer-entrapped enzymes and microorganisms in microbial transformation processes with special reference to steroid 1 I -phydroxylation and A'-dehydrogenation. Biotechnol. Bioeng. XII, 19-27. Mosbach, K., & Mattiasson, B. (1970). Matrix-bound enzymes. Part 11. Studies on a matrix-bound two-enzyme-system. Acta Chem. Scand. 24,2093-2 100. Mosbach, K., Guilford, H., Ohlsson, R., & Scott, M. (1971). General ligands and (co)substrate elusion in affinity chromatography. Biochem. J. 127, 12-13. Mosbach, K., & Danielsson, B. (1974). An enzyme thermistor. Biochim. Biophys. Acta 364, 14&145. Nilsson, K., Scheirer, W., Merten, O.W., Ostberg, L., Liehl, E., Katinger, H.W.D., & Mosbach, K. (1983). Entrapment ofanimal cells for production ofmonoclonal antibodiesand other biomolecules. Nature 302,629-630.

KLAUS MOSBACH

This Page Intentionally Left Blank

INTRODUCTION In December 1992, the Department of Pure and Applied Biochemistry at the Chemical Center in Lund, Sweden, organized an international meeting, the Mosbach Symposium on Biochemical Technology, to celebrate the 60th birthday of professor Klaus Mosbach, one of the founders of modern biotechnology. The history of Pure and Applied Biochemistry had its start in 1970, a couple of years after the foundation of the Chemical Center. Klaus Mosbach has been its professor and head of Pure and Applied Biochemistry since its start. During the 1980s he also maintained a professorship at the ETH in Ziirich, Switzerland.Professor Mosbach has, so far, supervised a remarkable number of Ph.D. studentsapproximately 60! Professor Mosbach is internationally well-known and he has world-leading position within the field of immobilization of bioactive substancesand cells as well as affinity chromatography. He has published about 300 original articles (not including abstracts or conference reports), about 110 review articles and he holds about 35 patents. In 1990, Professor Mosbach was awarded the gold medal by the Royal Swedish Academy of Engineering Sciences for his contributionsto biotechnology, especially on the immobilization of bioactive substances. The research activities of the Department of Pure and Applied Biochemistry cover a broad area, such as affinity and separation techniques, bioprocess control, biosensors, development of new carriers and new immobilization procedures for small molecules as well as proteins and cells, including animal and plant cells, gene technology, processes based on immobilized biocatalysts, and construction of xxxiii

INTRODUCTION

xxxiv

organic polymers with enzyme-like properties. The hallmark of the department is its diversified research that generates considerable synergistic effects that are manifested by many new techniques and concepts emanating from the laboratory during the last 20 years. Several of these are marketed by various biotechnology companies. At this meeting we therefore arranged for some of the world’s leading experts in biochemistry and biotechnology to give lectures. The topics covered comprise enzyme technology, immobilization of enzymes and cells, abzymes, metabolic engineering, biosensors, and molecular recognition. No doubt much of Professor Mosbach’s success depends on his astounding profusion of ideas and his ambition to test the “impossible.” This wealth of ideas in combination with a cheerful temperament is the foundation of creativity. However, the development of half-crazy ideas into functional procedures and industrial processes can be excruciating. Everyone of his graduate students has experienced this difficult and often frustrating route to learn independence of thought. Nevertheless, their satisfaction when after years of exertion things are operating properly can be indescribable. Without such efforts the development of society would cease and life would be no fun . . . The official gift from the symposium committee and the participants is this ‘‘Festschrift”which covers several important fields of research within the area of biochemical technology. We have made a very unusual approach and have let the “hero of the occasion” present the history of his research. On behalf of all participants it is an honor to congratulate Mosbach to an extremely successful scientific career. We all hope for many more years of original ideas, laughter, enthusiasm, and thought provoking comments in the laboratory, in Mosbach’s home, at conferences and of course at night clubs.

ACKNOWLEDGMENTS The organizing committee gratefully acknowledgesthe financial support from the following companies and organizations: Wenner-Gren Center Foundation for Scientific Research The Biotechnology Research Foundation Perstorp Biotec Pierce Chemical Company Hyclone Laboratories Pernovo

Novo Nordisk A/S Norsk Hydro as. Bioinvent International AB Ferring AB Pharmacia Lund University

xxxv

Introduction

ORGANIZING COMMITTEE Staffan Birnbaum Elisabet Bulow (conference secretary) Leif Bulow (chairman) Bengt Danielsson Per-Olof Larsson Mats-Olle Mhnsson Leif Bulow and Ben@ Danielsson

This Page Intentionally Left Blank

PART IV

B IOSENSORS

This Page Intentionally Left Blank

B IOSENSORS: AN INTRODUCTION

Bengt Danielsson

Biosensors have always played an important role in research in the Department of Pure and Applied Biochemistry. The enzyme thermistor and several other contributions of fundamental importance to the biosensor field were developed here. Biosensors are very suitable to demonstrate and quantitate interactions related to various biochemical phenomena such as molecular recognition or biocatalysis. Unfortunately,we have not yet seen the biosensor which lives up to all expectations. This is at least in part due to advancements in competing technologies that have been characterized by more successful instrumental developments such as high performance liquid chromatographyand high performance capillary electrophoresis. Nevertheless, many workers in this area including the session chairman, Professor G. Guilbault, hold that the biosensor market will now take off to become a multibillion dollar market before the turn of the century. Guilbault rounded off the session with a summary of the current status of biosensors worldwide. Referringto a recent survey ofbiosensor research in Europe, comparisons were made between the situation in the USA and Japan where the achievements to date have been much more significant. The situation is, however, Advances in Molecular and Cell Biology Volume 15B, pages 34e352. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

349

350

BENGT DANIELSSON

rapidly changing, as exemplified by the commercial exploitation of the Exactech pen-sized glucose meter and the BIAcore instrument from the PharmaciaBiosensor

co. Professor F. Scheller presented an overview of modem biosensor technology including aspects of molecular recognition, transducers, immobilization, signal transfer, organic phase operation, and applications in various fields. Enzymes are the dominating type of biocomponents used for molecular recognition, whereas intact cells are suitable for quantifying biological effects. Antibodies are also used in various immunosensing situations; for example, the use of different antibody fragments including single chain antibodies or the binding domain alone, is becoming increasingly popular. Miniaturization is a common feature of modem transducer design, from field-effect transistors to thin-film electrodes. Miniaturization poses special problems concerning the immobilized biolayer. Microelectronic production technology is employed in the fabrication of such sensors. Efficient transfer of the chemical signal to the transducer is especially important in connection with amperometric biosensors. A great deal of effort has been devoted to designs that eliminate interference from disturbing electroactive substances. In order to improve electron transfer various mediators or mediator-free electrodes have been employed as well as alternative enzymes with better electron transfer properties, such as PQQ-containing dehydrogenases. Biosensors operational in organic media offer special qualitiesthat may be of special interest in environmental analysis. Finally, typical biosensor applications were presented together with a display of commercial biosensor devices. Micromachining of biosensors was the focus of the paper presented by Professor I. Karube. A 20 nL electrochemical flow cell with an integrated enzyme column etched into the silicon substrate and an integrated enzyme reactor with a chemiluminescence detector were described. The biosensor research in Japan has always been directed towards very practical applications, and one large project that may seem somewhat amusing but has serious and important implications is the mounting of (bio)sensors in toilets for automatic health control. The age distribution of the Japanese population is shifting towards ever higher average ages, which makes the costs for health care skyrocket. Areliable system for automatic health control could help keep costs under control. Sensors for some key metabolites and proteins connected to a computer at the health center could keep track of the health status of an aging population (or anybody who wants to be connected)and bring abnormal values to immediate attention. An advantage of such a system would be that old people could be treated in their homes longer and transferred to hospitals later and still feel safer than without the automatic health control. Professor A. Heller is well known for his work on “wiring” redox enzymes to electrodes at the molecular level resulting in amperometric probes with higher sensitivities.For this purpose electron relaying polymers complexed with osmium compounds have been designed to efficiently lead electrons from the enzyme to the electrode. Several groups of enzymes have been connected through such electron-

Biosensors: An Introduction

351

conducting hydrogels including flavoenzymes such as glucose oxidase, quinoprotein (PQQ) enzymes, and heme peroxidases. High current densities were realized by immobilizing the enzymes in hydrogels with a polymer skeleton that permitted efficient electron diffusion at the same time that water soluble reactants could rapidly difise through the gel. The peroxidase electrodes were also found useful for sensitive detection of NAD(P)H using N-methyl phenazonium salts as mediators that could be oxidized with oxygen to produce hydrogen peroxide. U. Wollenberger presented a paper on recycling sensors based on kinases in various analytical arrangements with electrochemical, optical, and thermometric detection. The biosensor group in Berlin has developed considerable expertise in exploitingrecycling systems to increase the analytical sensitivity.Up to 48,000-fold signal enhancement has been noted for an enzyme electrode with a lactate oxidasellactate dehydrogenase membrane. An amplification factor of 5000 was obtained using the same couple in an enzyme thermistor reactor. P. Gemeiner and co-workers at the Slovak Academy of Sciences in Bratislava have developed methods for the characterization of the micro-kineticproperties of immobilized biocatalysts using enzyme flow calorimetry. The immobilized biocatalyst is placed in an enzyme thermistor column and the catalytic activity can be directly derived from the heat signal. Coupling of the mathematical description of the reaction-diffusion phenomena in the reactor with the heat balance provides kinetic data about the biocatalyst. The method has been tested on various enzymes, including invertase, D-amino acid oxidase and penicillin amidase. In the last two cases, immobilized cells have been used as well. The main advantages of the technique are the simplicity, general applicability, and reagentless operation (no chromogenic substrates, etc. are required). L. Gorton and co-workers provide an exhaustive account of the state of amperometric electrodesmodified with redox enzymes. Sensors for L-lactate, aldoses, alcohol, D-amino acid, and L-glutamate were produced by immobilizingthe appropriate enzymes in carbon paste electrodes operating at about zero mV versus SCE. In the two first cases, dehydrogenase enzymes were used while in the others oxidases coimmobilized with peroxidases were used. R. Feldbrtigge, K.P. Rindt, and A. Borchert describe an enzyme-based diffusion badge for the detection of formaldehyde. The badge is intended for the detection of formaldehyde in the atmosphere in working places and at home. It uses formaldehyde immobilized on sintered glass with diaphorase that transfers the NADH formed and a tetrazolium salt into colored formazan. The device is simple to operate and capable of detecting 100 ppb of formaldehyde within 15 minutes. It can be stored at least 18 months. I. Satoh describes biosensing of heavy metal ions based on their specific interactions with apoenzymes. Metalloenzymes are made inactive by removing the metal ion coordinated in the active center with use of strong chelating agents. The apoenzyme can be reactivated by exposure to samples containing the proper metal(s). The degree of activation depends on the metal ion concentration. Ther-

352

BENGT DANIELSSON

mometric as well as electrochemical and photometric detection has been used. This technique is highly selective and specific. The enzyme reactor operates under mild conditions and is reusable. M. Mecklenburg who investigates developments of the PCR technology has designed high-annealing-temperature(HAT) primers that drasticallyreduce cycling times and nonspecific amplification products. Other modifications lead to more efficient use of the polymerase enzyme and reduced primer oligomer formation. HAT primers were used to develop a highly specific, low-copy-number PCR assay for human cytomegalovirus. M. Rank and B. Danielsson have adopted and adapted a split-flow thermal biosensor system for on-line monitoring of industrial fermentations. The system has been thoroughly tested in real industrial environments at pilot-plant and production-scale fermentations. Excellent results were obtained for penicillin V, ethanol, glucose, and lactate. B. Xie has developed different designs of miniaturized thermal biosensors with surprisingly good performance, including devices made by micromachining of silicon or quartz. The paper submitted by U. Hedberg and B. Xie focusses on the use of such sensors for the measurement of glucose in whole blood. The present state of the technology permits fabrication of portable instruments for home monitoring of glucose.

OVERVIEW OF BIOSENSOR TECHNOLOGY

Frieder W. Scheller, Ulla Wollenberger, Dorothea Pfeiffer, and Florian Schubert

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

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353 INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354 BIOMOLECULESUSEDFORMOLECULARRECOGNITION . . . . . . 354 OPTIMIZATION OF TRANSDUCERSAND OF IMMOBILIZATION . . . 355 TRANSFER OF THE CHEMICAL. SIGNAL TO THE TRANSDUCER . . . 356 ORGANIC PHASE BIOSENSORS . . . . . . . . . . . . . . . . . . . . . . . 358 APPLICATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361

ABSTRACT Biosensors for about 140 different analytes have been described in the literature. Among them are low molecular weight substances (metabolites, drugs, nutrients, gases, metal ions, coenzymes, enzyme activators, and vitamins) as well as macromolecules (enzymes, lectins, nucleic acids, polymeric carbohydrates like starch, and cellulose), viruses, and microorganisms. Advances in Molecular and Cell Biology Volume 15B, pages 353-363. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

353

SCHELLER, WOLLENBERCER, PFEIFFER, and SCHUBERT

354

1. INTRODUCTION Biosensors are in the forefront of current bioanalytical chemistry. Even though a definition by the IUPAC has not yet been established, the spatial unity of biomolecules with a signal transducer has been accepted to be the main feature of biosensors (Lowe, 1985). According to L. Clark, biosensorics is a combination of science and art. This statement is obviously true for the initial phase of research, but mass production for widespread application requires developments in technological areas. Two major branches-biotechnology and microsystem technology will contribute to future progress in biosensorics.

II. BIOMOLECULES USED FOR MOLECULAR RECOGNITION In order to cover the whole spectrum of analytes within an extremely large concentration range, different biomolecules have been used for molecular recognition. These include: (i) Enzymes that dominate in the determination of low molecular weight analytesin the micro and millimolar range. (ii) Intact cells that are suited to quantify biological effects, e.g., nutritivity, taste, odor, and mutagenicity. And (iii) Antibodies for the analysis of both high and low molecular weight compounds down to subnanomolar concentrations (see Table I).

Table 1. Fields of Applications Medical Diagnostics [g c

0

-2

Bioprocess Control

-1 2

Meiabolites

Arninoacids Inorg. Salts

Drugs Salicylate

Glucose Urea Lactate Creatinine

Antibiotics

Inorg. Cornp.

Theophylline

aP GOT Cortisone

Antibodies Enzymes

-8 -10

Enzymes/ Hormones

Organic Acids Carbohydrates Degradable Org. Cornp.

4 -6

Environmental Control

Digoxin

Toxins Hormones Vitamins Toxins

Insulin T4

HCG

Overview of Biosensor Technology

355

Considerable efforts in industry and academia are being made to discover novel enzymes.Both oxidases (e.g., for glutamate,NADH, phenylalanine,acylcoenzyme A, bilirubin, glycerol, and theophylline)and PQQ-containing dehydrogenases (e.g., for oligosaccharides, aldoses, fructose, methylamine, and aliphatic alcohols) have been used successfully for biosensor development. On the other hand, the arsenal of catalytic antibodies (abzymes) occupies only marginal interest in biosensors. Significant alternatives to molecular recognition by biomolecules are offered by “biomimetic” (binding) molecules and “molecular imprinted” polymers. In immunosensor development monoclonal antibodies allow screening for high affinity and appropriate kinetic parameters. Different fragments (e.g., the Fab part) and the sole binding domain (“mini-antibodies”) or natural binding proteins (e.g., for the low density lipoprotein) are promising candidates for reducing the nonspecific binding effects.

111. OPTIMIZATION OF TRANSDUCERS AND OF IMMOBILIZATION Chemical sensors, that is, potentiometric, amperometric, and impedimetric electrodes, various optical detectorsusing colored indicators as well as physical sensors, that is, piezoelectric crystals, thermistors, and reagent-free optical sensors, have been combined with appropriate biocomponents. Potentiometric and amperometric enzyme electrodes are at the leading edge of biosensors when considering the body of scientific publications and commercially available devices. Only few conductometric enzyme electrodes have been described, but the relevance of this sensor type may increase because of the relative ease of their production (Watson et al., 1988). The goals of miniaturizing biosensors are the reduction of the required sample volume, the multicomponent analysis of complex chemical substances by multiple microsensors, and cost reduction by mass production. Three basic types of microsensors have been used: l . pH- or pF-sensitive field-effect transistors (ISFETs). 2 . gas-sensitive metal-oxide-semiconductor (MOS) capacitors. And 3. thin-film electrodes. Besides the common antimony oxide electrode,platinum oxide and iridium oxide probes have been coupled to immobilized enzymesto give miniaturized biosensors. The light-addressable biosensor provides a measurement of the photocurrent generated by illumination of the back side by an array of light-emitting diodes. The signal reflects the local pH or redox potential at the surface. In this way changes of the surface potential by enzyme reactions, microorganisms, or the action of receptors can be measured with high sensitivity(Owicki and Parce, 1990).More recently, thermistors, integrated optical sensors, and surface acoustic devices have also been included in the development of microbiosensors (Karube et al., 1990). These sensors are fabricated by microelectronic productiontechnology, such as thick- and thin-film deposition, photolithographic reduction, and chemical and plasma etch-

356

SCHELLER, WOLLENBERGER, PFEIFFER, and SCHUBERT

ing, which permit well-delineated patterning of metallic, insulating, and semiconducting surface layers. In this way arrays of identical or different sensors can be produced, thus enhancing the reliability, repeatability, and versatility of the sensor. Most problems arise from the manufacturing technique for the formation of the immobilized biolayer. Layers of polymers, for example, polypyrrole or polyaniline, are deposited on conducting areas by electropolymerization. The biomolecules are either entrapped in the polymer matrix during the layer formation (Foulds and Lowe, 1988) or coupled to the layer via typical chemical reactions. Structuring of uniformly deposited biocomponent-containing layers is also possible when photodeactivation of passive regions is performed. For this purpose, illumination with UV light is applied (Van der Schoot and Bergveld, 1987).Alternatively, enzymes are deposited only at the sensitive region by using an enzyme solution in a negative photoresist. Using glow discharge in combination with the mask technique, plasma polymers can be deposited on the active sensor region. In a second procedure the enzymes or antibodies are covalently bound via bifunctional reagents.

IV. TRANSFER OF THE CHEMICAL SIGNAL TO THE TRANSDUCER Enzyme electrodes based on oxidases combined with amperometric H,O, measurement are the most common among biosensors. However, an electroactive substance being converted at lower potential contributes to the total current. To eliminate these interferences by co-oxidizable sample constituents, the electrode potential is kept as low as possible. Therefore,one chooses a reaction partner which is converted at the lowest possible potential. The natural electron acceptors of many oxidoreductases, for example methanol dehydrogenase, alcohol dehydrogenase, cytochrome b,, and the oxidases of glucose, lactate, pyruvate, glycolate, sarcosine, and galactose can be replaced by redox-active dyes or other reversible electron mediators. With these mediators, an electrode potential around +200 mV can be applied; this decreases interferences by ascorbic acid and enables one to couple such enzymes with electrodes in oxygen-free solution. For this reason chemically modified electrodes where the mediator is integrated with the amperometric electrode have been combined with immobilized, mediator-dependent enzymes. In the simplest approach the solid mediator was introduced in crevices of the carbon electrode body, and a dialysis membrane prevented it from leaching out. In the next stage the mediator was admixed to the carbon paste ofthe electrode body and finally all reagents-enzyme, mediator, and cosubstrate-were integrated into the top of the electrode (Figure 1). Similar to mediators, enzymes may be covalently bound to the electrode surface, thus giving chemically modified enzyme electrodes.Adsorption of redox polymers containing benzoquinone or heavy metal ion complexes at carbon electrodes resulted in the catalysis of the electron transfer by wiring the enzyme molecules to

Overview of Biosensor Technology

357

I k d n nnd Sendo ( 1

Kulys ond Svirnlickar (198

G u s .Hill I Higgin

Racinc and M n d f (I97

Torslensson, Johansson I Morboch el Drgoni ond lleller ( 1 9 8 7 )

Figure 1. Types of mediated enzyme electrodes.

the electrode. When enzymes and mediators are coimmobilized at the surface or within the electrode, addition of auxiliary substances during the measuring process can be avoided, thus a reagentless measuring regime becomes feasible. An alternative to the application of mediators is the direct electron transfer between the prosthetic group ofthe enzyme and the redox electrode.Heterogeneous electron transfer reactions have been realized with more than 30 different proteins, mainly electron transferases, but also substrate-convertingoxidoreductases.At bare metal electrodes irreversible adsorption accompanied with denaturation prevents a fast electron transfer to the protein molecules. Adsorption of modifiers that promote an appropriate orientation of the protein resulted in a facilitated direct electron transfer with different redox enzymes, for example, cytochromes and ferrodoxins. The mediator-free electron transfer within bulk modified carbon paste electrodes has been used for sensor application. Peroxidase and the PQQ-containing fructose dehydrogenase have been applied in reagentless sensor arrangements for hydroperoxides and fructose, respectively (Ikeda et al., 1991; Gorton et al., 1992). Dual sensor systems have been developed both with amperometric and potentiometric electrodes. These dual sensor systems use one enzyme loaded sensor and one blank sensor as a reference in order to eliminate erroneous readings because of differences in the background current and sample pH. Problems arise from the independent fluctuations of the sensor's characteristics. Changes in buffer capacity ofthe sample cannot be compensated for by this arrangement. ThepHstatic enzyme sensor (van der Schoot and Bergveld, 1987/88) circumvents this disturbance: The Hf or OH- ions generated in the enzyme reaction are consumed in the reaction of water electrolysis. The charge necessary for maintaining a constant pH represents

358

SCHELLER, WOLLENBERGER, PFEIFFER, and SCHUBERT

the amount of ions formed and is independent of the buffer capacity. The device contains in addition to the enzyme covered pH-FET, a pair of amperometric electrodes for 0, or H, generation which are controlled by an on-chip pH static coulometer system. Coulometric generation of the cosubstrate oxygen is the basis of the ovgen-stabilized glucose electrode. The oxygen consumed in the enzyme reaction is regenerated by water electrolysis at an auxiliary electrode. The consumed charge reflects the amount of the converted substrate and this principle allows measurement even in solutions of changing oxygen content (Cleland and Enfors, 1984).

V. ORGANIC PHASE BIOSENSORS The potential advantages of using enzymes in organic as opposed to aqueous solutions include increased solubility of many organic substrates, reversion of hydrolytic reactions, relative ease of enzyme immobilization based on their insolubility in organic solvents, elimination of microbial contamination,and reduction of side reactions. The employment of immobilized enzymes directly in the organic phase for analytical purposes began with a study of a system consisting of horseradish peroxidase (HRP) and cholesterol oxidase (COD) for cholesterol analysis in toluene (Kazandjian et al., 1986).The enzymes were adsorbed onto glass beads and detection was carried out spectrophotometrically. Danielsson and Flygare (1 989) found that the sensitivity of a HRP thermistor operating in organic solvents increased compared with operation in water because of the lower heat capacity of organic solvents. A two-channel enzyme thermistor configuration has been em-

Table 2. Organic Solvent Biosensors Sensor-type

Transducer

Enzyme sensor

electrode

thermistor

Microbial sensor

electrode

Immunosensor

fluorescence detector

Notes: AAO - Aminoacid Oxidase

AOD - Alcohol Oxidase

Receptor

chymotrypsin tyrosinase COD HRP AOD HRP AAOlchymohypsin Alkaligensis spec. Corynebacter MB 1 Pseudomonas spec. polyclonal AB

Analyte

peptides phenols cholesterol peroxides CI-C4alcohols enantiomeric aminoacid ester xenobiotics. PCB

atrazine

Table 3. Enzyme-Electrode Based Analyzers Sample Frequency Company

Yellow Springs Instruments, USA

Model

Analyte

Linear Range (mM)

23 A 23 L 27 2000

glucose lactate ethanol lactose galactose sucrose

145 0.1-15

0-l)

40

Serial

cv PA)

Stabilig

<2

300 samples

M O

40 20

C-55

20

2

0-27

M

8C-90 30 3

4-5

UA-300A

glucose a-amylase uric acid

Auto-STAT

glucose

140

6C-120

1

Stat Analyzer S 80

glucose

C-50

120

2

STAT-Profile 5

glucose lactate

1-25 C-15

3%38 3%38

4.1

Satellite G

glucose

2-33

1s-17

ABD Amperometric Biosensor Detector

glucose urea uric acid lactate

0.003-3.0

Inst. Biochem., Lithuania

EXAN-G

glucose

2.530.0

20

Omron Tateisi, Japan Seres, France

HER- 100 Enzymat

lactate glucose

w.3 0.%22

60

Fuji Electric, Japan

Daiichi, Japan

GLUCO 20

1.7

> 500 samples

GA-112

Analytical Instruments, Japan NOVA Biomedical. USA

2

7d 7d

Analysator MediSense, England Universal Sensors. USA

disposables 500 samples SO0 samples 500 samples 500 samples

0.1-10.0

O.OOW.6 0.01-2.5

> IOd (continued)

Table 3. (Continued) Company

Model

Sample Frequency

Serial

Analyie

Linear Range (mM)

Wi)

cv PA)

Stability

choline

1.0-29 0.1-2 0-1.5

60 60 3s

<2

7 200 samples

2.5-25

35 90 120-130 120 80 20 20

<2 <2 < 1.5 <2 <2 5%

(15 d) samples > 2000 samples 10 d 14 d IOd 14 d 30 d

L-1ysine

SGI Setric Genie Industriel, France

Tacussel, France Priifgefite-Werk Medingen, FRGEppendorf, FRG

EKF Magdeburg, FRG

DOSIVIT Nantes, France

Microzym-L

Gluco-processeur ESAT 6660/61 EBIO Biosen. 6020 G Biosen. 5020 L

lactate glucose glucose glucose lactate uric acid glucose lactate glucose lactate disaccharides

0.055 0.MO

1-30 0.1-1.2 0.S25 1-25

5%

7 400

disposables

Overview of Biosensor Technology

361

ployed for the enantioselectiveanalysis of D- and L-amino acid esters in mixtures containing up to 40% dimethylformamide (Hundeck et al., 1992). The first organic phase enzyme electrode was based on polyphenol oxidase for the measurement of phenolic compounds in chloroform (Hall et al., 1988). The sensor could be employed for the detection of very small amounts of phenolic contamination of waste water by using the extraction of the analyte from a large aqueous volume into a small organic volume. A potentiometric chymotrypsin sensor has been devised for monitoring ester synthesis in organic media (Miyabayashi et al., 1989). It was found that the enzyme-associated hydration was the most important parameter affectingthe response of the sensor.Applicability in the organic phase has also been demonstrated for microbial and immunosensors (Table 2).

VI. APPLICATION The most relevant fields of practical application of enzyme electrodes are medical diagnostics, process control, food analysis, and environmental monitoring (Table 1). To meet the high demand for glucose determination in the blood of diabetic patients, different types of analyzers have been developed. In total, analyzers for about 12 different analytes have been made commercial with an increasing importance of nonmedical applications (Table 3). Compared with conventional enzymatic analysis, the main advantages of such analyzers are the extremely low enzyme demand (a few milliunits per sample), the simplicity to operate them, the high analysis speed, and the high analytical quality. In the medical field, a trend towards hand-held devices based on either reusable or disposable glucose and lactate enzyme electrodes for home control of diabetics and for on-site monitoring of surgery and exercise is evident (Table 4).

Table 4. Pocket Devices Based on Enzyme Electrodes

Company

Medi Sense, England Eli Lilly/ELCO, USA Med Test Systems, USA EKF Industrie Elektronik, FRG I-STAT Corp.

Model

Analyte

Sample Linear Frequency Serial CV Range (mM) (h-‘) PA)

Stability

ExacTech

glucose

1.1-33.3

2C30

3.3-8.1

disposables

Direct 30/30

glucose

2.5-28

30

2-8

30 d 500 samples disposables

<5

10 d

medisensor” glucose 200 1 urea BIOSEN 2000 lactate i-STAT

glucose urea

60

1-15

> 30 40

disposables

362

SCHELLER, WOLLENBERGER, PFEIFFER, and SCHUBERT

In process control and the food industry, enzyme electrodesare employed mainly for the determination of carbohydrates, ascorbic acids, amino acids, and penicillin. However, the application of sensors in situ in bioreactors is associated with significant difficulties. Environmental control will benefit from enzyme electrode based measurements of toxic compounds such as heavy metals, fumes from organic wastes, and pesticides in water, air, and soil.

REFERENCES a s s , A.E.G., Davis, G., Francis, G.D., Hill, H.A.O., Aston, W.J., Higgins, I.J., Plotkin. E.V., Scott, L.D.L., & Turner, A.P.F. (1984). Ferrocene-mediated enzyme electrode for amperometric determination of glucose. Anal. Chem. 46,667-671. Cenas, N.K., Pocius. A.K., & Kulys, J.J. (1983). Biosensors based on redox polymers. Bioelectrochern. Bioenerg. lI,61-73. Cleland, N. & Enfors, S O . (1984). The oxygen stabilized enzyme electrode. Anal. Chem. 56, 188G1886. Danielsson, B. & Flygare, L. (1989). Biothemal analysis performed in organic solvents. Anal. Lett. 22, 1417-1428. Degani, Y.& Heller, A. (1987). Direct electrical communication between chemically modified enzymes and metal electrodes. J. Phys. Chem. 91, 12851289. Durliat, H. & Comtat, M. (1984). Amperometric enzyme electrode for determination of glucose based on thin-layer specnoelectrochemistry of glucose oxidase. Anal. Chem. 56, 148-1 52. Foulds, N.C. & Lowe, C.R. (1988). Immobilization of glucose oxidase in ferrocene-modified pyrrole polymers. Anal. Chem. 60,247S2478. Gorton, L., Jonsson-Pettersson, G., Csoregi, E.. Johansson, K., Dominguez. E., & Marko-Varga, G. ( 1992).Amperometric biosensors based on an apparent direct electron transfer between electrodes and immobilized peroxidases. Analyst 117, 12351241. Hall, G.F., Best, D.J.. &Turner. A.P.F. (1988). Amperometric enzyme electrode for the determination of phenols in chloroform. Enzyme Microbiol. Technol. 10,542-546. Hundeck, H.G., Scheper, T., Lubbert, A., Schmidt, J., & Schubert, F. (1992). Application of a 4-channel enzyme thermistor in biotechnology. In: Biosensors: Fundamentals, Technologies and Applications (Scheller, F. & Schmidt. R.D., Eds.), GBF Monogr. 17, VCH, Weinheim, New York. Ikeda, T., Katasho, I., Kame, M., & Senda, M. (1984). Glucose oxidase-immobilized benzoquinonemixed carbon paste electrode. Agric. Biol. Chem. 48. 196S1975. Ikeda, T., Matsushita. F.. & Senda, M. (1991). Amperometric fructose sensor based on direct bioelectro-catalysis. Biosensors Bioelectronics 6,29%304. Karube, I., Sode, K.. & Tamiya, E. (1990). Microbiosensors. J. Biotechnol. 15,267-282. Kazandjian, R.Z., Dordick, J.S., & Klibanov, A.M. (1986). Enzymatic analyses in organic solvents. Biotechnol. Bioeng. 28,417421. Kulys, J.J. & Svirmickas, G.-J.S. (1980). Reagentless lactate sensor based on cytochrome b2. Anal. Chim. Acta 117, 115120. Lowe, C.R. (1985). An introduction to the concepts and technology of biosensors. Biosensors 1, S 1 6 . Miyabayashi, A., Reslow, M., Adlercreutz, P., & Mattiasson, B. (1989). A potentiometric enzyme electrode for monitoring in organic solvents. Anal. Chim. Acta 219,27-36. Owicki, J.C. & Parce, J.W. (1990). Bioassays with a microphysiometer. Nature 344,271-272. Racine, P. & Mindt, W. (1971). Enzyme Electrode, Swiss Patent 13,211/71, US. 3,838,033. Scheller, F., Stmad, G.,Neumann, B., Kiihn, M., & Ostrowski, W. (1979). Polarographic reduction of the prosthetics groups in flavoproteins. Bioelectrochem. Bioenerg. 6. 117-122.

Overview of Biosensor Technology

363

Torstensson, A,, Johansson, G., Mansson, M.-0.. Larsson, P.-O., & Mosbach, K. (1980). Electrochemical regeneration of NAD covalently bound to liver alcohol dehydrogenase. Anal. Lett. 13, 837-850. van der Schoot, B.H. & Bergveld, P. (1987/88). ISFET based sensors. Biosensors 3, 161-186. Watson, L.D., Maynard, P., Cullen,D.C., Sethi, R.S., Brettle, J., & Lowe, C.R. (1988). Amicroelectronic conductimetric biosensor. Biosensors 3, 101-1 16. Wingard, L.B., Jr. (1984). Cofactor modified electrodes. Trends Anal. Chem. 3,235-238.

This Page Intentionally Left Blank

CURRENT STATE OF BIOSENSORS

George G. Guilbault

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. NEW AREAS OF BIOSENSORS . . . . . . . . . . . . . . . . . . . . . . . . 111. STATUS OF BIOSENSOR SALES AND RESEARCH . . . . . . . . . .

.

365 366 366 . 370

ABSTRACT Biosensors, first described in the early 1960s,have advanced in 30 years to an age of sophistication. Electrochemical biosensors, the first of such sensors, have become with improved methods of electron transfer, better immobilization procedures, and miniaturization a 500 million dollar business. New immuno-electrobiosensors and noninvasive sensing methods have brought the technology to a new frontier. In addition, other useful biosensors based on highly sensitivepiezoelectriccrystal-based biosensors and fiber optrodes have gained importance.

Advances in Molecular and Cell Biology Volume 15B, pages 365-373. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

365

366

GEORGE G . GUILBAULT

Table 1. History of Biosensor Development Enzyme Electrodes

Amperometric Biosensor Potentiometric Biosensor

- Clark, 1962 - Guilbault, 1969

Immobilized Enzyme Systems in Flow Analysis

- Hornby, Miles Catalinks

Enzyme Coils

- Horvath, Technicon Enzyme Thermistor Probes Mass Biosensors Fiber Optic Sensors

- Cambiaghi. Carlo Erba - FIA - Ruzicka, Hansen - Danielsson - Shons, 1971 - Guilbault, 1986 - 1985

1. INTRODUCTION Biosensors, consisting of a biologically active compound (an enzyme, antibody, antigen, whole cell suspension, or receptor) coupled to a transducer (which converts the biological signal into an electronic signal) were first described in the early 1960s.Table 1 gives a history of biosensor development. The first transducer used was an amperometric electrode, and the biological compound was soluble glucose oxidase in the construction of a glucose biosensor (Clark and Lyons, 1962).In 1965 Guilbault et al. described the first biosensor, a cholinesterase based nerve agent detector using an amperometric readout. In 1969 Guilbault and Montalvo built the first potentiometric biosensor for urea using a cation electrode and immobilized urease. Immobilized enzyme systems were used in flow analysis, first described by Homby using the Technicon Auto Analyzer approach and then flow injection analysis as introduced by Ruzicka and Hansen.

11. N E W AREAS OF BIOSENSORS Electrochemical sensors are still the most popular, with rapid growth in the fields of mass-sensitive sensors (Ngeh-Ngwainbi et al., 1989; Guilbault and Schmid, 1990) and fiber optrode sensors (Arnold, 1990).It is predicted by many that optical based biosensors and mass sensors will occupy a significant portion of the current market in biosensors in the next decade. New areas ofresearch in biosensors are listed in Table 2. Immunological sensors, in which antigens or antibodies are used to assay analytespresent in a sample, offer high selectivity and high sensitivity. Table 3 lists the principles underlying two

Current State of Biosensors

367

Table 2. New Areas of Biosensors A . Immunological Sensors

-Antigens -Antibodies B. Receptor Sensors C. Micro Electrodes

- Miniaturization (1-50 Microns) - Direct Sampling in Tissues, Brain Cells, etc. D. Noninvasive Sensors

- Saliva (Lactate, Alcohol) - Sweat (Lactate) - Transbuccal Mucosa (Glucose) E. Nonclinical Applications

- Food and Agricultural Analysis - Environmental Testing For: Pesticides, Bacterial Contamination, Toxins

approache-lectrochemical and piezocrystal, while in Table 4 some of the applications of immunobiosensors are listed. Applications to pesticides are now fairly well worked out, as illustrated by a recent paper describing the assay of atrazine at concentrationsas low as 0.0 1 ppb in water (Guilbault et al., 1992b). We have done preliminary work on the assay of toxins-such as brevitoxin and saxitoxin-at part per billion levels in food and environmental samples (e.g., linearity for saxitoxin 2 ppb to 1 ppm). We can certainly expect much more in this area in the very near future. Another very exciting area of research is bacterial detection in food samples-E. coli, Listeria, and Salmonella are organisms that are of extreme importance to assess the quality of food against contamination. Current methods are long (72 hr) and require extensive manpower.

Table 3. lmmunobiosensor Principles:

A.

B.

Electrochemical - Enzyme linked to antibody is attached to surface of Pt or C electrode. Follow change in current. Piezoelectric Crystal - Look at direct reaction of antigen and antibody by weight change and resulting AF.

3 68

GEORGE G. GUILBAULT Table 4. lmmunosensors A.

Pesticides

- Organophosphorus

- Carbamates - Herbicides B.

Toxins

- Saxitoxin

- Brevitoxin - Aflatoxin C.

Drugs

- Cocaine - Pharmaceuticals D.

Bacteria

- E. coli - Listeria - Salmonella We have developed a rapid PZ based immunosystem capable of detecting E. coli at low concentrations(Guilbault et al., 1992a). In current research we have shown that as little as 100 celldwell of E. coZi can be detected in food by an ELISA based immunosensor (Figure 1).

1.2 h

E C

v) 0 Q,

9

__

/=

1. -

0.8

a-

d.'

,.-

0.6

1

0.4

!

0

2 -E g

-~ ~

r

J'

i

0.2

1 0

01 -0.2

!

I

100

. . 1E4 1E6 1E8 I E I O 1E12 cfulwel I

Figure 1. The relationship between absorbance and cfdwell obtained with an ELISA immunosensor.

Current State of Biosensors

369

Another very interesting area of current research is the use of microelectrodes (1-50 p) for direct sampling in tissues and brain cells. We have demonstrated that glutamate biosensors (based on glutamate oxidase), choline and acetylcholine biosensors (using choline oxidase and acetycholinesterase),and L-alanine sensors (based on L-alanine dehydrogenase and NAD) can be built as small as 25 microns using C fibers and Pt wires for direct insertion into the brains of rats and mice. Finally, noninvasive sensors, in which the sensor is placed on external parts of the body, are capable of performing the same functions as invasive sensors (i.e., blood). A direct correlation of lactate and alcohol in saliva compared with blood exists, making possible a very easy assay of these metabolites. In meat we have been able to assay lactate, a metabolite of great use in sports medicine because it indicates the physical condition of the body (Guilbault et al., 1991). In the transbuccal mucosa we have shown that glucose can be measured as reliably as it can in blood. Another approach to noninvasive sensing of glucose is infrared (IR) spectroscopy, where a near IR scan of the glucose content in the finger of an individual was correlated with the blood level (Arnold, 1990).

50

40

30

20

1c

c

SW

AU

Figure 2. Number of centers dealing with biosensor research. From Mascini, 1992

GEORGE G. GUILBAULT

3 70

111. STATUS OF BIOSENSOR SALES AND RESEARCH A recent survey was conducted by the European Economic Community (EEC) of biosensor research in the USA, Japan and the EEC countries. The entire study was published (Mascini, 1992). The number of centers dealing with Biosensor Research are listed in Figure 2. The total number, 264, represents 70 centers in the United States, 62 in Japan, and 132 in the EEC countries-breakdown of work in each of the EEC countries is given in this figure. Figure 3 showsthe total number ofbiosensor projects in Europe, the United States, Canada, and Japan. Figures 4 and 5 show a breakdown of the total publications in biosensors in each of the EEC countries the United States and Japan. From this objective search (conducted in the period January, 1987 to February, 1992 by Professor R. Schmid of the GBF in Braunschweig) (Mascini, 1992), we can see how the United States and Japan are important as single nations, but, also, how the EEC is highly productive as a whole (1244 entries vs. 1284 in Japan and 864 in the United States). In a comparison of patent productivity in the EEC, Japan, and the United States, only 183 patents in this period were obtained by the whole of Europe vs. 551 in Japan. Germany, the United Kingdom, France, and Italy dominate the scene, but if the figures are divided by the number of inhabitants the results show that some small countries have excellent centers that are very active in biosensor technology.

-

EEC

USA

JAP

Figure 3. Total number of biosensor projects in Europe, USA, Canada, and Japan. From Mascini, 1992

Current State of Biosensors

Switz. Italy F rance

\

371

Swed.

I

/Mhers

\ UK

JAPAN

Ger

USA figure 4. A breakdown of the total number of publications dealing with biosensors in the EEC, USA and Japan. From Mascini, 1992

Finally, let us consider the market for biosensors. The majority of commercially available biosensors have been designed specifically for the clinical market and mainly for the determination of glucose concentrations.

Figure5 A breakdown of the publications including papers and patents coming from Europe. From Mascini, 1992

3 72

GEORGE G. GUILBAULT

Table 5. Growth Areas within lmrnunodiagnosticMarket Segments in the U.S. (in millions of dollars) I985 Diagnostic Category

Therapeutic drug monitoring Infectious diseases Thyroid function Immunoproteins Tumor markers Drugs of abuse Reproductive hormones Allergy Total

Doctor i Office

5 9 2 1 1 1

1990

Clinical Laboratory

Doctor's Office

Clinical Laboratory

1 4

34 24 20 24 6

75 55 10 2 1 15 5 12

325 196 70 49 49 40 39 13

24

427

175

78 1

165 106 106

From Mascini, 1992

In 1987, a new concept in biosensor design was developed in England and commercially exploited principally in the United States. The ExacTech pen-sized glucose sensor may be considered to be a second generation amperometric biosensor. It was developed especially for self monitoring of blood glucose and addresses a home test market worth about US 800 million dollars worldwide. The total world market for diagnostics is currently estimated to be about five billion dollars of which 2.1 to 2.8 US billion dollars falls into the immunochemical sector. Table 5 reports the predicted growth areas within immunoassay testing in the USA, many of which may be addressed with biosensor technology. Another significantmarket sector is that of rapidmicrobiologicaltesting. In 1983, the market was valued at 542 million dollars, by 1988 the same report stated that the number oftests had increased from 996 million tests to 1347million, an increase of 35% over that period. Japan and the United States have dominated the biosensor industry in the Western World. A U.S. company (Yellow Springs Instrument Co.) held the biggest market share prior to the introduction of the Exactech, but the Japanese have the greatest number of manufacturers. The market for biosensors in Europe in 1984 was reported as only 362,000 dollars. This figure was predicted to grow steadily in the early years of the forecast period ( 1984-1 990) before being replaced by explosive growth in the later years, resulting in a market of over 8.6 million dollars by 1990. The European market is far less significant than the U S . market over a similar period (19861991). A report entitled The Biosensor Market in the United States estimated the 1986 market size at 14.4 million dollars but realized a huge market increase to 365 million dollars by 1991 (actually about 400 million dollars), followed by explosive growth thereafter (Figure 6).

Current State of Biosensors

3 73

350 ...........

.......

"

300 ..................

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

250

z

9 -

200

i e

150

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

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

2

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

100

50 0 1986 1987 1988 1989 1990 1991 Figure 6. The current growth rate of the biosensor market in the United States.

REFERENCES Arnold, M. (1990). Fibre Optic Biosensors. Journal of Biotechnology 15,21%228. Clark, L. & Lyons, L. (1962). Glucose enzyme electrode. Ann. NY Acad. Sci. 102,582-585. Guilbault, G., Kramer, D., Goodson, L., & Bauman, E. (1965). The preparation of immobilized cholinesterase for use in analytical chemistry. Analyt. Chem. 37, 1378-1381. Guilbault, G. & Montalvo, J. (1969). A urea specific enzyme electrode. J. Amer. Chem. Soc. 91, 2 164-2 167. Guilbault, G. & Schmid, R.D. (1990). Electrochemical,piezoelectric and fibre optic biosensors. Adv. Biosens. 1,257-289. Guilbault, G.G., Faridnia, M., Palleschi, G., & Lubrano, G. (1991). Determination of lactate in human saliva with an electrochemicalenzyme probe. Analyt. Chim. Acta 245, 151-157. Guilbault, G.G., Hock, B., & Plomber, M. (1992a). Development of a piezoelectric immunosensor for the detection of enterobacteria.Enzyme and Microbial Technology 14,23&235. Guilbault, G.G., Schmid, R., & Hock, B. (1992b). A piezoelectric immunobiosensor for atrazine in drinking water. Biosens. Bioelectronics 7,411419. Mascini, M. (1992). Biosensor technology in the USA, Japan and Europe. EEC Special Report, EEC, Brussels, Belgium. Ngeh-Ngwainbi, J., Suleiman, A., & Guilbault, G.G. (1989). Piezoelectric crystal biosensors. Biosens. Bioelectronics 5, 1326.

This Page Intentionally Left Blank

BIOSENSORS AND MICROMACHINING

lsao Karube, Kenji Yokoyama, Yuji Murakami, and Masayuki Suda

I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375 11. ELECTROCHEMICAL FLOW CELL . . . . . . . . . . . . . . . . . . . . . 376 111. INTEGRATION OF ENZYME IMMOBILIZED COLUMN AND ELECTROCHEMICAL FLOW CELL . . . . . . . . . . . . . . . . . . 376 IV. INTEGRATION OF ENZYMATIC REACTOR AND CHEMILUMINESCENCE DETECTOR . . . . . . . . . . . . . . . . . . . . 377 V. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378

1. INTRODUCTION In recent years, microfabrication techniquesbased on integrated circuit technology, such as photolithography and etching, have been applied to other fields. These techniques, which are used to make some small and efficient three-dimensional devices, are called micromachining (Borky and Wise, 1979; Roylance and Angell, Advances in Molecular and Cell Biology Volume 15B, pages 375-379. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

375

376

I . KARUBE, K. YOKOYAMA, Y. MURAKAMI, and M. SUDA A'

PYREX

A

A'

A

figure 1. Structure of the electrochemical flow cell. From Suda et al., 1990.

1979; Fan et al., 1989; Shoji et al., 1990). Furthermore, some studies to make miniaturized chemical analysis systems were already reported (Terry et al., 1979; Manz et al., 1990). These analysis systems have many advantages such as fast response, small amount of sample, and low consumption of reagents as compared With the conventional one. We applied the micromachining technique to make a miniaturized enzyme-based sensor system.

II. ELECTROCHEMICAL FLOW CELL An electrochemical flow cell that has a very small inner volume of about 20 nl was

fabricated (Suda et al., 1990). Figure 1 shows a structure of the micro electrochemical flow cell. This flow cell can be used as an electrochemical detector for liquid chromatography or flow injection analysis (FIA). The enzyme-immobilized flow cell can be employed as an electrochemical biosensor. Glucose oxidase was immobilized onto the sample inlet hole of the cell using glutaraldehydeand bovine serum albumin. Glucose was calibrated in the range of 30 to 1000 mg/dl when 0.2 pl of the sample was injected.

111. INTEGRATION OF ENZYME IMMOBILIZED COLUMN A N D ELECTROCHEMICAL FLOW CELL A long open-tubular column was fabricated on the silicon substrate (Murakami et al., 1993). A glucose sensor was integrated with both an enzyme immobilized column and an electrochemical flow cell (Figure 2). The column was made by anisotropic silicon etching to be 100 pm wide, 70 pm deep, 1 m long, and a total volume of 5 yl. Four gold electrodes were formed on the glass substrate. Both of the two substrates were anodically bonded. Connecting unions to the pump and

Biosensors and Micromachining

377 A

PYREX A

5

Slllcon

I

Inlet

,

,I E

I

R

L

E

#

I

P Column

Outl&

I

-7

/

A'

Electrodes

A'

figure 2. Electrochemical detector integratedwith enzyme column. From Murakami et al., 1993.

sample injector were glued on the inlet and outlet holes with epoxy resin. Glucose oxidase (GOD) was immobilized on the inner wall of the column using 3-aminopropyltriethoxysilane and glutaraldehyde. This device was applied to a conventional FIA system.

IV. INTEGRATION OF ENZYMATIC REACTOR AND CHEMILUMINESCENCE DETECTOR An enzymatic reactor and a chemiluminescence detector were integrated on the

same chip (Figure 3) (Suda et al., 1992). The reactor consists of a silicon and glass substrate. On the silicon substrate, an enzymatic reaction column, a mixing chamber, a spiral flow cell were added by anisotropic etching. The total size of the measuring unit was 15 mm x 20 mm, and the total internal volume of the device was about 15 yl. Enzyme-immobilized glass beads were packed into the column, and a photodiode was placed onto the spiral flow cell. Using GOD-immobilized glass beads, determinationof glucose concentration was carried out in the range of 10 to 300 mg/dl. Glucose in human serum and urine was measured by a chemiluminescence detector. The correlation coefficient between this chemiluminescence

3 78

I. KARUBE, K. YOKOYAMA, Y. MURAKAMI, and M. SUDA Splral flow cell

T E

E

z

I-

I

\ A'

/

Mlxlng chamber

ErKyWWk mctor

Enzyme lmmobillzed beads

, , PYREX

/,'

J

A

/

\******.**.,~*J-

v

A

.

Figure 3. Chemiluminescence detector integrated with enzyme reactor. From Suda et at., 1992.

method and the conventional method was 0.99. Lactic acid contained in human serum was quantitated using the same procedure as the glucose determination. Samples containing L-lactic acid at concentrations from 4 to 50 mg/dl could be measured. The correlation coefficient between this chemiluminescencemethod and the conventional method was 0.98.

V. CONCLUSION Detection units for enzyme-based FIA were fabricated using micromachining techniques. Since these micromachined devices are batch-processed, they can be made at low cost and with good reproducibility. The signal from the detector decreases as the detector size decreases. However, the electrochemical and chemiluminescence methods are more sensitive than the spectroscopic method. Hence, the measurable range of the micromachined detector is almost the same as for the conventional method. Additionally,the measurement could be carried out at a flow rate of less than 50 pl/min and yet the pressure drop over the column was less than 0.1 atm. This suggests the possibility of applying a micromachined pump. Thus, the conventional plunger pump and sample injector used in the experiments described above will be replaced by micromachined devices in the near future.

Biosensors and Micromachining

3 79

REFERENCES Borky, J.M. & Wise, K.D. (1979). Integrated signal conditioning for silicon pressure sensors. IEEE Trans. Electron Devices ED-26, 1906-191 1. Fan, L.S., Tai, Y.C., & Muller, R.S. (1989). IC-processed electrostatic micromotors. Sensors and Actuators 2 0 , 4 1 4 8 . Manz, A,, Graber, N., & Widmer, H.M. (1990). Miniaturized total chemical analysis system: a novel concept for chemical sensing. Sensors and Actuators B1,244-248. Murakami, Y., Suda, M., Takeuchi, T., Yokoyama, K.. Tamiya, E.. & Karube, I. (1993). Integration of enzyme-immobilized column with electrochemical flow cell using micromachining techniques for a glucose detection system. Anal. Chem. 65,273 1-2735. Roylance, L.M. & Angell, J.B. (1979). A batch-fabricated silicon accelerometer. IEEE Trans. Electron Devices ED-26, 1911-1917. Suda, M., Muramatsu, H., Sakuhara, T., Ataka, T., Uchida, T., Murakami. Y.. Suzuki, M.. Tamiya, E.. Masuda, Y., & Karube, I. (1990). Glucose sensor processed with IC technology for microanalysis. Proc. Electrochem. Conf. Sept. 2%30,33. Chiba, Japan. Suda, M., Sakuhara, T., Suzuki, M., Tamiya, E., & Karube, I. (1992). Micromachined biosensors. Proc. 2nd World Congr. Biosensors, 4 0 M 0 6 . Shoji, S., Nakagawa, S., & Esashi, M. (1990). Micropump and sample-injector for integral chemical analyzing systems. Sensors and Actuators A2 1-A23, 18W92. Teny, S.C., Jerman, J.H., & Angell, J.B. (1979). Gaschromatographicairanalyzer fabricatedon a silicon wafer. IEEE Trans. Electron Devices ED-26, 1880-1 886.

This Page Intentionally Left Blank

RECYCLING SENSORS BASED ON KINASES

Ulla Wollenberger, Florian Schubert, Dorothea Pfeiffer, and Frieder W. Scheller

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. THEREACTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. THE ENZYME ELECTRODES . . . . . . . . . . . . . . . . . . . . . . . . IV. THE ENZYME OPTODE . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. THE ANALYTICAL REACTOR . . . . . . . . . . . . . . . . . . . . . . . . VI. THE ENZYME THERMISTOR . . . . . . . . . . . . . . . . . . . . . . . . VII. CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

381 382 383 384 385 386 388 388

ABSTRACT The sensitivity of enzyme sensors can be substantially increased by using substrate recycling. This paper summarizes studies on highly sensitive measurements of ADP and ATP based on immobilized hexokinase and pyruvate kinase in various types of analytical arrangements. The analytical techniques applied include electrochemical, optical, and thermometric methods. Advances in Molecular and Cell Biology Volume 15B, pages 381-390. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

381

382

WOLLENBERGER, SCHUBERT, PFEIFFER, and SCHELLER

1. INTRODUCTION Analyte recycling is a well studied way to evaluate enzymatic systems that can serve as biochemical amplifiers in enzymatic analysis of very low analyte quantities. The sensitivity enhancement is provided by shuttling the analyte between enzymes acting in a cyclic series of reactions (Lowry and Passonneau, 1972). The basic system consists of an enzyme couple where one enzyme catalyzes the regeneration of the substrate for the other enzyme (see Figure 1). S1 and S2 are depicted as substrates or coenzymes of the respective enzyme. Assuming a sufficiently high activity of enzyme 1 in the presence of its cosubstrate (Cl) and an analyte (Sl) (the concentration of which is far below its Michaelis constant), an amplification is achieved by switching on enzyme 2 by addition of its cosubstrate, C2. The analyte is then cycled between the two enzymes. Consequentlymuch more of the cosubstrates will be converted to the respective products (P1 and P2) than the analyte present, with the overall reaction being: c 1 + C 2 + P 1 +P2 The number of cycles in which the substrate is turned over within a given time is a function of the substrate concentration. While the total concentration of the substrate(s) (S1 + S2) in the reaction cycle is constant, the concentrations of the coreactants increase (P 1, P2) or decrease (C 1, C2) linearly with time. By measuring the concentration change of one of the coreactants directly or in an additional analytical step, the recycling system is used as a biochemical amplifier for the analyte (S 1 or S2). With soluble enzymes at steady state, the overall cycling constant that represents the amplification is that given by Lowry and Passonneau (1972). That is,

k = k,k,/(k,

+ k2)

with ki being the apparent first order constant Vm,i/KM,i. When the enzymes are used in membrane-immobilized form (as in an electrochemical biosensor), the ratio of the sensitivities in the linear measuring range of the amplified and unamplified

enzyme 1 P1

P2

*

enzyme 2

9,.

figure 7. Schematic illustration of enzymatic substrate recycling.

Recycling Sensors Based on Kinases

3 83

regime is called amplification factor, G, which is under steady state conditions (Kulys et al., 1986) expressed by:

G = L2k,k,/ 2D(k, + k2) where L is the membrane thickness and D the diffusion coefficient. The possible amplification is very large. In solution, an amplification rate of 20,000 per hour was obtained (Lowry and Passonneau, 1972), whereas in a membrane of an enzyme electrode up to 48,000-fold signal enhancement was achieved (Scheller et al., 1992). This concept of enzymatic signal amplification has been applied to a number of bioanalytical systems. For example, Lowry et al. (196 1) used enzymatic coenzyme recycling in solution for the determination of pyridine dinucleotides in the lower nanomolar range. In 1974, Davies and Mosbach utilized the regeneration of NAD(H) in enzyme electrodes for the determination of glutamate and pyruvate. Several enzyme electrodes based on recycling of the substrate to be measured have been described (e.g., Schubert et al., 1985, 1986; Kulys et al., 1986; Mizutani et al., 1985; Wollenberger et al., 1987b; Yang et al., 1991a), but there have also been reports on the use of the enzymes in a column combined with a thermistor (Scheller et al., 1985; Kirstein et al., 1989) and in a flow injection system with amperometric detection (Asouzu et al., 1990; Hansen et al., 1990; Yang et al., 1991b). Kinase reactions, owing to their mostly favorable equilibrium constants, are well suited to recycling experiments. Recycling of ATP with immobilized enzymes was first achieved using coencapsulated hexokinase and pyruvate kinase (Campbell and Chang, 1975). However, these investigations were aimed at the regeneration of cofactors. This particular paper summarizes studies on highly sensitive measurements of ADP and ATP based on immobilized hexokinase andpyruvate kinase in various types of analytical arrangements. The analytical techniques applied include electrochemical, optical, and thermometric methods.

II. THE REACTION For the recycling experiments hexokinase (HK) and pyruvate kinase (PK) have been coimmobilized. The basic reaction cycle is HK

ATP + glucose -+ ADP + glucose-6-phosphate PK

ADP + phosphoenolpyruvate + ATP + pyruvate In the unamplified mode only one ofthe reactions is used. For example, unamplified (usual) ADP measurements can be performed in the presence of phosphoenolpyruvate (PEP). One molecule of pyruvate and ATP are formed per molecule ADP converted.

3 84

WOLLENBERGER, SCHUBERT, PFEIFFER, and SCHELLER

Amplification is obtained when, in addition to PEP, glucose is introduced in the bulk solution. Now, HK is switched on; consequently, ATP is converted back to ADP, which again enters the PK reaction, and so forth. Any ADP or ATP diffusing to the enzymes is recycled many times during formation of one molecule of pyruvate (in the PK-reaction) and glucose-6-phosphate (in the HK-reaction) per cycle. The amplification can be detected via glucose-6-phosphate or pyruvate measurement or, more directly, with registration of the reaction heat.

111. THE ENZYME ELECTRODES The sensitive measurement of ADP/ATP with the enzyme electrode is based on the transformation of the tremendous amount of pyruvate formed in the recycling reaction involving oxygen consumption by the sequence of lactate dehydrogenase (LDH) and lactate monooxygenase (LMO) (Wollenberger et al., 1987a). Pyruvate + NADH 3 lactate + NAD+ LMO

Lactate + 0, -+ acetate + CO,

+ H,O

An appropriate biosensor has been constructed by combining an oxygen electrode and a gelatin layer comprising the four enzymes. In the presence of nonlimiting amounts of PEP and NADH (2.5 mM each), the sensor works as a three enzyme sequence electrode for ADP. The oxygen consumption depends linearly on ADP concentration up to 1 mM. When an excess of glucose (4.4 mM) is added, the sensitivity ofthe electrode to ADP is significantly increased, thus indicating analyte recycling between PK and HK. The corresponding current decrease is completed after 4 to 10 min. In a typical experiment 1.25 pM ADP caused the same current change in the amplification system as 300 pM did with the unamplified enzyme sequence. The detection limit with substrate amplification is 0.25 pM ADP and 0.1 pM ATP, whereas 50 pM is the limit of the ADP-sequence electrode without recycling. The calculated average amplification factor is 220 in the linear range of the ADP concentration dependence (Figure 2). The LDWLMO detector enzymes have been substituted with pyruvate oxidase (Schubert and Kehr, unpublished) or glucose-6-phosphate dehydrogenase (G6PDH). The latter has been utilized with an indication of NADH at a chemically modified graphite electrode (Yang et al., 1991a). For this purpose, heat treated graphite rods were immersed into an acetonic modifier solution [0.5% bis(benzophenoxazinyl) derivative of terephthaloic acid] for 5 s and covered with a layer of gelatin entrapped G6PDH, PK, and HK. The recycling process produces glucose6-phosphate, which is oxidized by G6PDH in the formation of NADH. Glucose-6-phosphate + NAD'

G6PDH

-+ gluconate-6-phosphate + NADH

Recycling Sensors Based on Kinases

385

current. nA

1 031

ATP ADP amplified 1

rnpIified 10

100

loo0

loo00

concentration, pmol/l Figure2. Relationship between the steady state current ofthe PWHWLDHILMO-fourenzymeelectrode and the concentrationof ADP (withand without PEP), and ATP (with PEP).

At a working potential of 0 mV vs. SCE the mediated oxidation of NADH is monitored. With all substrates present in sufficient amounts (3 mM PEP, 1.25 mM glucose, 1mM NAD'), a response to ATP was observed in 2 to 10min. The calibration graph is linear up to about 0.4 pM. ATP concentrations as low as 1 nM were detected. Comparing the slopes of the ATP calibration graph with and without (i.e., without PEP) recycling, a ratio of 1450 was calculated. This value represents the amplification factor of the given electrode.

IV. THE ENZYME OPTODE Because the combination of biochemical recognition elements with optical transducers has gained increasingimportance during the last years, it was of interest to couple the present kinase cycle with a fiber optic detector in order to obtain an enzyme amplification optode (Schubert, 1993).The reaction of G6PDH, which can be conveniently coupled to the kinases, lends itself to optical signal generation through the fluorescence of the formed NADH. The optode was constructed by fixing a polyurethane membrane containing coimmobilized PK, HK, and G6PDH to the tip of a fiber optic fluorescenceprobe, the Ingold FluorosensorTM.Optode function is based on the fact that NADH

386

WOLLENBERGER, SCHUBERT, PFEIFFER, and SCHELLER -3-

ADP, pmol/l

10

20

30

40

50

I

I

I

I

I

I

1

0.2

0.L

I

-x-

0.6 0.8 ATP, mmol/l

1

I

1.0

Figure 3. Calibration graphs for ADP (with PEP) and ATP (without PEP) using the PKIHWCGPDH optode.

fluoresces at 460 nm when excited at around 360 nm whereas its oxidized form does not. Fluorosensor light from a low-pressure mercury lamp is passed through interference filters to produce a light beam at 360 nm, which is guided through the probe tip into the surrounding medium by a quartz fiber bundle. The radiation emanating from the tip causes NADH to fluoresce, and the resulting emission is detected via a parallel fiber bundle, appropriate filters, and a photomultiplier tube. Operation of the enzyme optode as a linear two-enzyme sequence sensor for ATP results in a calibration curve that is linear up to 0.6 mM, the lower limit of detection being 10 pM. If, in addition to NAD' and glucose required for this measurement, PEP is present in the background solution (optimum concentration, 0.8 mM), the optode responds to ADP. The calibration graph (Figure 3) shows a short linear portion, up to 5 pM, with a slope about 90 times as large as that for ATP. This amplification permits ADP concentrations as low as 100 nM to be determined. Saturation for both analytes is reached at the same signal magnitude, thus giving evidence for diffusion limitation. The relative standard deviation for 10 successive assays of 2.5 pM ADP was determined to be 5.9%.

V. THE ANALYTICAL REACTOR An alternative way to use enzymatic substrate amplification is the separation of enzyme recycling and detection. For this purpose PK, HK, and G6PDH, which were also used in an enzyme electrode (see above), were coimmobilized on porous glass (Serva CPG-l0,0.75-0.125 mm, pore diameter 50 nm) and filled into a miniaturized column (Yang et al., 1991b). The enzyme column was then applied to a flow injection systemwith amperometricNADH indication using a phenazinium-modified graphite

RecyclingSensors Based on Kinases

387

working electrode. Oxidation of NADH is monitored at 0 V vs. Ag/AgCl(sat. KCl) reference electrode. In the presence of 1 mM NAD' the indicator reaction depends linearly on the concentration of glucose-6-phosphate up to 400 pM. The operation of the enzyme column under nonamplified conditions (presence of 1 mM glucose) was stable for more than 2 months. No significant variation of the sensitivity has been found at different flow rates. This lack of variation might be due to the compensation of two opposite effects, a reduced sensitivity at increasing flow rates, i.e., higher dispersion, and an increased sensitivity based on improved mass transfer. The recycling system is switched on by addition of PEP (3 mM). The response was linear over more than three decades, from 1 nM to 5 pM.In contrast to enzyme electrodes, the amplification factor for recycling in an enzyme reactor is not fixed, it depends on the residence time T of the substrate in the column (Yang et al., 1991b): G = Tk,k,/(k, + k2)

Obviously, the residence time plays the same role as the characteristic diffusion term L2/2D in enzyme electrodes with internal analyte recycling. The residence time for 25 p1 of 1 p M ATP has been increased to 70 s by decreasing the flow rate from 0.98 ml/min to 0.07 ml/min. This caused a linear increase of the oxidation current from 10 nA to nearly 150 nA. At this low flow rate the amplification factor became 350. The logical continuation of these experiments is the complete stop of the analyte flow within the column. A 5-min stop time resulted in a 12-times higher value than current, nA I

Figure 4. Effect of the controlled stop-time of the analyte flow (0.22 ml/min) on the response to 10 nM ATP of a substrate recycling analytical reactor in combination with a chemically modified electrode for NADH.

388

WOLLENBERGER, SCHUBERT, PFEIFFER, and SCHELLER

that obtained for 10 nh4 ATP using a continuous flow of 0.22 d m i n (Figure 4). An amplification factor of 1200 could be achieved at the controlled stop-time of 12 min.

VI. THE ENZYME THERMISTOR One enzyme thermistor probe experiment involved an enzyme column with coimmobilized kinases, a thermistor at the outlet of the reactor, and a reference thermistor, both arranged in a flow-through setup (Danielsson and Mosbach, 1988; Kirstein et al., 1989). The heat produced by the enzymatic reactions in the column was monitored. PK and HK were coimmobilized with high activities (44 U HK, 5 1 U PK per 1 g) on aminopropyl controlled pore glass. The PK/HK couple combines an exothermic and an endothermic reaction. The calorimetric signal of the HK catalyzed reaction is positive (about 75 kJ/mol in TRIS buffer) and depends linearly on the ATP signal up to 1 mM. The enthalpy change of the PK reaction is positive (about +30 kJ/mol). With cofactor recycling (Le., in the presence of PEP and glucose) at 10 pM ADP or ATP the signal is as high as that obtained for 300 pM ATP with HK alone, i.e., an amplification factor of 30, was reached. Since the absolute value of the negative enthalpy of the HK reaction is higher than that of the PK reaction, the net heat change, which is amplified is also negative. However, the calibration curves for ADP and ATP are nonlinear. Considerable enhancement of the sensitivity for the cofactors was achieved by double amplification in a two-reactor system. A lactate oxidase (LOD)/LDW catalase reactor, as has been described for pyruvate amplification (Scheller et al., 1985), was inserted into the flow between the PUHK reactor and the thermistor. For this further amplification, it is advantageous that both enzyme reactions are exothermic (LDH AH = -62.1 kJ/mol, LOD AH = -100 kJ/mol). The pyruvate leaving the first column is recycled in the second reactor. Pyruvate + NADH L3 lactate + NAD+ LOD

Lactate + 0,+ pyruvate + H,O, In the LDWLOD/catalase reactor (in the presence of 2 mM NADH) an amplification factor for the pyruvate response of about 55 was obtained. The amplification factor of the HKPK cycle was 30. Thus the combination of both recycling systems provided an overall amplification of the sensitivity for ADP/ATP up to 1700 and a detection limit of 10 nM.

VII. CONCLUSIONS The common principle for all the sensors presented here is the recycling of the analyte between pyruvate kinase and hexokinase. Apart from the different transduc-

Recycling Sensors Based on Kinases

389

tion schemes, a variety of immobilization matrices using different specific enzyme activities have been employed.Therefore it appears difficult to compare the sensors in terms of quantitative data. The differences in the amplification factors between the two enzyme electrodes rely mainly on different membrane thicknesses and the higher enzyme activities used in the three-enzyme electrode. However, the threeenzyme electrode and the enzyme optode, both employing the HWPWG6PDH system, exhibit very different recycling efficiencies. This may be due to the different membrane materials used. Furthermore, the lower overall sensitivity of the optode may be a result of the poor (large and flat) geometry of the sensor tip and the fact that, in contrast to the electrochemicalsensor, NADH is accumulated in the membrane, thereby slowing down the G6PDH reaction. In the reactor the immobilization of high kinase activities results in an extended operational stability. However, the efficiency of the enzymatic analyte recycling is restricted by the volume of the column and the residence time of the analyte within the column. Consequently, the accumulation of a product of the recycling process and its subsequent stripping is a way to improve the amplified sensor operation. The multiplication of the amplification by a second recycling enzyme pair yields a futher increase in sensitivity. It is quite clear that enzymatic recycling provides a potent and universal method for amplification of the response of biosensors that is applicable to all types of transducer. The coupling of the recycling system with additional enzymatic reactions by coimmobilizingall the required enzymes in the sensor membrane demonstrates that enzymatic amplification sensors do not necessarily require recycling systems that produce directly transducable species. This considerably extends the concept of enzymatic analyte amplification in electrochemical biosensors.

REFERENCES Asouzu, M.U., Nonidez, W.K.. & Ho, M.H. (1990). Flow injection analysis of L-lactate with enzyme amplification and amperometric detection. Anal. Chem. 62, 70g712. Campbell, J. & Chang, T.M.S. (1975). Enzymatic recycling of coenzymes by a multi-enzyme system immobilized within semipermeable collodion microcapsules. Biochim. Biophys. Acta 397, 101120. Danielsson, B. & Mosbach, K. (1988). In: Methods in Enzymology, Vol. 137 (Mosbach. K., Ed.), Academic Press, New York. Davies, P. & Mosbach, K. (1974). The application of immobilized NAD' in an enzyme electrode and in model enzyme reactors. Biochim. Biophys. Acta 370,329-338. Hansen, E.H., Amdal, A,, & Norgaard, L. (1990). Exploitation of the flow injection approach for analytical procedures based on enzymatic amplification reactions. Anal. Lett. 23,225-240. Kirstein, D., Danielsson, B., Scheller, F.W., & Mosbach, K. (1989). Highly sensitive enzyme thermistor determination of ADP and ATP by multiple recycling enzyme systems. Biosensors 4,231-239. Kulys, J.J., Sorochinskii, V.V., & Vidziunaite, R.A. (1986). Transient response of bienzyme electrodes. Biosensors 2, 135146. Lowry, O.H. & Passonneau, J.V. (1972). In: AFlexible System ofEnzymatic Analysis. Academic Press, New York.

390

WOLLENBERCER, SCHUBERT, PFEIFFER, and SCHELLER

Lowry, O.H., Passonneau, J.V., Schulz, D.W., & Rock, M.K. (1961). The measurement of pyridine nucleotides by enzymatic cycling. J. Biol. Chem. 236,274&2755. Mizutani, F., Yamanaka, T., Tanabe, Y., & Tsuda. K. (1985). An enzyme electrode for L-lactate with a chemically-amplified response. Anal. Chim. Acta 177, 153-1 66. Scheller, F.W., Schubert, F., Pfeiffer, D., Wollenberger, U., Renneberg, R., Hintsche, R., & Kiihn, M. (1992). In: Biosensors: Fundamentals, Technologies and Applications (Scheller. F.W. & Schmid, R.D., Eds.), GBF Monographs Voi. 17, 3-10, Scheller, F.W., Siegbahn, N., Danielsson, B., & Mosbach, K. (1985). High-sensitive enzyme thermistor determination of L-lactate by substrate recycling. Anal. Chem. 57, 174k1743. Schubert, F, (1993). A fiber-optic enzyme sensor for the determination of adenosine diphosphate using internal analyte recycling. Sensors Actuat. B. 11,53 1-535. Schubert, F., Kirstein, D., Schroder. K.L.. & Scheller, F.W. (1985). Enzyme electrodes with substrate and co-enzyme amplification, Anal. Chirn. Acta 169,391-396. Schubert, F., Kirstein, D., Scheller, F., Appelqvist, R., Gorton, L., & Johansson, G. (1986). Enzyme electrodes for L-glutamate using chemical redox mediators and enzymatic substrateamplification. Anal. Lett. 19, 1273-1288. Schubert, F. & Kehr, S., unpublished. Wollenberger, U., Schubert, F., Scheller, F.W., Danielsson, B., & Mosbach, K. (1987a). Abiosensor for ADP with internal substrate amplification. Anal. Lett. 20.657468. Wollenberger, U., Schubert, F., Scheller, F.W., Danielsson, B., & Mosbach, K. (1987b). Coupled reactions with immobilized enzymes in biosensors. Studia Biophys. 119. 167-170. Yang, X., Pfeiffer, D., Johansson, G., & Scheller, F.W. (1991a). Enzyme electrodes for ADPIATP with enhanced sensitivity due to chemical amplification and intermediate accumulation. Electroanalysis 3,659663. Yang, X., Pfeiffer, D., Johansson, G., & Scheller. F.W. (1991b). Nanomolar level amperometric determination of ATP through substrate recycling in an enzyme reactor in a FIA system. Anal. Lett. 24, 1401-1417.

ELECTRON CONDUCTING ADDUCTS OF WATER-SOLUBLE REDOX POLYELECTROLYTES A N D ENZYMES

loanis Katakis, Mark Vreeke, Ling Ye, Atsushi Aoki, and Adam Heller

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 392 I. INTRODUCTION AND SCOPE . . . . . . . . . . . . . . . . . . . . . . . . 392 11. ELECTRON DIFFUSION IN REDOX HYDROGELS . . . . . . . . . . . . 393 111. ELECTRON CONNECTION OF FLAVOENZYME REDOX CENTERS TO ELECTRODES . . . . . . . . . . . . . . . . . . . . . . . . . 396 IV. ELECTRICAL CONNECTION OF PYRROLOQUINOLINE 401 QUINONE REDOX CENTERS . . . . . . . . . . . . . . . . . . . . . . . . v. ELECTRICAL CONNECTION OF HEMEPROTEIN PEROXIDASES . . . 403 VI. ASSAY OF NAD(P)H THROUGH QUANTITATIVE TRANSLATION TO H202 . . . . . . . . . . . . . . . . . . . . . . . . . . . 405

Advances in Molecular and Cell Biology Volume 15B, pages 391409. Copyright 0 1996 by JAI Press h e . All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

391

392

I. KATAKIS, M. VREEKE, L. YE, A. AOKI, and A. HELLER

ABSTRACT Biochemical fluxes were transduced to electrical currents through catalytic centers of oxidoreductases, that were electrically connected to electrodes with crosslinked hydrogel forming redox polymers. The enzymes were covalently bound to the hydrogels through which the electrons diffuse. The electron difision coeficients, reaching lo4 cm2sec-' in the redox polymer gels, were high enough to allow efficient collection of the electronstransferred during the electrooxidationor electroreduction of substrates. The adequate electron diffusion through the polymer skeleton of the hydrogel and the rapid diffusion of water-soluble substrates and products in the gels led to high current density enzyme electrodes. These included electrodes with electrically connected flavoenzymes (e.g., sarcosine, glutamate, lactate and glucose oxidases), quinoprotein enzymes (e.g., PQQ glucose dehydrogenase) and heme enzymes (e.g. peroxidases). The peroxidase electrodes were useful in the sensing of NAD(P)H that reacted with oxygen in the presence of N-methyl phenazonium salts to form NAD(P)+ and H202.

1. INTRODUCTION AND SCOPE In this chapter, dedicated to Klaus Mosbach on his 60th birthday, we consider the non-diffusional relaying of electrons between redox enzymes and electrodes through hydrogels, built with crosslinked polymer networks. As will be seen, even though these gels contain predominantly water, their 3-dimensional redox-polymer network, to which the enzymes are covalently bound, conducts electrons. Electrochemical biosensors, such as amperometric biosensors based on such gels have some rather unique properties. On the one side, water soluble substrates and products permeate through the hydrogels. On the other, electrons, originating in an oxidizable substrate, or delivered to a reducible substrate, diffuse through the network. Because the electrons are efficiently collected or delivered and because the diffusion of water soluble molecules through the gel does not differ excessively from that through a plain aqueous solution, the current densities flowing through hydrogel-coated electrodesare high. The high current densities translate to substantial sensitivitiesin the transduction of the flux of biochemicals to electrical currents as reviewed by Heller (1992). The group of enzymes that has been electrically connected through electron-conducting hydrogels to electrodes includes now flavoenzymes (Foulds and Lowe, 1988; Gregg and Heller, 1990, 1991a,b;, Katakis and Heller, 1992),quinoprotein (pyrroloquinoline, PQQ) enzymes (Ye et al., 1993), and heme peroxidases (Vreeke et al., 1993). As will be seen, the hydrogels also allow transduction of NAD(P)H fluxes, after translation to hydrogen peroxide fluxes, to electrical currents (Vreeke et al., 1993). We shall review the electron diffusionin redox-polymer skeleton based hydrogels (Aoki and Heller, 1993),then discuss the electrical connection of flavoenzymes (Foulds and Lowe, 1988; Gregg

Electron Conducting Adducts

393

and Heller, 1990, 1991a,b; Katakis and Heller, 1992), heme enzymes (Vreeke et al., 1993), PQQ-enzymes (Ye et al., 1993) via these hydrogels to electrodes.

II. ELECTRON DIFFUSION IN REDOX HYDROGELS Cross-linked redox polymer films in which redox enzymes are immobilized relay electrons between oxidoreductases such as glucose oxidase and electrodes. An effective electron relaying polymer is poly(4-vinylpyridine)partially complexed with osmium bis(bipyridine) dichloride and quatemized with 2-bromoethylamine (POs-EA) (Figure 1) (Gregg and Heller, 199la,b). A key parameter in defining the electron relaying behavior of a redox polymer is its electron diffusion coefficient (De).We measured D, for POs-EA, cross-linked with poly(ethy1ene glycol) diglycidyl ether (PEGDE),using steady state voltammetry at interdigitated array (IDA) electrodes (Chidsey et al., 1986; Feldman and Murray, 1986, 1987; Dalton et al., 1990; Sunidge et al., 1992; Nishihara et al., 1991; Kittlesen et al., 1985; Belanger et al., 1987; Smith et al., 1988; Shu and Wrighton, 1988; Goss and Majda, 1991). D, depends on the ionic strength,anion, pH, and degree of cross-linking of POs-EA. D, was measured by the steady state voltammetry at interdigitated array (IDA) electrodes as follows: IDAgold electrodeson glass were fabricated by conventional photolithography and sputter-deposition of gold onto chromium primed glass

i

C&OCCH2CH2N PA2 (I)

3l3

l!3POA""y-v PEGDE (b)

Figure 1. Structure of the water soluble, enzyme complexing redox "wire" POs-EA. m = 1, n = 1, p = 3 or 2. The polymer-enzyme complexes are crosslinked with either: (a) a polyfunctional aziridine (PAZ), or (b) poly(ethy1ene glycol) diglycidyl ether (PEGDE)to yield hydrogel-forming networks.

394

I. KATAKIS, M. VREEKE, L. YE, A. AOKI, and A. HELLER

substrates. The electrodes consisted of 100 (N), 5.0 pm wide fingers (w), separated by 5.0 pm gaps (gap), that were 2.0 mm long. The narrow widths and gaps of the 100 fingers allowed measurement of electron diffusion coefficients, that were much lower than ion diffusion coefficients. When the potential at one of the IDA electrodes was swept from reducing to oxidizing, while the other electrode was maintained at a reducing potential (generator-collector experiment), an anodic steady state current (Is,) was observed upon redox cycling, because the electrode at the fixed reducing potential generated an oxidizable species. The total charge associated with the redox centers (Q) was obtained by integrating the area of the surface wave voltammogram when the potential of both of the IDA electrodes was scanned (generator-generatorexperiment). From I,, and Q, D, is obtained through equation 1 (Chidsey et al. 1986, Feldman and Murray, 1986, 1987). De = (IJQ gap (W )+ gap) NlCN-1)

(1)

The attractive feature of this technique was that D, was derived of I,,, Q and the IDA geometry, and did not require knowledge of either the film thickness or of the concentration ofthe redox species. Furthermore,because the technique was a steady state method, the current was not affected by diffusion of ions into and out of the redox polymer network, that necessarily takes place in transient experiments. Typical cyclic voltammograms for POs-EA, cross-linked with 5.0 wt% PEGDE, on IDA electrodes at a 1.O mV/s scan rate are shown in Figure 2a. The voltammograms in generator-collector experiments (solid line) had a sigmoidal shape; They showed the expected Os2+oxidation at the generator, transport of electrons through the redox polymer (by electron hopping or electron self-exchange between neighboring Os3+and 0s'' sites) and re-reduction of Os3+sites at the collector (so called redox cycling). The anodic limiting current was identical with the cathodic one, and the limiting current was independent of scan rate. These results indicated the ' ~ ratio ' between the existence of a linear concentration variation in the 0 ~ ~ 'site generator and the collector electrodes at the 1.O mVs-' scan rate, and established that charge transport through the POs-EA polymer limited the current. In contrast, the voltammograms in generator-generatorexperiment (dotted lines) showed welldefined surface waves. As the voltammetric wave was symmetrical and the peak current was proportional to the scan rate, the total charge of POs-EApolymer coated IDA electrode (Q) could be calculated by integrating the current of the voltammogram. From equation 1, D, at 0.1 M NaCl was calculated to be 4.8 x lo-' cm2s-'. The voltammograms of Figure 2 reveal a strong ionic strength dependence of D,. When the concentration ofNaCl was changed from 0.1 M to 1.O M, Q was constant but I,, decreased. Moreover, when the potential of one IDA electrode was swept, while the other of IDA electrode was disconnected (generator-open circuit experiment, dashed line), a shoulder, indicating lateral electron diffusion between fingers was seen. This shoulder was sharp in 0.1 M NaCl but had, in 1.0 M NaCl, a diffusional tail, leading us to conclude that De of POs-EA decreases upon increasing

Electron Conducting Adducts

395

b

a -0.1

0.1

0.3

0.5

Potentlat, V vs SCE

I

01

03

05

7

Potential. V vs SCE

Figure 2. Ionic strength dependence of cyclic voltammograms for POs-EA crosslinked with 5.0 wt% PEGDE coated on IDA electrodes, at an Os3+’*+site coverage of r = 1.93 x mol cm-2 in 20 mM phosphate buffer at pH 7.0 and at 1 .O rnV s-’ scan rate. Generator-collector (solid line), generator-open circuit (dashed line), generator-generator (dotted line) voltammograms are shown for (a), 0.1 M; (b), 1 .O M NaCI.

with the ionic strength. Evidently, dehydration of POs-EA at increasing ionic strengths made the cross-linked structure more rigid. Figure 3 presents the pH dependence of D, for POs-EAcross-linked with 5.0 wt% PEGDE. D, changes remarkably near pH 4, close to pKa of the pyridine nitrogen, from 4.5 x lop9to 1.6 x lo4 cm2s-’. Evidently, the polymer film was swollen and the backbone of the polymer was flexible when the pyridines were protonated at a pH below pKa,where D, was high. We conclude that D, is determined by the degree of swelling of the cross-linked POs-EApolymer. D, values of the POs-EApolymer at various concentrations of cross-linker are summarized in Table 1. For the network made with 5.0 wt% cross-linker D, was pH dependent. D, was, however, almost independent of pH at 25 wt% cross-linker. When the concentration of cross-linker reached 25 wt%, 32 % of the pyridines in the polymer were cross-linked with the diepoxide. Because the polymer could not swell even at low pH, D, became independent of pH. We conclude that in cross-linked POs-EA the electron diffusion coefficients increase upon hydration of the network and that electron transport through networks is controlled by segmental motion of the polymer backbone.

396

I . KATAKIS, M. VREEKE, L. YE, A. AOKI, and A. HELLER

1.5

.-

r

'cn

N5 m -

E

1 . 0 -.

X

ow 0.5

-.

0

0

2

6

4

8

10

12

PH

Figure 3. pH dependence of the electron diffusion coefficient of POs-EA cross-linked with 5.0 wt% PEGDE, = 4.35 x 1 O-' mol cm-2 in 20 mM phosphatebuffer containing 0.1 M NaCl at 2.0 mV s-' scan rate.

Table 1. Effect of the Extent of Cross Linking on the Electron Diffusion Coefficient (D,) ~

Cross Linker wt%

p H 3.0

p H 7.0

5

1.6 x 10-8

4.5 x 10-9

25

4.2 x lo4

3.2 x 10")

111. ELECTRON CONNECTION OF FLAVOENZYME REDOX CENTERS TO ELECTRODES We have described previously the synthetic procedure for hydrogel-formingredox polymers (Gregg and Heller, 1991b). Poly(viny1pyridine) can be derivatized with Os(bpy),Cl, in ethylene glycol or ethanol under reflux. The osmium derivatized polymer can be reacted with bromoethylamine at 60°C. It has been found that the polymer containing a molar substitutionratio near 1: 1:3 osmium :ethylamhe :non substituted pyridines, meets requirements relevant to enzyme electrodes, such as solubility in water, high charge density, high osmium loading, and crosslinkable

Electron Conducting Adducts

397

side chains. For effectiveelectricalconnectionof enzyme redox centers the polymer must be flexible so that it can penetratethe proteins. We have found that the polymer POs-EA (Figure 1) electrically “wires” redox enzymes. The hydrogel networks are formed by crosslinking POs-EA at or near ambient temperature and in aqueous solution with the diepoxide polyethylene glycol diglycidyl ether (PEGDE) (Figure 1b) or a polyfunctional aziridine (PAZ) (Figure 1a). The crosslinker is chosen so as to avoid reaction of amino-acids that are critical for the fknction of the enzyme. With polymeric “wires” virtually all flavoproteins,that can transfer one electron at a time from their reduced active centers to an electron acceptor, have been shown to be “wireable” and useful in the construction of enzyme electrodes. Selectivity of the electrodes requires absence of direct electrooxidationof the substrate on the electrode surface itself, or in the redox hydrogel, directly by the oxidized osmium complex, at the applied usually 0.45 V vs. SCE for POs-EA. High catalytic current density (1@-lod3Acm-’) enzyme electrodes have been constructedby “wiring” glucose oxidase, lactate, or L-a-glycerophosphateoxidase (Gregg and Heller, 1991b; Katakis and Heller, 1992); moderate current densities ( 104-1 OP5 AcmP2)resulted form “wiring”glycolate, theophylline, D-amino acid, sarcosine, L-glutamicacid, or cholesterol oxidase. The catalyticcurrents of L-amino acid and choline oxidase electrodes were very low. For xanthine oxidase the catalytic current was almost indistinguishable from the background current associated with nonenzymatic electrooxidation of xanthine by the oxidized redox polymer. Examination of the significance of complex formation between the redox polymers and the enzymes on the efficiency of current collection shows that electrostatic complexation is important for efficient communication between the enzymes and the polymers (Katakis et al., 1994), but other kinds of interactions may also be relevant. Figure 4 shows the response of a sarcosine oxidase electrode and the variation of its current density with the electrode construction parameters. The response of the electrode under anaerobic conditions and under oxygen are shown in Figure 4a. Figure 4b shows the variation of the current density with the enzyme content of the hydrogel. Figure 4c shows that by changing the enzyme content of the gel one can adjust the dynamic range (apparent Michaelis constant, Km) of the sensor. Figure 4d shows that increasing the thickness of the enzyme-polymer film results in an almost linear increase of the maximum current density obtained, up to about 200 pg cm-2 loading, where the film thickness reaches about 2 pm. Sarcosine oxidase has been used in bilayer electrodes for the detection of creatine (Motonaka et al., 1990). Creatine was converted in an outer creatine amidinohydrolase film into sarcosine, which was sensed by an inner layer of sarcosine oxidase-redox polymer hydrogel. The response of such an electrode is compared to one without the creatine amidinohydrolase in Figure 5. The electrode might be applicable in the direct determination of creatine phosphokinase activity in serum or whole blood.

150

a

20

0

60

40

[Sucosine],

80

100

120

mM

0.0

2 0 4 . 0.0

0.2

0.4

0.6

0.2

0.6

0.4

0.8

1.0

Enzyme Weight Wction in Hydrogel

0.8

1.0

Enzyme Weight Fraction in Hydrogel

0

I

100

.

,

200

.

, 300

.

I 400

Leading, pg em-'

Figure 4. (a) Dependence of the current density of a sarcosine oxidase electrode on sarcosine concentration. Open circles, Ar atmosphere; solid circles, 0 2 atmosphere; glassy carbon electrode; 20 weight OO/ enzyme; 200 pg cm-2 loading; 0.45 V (SCE); pH 7.2 phosphate buffer, 0.1 M NaCI; 1000 rpm; 22°C. (b) Dependence of the maximum current density of sarcosine oxidase electrodes on the enzyme content of the hydrogel at 150 pg crK2 loading. Conditions same as above. Jmax determined from Eadie-Hofstee plots. (c) Dependence of the apparent Michaelis constants of sarcosine oxidase electrodes on the enzyme content of the hydrogel. (d) Dependence of the maximum current density of sarcosine electrodes made with 20 weight YOenzyme on the thickness of the hydrogel-films.

Electron Conducting Adducts

0

399

10

20

D

[Creatine], mM

Figure 5. Response of a bilayer sarcosine - creatine amidinohydrolase electrode (open circles) to creatine. The solid circles show the creatine-responseof a sarcosine electrode made without the hydrolase. The hydrogel-forming polymer contained 26% sarcosine oxidase, 20% PAZ crosslinker, and 54% POs-EA. Loading was 50 pg cm-2.The bilayer electrode had an exterior film of 40 pg of creatine amidinohydrolase (0.5 units) crosslinked with PAZ and poly(viny1pyridine) quaternized with bromoethylamine.

In some of the flavoenzymes, for example in glycerophosphate oxidase, the semiquinone form of FADH. could not be observed during the normal catalytic cycle when dioxygen was the electron acceptor (Claibome, 1986; Jacobs and VanDemark, 1960). From the efficient electron transfer to Os(II1) relays that were observed, it was evident nevertheless that the substrate-reduced enzyme was reoxidized in two sequential electron-proton transfer steps. The rate limiting step in the multistep electron transfer from lactate oxidase to an electrode was the electron transfer between reduced FADH, centers and 0 s relays in the hydrogel. Figure 6 shows the pH response of a glucose oxidase electrode. Different pH maxima exist for the redox-hydrogel electrodeand the natural enzyme activity. This shift in pH is consistent with the proposal that electron transfer between FADH, or FADH. and Os(III), which is pH independent, is rate limiting. In the natural enzymatic reaction the pH dependent reduction of 0, to H,O, is rate limiting. The key determinants in realization of high current densities are the turnover number and specific activity of the enzyme, and the strength of the complex between the enzyme and the redox polymer (Table 2). In general enzymeswith high turnover numbers and high complex formation constants with POs-EA give the highest current densities. The unexpectedly high current density of sarcosine oxidase and low current density of glutamate oxidase show, however, that the two parameters are not the only ones that are important. It is conceivable that the aberrations result from a specific interaction of the polymer with the catalytically active site of the enzyme, or from particularly deep protein penetration of an electron relay, or from deactivation of the enzyme upon immobilization.

I. KATAKIS, M. VREEKE, L. YE, A. AOKI, and A. HELLER

400 1.1

1

5

i

6

0

7

9

10

PH Figure 6. pH dependence of the glucose-produced current of a glassy carbon electrode, modified with the hydrogel formed of glucose oxidase, POs-EA and PEGDE under argon (open circles) and of a hydrogen peroxide detecting platinum electrode modified with the same gel, except for having the POs-EA analog without the osmium complex under oxygen (solid circles). 10% weight glucose oxidase, 80 pg cm-2 loading.

The turnover number and specific activity of the enzyme define the number of electronsproduced per unit volume. In extreme cases of low turnover numbers, the volume fraction of protein necessary for the production of a detectable current is so high that the hydrogel no longer contains a hgh enough density of electron relaying centers and electron diffusion through the film becomes rate limiting. Table 2 shows Table 2. Response of "Wired" Flavin Oxidase Electrodes and its Relationship to Kinetics of the Enzyme, Strength of Enzyme-Polymercomplex and Crosslinker Used* Highest Current Density (PA cm-=,

Enzyme'

GOx LOX GPO GLUx

sox THOx CHOx DAAOx Notes:

290 120 83 6.5 90 2 1 7

% Enzyme Content of

Turnover Number

Complex at Sawration"

(s-')

73 >68 81

203 144 30 79 14 ? 8 3.5

75 73 >56

62 86

Crosslinke;"

PEGDE PEGDE,PAZ PEGDE PEGDE PAZ PAZ PEGDE PAZ

Electrodes so constructed as to be kinetically limited and have 0.1 k 0.02 units immobilized. Abbreviations : GOx, glucose oxidase; LOX,lactate oxidase; GPO, L-a-glycerophosphateoxidase; GLUx, L-glutamic acid oxidase; SOX,sarcosine oxidase; THOx, theophyline oxidase; CHOx, cholesterol oxidase; DAAOx, D-amino acid oxidase ** As determined by iwlectric focusing experiments. .I. PAZ: polyfunctional aziridine, PEGDE : poly(ethy1ene glycol) diglycidyl ether.

Electron Conducting Adducts

401

the strength of the complexes formed between POs-EA and enzymes on a scale built on detection by isoelectric focusing of migration of free enzyme from the complex (Katakis et al., 1994). The values in the table indicate the % enzyme content of the complex when unbound enzyme is first observed. Usually, the higher this number, the stronger the complex, and the stronger the complex, the higher the current density. Formation of a stronger complex can result in a shorter electron transfer distance between the redox center of the enzyme and a relay ofthe polymer, or an increase in the density of electrically connected enzyme centers in the hydrogel.

IV. ELECTRICAL CONNECTION OF PYRROLOQUlNOLlNE QUINONE REDOX CENTERS Several groups of quinonprotein enzymes have been isolated and characterized in recent years (Duine et al., 1987). These enzymes have either pyrroloquinoline quinone (PQQ), topaquinone (TPQ) or tryptophanyl-tryptophan quinone (TTQ) cofactors (Duine, 1991). Glucose dehydrogenase GDH (EC 1.1.99.17), that was successfidly wired through the POs-EA based hydrogel (Ye et al., 1993)belongs to the group of PQQ-containing quinoproteins (Duine et al. 1979). The apo-enzyme (Van der Meer et al., 1990)of GDH was reconstituted by incorporating PQQ (Duine et al., 1979) in the presence of Ca2' (Geiger and Goerish, 1989). PQQ and Ca2+are firmly bound in the reconstituted enzyme. The activity of the reconstituted holoenzyme, 250 units mg-', was measured spectrophotometrically by monitoring the decoloration of Wurster's Blue (Dokter, et al., 1986). The reconstituted GDH can be stored at 4°C for more than two months without measurable loss of activity. The dependence of the current density on glucose concentration for POsENPEGDE-wired GDH and similarly wired glucose oxidase (GOX) electrodes is shown in Figure 7. The current density ofthe GDH electrode substantiallyexceeded that of the GOX electrode, reaching 1.8 mA cm-2 at 70 mM glucose. This current density was about three times higher than that of the electrode made with GOX. The sensitivity of the wired GDH electrode at 5 mM glucose was 165 pA cmP2M-' . The higher current density of the GDH electrode, relative to that of the GOX electrode, derived from the faster rate of electron transfer from the PQQH, centers than from FADH, centers to the osmium complex in the redox polymer/enzyme network. That the 3-dimensional redox network based hydrogel is effective in collecting and carrying electrons from GDH is seen from the fact that in electrodesmade with the electron shuttling diffusional mediator 1,l '-dimethylferrocene, the current density was only 54 pA cmP2at 4 mM glucose (D'Costa et al., 1986) versus 724 pA cmP2 for the redox hydrogel at the same glucose concentration. The current density of the redox-hydrogel based GDH electrode changed relatively little between pH 6.3 and 8.8.

402

I. KATAKIS, M. VREEKE, L. YE, A. AOKI, and A. HELLER 2ooo 1500

lo00

500

0

0

20

40

60

80

100

glucose Concentration, m M

Figure 7. Comparison of the current output and the sensitivity for GDH and GOX electrodes. Open square: GDH electrode; Solid diamonds: GOX electrode. 0.4 V (SCE); 1000 rpm; air atmosphere.

The half-life of the dissolved enzyme in 10 mh4 HEPES pH 7.3 buffer at room temperature was 5 days (Ye et al., 1993). The decay of the current of the GDH electrodes in 10 mM glucose at 25”C, pH 7.3 was, however, faster (Fig. 8, open squares), dropping to one half of its initial value in about 8 hours under continuous operation. The operational stability of the “wired” GDH electrode was found to be glucose concentration dependent, the output declining more rapidly at higher glucose concentrations. Fig. 8 shows the rates of decline for GDH electrodes in an electrochemical cell where the glucose concentration was held at 10 mM, and in a flow injection analyzer where a computer controlled injection system injected fixed volumes of 10 mh4 glucose once every 60 minutes. Under the conditions of the

0

10

20

30

40

time, hours

Figure 8. Time dependence of the output of GDH electrode at 0.4 V (SCE). Continuous operation at constant 10 mM glucose concentration (open squares) and in a flow injection analyzer where 30 second 10 mM glucose pulses were injected once per hour (solid diamonds).

Electron Conducting Adducts

403

experiment, it took less than 3 minutes for the sample to pass through the cell, i.e., the electrode was in 10 mh4 glucose for only 3 minutes each hour. Evidently, the rate of current loss depended on the amount of glucose electrooxidized(Figure 8).

V. ELECTRICAL CONNECTION OF HEMEPROTEIN PEROXlDASES Most peroxidases have catalytic sites containing covalently bound iron porphyrins. In peroxidases the iron heme is positioned close to the surface of the enzyme. The near surfaceposition of the catalytic site allows direct, non-mediated, electroreduction of the enzymes on graphite electrodesand thus the construction of direct H202 sensors (Kulys and Schmid, 1990, Jonsson-Pettersson, 1991). Many diagnostic tests ultimately rely on amperometric or colorimetricdetection of H202.Although feasible, enzymatic amperometric H202 detection is not commercially used. The most common amperometric H202assays involve use of a Pt anode held at an oxidizing potential, where not only H202, but also a variety of interferants are electrooxidized. Electron conducting POs-EA and PEGDE (Figure 1) redox-hydrogel based H202 sensors, involving electroreduction of lactoperoxidase (LOP), Arthromyces ramosus peroxidase (ARP), or horseradish peroxidase (HRP) have been made (Vreeke et al., 1993). In these, the H202fluxes were connected to currents by electrocatalytically reducing H20, according to the scheme shown in Figure 9. These enzymes were covalently bound to the 3-dimensional redox epoxy based hydrogel. Upon such immobilization the ensemble of enzyme molecules in the gel volume was electrically connected to the electrode. The amperometric measurements were performed in a standard 3-electrode cell, with the working electrode maintained at 0.OV (SCE) and rotated at 1000 RPM. H202 was catalytically electroreduced at the operating potential. The dependence ofthe catalytic H202electroreduction currents on the H202 concentration in pH 7.4

Figure 9. Scheme of cathodic assay of hydrogen peroxide using POs-EA-based electron conducting hydrogels to which a peroxidase is covalently bound. 0 ~ ~ ' ' ~ ' denotes gel-bound redox centers.

I . KATAKIS, M. VREEKE, L. YE, A. AOKI, and A. HELLER

404 loo0

r . 4

‘ 06 100

a

E

G

10

1

1 .i! ,001

I

....-..., . . .4

. . . . . . . . I

.01

.l

1

H202,

Figure 10. Arnperometric response of peroxide sensing cathodes built with different peroxidases covalently bound to electron conducting hydrogels based on POs-EA. Closed circles, HRP; open circles, N a l 0 4 oxidized HRP; squares, LOP; triangles, ARP.

phosphate buffer is seen in Figure 10 for four enzymes A m , LOP, HRP, and periodate-oxidized HRP. Although immobilization of lactoperoxidase in the redox hydrogel resulted in substantialreduction currents,the currents rapidly decayed when the H,O, concentration exceeded 10 pM, because of reversible substrate inhibition of the enzyme (Polis and Shmukler, 1955). Wiring of the peroxidase from Arthromyces ramosus in the electron-conductinghydrogel led to the highest current density > ImA cm-’. The current density increased linearly with the H,O, concentration from 0.1 yM to 500 pM.The current was remarkably stable, decaying only a few percent per hour. The horseradish peroxidase (HRP)containing electrodes did not exhibit either the high current density nor the exceptional linear range of the Arthromyces ramosus peroxidase based electrodes. Nevertheless, when NaIO, oxidized HRP was used, the sensors’performance improved. The aldehydes formed upon NaIO, oxidation of HRP sugar residues formed Schiff bases with the POs-EA amines, coupling the enzyme’s redox centers with those of the hydrogel and shortening the electron transfer distances. The currents in electrodesmodified with hydrogels with NaIO, oxidized HRP exceeded by a factor of 5 those observed in electrodes with untreated HRP, and the sensitivity reached 1 AM-‘ crn-,. The NaIO, oxidized HRP based electrode was operable at potentials between -O.lSV(SCE) and +0.3V(SCE). The wide potential window allowed choice of potentials where non-enzyme specific electrode reaction rates were slow, i.e. where interferant caused currents were low. The useful pH range of the sensors was 3.8 to 9.6, remaining within 10% of the maximum response throughout this range. Transient exposure of the sensors to solutions that were strongly acidic (PH < 3.8) reduced their response, but upon returning to a neutral pH the sensors recovered. Exposure to strong bases @H > 9.6) caused irreversibledecomposition of their gel.

Electron Conducting Adducts

405

VI. ASSAY OF NAD(P)H THROUGH QUANTITATIVE TRANSLATION T O H 2 0 2 Quinones and molecules with quinoid structures have been reduced in two electron transfer reactions by NAD(P)H (Kitani et al., 1981; Ito et al., 1989; Ottaway, 1966; Jones and Taylor, 1976). Amperometric NAD(P)H sensors based on such reduction have been described by Degrand and Miller (1980) and Fukui et al. (1 982), Gorton et al. (1991) and Cenas et al. (1985) and Kulys (1986). An example of a water soluble quinoid that is readily reduced by NAD(P)H is the N-methylphenazonium ion NMP+: NAD(P)H + NMP'

-+

NAD(P)+ + NMPH

The reduced phenazine, NMPH is reoxidized by molecular oxygen to NMP' in the H,O, generating reaction: NMPH + 0, 4 NMP'

+ H,O,

The sum of these reactions is the catalytic reduction of 0, to H,O, by NAD(P)H: NAD(P)H + O , N z * NAD(P)++ H,O, Because the reaction is fast and stoichiometric, NAD(P)H sensors based on amperometric assays of the depletion of 0, (Polster and Schmidt, 1989; Huck et al., 1976) and the colorimetric assay of the H,O, generated (Williams and Seitz, 1976) have been built. The translation of NAD(P)H into H,O, through the above reaction sequence allowed the sensing of NAD(P)H with the peroxidase electrodes with 1 AM-' cmP2 sensitivity and with a 0.1 to 200 pM linear range, similar to those of the H,O, sensors (Vreeke et al., 1993). Figure 11 shows the steady state alcohol response of

I

-

i 100

0 O O

0

-

0 0

0

e

E

0 0O0

u

O O O O

0

--,

- .-..-. . ...._ . . ....

. .-

% ethanol

Figure 77. Steady state alcohol response for alcohol dehydrogenase with the "wired" peroxidase cathode. 0 V (SCE);1000 rprn; pH 7.4 phosphate buffer.

406

1

I. KATAKIS, M. VREEKE, L. YE, A. AOKI, and A. HELLER

homogeneous solution reaction

I

HRP modified electrode

I

Figure 12. Electron transfer sequence in the transduction of the concentration dependent alcohol flux to a cathodic current. Even though 7 electron transfer steps are involved in the transduction, these are so efficient that the current represents about 10% of the electron pairs transferred in the oxidation of ethanol to acetaldehyde.

a sensor with NAD+/NADH dependent alcohol dehydrogenase, pH 7.4 solution containing 7.1 x 1OdM N-methylphenazonium methosulfate, 13 units/mL alcohol dehydrogenase and 1.3 x 104M NAD+.Figure 12 summarizes the electron transfer sequence in the sensing of alcohol.

CONCLUSION Electrons difise in hydrogels built with redox polymer skeletons at sufficient rates to allow interception of electrons transferred between oxidoreductases and their substrates. Their transfer to electrodes results in the transduction of biochemical fluxes to electrical currents. Electrical currents are produced by substrates of flavoenzymes, heme enzymes and quinoprotein enzymes. Through catalytic reaction of NAD(P)H with 0, in an NAD(P)+ and H,O, producing reaction and electroreduction of H,O, in peroxidase-containing redox hydrogels NAD(P)H concentrations can also be translated to electrical currents.

ACKNOWLEDGMENT This work is supported by grants from Office of Naval Research, National Science Foundation, the National Institutes of Health Grant No. 1-R01-DK42015-01, and the Robert A. Welch Foundation.

REFERENCES Aoki, A. & Adam Heller, A. (1993). Measurement of electron diffusion coeflicients in redox polymers by steady state voltammetry with interdigitated array electrodes. In: Extended Abstracts of the Joint Meeting of the Electrochemical Society and the Electrochemical Society of Japan, Honolulu, Hawaii. Belanger, D. & Wrighton, M.S. (1987). Microelectrochemical transistors based on electrostaticbinding of electroactivemetal complexes in protonated poly(4-vinylpyridine): Devices that respond to two chemical stimuli. Anal. Chem. 59, 14261432.

Electron Conducting Adducts

407

Cenas, N.K., Kanapieniene, J.J., & Kulys, J.J. (1985). Electrocatalytic oxidation of NADH on carbon black electrodes. J. Electroanal. Electrochem. 189, 163-1 69. Chidsey, C.E., Feldman, B.J., Lundgren, C., & Murray, R.W. (1986). Micrometer-spaced platinum interdigitated array electrode: Fabrication, theory, and initial use. Anal. Chem. 58.60 1 4 0 7 . Claiborne, A. (1 986). Studies on the structure and mechanism of Streptococcusfaeciurn L-a-glycerophosphate oxidase. J. Biol. Chem. 261, 14398407. D’Costa, E.J., Higgins, I.J., & Turner, A.P.F. (1986). Quinoprotein glucose dehydrogenase and its application in an amperometric glucose sensor. Biosensors 2, 7 1-87. Dalton,E.F., Surridge, N.A., Jernigan, J.C., Wilbourn,K.O., Facci, J.S.,&Murray, R.W.(1990). Charge transport in electroactive polymers consisting of fixed molecular redox sites. Chem. Phys. 141, 14S157. Degrand, C. & Miller, L.L. (1980). An electrode modified with polymer bound dopamine which catalyzes NADH oxidation. J. Am. Chem. Soc. 102,572t332. Dokter, P., Frank, J., & Duine, J.A. (1986). Purification and characterization of quinoprotein glucose dehydrogenase from Acinatobacter calcoaceticus L.M.D.79.4 1. Biochem. J. 239, 16S167. Duine, J.A., Frank, J., & Jongejan, J.A. (1987). Enzymology of quinoproteins. Adv. Enzymol. 59, 169-2 12. Duine, J.A. (I99 1). Quinoproteins: enzymes containing the quinonoid cofactor pyrroloquinoline quinone, topaquinone or tryptophan-tryptophan quinone. Eur. J. Biochem. 200,27 1-284. Duine, J.A., Frank, J., & Van Zeeland, J.K. (1979). Glucose dehydrogenase from Acinatobuctor calcoaceticus. FEBS Lett. 1 0 8 , 4 4 W 6 . Feldman, B.J. & Royce W. Murray, (1986). Measurement of electron diffusion coefficients through prussian blue electroactive films electrodeposited on interdigitated array platinum electrodes. Anal. Chem. 58,2844-2847. Feldman, B.J. & Murray, R.W. (1987). Electron diffusion in wet and dry prussian blue films on interdigitated array electrodes. Inorg. Chem. 26, 1702-1 708. Foulds, N.C. & Lowe, C.R. (1988). Immobilization of glucose oxidase in ferrocene-modified pyrrole polymers. Anal. Chem. 60,247S2478. Fukui, M., Kitani, A,, Degrand, C., & Miller, L.L. (1982). Propogation of a redox reaction through a quinoid polymer film on an electrode. J. Am. Chem. SOC.104,2%33. Geiger, 0. & Goerisch, H. (1989). Reversible thermal inactivation of the quinoprotein glucose dehydrogenase from Acinatobacter calcoaceticus. Biochem. J. 261,415-421. Gorton, L., Csoregi, E., Dominguez, E., Emneus, J., Jonsson-Pettersson, G., Marko-Varga, G.. & Persson, B. (1991). Selective detection in flow analysis based on the combination of immobilized enzymes and chemically modified electrodes. Chim. Acta 250,203-248. Goss, C.A. & Majda, M. (1991). Lateral diffusion in organized bilayer assemblies of electroactive amphiphiles influence of the oxidation state of the amphiphile investigated by steady-state methods involving an interdigitated miro-electrode array device. J. Electroanal. Chem. 300, 377-405. Gregg, B.A. & Heller, A. (1990). Cross-linked redox gels containing glucose oxidase for amperometric biosensor applications. Anal. Chem. 62,25%263. Gregg, B.A. & Heller, A. (1991a). Redox polymer films containing enzymes. 1. A redox-conducting epoxy cement: synthesis. characterization, and electrocatalytic oxidation ofhydroquinone. J. Phys. Chem. 95,5970-5975. Gregg, B.A. & Heller, A. (1991b). Redox polymer films containing enzymes. 2. Glucose oxidasecontaining enzyme electrodes. J. Phys. Chem. 95,576-580. Heller, A. (1992). Electrical connection of enzyme redox centers to electrodes. J. Phys. Chem. 96, 3579-3587. Huck, H., Schelter-Graf, A., Danzer, J., Kirch, P., & Schmidt, H.L. (1984). Bioelectrochemical detection systems for substrates of dehydrogenases. Analyst, 109, 147-1 50.

408

I. KATAKIS, M. VREEKE, L. YE, A. AOKI, and A. HELLER

Itoh, S., Kinugawa, M., Mita, N.. & Ohshiro, Y. (1989). Efficient NAD’ regeneration system with heteroaromatic o-quinones and molecular oxygen. J. Chem. SOC.Chem. Commun. 694-695. Jacobs, N.J. & VanDemark, P.J. (1960). The purification and properties ofthe a-glycerophosphate-oxidizing enzyme of StreptococcusJuecal. Arch. Biochem. Biophys. 88,25&255. Jones, J.B. & Taylor, K.E. (1976). Nicotinamide coenzyme regeneration. Flavin mononucleotide (riboflavin phosphate) as an efficient, economical, and enzyme-compatible recycling agent. Can. J. Chem., 54,2969-2973. Jonsson-Pettersson, G. ( 199 I). Reagentless hydrogen peroxide and glucose sensors based on peroxidase immobilized on graphite electrodes. Electroanalysis 3,741-750. Katakis, I. & Heller, A. (1992). L-a-Glycerophosphate and L-lactate electrodes based on the electrochemical “wiring” of oxidases. Anal. Chem. 64,1008-lO13. Katakis, I., Ve, L., & Heller, A. (1994). Electrostatic control of electron transfer enabling binding of redox polyelectrolytes and glucose oxidase. J. Am. Chem. SOC.116,3617-3618. Kitani, A., So, Y.-H., & Miller, L.L. (1981). An electrochemical study of the kinetics ofNADH being oxidized by diimines derived from diaminobenzine and diaminopyrimidines. J. Am. Chem. Soc. 103,76367641. Kittlesen, G.P., White, H.S., & Wrighton, M.S. (1985). A microelectrochemical diode with submicron contact spacing based on the connection of two microelectordes using dissimilar redox polymers. J. Am. Chem. Soc. 107,7373-7380. Kulys, J.J. (1986). Enzyme electrodes based on organic metals. Biosensors 2.3-13. Kulys, J.J. & Schmid, R.D. (1990). Mediatorless peroxide electrode and preparation of bienzyme sensors. Bioelectrochem. Bioenerg. 24,305-31 1 . Motonaka, J., Takabayashi, H., Ikeda, S., & Tanaka, N. (1990). Preparation and properties of a micro enzyme sensor for creatine. Anal. Lett. 23, 1981-1991. Nishihara, H., Dalton, F., &Murray, R.W. (1991). Interdigitated array electrode diffusion measurements in donor/acceptor solutions in polyether electrolyte solvents. Anal. Chem. 63,2955-2960. Ottaway, J.H. (1966). Some rate constants for the phenazine methosulfate-catalysed oxidation of reduced nicotinamide-adenine dinucleotide. Biochem. J. 99,253-256. Polis, D. & Shmukler, H.W. (1955). Lactoperoxidase. Meth. Enzymol. 2, 813417. Polster, J. & Schmidt, H.L. (1989). Determination of dehydrogenase substrates by Clark type oxygen electrodes and photosensitized coenzyme oxidation. Talanta, 36,864-866. Shu, C.-F. & Wrighton. M.S. (1988). Synthesis and charge-transport properties of polymers derived from the oxidation of I-hydro-I ‘-(6-(pyrrol-l-yl)hexyl)-4,4’-bipyridinium bis(hexafluorophosphate) and demonstration of apH-sensitive microelectrochemical transistor derived from the redox properties of a conventional redox center. J. Phys. Chem. 92,5221-5229. Smith, D.K., Lane, G.A., & Wrighton, M.S. (1988). Charge-transport properties ofan electrode-confined redox polymer derived from a monomer consisting of a quinone flanked by two benzylviologen subunits. J. Phys. Chem. 92,2616-2628. Surridge, N.A., Zvanut, M.E., Keene, F.R., Sosnoff, C.S., Silver, M., & Murray, R.W. (1992). Effects of mixed-valent composition and bathing environment on solid-state electron self-exchanges in osmium bipyridine redox polymer films. J. Phys. Chem. 96,962-970. Tse, D.C.-S. & Kuwana, T. (1978). Electrocatalysis of dihydronicotinamide adenosine diphosphate with quinones and modified quinone electrodes. Anal. Chem. 50, 13 15-13 18. Van der Meer, R.A., Groen, B.W., Van Kleef, M.A.G., Frank, J., Jongejan, J.A., & Duine, J.A. (1990). Isolation, preparation, and assay of pyrroloquinoline quinone. Meth. Enzymol. 188,260-283. Vreeke, M., Maidan, R., & Heller, A. (1993). Hydrogen peroxide and P-nicotinamide adenine dinucleotide sensing amperometric electrodes based on electrical connection of horseradish peroxidase redox centers to electrodes through a three-dimensional electron relaying polymer network. Anal. Chem. 64,3084-3090.

Electron Conducting Adducts

409

Williams, D.C. I11 & Seitz, R.W. (1976). Automated chemiluminescence method for determining the reduced form of nicotinamide adenine dinucleotide coupled to the measurement of lactate dehydrogenase activity. Anal. Chem. 48, 147S148 1. Ye, L., Hammerle, M., Olsthoom, A.J.J., Schuhmann, W., Schmidt, H.-L., Duine, J.A., & Heller, A. (1993). High current density “wired” quinoprotein glucose dehydrogenase electrode. Anal. Chem. 65,23%241.

This Page Intentionally Left Blank

SCREENING A N D DESIGN OF JMMOBlLlZED BIOCATALYSTS BY MEANS OF KINETIC CHARACTERIZATION ON ENZYME TH ERMISTOR/THE RMAL ASSAY PROB E

Peter Gemeiner, Vladimir Stefuca, and Bengt Danielsson

I. 11. 111. IV.

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . MATHEMATICAL MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . RESULTS AND DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Advances in Molecular and Cell Biology Volume ISB,pages 411-419. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

41 1

412 412 4 13 4 15 418

41 2

P. GEMEINER, V. STEFUCA, and B. DANIELSSON

ABSTRACT Screening and design of immobilized biocatalysts (IMB) is a time-consuming process; ideally, the process should be universal, fast, convenient, precise, and sufficiently reproducible. These requirements are met by enzymic flow (micro)calorimetry (EFMC) also known as “enzyme thermistor” or “thermal assay probe.” Adaptation of EFMC to real measurements of the reaction rates requires coupling of the mathematical description of the reaction-diffusion phenomena in the EFMC column with heat balance and, subsequently, experimental verification of the mathematical model. This paper presents a brief review of the results obtained during the adaptation of EFMC for the characterization of the microkinetic properties of IMB and their further application for screening and design of IMB.

1. INTRODUCTION Complete design of the optimum immobilized biocatalyst (IMB) seems still to be a matter of the future. To be successful, it will require numerical determination of all significant parameters at each enzyme engineering phase, for example, at the design of the carriers, immobilized biocatalysts, and immobilized reactors. Therefore, there is an urgent need for procedures that rationalize the design of optimum IMB in the individual phases (Gemeiner, 1992; Gemeiner et al., 1993a). One procedure is a uniform characterization of the microkinetic properties of IMB. Primary demands for its application are universality of the detection principle, definite system configuration, and facile adaptability to different IMB, as well as a transparent software for the transformation of the experimental data into the real kinetic constants. The majority of the above demands are met by the EFMC in the so-called enzyme thermistor (Danielsson and Mosbach, 1987; 1988; Danielsson, 1990; 1992) originally designed for measurement of metabolite content in the flow-injection analysis mode. Adjustment of EFMC for the direct measurement of the kinetic properties of IMB required, in particular, coupling of the mathematical description of the reaction-diffusion phenomena in the IMB minicolumn with heat balance. A mathematical model has been put forward and gradually verified experimentally (Stehca et al., 1990; Gemeiner et al., 1993b; DoEolomanskf et al., 1994). Adaptation of EMFC to the screening and design of IMB via determination of the microkinetic properties of IMB thus demanded (a) transformation of the experimental thermochemical data into the lunetic data, (b) determination of the real kinetic constants of IMB, (c) screening, and (d) design of IMB by means of the kinetic parameters and constants. This paper presents the results obtained from an adaptation of the flow (micro)calorimetry.

Design of Immobilized Biocatalysts

41 3

II. MATHEMATICAL MODEL Mathematical modeling of microkinetic phenomena in an EFMC is based on chemical engineering principles for the modeling of the reaction-diffusion phenomena in heterogeneous catalytic reactions. Thus, in the case of a reaction catalyzed within the pores of the particles containing IMB, the substrate steadystate mass balance can be written in the following form:

where cs, r, and De are substrate concentration,particle coordinate and the effective difision coefficient of the substrate, respectively. Parameter n depends on the particle geometry and equals 0, 1, or 2 for planar, cylinder and spheric geometry, respectively. The right side of the equation expresses the kinetic form. Solution of Equation 1 enables one to calculate a so-called effectiveness factor, 7, defined as: q=-'obs 'kin

[z)

where the observed reaction rate, vobs,is defined as vobs= A . D, VL

r=R

and represents the total rate ofthe reaction catalyzed by the particles of surface area A in a reactor with liquid volume V,. The kinetic reaction rate is the rate in the reactor with no diffusional limitation present. The particle substrate concentration in this case is therefore the same as the bulk substrate concentration,C,b, and hence Vkin =

V

VL

. V(C,)

(4)

where V, is the total IMB particle volume. Once the effectiveness factor value is known, the observed reaction rate can be calculated from 'obs

= q . 'kin

(5)

When the process in the EFMC is modeled, the introduced mass balance equations should be coupled with mass and heat balances in the EFMC column. The mass balance equations in the EFMC column have been derived, verified experimentally (Stefuca et al., 1990), and coupled with the particle mass balance equations for the case of first order kinetics (Gemeiner et al., 199313). The mathematical modeling was based on the following assumptions:

P. GEMEINER, V. STEFUCA, and B. DANIELSSON

41 4

1. Changes in the substrate concentration and temperature change along the reactor are so small that the reaction rate change is insignificant and the molar reaction enthalpy is constant, 2. plug flow occurs in the reactor, 3. the superficial flow rate is high enough to prevent reaction rate limitation by external mass transfer, and 4. heat losses from the reactor are negligible and the reactor is considered to be adiabatic. Under these assumptions, the steady state substrate balance in the reactor can be presented as follows:

where z, w, and E are the axial coordinate, superficial velocity, and void fraction, respectively; whereas, p, cp, T, and AHf represent fluid density, heat capacity, temperature, and molar reaction enthalpy, respectively. Since the reaction rate in the reactor is considered to be constant,differential equation (6) can be transformed into the following difference equation: dT

AT

dz-Az

(1 - E l 1 1 (-AHf) -rl

w P cp

(8) . rkin

After introduction of the parameter, v,, defined as V, (- AHf) (1 - E) AZ , V

=

w P cp Equation 8 can be simplified to AT=-

rlnr'

rkin

(10)

vln

The meaning of the last equation can be illustrated by an example in which the kinetic term is substituted for the enzyme kinetic equation. Thus, in the case of substrate inhibition kinetics, Equation 10 will appear as follows: AT=q

Vm

S

I<,+ c, + (c%/Ki)

Equation 10 or 1 1 represents a mathematical basis for the formulation of EFMC applications for the investigation of microkinetic properties of an IMB. This

Design of Immobilized Biocatalysts

41 5

requires calibration of the EFMC column, that is, establishment of the relation between the reaction rate, robs,and the thermometric signal, AT. One possible method for processing experimental data is described below.

111. RESULTS AND DISCUSSION One example of the application of the foregoing model is a study of the kinetics of the enzyme immobilized on the surface of the spherical particles of bead cellulose (Stefuca et al., 1990) (Table 1) where no internal diffusion was present (q = 1). Calibration of the EFMC enabled rapid screening of the preparations of the immobilized invertase using its specific activity on a carrier (DoEolomanskg et al., 1993). Incorporation of the internal diffusion term into the model for cases with first order kinetics has been verified using a cell biocatalyst possessing a substantial D-amino acid oxidase activity entrapped in calcium pectate gel (CPG) particles (Gemeiner et al., 1993b) (Table 1). Improved mathematical solution (Satterfield, 1976) using Equation 11 led to a more detailed description of the kinetic data for gel-entrapped cells. This approach was used for the evaluation of kinetic data for a cell biocatalyst using penicillin G acylase activity immobilized in CPG beads (Stehca et al., 1994) (Figure 1). Kinetic data shown in Figure 1 were described by means of Equation 11, whereas the effectiveness factor was calculated by solving the particle mass balance equation 1. Kinetic parameter values were estimated by means of mathematical optimization. The experimental data represented by the pairs (cSbi,ATi)were used and parameter values optimized using a Gauss-Newton nonlinear regression procedure (Ramachandran, 1975; KubiEek, 1983). Once kinetic data are estimated, the mathematical design of the biocatalyst particle can be provided. Plotting the effectiveness factor values as the fhction of

Table 7. Screening of Immobilized Biocatalysts by Means of Microkinetic Characterization Using the Enzyme Flow (Microkalorimeter Biocatulyst

Form

Mode of Immobilization

Origin ~~

Invertase enzyme D-Amino enzyme, acid oxidase cells

~~

Reference

~

Ddolomanskj et al., 1993 S. cerevisiae biospecific 2: variabilisa entrapment + cross-linking Gemeiner et al., 1993b

I: variabilis

cross-linking

Gemeiner et al., 1993b Gemeiner et al., 1993b entrapment + cross-linking Gemeiner et al., 1993b entrapment + cross-linking Welwardova et al., 1993

T variabilis entrapment Penicillin G acylase

2: variubilis enzyme cells E. coli

E. coli E. coli Notes: ‘Trigoflopsisvariabilis

entrapment + cross-linking Welwardova et al., 1993 entrapment Stefuca et al., 1994

P. GEMEINER, V. STEFUCA, and B. DANIELSSON

41 6 I..0 1

8

h

W

6

E

W

4

2

O ! 0

1

I

I

I

I

20

40

60

80

100

Substrate concentration (mM) Figure 7. Measurement of the kinetics of penicillin G hydrolysis catalyzed by the immobilized biocatalyst cells performed in the EFMC column. Cell loading concen39.2; (A) 29.6; (0) 23.7; (0) tration (mg of dry weight per ml of gelling suspension):(0) 11.9.

the parameters that can be modified, that is, particle dimension (R) and biocatalyst concentration the value of which is integrated in the parameter V,, is a powerful tool. This procedure is demonstrated in Figures 2a and 2b. Thus, the optimum particle dimension and the magnitude of the immobilized activity can be designed for a chosen substrate concentration. From Figure 2b one can conclude that for a low substrate concentration, the smaller a particle, the higher its efficiency (optimum Thiele modulus = 0). In the case of a substrate inhibition reaction, however, overstepping of a definite bulk substrate concentration causes inhibition of the reaction by the substrate. Consequently, the mass transfer resistance leads to a decrease in the effective substrate concentration in the IMB particle. Thus, the catalyst effectiveness is increased when a suitable value for the Thiele modulus is generated. This can be obtained by controlling the particle dimension (Figure 2b). This type of study requires reliable experimental data if the results are to correspond to a real situation. However, it is possible to characterize the kinetic properties of the non-immobilized catalyst that enable it to avoid problems associated with the mass transfer interferences. From our experience, however, when results obtained in this manner have been applied to describing the kinetic data of IMB, some discrepancies owing to losses during the immobilization by deactiva-

Design of Immobilized Biocatalysts

41 7

a)

0.0 0

I

I

20

40

1

$0

80

Thiele modulus

Substrate concentration (mM) Figure 2. Reaction - mass transfer phenomena analysis of the spherical biocatalyst particle for substrate inhibited reaction - kinetic model represented by Equation 11. Parameter values were estimated from the experimental data shown in Figure 1 : Km = 1.3 mM, Ki = 570 mM. a. Effectiveness factor as a function of Thiele modulus = R (Vm / Km De)0.5at the substrate concentration (mM) 1, 10, 20, 30,40, 50, 60, 70, 80 ( i i upward direction); b. Optimum Thiele modulus corresponding to the maximum cffectivenessfactor values at varying bulk substrate concentration.

41 8

P. GEMEINER, V. STEFUCA, and B. DANIELSSON

tion, inclusion of the biocatalyst into the inaccessible parts of the gel structure, changes of the cell wall permeability, or biocatalyst modification by the covalent linkage were observed. For this reason a more preferable method is to study the biocatalyst properties by means of the kinetic measurements after the immobilization. The EFMC in connection with the mathematical model described above provides a powerful technique for obtaining essential experimental data extremely rapidly and easily.

IV. CONCLUSION The EFMC provides a powerful tool for rapid and convenient investigation of the kinetic properties of IMB. Such investigations include measurement of the reaction rate that depends on the substrate concentration with the aim to determine kinetic parameters. As soon as the reaction rate is established, many very precise kinetic measurements can be carried out in one day. This procedure enables optimum biocatalyst configuration design.

ACKNOWLEDGMENT We are indebted t o Professor Klaus Mosbach for inspiring discussions. O n e of us (P.G.) is indebted to The Weizmann Institute of Science, Rehovot, Israel, for the Max Varon Professorship that gave him the excellent opportunity to consult on problems and gain further insight into the literature. This work w a s supported, in part, by the Slovak Grant Agency for Science (Grant No. 2/1238/95).

REFERENCES Danielsson, B. & Mosbach, K. (1987). In: Biosensors: Fundamentals and Applications (Turner, A.P.F., Karube, I. & Wilson, G.S., Eds.), pp. 575595. Oxford University, Oxford. Danielsson, B. & Mosbach, K. (1988). In: Methods in Enzymology: Vol. 137, Pt. D: Immobilized Enzymes and Cells (Mosbach, K., Ed.), pp. 181-197, San Diego. Academic Press. Danielsson, B. (1990). Calorimetric biosensors. J. Biotechnol. 15, 187-200. Danielsson, B., Hedberg, U., Rank, M., & Xie, B. (1992). Recent investigations on calorimetric biosensors. Sensors & Actuators B 6, 13S142. DoEolomanskg, P., Gemeiner, P., MisloviEovB, D., Stefuca, V., & Danielsson, B. (1994). Screening of Concanavalin A - bead cellulose conjugates using an enzyme thermistor with immobilized invertase as the reporter catalyst. Biotechnol. Bioeng. 43,286-292. Gemeiner, P. (1992). In: Enzyme Engineering. Immobilized Biosystems (Gemeiner, P., Ed.), Chapter 4.2, Ellis Horwood Series in Biochemistry and Biotechnology (Wiseman, A., Ed.), Ellis Horwood, Chichester & Alfa Publishers, Bratislava. Gemeiner, P., Stefuca, V., & Bale;, V. (1 993a). Biochemical engineering of biocatalysts immobilized on cellulosic materials. Enzyme Microb. Technol. 15, 55 1-566. Gemeiner, P., Stefuca, V., Welwardova, A,, Michalkova, E., Welward, L., Kurillova, L., & Danielsson, B. (1993b). Direct determination of the cephalosporin transforming activity of immobilized cells with use of an enzyme thermistor. I. Verification of the mathematical model. Enzyme Microb. Technol. 15, 30-35.

Design of Immobilized Biocatalysts

41 9

KubiEek, M. (1983). Numerickt Algoritmy ReSeni Chemicko-Inienirskjch Uloh (Numeric Algorithms for Solutions of Problems in Chemical Engineering) (in Czech), Stitni nakladatelstvi technicke literatury (SNTL-Alfa), Prague. Ramachandran, P.A. (1975). Solution of immobilized enzyme problems by collocation methods. Biotechnol. Bioeng. 17.21 1-226. Sattertield, Ch.N. (1976). MassoperedaEa v geterogennom katalize (Mass Transfer in Heterogeneous Catalysis) (in Russian), p. 133, Izdatelstvo Chimija, Moskva. Stehca, V.. Gemeiner, P., Kurillova, i.,Danielsson, B., & Bales, V. (1990). Application ofthe enzyme thermistor to the direct estimation of intrinsic kinetics using the saccharose-immobilized invertase system. Enzyme Microb. Technol. 12,83&835. Stefuca, V., Welwardova, A., Gemeiner, P., & Jakubova, A. (1994). Application of enzyme flow microcalorimeter to the study of microkinetic properties of immobilized biocatalyst. Biotechnol. Tech. 8,497-502. Welwardova, A,, Gemeiner, P., Michalkova, E., Welward, L.. & Jakubova. A. (1993). Gel-entrapped penicillin G acylase optimized by an enzyme thermistor. Biotechnol. Tech. 7, 80- 14.

This Page Intentionally Left Blank

AMPEROMETRIC B IOSE NSORS BASED ON CARBON PASTE ELECTRODES C HEMICA L LY MODIF IE D WITH REDOX-ENZYMES

L. Gorton, G. Marko-Varga, B. Persson, Z. Huan, H. Lindbn, E. Burestedt, S. Ghobadi, M. Smolander, S. Sahni, and T. Skotheim

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. EXPERIMENTAL.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Carbon Paste Electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . B. L-Lactate Sensor Based on L-Lactate Dehydrogenase . . . . . . . . . C. Aldose Sensor Based on Aldose Dehydrogenase . . . . . . . . . . . D. Alcohol Sensor Based on Alcohol Oxidase Coimmobilized with Horseradish Peroxidase . . . . . . . . . . . . . . . . . . . . . . . . E. D - h i n 0 Acid Sensor Based on Coimmobilized D - h i n o Acid Oxidase and Peroxidase . . . . . . . . . . . . . . . . . . . . . . . . . . Advances in Molecular and Cell Biology Volume 15B, pages 421-450. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 1-7623-0114-7

42 1

. .

422 422 423 423 . 424 . 424 425 425

L. GORTON ET AL.

422

F. L-Glutamate Sensor Based on Coimmobilized L-Glutamate .421 Oxidase and Peroxidase . . . . . . . . . . . . . . . . . . . . . . . . . 111. RESULTS AND DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . ,427 A. L-Lactate Sensor Based on L-Lactate Dehydrogenase . . . . . . . . . . .427 B. Aldose Sensor Based on Aldose Dehydrogenase . . . . . . . . . . . . . .432 C. Sensors Based on Coimmobilized Hydrogen Peroxide Producing .434 Oxidases and Peroxidases . . . . . . . . . . . . . . . . . . . . . . . . . D. Alcohol Sensor Based on Coimmobilized Alcohol .438 Oxidase and Peroxidase . . . . . . . . . . . . . . . . . . . . . . . . . . E. D-Amino Acid Sensors Based on Coimmobilized D - h i n O Acid ,440 Oxidase and Peroxidase . . . . . . . . . . . . . . . . . . . . . . . . . . F. L-Glutamate Sensor Based on Coimmobilized L-Glutamate .445 Oxidase and Peroxidase . . . . . . . . . . . . . . . . . . . . . . . . . .

ABSTRACT A number of redox enzymes have been immobilized in carbon paste electrodes operating around 0 mV versus saturated calomel electrode (SCE). Examples will be given of an L-lactate sensor based on L-lactate dehydrogenase, an aldose sensor based on aldose dehydrogenase, alcohol sensors based on coimmobilized alcohol oxidase and peroxidase, D-amino acid sensors based on coimmobilized D-amino acid oxidase and peroxidase, and L-glutamate sensors based on coimmobilized L-glutamate oxidase and peroxidase. The positive effects on the sensor performances upon the addition of various amine-containing polymers will be demonstrated.

1. INTRODUCTION Enzymes have been used in conjunction with various electrodesfor the construction of enzyme electrodes for more than two decades since the first enzyme electrode was reported by Updike and Hicks (1967). The use of amperometnc enzyme electrodes is rationalized by the proposed use of the inherent selectivity shown by the enzyme to promote a selective detection of the enzyme substrate. However, in most instances the necessary applied potential of the enzyme electrode is either too low o r too high to allow the electrochemical reaction to occur without interfering reactions or excessive background currents. The optimal potential range for an amperometric biosensor to promote sensitive and selective detection should be between -200 and 0 mV versus SCE where the background current switches signs and thus takes its lowest value. Electrochemical reduction of molecular oxygen and oxidation of easily oxidizable species (e.g., ascorbate,urate, paracetamol, catecholamines, etc.) are negligible and do not contribute to the response signal. Oxidoreductases or redox-enzymes are of particular interest for the construction of amperometric enzyme electrodes because an electron transfer reaction takes place in the enzymatic conversion of the substrate. Many investigationshave been

423

Amperometric Biosensors

performed to try to obtain direct electron transfer between the redox cofactor of the oxidoreductase and an electrode at a low overpotential (Frew and Hill, 1987; Armstrong et al., 1988). In most instances, however, a direct electron transfer is hindered because steric or kinetic barriers prevail. To circumvent these effects, small molecules acting as charge transfer mediators can be used to shuttle the charge to/fiom the active site ofthe redox-enzyme fiom/to the electrode. The great interest shown in the construction and studies of chemically modified electrodes (CMEs) has provided an impetus to recent developments in the immobilization of active mediators and enzymes on electrodes (Murray et al., 1987; Baldwin and Thomsen, 1990;Labuda, 1992;Wing and Hart, 1992).The use of mediators and mediator-modified electrodes for amperometricbiosensors was reviewed by Bartlett et al. (199 1). Immobilization of the enzyme in close conjunction either directly on the electrode surface or in a composite electrode are recent ways to promote fast overall reaction kinetics in contrast to immobilizing the enzyme in a separate membrane, which is after fabrication put in place on the electrode surface. In most cases, however, these electrodes suffer from not having long-term stability. Some attempts have been made to stabilize the immobilized enzyme in amperometric biosensor configurations(Bahulekar et al., 1991;Gorton et al., 1992a;Kacaniklic et al., 1994). However, only recently has systematic research been focused on how to stabilize the immobilized enzyme (Gibson et al., 1992; Gibson and Woodward, 1992). Similarly, little attention has been paid to how to fbrther speed up the overall reaction kinetics of the enzyme probe. This paper gives details on some recent results obtained for a number of enzymes of different classes of redox-enzymes immobilized either on or in graphite paste electrodes, all operating within the optimal potential range. Some results are also presented on the effect of adding polymeric stabilizers to the paste, thereby also enhancing the reaction kinetics of the probes.

II. EXPERIMENTAL A. Carbon Paste Electrodes

Plain graphite-paraffin oil paste was prepared by thorough mixing of 40 pl of paraffin oil (Fluka, cat. no. 76235) with 100 mg of graphite powder (Fluka, cat. no. 50870). Plastic syringes (1 .Oml syringe, Brunswick 81/79503,with a tip of 7.0 mm OD and 1.8 mm ID) were filled with the graphite-paraffin oil paste leaving about 3 to 4 mm empty in the top to be filled with enzyme modified carbon paste for producing the final electrode. Electric contact was provided by inserting a gold or a silver thread into the paste. After aliquots of enzyme modified pastes (see below) were filled into the ends of the syringes, the end was gently rubbed on glass to produce a flat shining electrode surface with an area of about 0.024 cm2.The final electrodes were mounted in a flow-through amperometric cell of the wall-jet type (Appelqvist et al., 1985) under three electrode potentiostatic control using a

potentiostat (Zata Electronics, Lund, Sweden). When not in use the enzyme electrodeswere stored in a dry state at 4 "C. Aplatinum wire and a saturatedcalomel electrode (SCE) served as the counter and reference electrodes, respectively. The cell was connected to a single line flow injection (FI) system using either an Automated Sample Injection Analyser (Ismatec, SA, Glattburg-Ziirich, Switzerland) or an in-lab made system described earlier (Appelqvist et al., 1985).Connections between the various parts of the FI systems were made with Teflon tubings, ID 0.5 mm, and Altex screw couplings. In one instance a liquid chromatographic system was used. In this case a liquid chromatographic pump (model 2 150, LKB, Bromma, Sweden) delivered a 0.05 M phosphate buffer at pH 8.0 with a flow rate of 0.5 ml m i d . An injector (Reodyne, model 7000, injection volume 15.4 pl) and a polymer separation column (PLRP-S, 5 pm, 100 A, 100 x 0.4 mm ID, Church Stretton, Shropshire, UK) were used. All solutions were degassed before use to prevent microbubble formation in the flow system. B. L-Lactate Sensor Based on L-Lactate Dehydrogenase

L-lactate dehydrogenase, EC 1.1.1.27, type XI, from rabbit muscle (LDH, Sigma Chem. Co., cat. no. L-1254) was obtained as a lyophilized powder with an activity of 7 10 U mg-' solid and 729 U mg-' protein. The chemically modified carbon paste was manufactured by first thoroughly mixing 4% (wlw) of the redox polymer (microcrystalline powder) containing covalently bound Toluidine Blue 0 as the mediating active moiety with graphite (see Results and Discussion). One hundred mg of this powder was added to 1.O ml of 0.1 M sodium phosphate buffer at pH 7.0 containing 2.3 mg of LDH, 50 mg of NAD' (Sigma Chem. Co., Grade 111, cat. no. N-15 1l), and 100 pl of a 2.6 %aqueous polyethylenimine (PEI) solution (Sigma Chem. Co., cat. no. P-3 143, 50% aqueous solution). After stirring for 2 h at 4 "C, the mixture was dried in a desiccator under reduced pressure. Finally, a paste was obtained by mixing the dry chemically modified graphite with 40 pl of paraffin oil. Some properties of the Toluidine Blue 0 (TBO) containing polymer for the electrocatalytic oxidation of NADH (Sigma Chem. Co., cat. no. N-8129) were investigated when used as a modifier on solid graphite electrodes (RingsdorffWerke, GmbH, type RW 00 1). Cyclic voltammetry (BAS Electrochemical Analyzer, 100 B) and a rotating ring disk device (Tacussel,model EDI) using a platinum gauze and an SCE as the counter and reference electrodes, respectively. The solid graphite electrodes were pretreated as reported previously (Persson, 1990) and the polymer coating was obtained by drop coating the electrode surface with a dimethylsulfoxide polymer solution and allowing the solvent to evaporate at reduced pressure for 4 h. C. Aldose Sensor Based on Aldose Dehydrogenase

D-aldose dehydrogenase (ALDH, from Gluconobucter oxydans. ATCC 62 1) (no EC number given yet) was purified as described previously (Smolander et al.,

Arnperometric Biosensors

42 5

1993a). The enzyme was stored in 10 mM sodium acetate at pH 5.0 containing at least 0.1% Triton X- 100 (iso-octylphenoxypolyethoxyethanol,BDH, cat. no. 30632). Plain carbon paste electrodes were prepared as above by mixing 100 mg of heat treated graphite powder with 40 p1 of parafin oil. Some of the electrodes were also modified with the addition of an electron transfer mediator by adding 2.0 mg of ferrocene monocarboxylic acid (Sigma Chem. Co., cat. no. F-2641) to 100 mg of graphite prior to the addition ofthe pasting liquid. Aportion of 10 pl ofALDH (glucose oxidizing activity; 7 U ml-') was added to the surface of a carbon paste electrode and allowed to dry in air. Some electrodes were prepared by mixing 400 p1 of ALDH (6 U ml-') with 100 mg of heat treated graphite powder and 2 mg of ferrocene monocarboxylic acid and letting the enzyme immobilize for 2 h at 4 "C. After that, the mixture was allowed to dry in a desiccator under reduced pressure before the addition of 40 p1 of parafin oil as the pasting liquid. D. Alcohol Sensor Based on Alcohol Oxidase Coimmobilized with HRP

Three.differentalcohol oxidases were investigated. The amount of enzyme used was adjusted so that each graphite paste contained 137 U of alcohol oxidase (AOD, EC 1.1.3.13).Horseradish peroxidase (1.37 mg) (HFW, Sigma Chem. Co., cat. no. P-8375,288 U mg-I solid) was dissolved in 200 pl of a 0.1 M phosphate buffer at pH 7.0. Alcohol oxidase (1 13.0 p1) (from Cundidu boidinii, obtained as a solution in phosphate buffered 60% sucrose from Serva, cat. no. 12085,30.3 U mg-l protein) and 100 mg of heat treated graphite powder were added to the solution. While stirring, 200 pl of a 0.32% aqueous solution of PEI in 0.1 M phosphate buffer at pH 7.0 were added. The mixture was stirred for 16 h at 4 "C and thereafter dried in a desiccator under reduced pressure for 4.5 h. Finally, 40 pl of phenylmethylsilicon oil (Silicone DC 710, Alltech Associates, Arlington Heights, IL, USA) (Gorton et al., 1992a)were thoroughly mixed with the enzyme-modified graphite. To a second batch, 125 pl of AOD from Pichia pastoris (obtained as a solution in phosphate buffered 60% sucrose from Sigma Chem. Co., cat. no. A-2404, 24 U mg-') were used with the same immobilization procedure. To a third batch, 115 mg of AOD from Candida boidinii (obtained as a lyophilized powder from Sigma Chem. Co., cat. no. A-0763, 10 U mg-') were first dissolved in 100 p1 of the phosphate buffer before addition to the HRP, which was dissolved in 150 p1 of the same buffer. The rest of the immobilization procedure was identical to the first one described above. E. D-Amino Acid Sensor Based on Coimmobilized D-Amino Acid Oxidase and Peroxidase

D-aminO acid oxidase (1.0 mg) (D-AAOD, EC 1.4.3.3, from porcine kidney, Sigma Chem. Co., cat. no. A-1789) was dissolved in 1.0 ml of 0.1 M phosphate buffer at pH 7.0. The activity of the D-AAOD was determined according to the following: To a cuvette were added 1.4 ml of a solution containing 4-minoantipyrin (4-AP) and phenol (2.5 mM 4-AP, Merck cat. no. 27128 and 0.17 mM

426

L. GORTON ET AL.

phenol, Merck cat. no. 206) dissolved in 0.1 M Tris buffer (Tris[hydroxymethyl]aminomethane, Merck cat. no. 8382) at pH 8.0,25 pl of a HRP solution [ 1 mg (270 U) were dissolved in 250 p1 of 0.1 M phosphate buffer at pH 7.01, and 1.5 ml of a solution containing 2.5 mM D-phenylalanine (D-Phe, Sigma Chem. Co., cat. no. P-1751, in 0.1 M Tris buffer at pH 8.0). This reaction mixture was incubated in a spectrophotometerat 25 OC for 3 min to reach temperature equilibrium. Thereafter, 35 pl of the D-AAOD solution were added to the cuvette and the increase in the absorbance at 5 10 nm (E = 6 580 1 mol-' cm-') was followed for 5 min. The activity was calculated from the linear part of the slope of the increase in absorbance. The results obtained with four different sets of immobilizations were as follows: Set 1: The following two different D-AAOD carbon paste electrodes were made. To one batch, 1 mg (288 U) of HRP dissolved in 2 16 pl of 0.1 M phosphate buffer at pH 7.0, 1.O ml(2.7 U mg-') of a D-AAOD solution, and 553 p1 of a 0.32% PEI solution were added to coimmobilize on 100 mg of heat treated graphite powder for 2 h by stirring the reaction mixture at 4 "C. Thereafter the reaction mixture was allowed to dry in a desiccator under reduced pressure before addition of 40 p1 of paraffin oil as above. To a second batch, an equivalent enzyme paste was prepared but exchanging HRP for fungal peroxidase (ARP, from Arthromyces ramosus, EC 1.11.1.7, kindly provided as a gift from Suntory Ltd., Japan, lot no. 9005 11, obtained as a lyophilized powder with an activity of 250 U mg-') using 1 mg of the enzyme. Set 2: The following 8 D-AAOD containing pastes were prepared using paraffin oil as the pasting liquid. To 6 batches 1.0 mg (250 U) of ARF' and 1.0 mg (2.7 U mg-') of D-AAOD were added per 100 mg of heat treated graphite powder. To the first batch, 230 p1 of 1.3% PEI were also added; to the second batch, 227 p1 of 1.3% quaternized PEI (kindly obtained from Dr. M. G. Venegas, the Proctor and Gamble Distribution Company, 5299 Spring Grove Avenue, Cincinnati, OH 4521 7-1087, USA); to the third batch, 195 pl of 1.5% ethoxylated PEI (kindly obtained from Dr. M. G. Venegas); to the fourth batch, 220 p1 of Gafquat 734 (a quaternized copolymer of vinylpyrrolidone and dimethylaminoethyl-methacrylatewith an average molecular weight of lo5 Da, CAS reg. no. 53 633-54-8, obtained as a 50% active solution in water, kindly provided as a gift from Dr. T. D. Gibson, University of Leeds, UK); to the fifth batch, 219 p1 of 1.4% chitosan glutamate (Pronova LV chitosan glutamate, Protan Biopolymer NS,Drammen, Norway); to the sixth batch, 230 pl of a 1.3% mixture of 5050 chitosan glutamate and Gafquat 734. To the remaining 2 batches 1.0 mg of ARP and 2.1 mg (1.3 U mg-I) of D-AAOD were added and additionally to the first 97 p1 of 5% gentamicin (Sigma Chem. Co., cat. no. G-1397) and to the second 584 p1 of 1% neomycin (Sigma Chem. Co., cat. no. N-1142). Set 3: The following paste was prepared. 1.0 mg of ARP (250 U), 1.0 mg of D-AAOD (2.2 U), and 300 pl of 1% neomycin added to 100 mg of graphite. Set 4: The following pastes were made: to one batch, 1.0 mg of ARP (250 U), 1.O mg of D-AAOD (2.2 U) and 300 p1 of 1% neomycin were taken to 100 mg of graphite;

Amperometric Biosensors

42 7

and to a second batch, 1 mg of HRP (288 U), 1 mg of D-AAOD (2.2 U) and 300 p1 of 1% neomycin added to 100 mg of graphite.

F. L-Glutarnate Sensor Based on Coimrnobilized L-Glutamate Oxidase and Peroxidase 1.O mg of L-glutamate oxidase (GluOD, from Streptomyces sp., EC 1.4.3.11, kindly provided as a gift from Yamasa Shoyu Co., Ltd., Japan, cat. no. 7804, obtained as a lyophilized powder with an activity of 8 U mg-') and 1.O mg of HRP (EC 1.11.1.7, Sigma Chem. Co., cat. no. P 8375, obtained as a powder with an activity of 270 U mg-') were dissolved in 200 plO.1 M phosphate buffer at pH 7.0. A0.32% solution ofPEI (200 p1) were added to the enzyme solutionwhile allowing the mixture to immobilize on 100 mg of heat treated (700 "C, 15 s) graphite for 2 h at 4 "C. The enzyme-graphite was then allowed to dry under vacuum before addition of the pasting liquid, 40 p1 of paraffin oil. In a second immobilization experiment, L-glutamate oxidase was coimmobilized with ARP (1 mg, 250 U) in an equivalent procedure as described above with HRP.

111. RESULTS AND DISCUSSION Dehydrogenases constitute one major class of redox-enzymes, characterized by catalyzed reactions that are unaffected by molecular oxygen. Dehydrogenases can be further differentiated into two main groups, those with the cofactor strongly bound within the enzyme structureand those depending on a solublecofactor acting as a soluble cosubstrate. Enzymes of the former group contain either a flavin based cofactor alone or in combination with other redox-active compounds such as heme groups, e.g., L-lactate dehydrogenase (cytochrome b2)and diaphorase, or a pyrroloquinolinequinone (PQQ) group. These enzymes were recently discovered and are usually denoted PQQ-dehydrogenases (Duine et al., 1987; Davidson, 1992). Today, several enzymes of this group have been characterizedand are known to be active on glucose, methanol, fructose, L-lactate, and aldose. Some PPQ-dehydrogenases have also been investigated in biosensor configurations (Bartlett. 1991; Gorton et al., 1993; Smolander et al., 1992). Enzymes of the latter group are all dependent on a soluble nicotinamide dinucleotide cofactor for activity with NAD' or NADP+acting as a soluble cosubstrate. They include some 400 enzymes which is the largest group of redox-enzymes known today. A. L-Lactate Sensor Based on L-Lactate Dehydrogenase

The construction of a reagentless amperometric biosensor based on a NAD+dependent dehydrogenase faces three major obstacles. First, as the cofactor is soluble it needs to be immobilized in close proximity to the enzyme so that its continuous addition is ruled out. Second, the electrochemistry of both the oxidized

(NAD') and reduced (NADH) form of the cofactor suffers from high overvoltages and side reactions leading to electrode fouling (Chenault and Whitesides, 1987; Gorton et al., 199la; 1991b; 1992b).Third, the formal potential (E"') of the cofactor redox couple (NAD'NADH) is relatively low, -560 mV vs SCE at pH 7.0, indicating the low oxidizing power of NAD'. This results in that the equilibrium of many dehydrogenase-catalyzed reactions favors the substrate rather than the product formation. The oxidation of L-lactate with LDH is such an example: L-lactate + NAD'

LDH

(>) pyruvate + NADH

(1)

The equilibrium constant for Reaction 1 is 2.76 x lo4 M at pH 7.0 (Barman, 1969). By coimmobilizing both the enzyme and the cofactor in close proximity to the electrode surface, the NADH initially formed in Reaction 1 will be electrochemically oxidized and drive the reaction to the product side. This was demonstrated by Blaedel and Engstrom (1980). We have focused on solving this problem by coimmobilizing the dehydrogenase and the cofactor in carbon paste electrodes and by use of an immobilized redox-mediator added to the paste electrode to facilitate the electrochemical oxidation of NADH to decrease the overvoltage and prevent electrode fouling. The redox mediator in its oxidized form, Med,,, will react with the NADH initially formed in the enzymatic process as follows: NADH + Med,,

+NAD' +

MedEd

(2)

Next, the reduced form of the mediator, Med,,, will be rapidly electrochemically reoxidized if the Eapp,is higher than the E"' of the mediator: Eapp,>Eo'

Med,

+

Med,, + H' + 2e-

(3)

Bycoimmobilizingtheenzyme, thecofactor, and themediator in thesamepaste electrode, the reaction sites for the enzymatic, redox-mediated, and the electrochemical reactions are very closely situated, and apossible unfavorable equilibrium of the enzymatic reaction will be counteracted by the immediate reoxidation of the reduced cofactor initially produced. Thus, a net driving force will be established shifting the enzymatic reaction to the product side. Electrode fouling is circumvented by the use of mediators. The mediators studied by the authors for electrocatalytic NADH oxidation are based on phenoxazine or phenothiazine skeletons incorporating a charged paraphenylenediamine structure essential for their catalytic and reversible electrochemical properties (Gorton, 1987; Gorton et al., 1991a; 1991b; Gorton et al., 1992b). Previous work in this field has included studies on the mechanism of reaction (Reaction 2), on how to increase the reaction rate between the mediator and NADH, and on how to accomplish a stable immobilization of the mediator on or in electrode configurations with retained catalytic properties (Gorton, 1987;

Amperometric Biosensors

429

Bremle et al., 1991; Gorton et al., 1991a; 1991b; 1992b). It was concluded that NADH and Med,, first form a charge transfer complex (CT), which rapidly decomposes into NADf and Me4,,. With these mediators Reaction 2 should therefore be written as: NADH + Med,,

G+

CT -+NAD+ + Med,,

(4)

The E"' values of the mediators fall in the range of -200 and 0 mV vs SCE at pH 7.0. The second order rate constant between NADH and the immobilized mediator is in the order of lo4 to lo5 M-'s-', thus allowing the electrocatalytic oxidation of NADH to proceed in a near mass transfer limited step and below 0 mV versus SCE (Gorton, 1987; Gorton et al., 1991b; 1992b). Most of our studies have been focused on irreversibly adsorbed mediators (Gorton et al., 199lb; 1992b). The phenoxazine and phenothiazine mediators strongly adsorb on carbon electrodes. Synthetic work was done to increase the number of aromatic rings of the mediator to promote adsorption on solid graphite electrodes to reach a state where it can be considered fully irreversible (Gorton et al., 1991b; 1992b). In carbon paste electrodes, however, one portion ofthe mediator seems to be solubilized in the pasting liquid and will slowly leak into the contacting solution, thus decreasing the catalytic activity for NADH oxidation with time (Bremle et al., 1991; Gorton et al., 1991a). We were, however, able to show that phenoxazine and phenothiazine were incorporated into virtually aqueous insoluble polymeric backbones, which could be used as electrode modifiers with retained electrocatalytic activity for NADH oxidation (Gorton et al., 1992b). A new TBOconthining polymer (see Figure 1) has more recently been synthesized. A preliminary E"' value of -100 mV versus SCE at pH 7.0 was obtained from cyclic voltammetry of films cast on solid electrodes, the E"' value being similar to what has been found for other TBO-containing polymers (Persson et al., 1993). The ability of this polymer to electrocatalytically oxidize NADH is shown in Figure 2. Because both the enzyme and the cofactor are aqueous soluble species when added to the paste, precautions have to be taken to prevent them from leaking out into the contacting solution. In some studies the electrode surface was covered by an insoluble membrane (Bremle et al., 1991; Gorton et al., 1991a). Here we report on the possibilities to construct L-lactate sensors based on carbon paste chemically modified with LDH, NAD', PEI, and a phenothiazine based mediator covalently bound to a polymeric backbone. Figure 1 shows a representation of the chemically modified carbon paste electrode, the reaction cycle, and the structural formula of the polymer-bound mediator. The current response for L-lactate and NADH in the concentration range 0.1 to 12 mM using two electrodes prepared from the same batch of chemically modified graphite paste is shown in Figures 3a and 3b. The current response is presented as the average peak height for triplicate FI recordings for electrodes not covered with an outer membrane. The validity of a stable response current for shorter periods of

L. GORTON ET AL

430

PYRUVATE

2e-

NAD+

Polymer red

Enzyme and Cofnclur Modified Carbun Pnste Electrode

+-

+CH,CH+CH,CH-+CH,CH

$ 0 8 +CHz '

CI FHZ

'

CHz-NH c H ' a ; ' n N ( C H , ) ,

c~H,+C,H, c,- CzHs

CI-

Figure 1. Schematic representationof the reaction cycle in an L-lactate sensor based on LDH and NAD' coimmobilized in a mediator-polymer modified carbon paste electrode. The structural formula of the mediator polymer is also given x: 0.28, y: 0.60, and a:0.1 2.

40

30

lo 0 0

1

2 3 [NADH] I m M

4

5

Figure 2. The NADH response current at a rotating polymer modified solid graphite disk electrode (area 0.0731 cm2) in batch of 0.25 M phosphate buffer at pH 7.0. Polymer coating, 1.7 pg (obtained by evaporation of a dimethylsulfoxide polymer solution). Rotation rate: 100 rad s-' . Applied potential: 0 mV vs SCE.

431

Arnperometric Biosensors

0.4

0.3

4 .

.-

0.2

0.1

0

4 6 8 [L-lactate] I mM

2

0

10

12

. .4

0

2

4

6 8 10 [NADH] I mM

12

14

Figure 3. The current response for various concentrations of (a) L-lactate and (b) NADH for two equivalently prepared L-lactate sensors in a FI system. Carrier: 0.25 M phosphate buffer at p H 7.0. Flow rate: 1.1 rnl rnin-'. Injection volume; 50 pl. Applied potential: 0 mV vs SCE.

time using electrodes without a covering membrane was shown earlier using alcohol and glucose sensors based on carbon paste and alcohol and glucose dehydrogenase,respectively (Dominguez et al., 1993; Persson et al., 1993). From Figures 3a and b a linear current response is suggested in the range 0.1 to 1 mM and 0.1 to 2 mM for L-lactate and NADH, respectively. In these ranges, the average sensitivity for two electrodes was 104 nA/mh4 for L-lactate and 1190 nA/mM for NADH, that is, the L-lactate response is about 9% of that for NADH. The peak width at 10% of full peak height, t w,o.l, differed substantiallyfor the two analytes. Typical values in the 5 mM range was 12 s and 40 s for NADH and L-lactate, respectively.

L. GORTON

432

FT AL

As shown earlier (Dominguez et al., 1993; Gorton et al., 1993), the addition of cationic polymers to the paste can increase not only the sensor stability but also the overall observed kinetics of both the electrocatalytic oxidation of NADH and the reaction catalyzed by the dehydrogenase. It should be mentioned that mass transfer effects and the intrinsic kinetics at the paste solution interface cannot be distinguished in our investigations. The redox polymer used for the electrocatalytic oxidation of NADH incorporates positively charged moieties. However, the cationic charge of the redox polymer seems not to be sufficient to activate the reactions because attempts to compare the response current when PEI was not added failed as the L-lactate current response was almost absent. Both L-lactate and NADH have net negative charges at pH 7.0. The different values of tw,O.,observed in the FI recordings for L-lactate and NADH indicate kinetic limitations mainly in the enzymatic reaction. Substrate inhibition by too high an NAD' concentration in the paste may be one reason for this limitation. These results show that LDH can be incorporated into carbon paste electrodes with retained activity, as was previously shown for other NAD+-dependent dehydrogenases, glucose, and yeast alcohol dehydrogenase (Bremle et al., 199 1; Gorton et al., 1991a; Dominguez et al., 1993; Gorton et al., 1993; Persson et al., 1993). B. Aldose Sensor Based on Aldose Dehydrogenase

A PQQ-containing ALDH capable of oxidizing sugars, especially glucose, xylose, galactose, and mannose was reported by Buchert (199 1). This ALDH has been purified with a simple large-scale procedure and characterized hrther by Smolander et al. (1993a). ALDH catalyzes the oxidation of aldoses in the reaction where the cofactor of the enzyme, PQQ, is reduced. The reaction is exemplified by the oxidation of P-D-glucose (as with other glucose oxidizing enzymes we found that ALDH is specific for the p-anomeric form of glucose). ALDH

P-D-glucose + ALDH-PQQ

+

6-gluconolactone + ALDH-PQQH,

(5)

The reoxidation of the reduced enzyme (ALDH-PQQH,) into the oxidized form (ALDH-PQQ) can be accomplished by the use of various electron acceptors, for example, dichlorophenol indophenol, N-methylphenazine, benzoquinone, and ferrocene derivatives; that is, ALDH-PQQH, + Med,,

-+

ALDH-PQQ + Med,

(6)

The charge will next be delivered to an electrode with the necessary applied potential, compare Reaction 2 above. Contrary to the NADH system, both one and two electron acceptors work well with the reduced form of the PQQ-cofactor of these dehydrogenases (Bartlett et al., 1991; Davidson, 1992).

Amperometric Biosensors

433

a

M

bP

50nA

250nA

-

0

10 20 t i min

Figure 4. Peak arising on the injection of 5 m M glucose into the FI system containing an ALDH electrode as a working electrode. 50 m M sodium phosphate buffer at pH 6.5 was used as the carrier and 200 mV versus Ag/AgCI was used as the reference electrode. (a) response obtained with ALDH adsorbed on the carbon paste electrode with no mediator; (b) response obtained with ALDH mixed into the carbon paste containing 2% ferrocene carboxylic acid, and (c) response obtained with ALDH adsorbed on the carbon paste electrode containing 2% ferrocene carboxylic acid.

ALDH has been previously used for the construction of enzyme based detection devices, either contained in an immobilized enzyme reactor (Smolander et al., 1993b) or in amperometric biosensors by its covalent immobilization on a graphite electrode using adsorbed dimethylferrocene or soluble N-methylphenazine as mediators (Smolander et al., 1992). Here, in a first attempt, we tried to immobilize ALDH in plain graphite paste, but no direct electron transfer could be obtained in the absence of a mediator. However, when ferrocene monocarboxylic acid was included in the paste, a current response could be obtained by the injection of glucose into the FI system (see Figure 4). It was previously discovered that a membrane-bound fructose dehydrogenase is not catalytically active when immobilized into the carbon paste, but is capable of a direct communication with the electrode when adsorbed on the surface of a plain carbon paste electrode (Gorton et al., 1993). The same phenomenon was seen with ALDH. A concentration-dependentdirect catalyticcurrent for glucosewas obtained with ALDH immobilized on the surface of a plain carbon paste electrode (Figures 4 and 5). Apotential of +200 mV versus Ag/AgCl was required to obtain this current response indicating a difference in the electron transfer mechanism compared with fructose dehydrogenase for which the response for fructose was noticed previously at around 0 mV (Gorton et al., 1993). When ALDH was adsorbed on a graphite paste containing 2% ferrocene monocarboxylic acid, the catalytic currents obtained were around 100-fold higher than

relative response (YO) *0°

? _ _ _.../.. _.

0.2

_ _. _ .

. . .. . .. . . . , . . _ . . . . . . .. . . . . .

. . . ..I

I

I

I

I

I

I

I

I

0.5

1

2

5

10

20

50

100

glucose (mM) Figure 5. Direct electron transfer between PQQ-dependent ALDH and carbon paste electrode; current response as a function of glucoseconcentration. The measurements, were performed in an FI system and the electrode potential was +200 mV versus Ag/AgCI and a 50 mM phosphate buffer at pH 6.5 was used as the carrier. (0) and (A) denote peak height and peak area, respectively.

those obtained without a mediator (Figure 4). This clearly indicates the nature of the electron transfer as a limiting factor in the sensor performance. The dependence of the current on the concentration of glucose and xylose is shown in Figure 6. In Figure 7 is shown the response current to glucose injections (50 FL, 0.5 mM) with the applied potential of the ferrocene monocarboxylic acid containing carbon paste electrode with adsorbed ALDH on the surface. The E"' of the mediator was evaluated with cyclic voltammetry and found to be + 300 mV versus Ag/AgCl. The figure shows that at +200 mV maximum response current is already obtained, even though not all of the mediator is in its active oxidized form. C. Sensors Based on Coimmobilized Hydrogen Peroxide Producing Oxidases and Peroxidases

All oxidases depend on a cofactor strongly bound within the enzyme structure. The nature of the cofactor may be of different chemical structure (e.g., flavins or copper-containing structures). What they all share, and in contrast to the dehydrogenases, is that they make use of molecular oxygen as the natural reoxidation agent in the enzyme catalyzed reaction. Depending on the class of oxidase, molecular oxygen is either reduced to form hydrogen peroxide or water. Well-known hydro-

Amperometric Biosensors

435

10,000 I

8,000

6.000

.-

Y

4,000

2,000

0

glucose (mM) 700

b 600

500 h

P.- 4oa v

300

200

100

0

I

I

I

I

I

10

20

30

40

50

60

xylose (mM) Figure 6. Current response as a function of aldose concentration: (a) glucose and (b) xylose. ALDH was adsorbed on the carbon paste electrode containing 2% ferrocene carboxylic acid as a mediator. Measurement potential was +300 rnV vs Ag/AgCI and a 50 rnM phosphate buffer at pH 6.5 was used as the carrier.

relative response (YO)

c .,

120 100 -

80

?-.

+3

-

60 -

40

-

0 0

I

I

I

I

I

I

50

100

150

200

250

300

350

measurement potential (mV) Figure 7. Current response for 0.5 mM glucose as a function of the applied potential. ALDH was adsorbed on the carbon paste electrode. A 50 mM phosphate buffer at pH 6.5 was used as the carrier.

gen peroxide producing oxidases are glucose, L- and D-amino acids, galactose, alcohol, and choline oxidase, and water producing oxidases, such as, tyrosinase and ascorbate oxidase (Bartlett et al., 1991; Gorton et al., 1991b). As molecular oxygen is a strong oxidizing agent, all oxidase catalyzed reactions can be regarded as chemically irreversible (c.f. above Reaction 1). A general reaction for an oxidase catalyzed reaction can thus be written: oxidase

substrate + 0,

+

product + H,O, (or H,O)

(7)

From Reaction 7 it is obvious that the reaction can be followed electrochemically either by sensing the decrease in oxygen tension or by the increase in hydrogen peroxide concentration. The necessary applied potentials to either of the electrochemical detection reactions (reduction of molecular oxygen at 4 0 0 mV or oxidation of hydrogen peroxide at +600 mV) are, however, far too low or too high to allow very sensitive and/or interference free detection of the substrate for the oxidase. Much research has therefore been focused on trying to amperometrically follow oxidase based reactions at more optimal potentials (Bartlett et al., 1991). This can be achieved for example by the use of chemically modified electrodeswith electrocatalytic activity for hydrogen peroxide oxidation (Baldwin and Thomsen, 1990; Gorton et al., 1991b; 1992b) or with mediating functionalities acting as alternative electron acceptors to molecular oxygen (Bartlett et al., 1991; Gorton et al., 1991b).

Amperometric Biosensors

437

Horseradish peroxidase and other peroxidases have often been used in conjunction with oxidase based reactions to promote selective and sensitive detection of the substrate for the oxidase, be it spectrophotometrically or electrochemically (Gorton et al., 1991b). Recently it was found that an apparent direct and very efficient electron transfer could be obtained primarily between various carbon electrode materials and immobilized peroxidases in the presence of hydrogen peroxide (Yaropolovet al., 1979; Jonsson and Gorton, 1989; Paddock and Bowden, 1989;Kulys and Schmid, 1990; Kulys et al., 1991;Wollenbergeret al., 1990,1991; Gorton et al., 1991a, 1991b; Tatsuma and Watanabe, 1991; Csoregi et al., 1993a, 1993b). In the presence of hydrogen peroxide a bioelectrocatalytic reduction current of hydrogen peroxide starts to appear at about +600 mV versus SCE at pH 7.0. As the Eapp,is made more negative, the response current increases and reaches a steady state plateau at about -150 mV. The mechanism for this behavior, a truly direct electron transfer or a mediated one brought about by oxygen containing functionalities (quinones) is much debated (Yaropolov et al., 1979; Paddock and Bowden, 1989; Kulys and Schmid, 1990; Wollenberger et al., 1990; Tatsuma and Watanabe, 1991; Csoregi et al., 1993a). Peroxidases contain a heme group, most often ferriprotoporphyrin IX, as the redox-active prosthetic group contained in the active site. When the native form of the peroxidase reacts with hydrogen peroxide, the prosthetic group is oxidized in a single two-electron step, transforming the native form into compound-I, that is,

+ compound-I

native form + H,O,

(8)

The rereduction of compound-I to the native form occurs in two separate oneelectron steps with the intermediate formation of compound-11: compound-I + 1 e-

-+

compound-I1+ 1 e-

compound-I1

(9)

+ native form

(10)

Obviously Reactions 9 and 10 can occur at the electrode surface without the deliberate attachmentof mediating fimctionalities. In the presence of high concentrations of hydrogen peroxide there is a risk of an irreversible deactivation of the peroxidase due to the transformation of the prosthetic group into a higher oxidation stage. This form of the peroxidase is denoted compound-

m.

Peroxidase-modified electrodes were shown to work successfully as electrochemical sensors for hydrogen peroxide monitoring within the optimal potential range for an amperometric sensor. Several papers have appeared regarding the construction of such sensors based on a solid (Jonsson and Gorton, 1989; Paddock and Bowden, 1989; Kulys and Schmid, 1990; Kulys et al., 1991; Wollenberger et al., 1990; Tatsuma and Watanabe, 1991; Csoregi et al., 1993a), carbon fibers (Gorton et al., 1992a; Csoregi et al., 1993b), carbon paste (Gorton et al., 1991b,

Figure 8. Schematic representationof the reaction sequence of a sensor based on an oxidase coimmobilized with peroxidase in carbon paste.

1992a; Wollenberger et al., 1991), or composite electrodes (Wollenberger et al., 1991). A number of papers also reported the coimmobilization of a hydrogen peroxide producing oxidase with a peroxidase on a solid (Kulys and Schmid, 1990; Kylus et al., 1991; Gorton et al., 1991a; Jonsson-Pettersson, 1991) or in carbon paste electrodes (Gorton et al., 1992a, 1993; Johansson et al., 1993; Kacaniklic et al., 1993). Figure 8 shows the reaction cycle for a carbon paste electrode containing a hydrogen producing oxidase coimmobilized with HRP.From the group in Lund, glucose (Gorton et al., 1992a), alcohol (Gorton et al., 1992a; Johansson et al., 1993), L-amino acid, and D-amino acid (Gorton et al., 1992a;Kacaniklic et al., 1993) sensors based on the corresponding oxidase coimmobilized with HRP in carbon paste electrodes were reported. As in the case of LDH, the addition of PEI was shown to very efficiently promote stability and high reaction kinetics. Here we report sensors for alcohols, D-amino acids, L-glutamate, on the reaction cycle shown in Figure 8 and using alcohol, D-amino acid, and L-glutamate, coimmobilized with either H R P or peroxidase from Arthromyces rumosus (ARP)(Shinmen et al., 1986), respectively. A series of analytes, of high interest in food, medical, and biotechnological areas, can be selectively determined with these sensors. The examples chosen in this paper illustrate that D-amino acid oxidase and alcohol oxidase are group selective rather than one-substrate specific enzymes. Electrodes based on these enzymes should find use as semiselective bioelectrochemical detectors in conjunction with column liquid chromatography (Yao and Wasa, 1988; Johansson et al., 1993; Marko-Varga et al., 1993). Enzyme electrodes based on L-glutamate oxidase, however, are much more one-substrate specific sensors that need no separation step prior to detection and should find use as sensors in flow injection. D. Alcohol Sensors Based on Coimmobilized Alcohol Oxidases and Peroxidases

Alcohol oxidase (AOD) is a flavin containing enzyme active for a variety of alcohols and similar substrates containing hydroxyl groups. When the enzyme oxidizes the substrate, the corresponding aldehyde and the reduced form of the

Amperometric Biosensors

439

enzyme are produced. Molecular oxygen serves as the natural electron acceptor for the reduced form of the enzyme returning it to its oxidized state whereby hydrogen peroxide is produced. The net reaction, exemplified below by the oxidation of ethanol, is as follows (c.f Reaction 7): AOD

ethanol + 0, + acetaldehyde + H,O,

(11)

We previously showed that AOD from Candida boidinii (obtained from Serva) can be coimmobilized with HRP in carbon paste electrodes (Gorton et al., 1992a; Johansson et al., 1993; Marko-Varga et al., 1993). Of great importance for the performance, and both sensitivity and long-term stability, was the addition of PEI to the immobilization reaction mixture (Gorton et al., 1992a).It was also found that using a combination of covalent reagents, carbodiimide and glutaraldehyde, was beneficial for the stabilization of the sensor. As stated above, the enzyme is group rather that strictly substrate selective, and a sensor based on coimmobilized AOD/HRP was shown to be a very good semiselective sensor in liquid chromatography for the determination of alcohols and related analytes. In earlier reports (Johansson et al., 1993; Marko-Varga et al., 1993; Buttler et al., 1993), we showed that with this type of sensor, the production of ethanol in a biotechnological process is selectively followed chromatographically on-line for 24 h without fouling the electrode. In these reports, the AOD used was obtained from one manufacturer. To further investigatethe selectivityof sensorsbased on the combination ofAOD and a peroxidase, we report here on results obtained with two isoenzyme preparations from Candida boidinii but from different sources and one preparation from Pichia pastoris. In this paper, the use of covalent immobilization reagents are omitted and the only reagent added to the graphite, except the pasting liquid and the enzymes, is PEL Table 1 shows the relative responses of a set of three different electrodesprepared with the different isoenzymes for a variety of possible substrates. As can be seen, there are a number of differences in performance of these equivalently prepared electrodes. A difference in selectivity between the isoenzymes from Candida and Pichia is expected. The two sets of preparations based on enzymes from Candida biodinii reveal, however, unexpectedly large differences. No explanation for this difference can be given at this time, unless different strains of Candida boidinii were used by the two manufacturers. Figure 9 shows a chromatogram obtained when the sensor based on AOD from Pichia pastoris was used. The separation of methanol, ethanol, allylalcohol, and 2-chloroethanol is an example of an enzyme-based detection system in LC. A polymer-based separation column was used (PLRP-S) in reversed phase separations with phosphate buffer as the mobile phase. The separation was optimized on a small analytical column (50 x 4.0 mm I. D.) resulting in fast chromatographic separations. Ion exchange phases with similar sizes can be used for the separation

440

Table 1. Comparative Responses of a Number of Substrates and Possible lnterferants with Electrodes Containing Coimmobilized HRP and Three Different lsoenzyrnesof AOD Relative Response 96 Sample (0.5 mM)

ethanol H 2 0 2 (0. I mM) methanol formaldehyde acetaldehyde propionaldehyde butyraldehyde formic acid acetic acid propionic acid butyric acid monochloracetic acid citric acid lactic acid malic acid pyruvic acid inositol triphosphate allylalcohol 2-butanol 2-chloroethanol dihydroxyacetone

Candida boidinii (Serva) Pichia pastoris Candida boidinii (Sigma) 100

100

75 540 22 15 29 <1
189 282 99 c5 94 652 c5 19 56 41 11

151 68 79 22 4.6

159 55 470 24 57


100 866 252 58 10 163 378 <7 34 51 44 60 <7 <7 31 <7 217 46 377 15 57

of other solutes (shown in Table 1). Selective and sensitive separations in combination with catalytic reaction detection can be performed on a large number of analytes. The selectivity, but mostly the sensitivity, can be governed by the choice of sensor listed in Table 1. The choice of possible substrates for these three isoenzymes was chosen based upon the possible applications in the biotechnological and clinical fields. Current studies are being performed for the analysis in many of the listed analytes present in biological fluids like urine and plasma and in fermentation substrates and broths. E. D-Amino Acid Sensors Based on Coimmobilized D-Amino Acid Oxidase and Peroxidase Similar to the case for AOD, D-amino acid oxidase (D-AAOD) is an enzyme with a rather broad specificity. However, it is believed to be an absolute enantiomer specific for the D-amino acids, just as the corresponding L-amino acid oxidase

Amperornetric Biosensors

44 1 1

-

0

5

TIME /min

Figure 9. Chromatogram of 0.5 mM solutions of (1) methanol; (2) ethanol; (3) allylalcohol; and (4) 2-chloroethanol, using an AOD (Pichia pasforidlHRP coimmobilized carbon paste electrode as the sensing device. Chromatographic conditions; mobile phase, 0.05 M phosphate buffer at pH 8.0, flow rate, 0.5 ml min-’, injection volume, 15.4 pl, and an applied potential of -50 mV vs SCE (500 nA full scale).

(L-AAOD) is for the L-amino acids. The net enzymatic reaction can be described as follows: D-AAOD

D-amino acid + H,O

+ 0, -+ 2-0x0-acid + H,O, + NH,

(12)

In previous investigations of the immobilization of both L-AAOD and D-AAOD (Gorton et al., 1992a; Kacaniklic et al., 1994), it was noticed that when coimmobilizing these enzymes in carbon pastes with HRP,the use of covalent immobilization reagents such as carbodiimide and/or glutaraldehyde, resulted in an almost total inactivation of the oxidase. However, as found for many different redox-enzymes (c.f. above) the addition of PEI was beneficial not only for the stability but also for the response for the substrate of the oxidase (Gorton et al., 1992a, 1993; Kacaniklic et al., 1994). An interesting effect was also noticed for the selectivity of L-AAOD. When immobilized within the organic nature of carbon paste, the enzyme is much more active for a number of L-amino acids shown to have no or little reactivity with the enzyme in aqueous solutions or when immobilized on an inert hydrophilic carrier (Kacaniklic et al., 1994). In aqueous solutions or when immobilized on an inert carrier, D-AAOD is reported not to be highly active for all the biologically most important 20 amino acids (Dominguez et al., 1990). Previous preliminary investigations on the immobilization of D-AAOD in carbon pastes were not as successful as with the corresponding L-AAOD in terms ofresponse and stability (Kacaniklic et al., 1994). With the prospect of constructing a highly responding and long-term stable D-amino acid selective sensor based on coimmobilization ofD-AAOD and a peroxidase in carbon

Table 2. FI responses to 0.1 mM hydrogen peroxide (i (H,02)/nA) and 1 .O m M D-phenylalanine (i(D-Phe)/nA)for coimmobilized D-AAOD/HRP and D-AAOD/ARP, respectively. Each value is the average response obtained for two equivalent electrodes. Flow parameters: carrier, 0.1 M Tris buffer at pH 8.0, 0.6 ml min-’, injection volume, 50 PI, applied potential -50 mV versus Ag/AgCI. The electrodes used are those of set 1 described on p. 424. Coimmobilization

D-AAOD/HRP D-AAOD/ARP

i (H202)/nA

i(D-Phe)/nA

562 503

88 154

paste, we here report on the effects of the addition of a number of different cationic polymeric additives to the immobilization reaction mixture and on the choice of peroxidase. The choices of additives and peroxidases are based on previous results by others as well as by the authors (Atmstrong and Lannon, 1987, 1988; Kulys and Schmid, 1990; Wollenberger et al., 1991; Gibson et al., 1992; Gibson and Woodward, 1992; Gorton et al., 1992a, 1993). Table 2 shows the effect of the chosen peroxidase (HRP or ARP) on the response to H202and D-phenylalanine, a substrate with a high turnover rate with D-MOD, when PEI was used as the additive. As can be seen, almost equal amounts (in units) of HRP or ARP applied to the paste reveal similar responses to H20,, whereas the response

Table 3. FI responses to 0.1 mM hydrogen peroxide (i (H,O,)/nA) and 1.O m M D-phenylalanine (i(D-Phe)/nA)for differents additives. Each immobilization is an average of three equivalent electrodes. All electrodes are based on 1 .O mg of D-AAOD, 1.O mg of ARP and 3.0 mg of additive except the ones containing gentamicin and neomycin. The added amounts in these cases were 2.1 mg of D-AAOD, 1 .O mg of ARP, 5.0 mg of gentamicin, and 6.0 mg of neomycin. For further details, see electrodes of set 2 in Experimental. Flow parameters: carrier, 0.1 M Tris buffer at pH 8.0, 0.6 ml min-’, injection volume, 50 pl, applied potential -50 mV vs AglAgCI. Additive

polyethyleneimine ethoxylated polyethyleneimine quaternized polyethyleneimine Gafquat 734 chitosan glutamate gentamicin neomycin mixture of 50% chitosan glutamate and 50% Gafquat 734

i (H202)/nA

i(D-Phe)/nA

313 178 147 442 522 I93 222 191

84 57 43 95 66 89 106

53

Polyethylenlmlne

Gsfaust 734

400

400

350

350

300 4

5

2 - 300 250

250

4 200 I,

0; 200

E

150

'-

150

100 50

loo 50

0

0 0

20

P

P

40

60 80 Time / min

100

120

140

1 0

20

40

60 80 Time I min

100

120

Gentamicin

w

240

100

200

80

.

2

_. I

160

60

p

-3

0 : 150 I .-,

2 lD

-

$120

.-

40

80

.

0

200

100

20

40

0

0

20

60 Time / min

40

80

0 100

0

20

40 60 Time I rnin

80

100

Figure 10. Stability of a series of D-amino acid sensors using different promoters. The electrodes used were those described as set 2 additives on p. 424. Responses for (0)hydrogen peroxide and (0) D-phenylalanine are shown.

L. GORTON ET-AL

444

to D-phenylalanine is twice as high when using ARP. ARP, therefore, in a first evaluation is a more promising candidate than HRP for further experiments. Table 3 shows the effect with different types of additives obtained with newly made electrodes on the response to H202 and D-phenylalanine. As shown in the table, the additive yielding the highest response to H202(chitosanglutamate)does not necessarily give the highest response to D-phenylalanine. Neomycin was previously shown to be a strong promoter for obtaining good electrochemical contact between cytochrome c peroxidase and carbon electrodes (Armstrong and Lannon, 1987,1988) and is the best choice to achieve a high response to D-phenyl-alanine. In Figure 10 one can also see that neomycin has a superior effect on the long-term stability of the response to the amino acid compared with other additives. For Gafquat 734, gentamicin, and neomycin the stability of the responses for H202and D-phenylalanine follow each other whereas for PEI the response for H202 is stable and the response to D-phenylalanine gradually declines. Figure 11 shows the log-log plots of the calibration curves obtained for H202 and D-phenylalanine obtained at pH 8.0 using coimmobilized D-AAOD/ARP with neomycin as the additive. Near linear response behaviors were found between 5 and 1000 pM for hydrogen peroxide and between 0.1 and 14 mM for D-phenylalanine, respectively. The detection limit of D-phenylalanine was found to be 0.1 mM with a signal to noise ratio of 2: 1.

0.5

0

i

L

0

Figure 11. Log-log plots of calibration curves for (0) hydrogen peroxide and (0) D-phenylalanine. The electrodes used were prepared using 1 .O mg of ARP, 1 .O mg of D-AAOD, and 3.0 mg of neomycin er 100 mg of graphite (set 3, p. 424). Carrier; 0.1 M Tris buffer at pH 8.0, 0.6 ml n ,i-! Applied potential -50 mV vs Ag/AgCI. Injection volume 50 pl.

Amperometric Biosensors

445

.

E

g

1

Q, 0.8

0.8

0.6

0.6

0.4

0.4

0.2

1

4

m

T

0.2

1.2 1

-. z 0.8 0.4 0.2 0

5.5

6

65

7

75 PH

8

85

9

9.5

Figure 72. FI responsesfor (0)0.1 rnM hydrogen peroxide and (0)1.O rnM D-phenylalanine as a function of pH obtained for (a) a D-AAOD/HRP and (b) a D-AAOD/ARP coirnrnobilized carbon paste electrode prepared according to set 4 described on p. 424. Carrier (0.6 rnl min-'); at pH 6.0 and 7.0 a 0.1 M phosphate buffer, and at pH 8.0 and 9.0 a 0.1 M Tris buffer. Injection volume 50 pl. Applied potential, -50 mV vs AgIAgCI.

Figures 12a and 12b show the variation of the responses to H,O, and D-phenylalanine with two electrodes, one with HRP (Figure 12a)and one with ARP (Figure 12b) and with the use of neomycin as the additive. The responses to H,O, follow each other irrespective of the choice of peroxidase, having for both peroxidases a pH optimum at pH 7. The responses to D-phenylalanine, however, drastically differ above pH 8 and is much higher at pH 9 when ARP is used. F. g glut am ate Sensor Based on Coimmobilized L-Glutamate Oxidase and Horseradish Peroxidase

L-glutamate oxidase (L-GLUOD) is a much more substrate specific oxidase than L-AAOD. It oxidizes selectively L-glutamate in the presence of oxygen to form a-ketoglutarate and hydrogen peroxide in a reaction analogous with Reaction 12:

L-GLUOD

L-glutamate + H,O + 0, +

a-ketoglutarate + H,O,

+ NH,

(13)

Due to its importance in food, biotechnology, and medicine, several papers have appeared on the making of selective biosensors for L-glutamate based on LGLUOD (Wollenberger et al., 1989; Yao et al., 1990; Villarta et al., 1991). Figures 13a and 13b show calibration curves (loglog plots) for both L-glutamate and hydrogen peroxide obtained with one GLUOD/ARP (Figure 13a) and one GLUOD/HRP(Figure 13b)modified carbon paste electrode. As can be seen, almost strictly linear response ranges were obtained for both L-glutamate and hydrogen peroxide with the GLUOD/HRP electrode, whereas with the GLUOD/ARP electrode deviations from strict linearity are obvious. The slopes of the loglog plots

1.5

0.5

,-l

0.5

L__-

1

i

>

1.5

2

2.5

3

35

4

3

3.5

4

Log (conc /vM)

0.5

1

1.5

2

2.5

Log (conc./uM)

Figure 13. Log-log plots of calibration curves for (a) (0)hydrogen peroxide and (0) L-glutamate obtained with a GLUOD/HRP-PEI modified carbon paste electrode and (b) (0) hydrogen peroxide and (0)L-glutamate obtained with a GLUOD/ARP-PEI modified carbon paste electrode. Carrier (0.6 ml min-'); a 0.1 M phosphate buffer at pH 7.0. Injection volume 50 pl. Applied potential -50 mV vs Ag/AgCI.

Arnperometric Biosensors

447

for hydrogen peroxide and L-glutamate for the GLUODHRF' electrode were close to identical (0.80 and 0.78, respectively). In this case, the response to the substrate of the oxidase is somewhat closer to the response to hydrogen peroxide than those obtained for other oxidase/peroxidaseelectrodesreported from this laboratory,even though comparable or even higher amounts of the oxidases were added to 100 mg of graphite compared with the amount of GLUOD (8 U mg-').

ACKNOWLEDGMENT Financial support from the Swedish Natural Science Research Council (NFR), the Swedish Research Council for the Engineering Sciences (TFR), the Swedish National Board for Technical and Industrial Development (NUTEK), Direktor Albert Phhlsson's Foundation for Scientific Research is gratefully acknowledged. The authors also thank Suntory Ltd., Japan, and Yamasa Shoyu Co., Ltd., Japan; Dr. T. D. Gibson, UK,Dr. M. G. Venegas, USA for their generous gifts of t h e k t r o r n y c e s rarnosus peroxidase, L-glutamate oxidase, Gafquat 755N,and quaternized a n d ethoxylated polyethyleneimines, respectively; and Labinett AB, Sweden for the loan of the Automated Sample Injection Analyser. Acta Chemica Scandinavica and the Foundation for Biotechnical and Industrial Fermentation Research (Finland) are thanked for personal grants to M. Smolander and the Swedish Institute for personal grants to Z. Huan.

REFERENCES Appelqvist, R., Marko-Varga, G., Gorton. L., Torstensson, A,. & Johansson, G. (1985). Enzymatic determination of glucose in a flow system by catalytic oxidation of the nicotinamide coenzyme at a chemically modified electrode. Anal. Chim. Acta 169,237-247. Armstrong, F.A., Hill, H.A.O., & Walton, N.J. (1988). Direct electrochemistry of redox proteins. Acc. Chem. Res. 21,407413. Armstrong, F.A. & Lannon, A.M. (1987). Fast interfacial electron transfer between cytochrome c peroxidase and graphite electrodes promoted by aminoglycosides: Novel electroenzymic catalysis of H202reduction. J. Am. Chem. SOC. 109,721 1-7212. Armstrong, F.A. & Lannon, A.M. (1988). Towards macromolecular recognition at electrodes: Fast interfacial electron transfer to cytochrome c peroxidase at graphite electrodes is promoted by aminoglycosides. Biochem. Soc. Trans. 16,842-844. Bahulekar, R., Ayyangar N.R., & Ponrathnam, S. (1991). Polyethyleneimine in immobilization of biocatalysts. Enzyme Microb. Technol. 13,85%868. Baldwin, R.P. & Thomson, K.N. (1990). Chemically modified electrodes in liquid chromatography detection: A review. Talanta 38. 1-16. Bartlett, P.N., Tebbutt. P., & Whitaker. R.G. (1991). Kinetic aspects of the use of modified electrodes and mediators in bioelectrochemistry. Progr. React. Kinet. 16, 55-156. Barman, T.E. (1969). Enzyme Handbook, Vol. 1, p. 47. Springer-Verlag, New York. Blaedel, W.J. & Engstom, R.C. (1980). Reagentless enzyme electrodes for ethanol, lactate, and malate. Anal. Chem. 52, 1691-1697. Bremle, G., Persson, B., & Gorton, L. (1991). Anamperometricglucose electrode basedon carbon paste, chemically modified with glucose dehydrogenase, nicotinamide adenine dinucleotide and a phenoxazine mediator, coated with a poly(ester-sulfonic acid) cation-exchanger. Electroanalysis 3,7746.

Buchert, J. (199 I). A xylose-oxidizing membrane-bound aldose dehydrogenase of Gluconobucrer oxyduns ATCC 621. J. Biotechnol. 18, 103114. Buttler, T.A., Johansson, K.A.J., Gorton, L.G.O., & Marko-Varga, G. (1993). On-line fermentation process monitoring of carbohydrates and ethanol using tangential flow filtration and column liquid chromatography. Anal. Chem. 65,262%2636. Chenault , H.K. & Whitesides, G.M. (1987). Regeneration ofnicotinamide cofactors for use in organic synthesis. Appl. Biochem. Biotechnol. 14, 147-197. Csoregi, E., Jonsson-Pettersson, G., & Gorton, L. (1993a). Mediatorless electrocatalytic reduction of hydrogen peroxide at graphite electrodes chemically modified with peroxidases. J. Biotechnol., 36,315-337. Csoregi, E., Gorton, L., & Marko-Varga, G. (1993b). Carbon fibre as electrode materials for the construction of peroxidase modified amperometric biosensors. Anal. Chim. Acta 273.59-70. Davidson, V.L. (1992). Principles and Applications of Quinoproteins. Dekker, New York. Dominguez, E., Marko-Varga, G., Carlsson, M., & Gorton, L. (1990). Liquid chromatographic separation and stereo detection of L- and D-amino acids with catalytic reaction detection using immobilized enzymes. J. Pharm. Biomed. Anal. 8, 82-30, Dominguez, E., Lan, H.L., Okamoto, Y., Hale, P.D., Skotheim, T., & Gorton, L. (1993). Acarbonpaste electrode chemically modified with a phenothiazine polymer derivative and yeast alcohol dehydrogenase for the analysis of ethanol. Biosens. Bioelectron. 8, 167. Duine, J.A., Frank, J., & Jongejan, J.A. (1987). Enzymology of quinoproteins. Adv. Enzymol. Relat. Areas Mol. Biol. 59, 165-212. Frew, J.E. & Hill, H.A.O. (1987). Electrochemical biosensors. Anal. Chem. 59,933A-944A. Gibson, T.D. (1992). University of Leeds, Leeds, UK, personal communication. Gibson, T.D., Higgins, I.J., & Woodward, J.R. (1992). Stabilization ofanalytical enzymes using a novel polymer-carbohydrate system and the production of a stabilized single reagent for alcohol analysis. Analyst 117, 1293-1297. Gibson, T.D. & Woodward, J.R. (1992). Protein stabilization in biosensor systems. In: Biosensors and Chemical Sensors. ACS Symp. Ser. (Edelman, P.G. & Wang, J., Eds.), Vol. 487, pp. 4&55. Am. Chem. Society, Washington, D.C. Gorton, L. (1986). Chemically modified electrodes for the electrocatalytic oxidation of nicotinamide coenzymes. J. Chem. Soc. Faraday Trans., I, 82, 1245-1258. Gorton, L., Bremle, G., Csoregi, E., Persson, B., & Jonsson-Pettersson, G. (1991a). Amperometric glucose sensors based on immobilised glucose oxidising enzymes and chemically modified electrodes. Anal. Chim. Acta 249,4%54. Gorton, L., Csoregi, E., Dominguez, E., Emneus. J., Jonsson-Pettersson, G., Marko-Varga, G., & Persson, B. (I991 b). Selective detection in flow analysis based on the combination of immobilised enzymes and chemically modified electrodes. Anal. Chim. Acta 250,203-248. Gorton, L., Jonsson-Pettersson, G., Csoregi, E., Dominguez, E., Johansson, K., & Marko-Varga, G. (1992a). Amperometric biosensors based on an apparent direct electron transfer between electrodes and immobilized peroxidases. Analyst 117,1231-1241. Gorton, L., Persson, B., Hale, P.D., Bouguslavsky, L.I., b r a n , H.I., Lee, H.S., Skotheim, T., Lan, H.L., & Okamoto, Y. (1992b). Electrocatalytic oxidation of nicotinamide adenine dinucleotide cofactor at chemically modified electrodes. In: Biosensors and Chemical Sensors, ACS Symp. Ser. (Edelman, P.G. & Wang, J., Eds.), Vol. 487, pp. 5 6 4 3 . Gorton, L., Dominguez, E., Marko-Varga, G., Persson, B., Jonsson-Pettersson, G., Csoregi, E., Johansson K., Narasaiah, D., Ghobadi, S., Kacaniklic, V., Skotheim, T., Hale, P., Okamoto, Y., & Lan, H.-S. (1 993). Amperometric biosensors based on immobilized redox-enzymes in carbon paste electrodes. Bioelectroanalysis, 2, Akademiai Kiddb, Budapest, Hungary. Johanson, K., Jonsson-Pettersson, G., Gorton, L., Marko-Varga, G., & Csoregi, E. (1993). A reagentless amperometric biosensor for alcohol detection in liquid chromatography based on co-immobilized peroxidase and alcohol oxidase in carbon paste. J. Biotechnol. 31, 301-316.

Amperometric Biosensors

449

Jonsson, G. & Gorton, L. (1989). An Electrochemical sensor for hydrogen peroxide based onperoxidase adsorbed on a spectrographic graphite electrode. Electroanalysis I , 46-68, Jonsson-Pettersson, G. (1991). Reagentless hydrogen peroxide and glucose sensors based on peroxidase immobilized on graphite electrodes. Electroanalysis 3, 74 1-750. Kacaniklic, V., Johansson, K., Marko-Varga, G., Gorton. L.. Jonsson-Pettersson, G., & Csoregi, E. (1994). Amperometric biosensors for detection of L- and D-amino acids based on coinmobilized peroxidase and L- and D-amino acid oxidases in carbon paste electrodes. Biosens. Bioelectron.. Electroanalysis 6, 38 1-390. Kylus, J., Bilitewski, U., & Schmid, R.D. (1991). Robust graphite-based bienzyme sensors., Sens.Act. B, 3,227-234. Kylus, J. & Schmid, R.D. (1990). Mediatorless peroxidase electrode and preparation of bienzyme sensors. Bioelectrochem. Bioenerg. 24,305-3 11. Labuda, J. (1992). Chemically modified electrodes as sensors in analytical chemistry. Selective Electrode Rev. 14, 3 3 4 6 . Marko-Varga, G. Johansson, K., & Gorton, G. (1993). A reagentless coimmobilized alcohol biosensor as a detection unit in column liquid chromatography for the determination of methanol and ethanol. Quim. Anal. 12,74-79. Murray R.W., Ewing, A.G., & Durst, R.A. (1987). Chemically modified electrodes. Molecular design for electrocatalysis. Anal. Chem. 59,379A-39OA. Paddock, R.M. & Bowden, E.F. (1989). Electrocatalytic reduction of hydrogen peroxide via direct electron transfer from pyrolytic graphite electrodes to irreversibly adsorbed cytochrome c peroxidase. J. Electroanal. Chem. 260,487494, Persson, B. (1990). A chemically modified graphite electrode for electrocatalytic oxidation of reduced nicotinamide adenine dinucleotide based on a phenothiazine derivative, 3-P-naphthoyl-toluidine Blue 0. J. Electroanal. Chem. 287, 61-80. Persson, B., Lan, H.L., Gorton, L., Okamoto, Y., Hale, P.D., Boguslavsky, L.I., & Skotheim, T. (1993). Amperometric biosensors based on electrocatalytic regeneration ofNAD' at redox polymer-modified electrodes. Biosens. Bioe1ectron.g 8 , 8 1-88, Shinmen, Y., Asami, S., Amachi, T., Stimizu, S., & Yamada, H. (1986). Crystallization and characterization of an extracellular hngal peroxidase. Agric. Biol. Chem. 50,247-249. Smolander,. M., Buchert, J., & Viikari, L. (1993a). Large-scale applicable purification and characterization of a membrane-bound PQQ-dependent aldose dehydrogenase. J. Biotechnol. 29,287292. Smolander, M., Cooper, J., Schuhmann, W., H h m e r l e , M., & Schmidt, H.-L. (1993b). Determination of xylose and glucose in a flow-injection system with PQQ-dependent aldose dehydrogenase. Anal. Chim. Acta 280. 119-127. Smolander, M., Livio, H.-L., & Rasanen, L. (1992). Mediated amperometric determination of xylose and glucose with an immobilized aldose dehydrogenase electrode. Biosens. Bioelectron. 7, 637-643. Tatsuma, T. & Watanabe, T. (1 99 I). Peroxidase model electrodes: Heme peptide modified electrodes as reagentless sensors for hydrogen peroxide. Anal. Chem. 63, 1580-1585. Updike, S.J. &Hicks, G.P. (1967). The enzyme electrode. Nature 214,986-988. Villarta, R.L., Cunningham, D.D., & Guilbault, G.G. (1991). Amperometric enzyme electrode for the determination of L-glutamate. Talanta 38,49-55. Wing, S.A. & Hart, J.P. (1992). Chemically modified carbon-based electrodes and their applications as electrochemical sensors for the analysis ofbiologically important compounds. A Review. Analyst 117, 12151229. Wollenberger, U.,Scheller, F.W., Bohmer, A,, Passarge, M., & Muller, H.-G. (1989). Aspecific enzyme electrode for L-glutamate development and application. Biosensors 4,38 1-391. V'ollenberger, U., Bogdanovskaya, V., Bobrin, S., Scheller, F., & Tarsevich, M. (1990). Enzyme electrodes using bioelectrocatalytic reduction of hydrogen peroxide. Anal. Lett. 23, 1 7 9 5 4808.

Wollenberger, U., Wang, J., Ozsoz, M., Gonzalez-Romero, E., & Scheller, F. (1991). Bulk modified enzyme electrodes for reagentless detection ofperoxides. Bioelectrochem. Bioenerg. 26.287-296. Yao, T. & Wasa, T. (1988). High-performance liquid chromatographic detection of L- and D-amino acid by use of immobilized enzyme electrodes as detectors. Anal. Chim. Acta 209,259-264. Yao, T., Kobayashi, N., & Wasa, T. (1990). Flow-injection analysis for L-glutamate using immobilized L-glutamate oxidase: Comparison of an enzyme reactor and enzyme electrode. Anal. Chim. Acta 231, 121-124. Yaropolov, A.I., Malovik, V., Varfolomeev, S.D., & Berezin, 1.V. (1979). Electroreduction of hydrogen peroxide on an electrode with immobilizedperoxidase. Dokl. Akad. Nauk SSSR 249, 1399-1401.

ENZYME BASED DIFFUSION BADGE FOR THE DETECTION OF FORMALDEHYDE

R. Feldbrugge, K. P. Rindt, and A. Borchert

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. PRINCIPLES OF OPERATION . . . . . . . . . . . . . . . . . . . . . . . . 111. STABILIZATION OF ENZYMES . . . . . . . . . . . . . . . . . . . . . . . IV. ELIMINATION OF NONSPECIFIC COLOR FORMATION . . . . . . . V. PRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

451 452 452 455

. . 456 457 457

ABSTRACT The development of economical, easy to handle enzyme sensors for the colorimetric detection of gaseous toxic compounds is based on a patent for enzymatic gas measurement by Dr. Rindt. The present investigations have resulted in a one-way biosensor for selective detection of formaldehyde in a few minutes. The problems

Advances in Molecular and Cell Biology Volume 15B, pages 451459. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

45 1

452

R. FELDBRUGGE, K.P. RINDT, and A. BORCHERT

with nonspecific color formation and storage stability over 12 months at room temperature were solved. Many possibilities have arisen for economic production on a large scale. We conclude that such an enzyme diffusion Badge can be conveniently applied for the detection of various gaseous substances.

1. INTRODUCTION Nowadays people are surrounded by formaldehyde in the working place, at home, and even during their fkee time, which can have a negative influence on their health (Bundesgesundheitsamt, 1984, 1987; Deutsche Forschungsgemeinschaft, 1987). Thus there is an interest in the detection of such toxic compounds in a very reliable and simple manner (report 1986). A lot of gas detection tubes for the measurement of different gaseous substances by chemical reaction and color formation are now on the market. But due to low specificity it is not possible to detect one single toxic substance. By using enzymes, the selectivity of the reaction is very high. The enzyme formaldehyde dehydrogenase allows specific and quantitative oxidation of formaldehydewith NAD' to formic acid and NADH. Diaphorase transfers NADH and colorless tetrazolium salt in NAD+ and colorful formazan can be observed. The main aims of this investigation are: 1. stability of the enzymes, 2. specific color formation, and 3. production of the sensor. Inexpensive components and simple production methods were needed for the preparation of a biosensor as a product for single use. The user thus has a tool for uncomplicated applications. To date, such an enzyme badge for colorimetric detection of toxic compounds of air pollutants has not been on the market.

11. PRINCIPLES OF OPERATION The main functional part of the diffusion enzyme badge is a porous, absorbing rod of sintered glass. At the surface, the catalytic conversion of the gaseous substance and color compounds take place. The underlying principle is shown in Figure 1. At the top of the sintered glass rod, the enzymes and stabilizer are absorbed and lyophilized on the opposite site for the color compounds and the reagents. This rod is placed in a recyclable case together with a glass ampoule containing the activating buffer (see Figure 2). For activation of the biosensor, the glass ampoule is broken and the buffer sucked onto the bottom of the porous rod. Capillary forces and evaporation produce a flow of activating solution to the top of the rod. With this flow, the reagent and color compounds are transported to the activated and concentrated enzymes. Formaldehydemolecules in the atmosphere reach the active surface by diffusion. This exposure starts the specific reaction with formaldehyde dehydro-

substrate molecules

s

S

s

s

--

sinterglass

K.-P. Rindt Dragorwork AG

11i"lli activating solution

Figure 7. Functional part of the enzyme badge.

ampoule support

sinterglass r o d activating solution

connecting channel

Figure 2. Construction of the badge. 453

454

R. FELDBRUGGE, K.P. RINDT, and A. BORCHERT

Biosensor

Enzym-Diffusion-Badge ___ (Example)

-

-+NAD+

HCHO

i

1

visual Analysis

1. 1.

Signal

Figure 3. Reaction scheme of an enzyme - diffusion badge and a biosensor.

genase (EC 1.2.1.46)that generatesthe interesting product NADH. Diffusion is the limiting step of the reaction and this is directly stoichiometric to the atmospheric concentration of the gas compound. In a second reaction the fixed enzyme diaphorase (EC 1.6.4.3) in the enzyme layer reoxidizes the NADH to NADf and leads to color formation. The resulting color intensity depends on the formaldehyde concentration and exposure time. The product of time and concentration yields a color intensity that can be visualized by color comparison against a color code. The reaction scheme is shown in Figure 3. Both products, NADH and dye as first and second signals, are formed as a result of an enzymatic reaction and therefore this formaldehyde badge represents a special modification of a biosensor (Scholtissek et al., 1990).

Enzyme Based Detection of Formaldehyde

455

111. STABILIZATION OF ENZYMES The technical demands of a biosensor are as follows: 1. storage stability for 6 to 12 months, 2. transportation without refkigeration, and 3. stability in case of a short temperature rise up to 60 "C. For these purposes, enzymes must be stabilized and nonspecific color change of the dye has to be prevented. Known stabilizers for dehydrogenases often contain mercapto groups. They block the oxidation of the active center of the enzyme and regenerate the activity of the enzyme by reduction (Ogushi et al., 1984, 1986). These known substances, however, show nonspecific interaction with the color compound tetrazolium salt. This is also the case with serum albumin as a standard protein stabilizer (Europaische Patentanmeldung).In principle, the substrates and products can have a stabilizingeffect. In this case only NAD' is applicable and it needs costabilizers to meet the requirements for a commercial product. Several proteins and polyethyleneglycol (PEG) showed sufficient stabilizationin long time storageexperiments. Both enzymes, formaldehyde dehydrogenase and diaphorase, were tested with such mixtures over a 100-day period or longer at temperatures of 4 to 60 "C. Figures 4 and 5 show the results thus obtained. The temperature dependence is visible in spite of stabilization and decreases the activity, especially in the case of formaldehyde dehydrogenase. The remaining FormaldehydeDHmU/sensor 140 130 120 110 100 90. 80. 70. 60. 4 50. 40. 0. 30. n -.. 20. - - ' '.-. ....... 10. -.

* 1: 6

>.

-4°C

-

27°C.

+

37°C.

-

60°C.

Figure 4. Stability of fixed formaldehyde dehydrogenase.

R. FELDBRUGGE, K.P. RINDT, and A. BORCHERT

456 140 130 120 110 100

Diaphorase mU/sensor

90. 80. 70. 60. 50. 40.

30. 20. 10.

0 0

10.

20.

30.

40.

50.

60.

70.

80.

90.

100

days

Figure 5. Stability of fixed diaphorase.

activities of the compounds were, however, sufficient for more than one year and could be reproduced.

IV. ELIMINATION OF NONSPECIFIC COLOR FORMATION Apart from the stabilization of delicate substances present in a measuring system, there is the problem of blank reactions. The problem involves the signal being present or emerging in the absence ofthe specimen to be measured. In this particular instance, this means color formation without formaldehyde being present in the surrounding air. To achieve a high measurement accuracy this source of error must be minimized, or better, eliminated. The reasons for this inaccuracy of the biosensor can be attributed to a limited number of sources. In general the enzymes (formaldehyde dehydrogenase, diaphorase), the reagents @ADf, stabilizer), the dye (formazan) or its precursor (INT), the carrier substance (sintered glass), and the buffer substance (phosphate)may be responsible for nonspecific side reactions. In addition, it is reasonable to suspect the existence of unknown additives or impurities in the substances. Other possible reasons for this blank reaction might be the operations during the manufacturing of the biosensor as well as physical conditions of the air like temperature,humidity, and concentration of formaldehyde. The natural background level of formaldehyde due to the photochemical degradation of natural organic substances is about 0.1 pg/m3 in maritime air, and about 1 pg/m3 in continental air.

Enzyme Based Detection of Formaldehyde

457

To eliminate pollution of the biosensor by formaldehyde, a laminar-flow box was used to which clean air was permanently supplied via a special formaldehyde filter. Within this box, the sensors were prepared and genuine blank reactions were measured. The considerably reduced blank reaction data of sensors produced in this way show that their production in air free of formaldehyde is of indispensable necessity. The purity of the applied batches of sintered glass was investigated. By soaking up organic solvents, a yellowish band at the solvent front was observed. This color reaction was distinct when arriving at the top area and provides considerableblank reaction data. The color reaction results from residues coming from the production of the sintered glass bars and probably also from the absorption of substances from the packaging. Methanol turned out to be a suitable eluent. Thus, methanol washing used before the subsequent treatment of the sintered glass provided optimum conditions. Another processing step leading to different blank reaction data was found to be lyophilization of the components on the sintered material. Therefore, different lyophilization parameters were tested. An optimization and a shorter drying time process eliminated this nonspecific color formation (Golker, 1987; Willemer, 1989).

V. PRODUCTION The main factor for economical use is the price of the sensor. Therefore, it was necessary to minimize the amount of coupled enzyme. Placing a high local concentration only at the top of the sintered glass rods allows reduction in enzyme costs, but the handling of very small volumes (20 pl) is not easy and, therefore, constitutes an expensive production step, especially in large scale production. One way of overcoming this problem is the preparation of the porous sintered glass on one end with buffer and the color compounds in a microtiter plate for 96 rods. The wells of a second titer plate were filled with the concentrated enzyme solution. By turning the 96 rods into this second plate, the enzyme was homogeneously adsorbed at the opposite site of the sintered glass rods. The pipetting can be performed automatically by commercial machines. After lyophilization,the active part of the enzyme badge can be stored separately or put into the case of the sensor. This method of preparation minimized enzyme consumption and production costs and improved the quality of the sensors.

VI. CONCLUSIONS The present work has led to a number of conclusions.First, the sensor can be used by laymen: no additional equipment is required, activation is simple, and evaluation is performed visually by comparison against a color code. Second, the color of the exposed area accumulates over more than 2 hours depending on the atmospheric

458

R. FELDBRUGGE, K.P. RINDT, and A. BORCHERT

concentration of the species to be measured (dose dependence). Third, the system was shown to be very sensitiveindicating 100ppb formaldehyde within 15minutes. Fourth, the substance to be measured is the substrate of the correspondingenzyme. Thus, the enzymatic reaction is highly selective. Some cross-reactions, however, based on the chemical indicator reaction may occur. Fifth, enzymes and reagents can be separately stored on one single structure. And sixth, under optimum conditions, color formation is immediate. Above a definite enzyme activity level that is characteristic of aparticular enzyme system,color formation dependent on atmospheric substrate concentration does not depend on enzyme concentration. Thus, enzyme can be added in excess, and partial loss of enzyme activity does not influence the result. Furthermore, the permanent solvent stream caused by capillary forces during activation and later by evaporation has several functions: (1) During the activation process it hrther concentrates the enzymes in a minute volume at the surface to be exposed, i.e. exactly in that place where the enzymes are required. (2) During measurement it prevents the enzymecontaining surface from drying and serves as a permanent reagent supply. And (3) It prevents colored products from diffusing back into the structure, thus allowing also the use of such chromogens that give rise to water-soluble color products. The same system is easily adaptable from spot to long-term measurement, for example, by covering the exposed area with a partially gas-permeable membrane. As for storage, use of a special stabilizing cocktail revealed that lyophilizedenzyme diffision badges for formaldehyde stored at 37 "C for 18 months remained fully functional. However, the activity of formaldehyde dehydrogenase dropped at the same time to 30% of its original activity.

ACKNOWLEDGMENT This work is part of a diploma work (Diplomarbeit) done at the Dragerwerk AG Liibeck. More details of the design and manufacturing of these enzyme badges and also a list of the specific literature can be found in the Diplomarbeit (Feldbriigge, 1991).

REFERENCES Bundesgesundheitsamt:Formaldehyd.Gemeinsamer Bericht des Bundesgesundheitsamtes, der Bundesanstalt fiirArbeitsschutz unddes Umweltbundesamtes. (1 984). Familie und Gesundheit. Vol. 148, Verlag Kohlhammer, Stuttgart. Bundesgesundheitsblatt 30 Nr. 8, (1987). Deutsche Forschungsgemeinschaft. Maximale Arbeitsplatzkonzentrationen und Biologische Arbeitsstofftoleramen (1987). Mitteilung XXIII. Europaische Patentanmeldung, Veroffentlichungsnuer: 0054689 B 1, StabilisierteZubereitung von Tetrazoliumsalzen. Feldbriigge, R. (1991). Diplomarbeit, WS 90/91, Fachhochschule Ostfriesland. Gdlker, C.: Trochung biotechnologischer Produkte (1987). Biotech-Forum 4,4. Ogushi, S., Ando, M., & Tsuru,D. (1986). Formaldehyde dehydrogenase from Pseudomonas putida. The role of a cysteinyl residue in the enzyme activity. Agric. Biol. Chem. 50,250%2507.

Enzyme Based Detection of Formaldehyde

459

Ogushi, S., Ando, M., & Tsuru, D. (1984). Formaldehyde. dehydrogenase from Pseudomonusputidu: A zinc metalloenzyme.J. Biochem. 96, 1587-1591. Report forecasts explosive growth of European market in biosensors. (1986). Biosensors 2, 145. Scholtissek, S., Rindt, K.P., & Schlfer, G. (1990). Biosensoren. Focus Medizinische Hochschule Lubeck, 7. Jahrg. 172-179. Willemer,H.: PhysikalischeGrundlagender GefiiertrocknungFortschritteund Entwicklungstendenzen. Firmenschrift, Leybold AG, Koln.

This Page Intentionally Left Blank

BIOSENSING OF HEAVY METAL IONS BASED ON SPECIFIC INTERACTIONS WITH APOENZYMES

lkuo Satoh Abstract . . . . . . . . . . . . . . . . . . . . . . . I. INTRODUCTION.. . . . . . . . . . . . . . . . . 11. PRINCIPLE . . . . . . . . . . . . . . . . . . . . . 111. CHARACTERISTICS. . . . . . . . . . . . . . . . IV. PROCEDURE.. . . . . . . . . . . . . . . . . . . V. BIOSENSING OF HEAVY METAL IONS . . . . . A. Apoenzyme Beads as Sensing Elements . . . . B. Apoenzyme Membrane as the Sensing Element VI. CONCLUSION . . . . . . . . . . . . . . . . . . .

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

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

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

. . . . . . . 461 . . . . . . . 462 . . . . . . . 462 . . . . . . . 463 . . . . . . . 466 . . . . . . . 467 . . . . . . . 467 . . . . . . . 470 . . . . . . . 471

ABSTRACT The present work led to a novel idea about the biosensing of heavy metal ions based on apoenzyme reactivation methods. Several kinds of immobilized metalloenzymes

Advances in Molecular and Cell Biology Volume 15B, pages 461472. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

46 1

IKUO SATOH

462

as recognition elements for each metal ion were applied in the microdetermination of heavy metal ions in flow streams. Regeneration of the metal-free enzymes (apoenzymes) was made by loading chelating agents to the metal-bound enzymes (holoenzymes). The apoenzymes were positively reactivated by adding trace amounts of heavy metals. The recovery depended on metals trapped in the enzymes, which was shown to be a practical index for microanalysis of heavy metal ions. The proposed method was demonstrated to be convenient, safe, specific, and highly sensitive.

1. INTRODUCTION A great number of metal-dependent enzymes have already been identified (Vallee, 1980).Heavy metal ions coordinated in the active site of these enzymes play a very important role in the catalytic process. These heavy metal ions are intrinsic to each of the enzymes and most of them can be reversibly removed from the catalytic site of the enzymes. Therefore, metal-free enzymes (apoenzymes) can be used as specific and selective recognition elements for heavy metal ions in an enzymatic analysis. Use of enzymes as highly sensitive reagents is common in food and clinical analyses. Currently,immobilized enzymes in combination with transducers for monitoring their enzymatic activity are used for the construction of biosensors, and many of them are commercially available (Satoh, 1989a). Analytes have been mostly limited to substrates, products, activators, and inhibitorsin the enzyme-catalyzed reactions. In sharp contrast, we have adopted flow-injectionmicroassay for cofactors, namely heavy metal ions, based on apoenzyme reactivation (Satoh et al., 1986, 1987, 1988, 1989b, 1990ste, 1991a-e, 1992, 1993). We have tried to regenerate cofactor-free enzymes from cofactor-bound enzymes (holotype of metalloenzymes) and use them as the recognition elements for heavy metal ions in flow streams. The apoenzyme reactivation methods for flow-injection microdetermination of heavy metal ions are summarized in this paper with a special focus on our current studies.

II. PRINCIPLE Cofactors such as heaG metal ions and nucleotides are generally complexed in the active site of the metalloenzymes and flavin enzymes, respectively, and the cofactors are directly responsible for the activity of the enzymes. Therefore, these enzymes need the cofactors for expressing catalytic activity. The cofactor-bound enzyme is usually called a holoenzyme, whereas the cofactor-free enzyme is called an apoenzyme. Figure 1 schematically shows the correlation between the holoenzyme and the apoenzyme. The metalloenzymes capture the metal ions so tightly (dissociation constant: Kd< lo4 M) that they hold them throughout the purification process. In contrast with the metalloenzymes, metal-activated enzymes bind the

Biosensing of Heavy Metals

Cofactor-bound enzyme catalytically

463

Cofactor-free enzyme catalytically

[ active

[ inactive

Figure 1. lnterconversion between the holoenzyme and the apoenzyme. Cofactor: Heavy metal ions; r(lr(lS catalytic site; 000 substrate-binding site.

metal ions rather weakly (Kd < 1C3to 1O4 M), and then, the enzymes tend to lose the metal ions during purification (Wagner, 1988). Preparation of the apoenzyme lacking its catalytic activity can be made by removing the metal ion from the corresponding holoenzyme with strong chelating agents. The apoenzyme is reactivated by exposing it to the metal-containing sample so that metal ions can be taken up and trapped in the active site. Thus, the amount of metals coordinated in the catalytic center of the enzyme molecules may be closely related to the enzyme activity expressed and, in turn, be proportional to the added amount of the metals. The content of the trace metals can eventually be determined by monitoring the induced activity attributable to the reactivation of the apoenzyme. The metal ions responsible for the catalytic activity of the metalloenzymes vary with the type of enzyme. Selective determination of the metal ions depends upon choosing the appropriate metalloenzyme in which its catalytic site fits well with each metal ion.

111. CHARACTERISTICS In practice, use of metalloenzymes immobilized on supporting materials such as small glass beads and a thin polymer membrane can make assays continuous and, moreover, enhance the feasibility of handling in the process between regeneration and reactivation of the apoenzymes. Reusability and long-term stability of the immobilized enzymes may be expected. Microdetermination of heavy metal ions based on spectrophotometric monitoring was tried (Townshend and Vaugham, 1970) and a microassay for zinc(I1) ions using high performance liquid chroma-

IKUO SATOH

464

tographic methods (Risinger et al., 1983) as well as an electrochemical method (Mattiasson et al., 1979) were reported. We have originally developed unique biosensing of heavy metal ions based on flow-injectioncalorimetry (Satoh et al., 1986).Thermometric flow-injectionanalysis of biorelated compounds in combination with the use of a high performance semiadiabatic calorimeter were pioneered by Mosbach et al. (Danielsson and Mosbach, 1988; Danielsson, 1990). The flow-measuring system, that is, a thermal bioanalyzer, is better known as an “enzyme thermistor.” We used the calorimeter for monitoring the enzymatic activity in the assay cycle including regeneration and reactivation. The calorimetric biosensing system is schematically presented in Figure 2. The column packed with immobilized enzyme beads is interchangeable and, therefore, different kinds of metals are readily determined. In addition, other monitoring methods such as amperometry, potentiometry, and spectrophotometry are available by exchanging the thermistor with other transducers, for example, electrodes,photomultipliers, etc. The proposed biosensing methods in microanalysis of heavy metal ions do not require expensive instruments based on atomic absorption spectrophotometryand inductively coupled plasma atomic emission spectrophotometry.Furthermore, this

chela3nt

9substrate lons

I If

A k m i n m bath

Figure 2. Flow-calorimetric biosensing system for heavy metal ions based on the apoenzyme reactivation method. Carrier reservoir (buffer solution), pump (flow rate 1.O ml min-’), heat exchanger (thin-walled acid-proof steel tubing: 0.8 mm, i.d.), bath (80Q x 250 mm; 303 K), enzyme column (packed with metalloenzymes immobilized on porous beads), thermistor (attachedto a gold capillary placed in a polymer holder), WB/Amp. (DC-type Wheatstone bridge with a chopper stabilized operational amplifier), Rec. (pen recorder).

Biosensing of Heavy Metals

465

novel method does not involve pollution problems due to exhaust fumes. The characteristics of the proposed methods can be summarized as follows: 0

0 0 0

Mild assay conditions High sensitivity High selectivity Reusability of the immobilized enzymes Continuous flow use

0 0

0

0

Compact assay system Feasibility of handling Feasibility of field work Low-priced assay Separability of the metaltrapping and the activitymonitoring process

Procedure

Phenomena

Substrate

response t o Iiolocnzymcs

t Carrior

washing

t ChelatinE t . Carrier

aRcnt

t

r c g c n c r a t ion o f apociixymos washing

Subs t r a t c

response t o apocnzymcs

t Carricr

washing

Heavy mctal i o n s

p a r t i a l a c t i v a t i o n of apocnzymes

t

washing

Carrier

t

Substrate

response a t t r i b u t a b l c t o metal i o n s c o o r d i n a t c d i n Clic enzymes

Carrier

c

washing

Complctc r c a c t i v d t i o i i of apocnzymcs ( c o n v c r s i o o t o Iiolocnzymcs)

Carrier

washing

Figure 3. Flow chart for the assay procedure.

466

IKUO SATOH

IV. PROCEDURE A flowchart for the assay procedure is shown in Figure 3. Three rotary valves for injection are connected in series between a damper and a drain valve to avoid their mutual contamination. At first the catalytic activity of the immobilized holoenzymes at the reactor (column or membrane) is measured by injecting its substrate. The change in output is a measure of the activity attributable to the holoenzymes, as schematically illustrated in Figure 4. After exposing the enzymesto the chelating agent, a drastic decrease in the activity is observed. This means that the cofactors, namely heavy metal ions complexed in the catalytic site of the holoenzymes, are mostly removed. The amount of chelating agent required for regenerating the

Scan L

60 min

Y

Figure 4. Schematic presentation of response curves for a reaction cycle.

Biosensing of Heavy Metals

467

apoenzymes varies with the kind of metalloenzyme and with the conditions of the chelating agents (variables such as concentration, pH, volume, etc.). Subsequent injection of a trace amount of the cofactors reactivates the apoenzymes. Thus, the recovery of the activity is determined by adding the substrate. While sufficient cofactors are introduced to the mixture of the cofactor-free and cofactor-bound enzymes, all of the enzymes may be completelyconverted to the holoenzymes. The system is then ready for another assay. A reaction cycle generally takes 40 to 60 minutes, but application of a membrane-type reactor reduces the cycle time (within 30 min).

V. BIOSENSING OF HEAVY METAL IONS A. Apoenzyme Beads as Sensing Elements Alkaline Phosphatase as the Recognition Element

Alkaline phosphatase as the recognition element in combination with a couple of monitoring devices was feasible in the most sensitive assay of zinc(I1) and cobalt(I1) ions. The enzyme purified from Escherichia cofi (EC 3.1.3.1., Asahi Chemical Industry Co., Ltd., Ohito-cho, Tagata-gun, Shizuoka-ken, Japan) was immobilized on epoxide acrylic beads (Eupergit C: 100-200 pm particle diameter; 40 nm pore diameter; 180m2g-' surface area; Rohm Pharma, Darmstadt, Germany) and then packed into a small polymer column (0.3 ml). The enzyme is often employed as a labeling reagent for enzyme immunoassay. Hydrolysis of p-nitrophenyl phosphate top-nitrophenol and orthophosphate was monitored for measuring the enzyme activity: p-nitrophenyl phosphate + H,O

+ p-nitrophenol + orthophosphate

(1)

Tris-HC1 100mM buffer @H 8.0, containing 1.OM NaC1) was used and the catalytic activity was calorimetrically determined by injecting 0.1 ml of 100 mM substrate (p-nitrophenylphosphate) (Satoh et al., 1991d). Exposing the column to 5 ml of20 mM 2,6-pyridine dicarboxylate solution (PH 6.0) as the cofactor-complexingagent almost virtually converted the holoenzymes to the apoenzymes. The effect of pH on the chelating agent in the regeneration reaction was tested. No noticeable variation in the level of the regeneration was observed over the pH range of 4.0 to 8.0. The recovery of the activity attributable to the partial reactivation of the apoenzymes was a function of the added amount of zinc(I1) ions. The calibration graph for zinc(I1) ions demonstrated a sigmoidal curve. Zinc(I1) ions ranging from 0.01 to 1.O mM were calorimetrically determined for a 0.5 ml injection. The effect of pH on the reactivation of the apoenzyme was investigated in the weakly acidic pH region (from 4.0 to 6.0 in steps of 0.5). The recovery was unaffected by variations

IKUO SATOH

468

in pH; this means the proposed biosensing using alkalinephosphatasedoes not need any critical pH adjustment, which is very practical. The column retained much more long-term stability than that of a column packed with immobilized bovine carbonic anhydrase. The reactor was repeated over 120 times during the two months of operation. The spectrophotometric approach to measuring enzyme activity based on detecting changes in absorbance at 405 nm attributable to p-nitrophenol formation provided similar results, except for the highly sensitive assay for zinc(I1) ions in submicromolar levels (Satoh et al., 1990b, c). Potentiometric monitoring of the activity with a flow-through ISFET for detecting pH shifts attributable to orthophosphate released in the hydrolytic reaction was also carried out to determine zinc(I1) ion concentrations in the range of 0.01 to 1.O mM (Satoh et al., 1990d). Cobalt-substituted alkaline phosphatase is also capable of hydrolyzing the orthophosphate ester. Therefore, the same immobilized preparations were applied in the microdetermination of cobalt(I1) ions. Biosensing based on spectrophotometry gave excellent results (Satoh, 1992b),and cobalt(I1) ions ranging in concentration from 1-200 pM were determined. Use of a calorimetric sensing system was possible in the concentration range of 0.04 to 1.O mM. Ascorbate Oxidase as the Recognition Element

Ascorbate oxidase is one of the most typical copper-dependentenzymes involved in the oxidation of L-ascorbate to dehydroascorbate. The reaction is described by 2L-Ascorbate + 0, + 2dehydroascorbate + 2H,O

(2)

Ascorbate oxidase from cucumber (EC 1.10.3.3, Asahi Chemical Industry Co., Ltd.) immobilized onto porous glass beads with controlled pore size (CPG; 5 1.5 nm pore diameter, 120-200 mesh, 44 m'g-', Electronucleonics Inc., Fairfield, NJ, USA) was packed into a column and then mounted in the proposed flow-calorimetric system (Satoh et al., 1987). Since oxidative reactions involving molecular oxygen are usually accompanied by considerable heat generation, highly sensitive and precise biosensing of copper(I1) ions attributable to the significant exothermic reaction was anticipated. Thus, micromolar levels of copper(I1) ions were calorimetrically determined. Regeneration of the apoenzymes was achieved with exposure of 20 mM NJV-diethyldithiocarbamate solution (PH 8.0) to the column. The apoenzymes derived from the ascorbate oxidase was selectively responsive to copper(I1) ions and not to divalent cations in 1 mM level such as Ca(II), Co(II), Mg(II), Ni(II), and zinc(I1). Trace amounts of copper(I1) ions in human blood sera were analyzed by the calorimetric method and compared with those obtained by atomic absorption spectrophotometry. There was satisfactory agreement between these methods. Amperometric monitoring of the enzyme activity with a polarographic oxygen electrode showed a more rapid and sensitive determination of copper(I1) ions. The

Biosensing of Heavy Metals

469

assay covered concentrationsranging from 0.5 to 2.0 pM and required about half an hour. L-Ascorbate can be determined by detectingabsorbance at 265 nm, which is used for monitoring the oxidase activity.Thus, copper(I1)ions measured photometrically ranged from 0.1 to 10 pM. A series of these monitoring methods validated the use of immobilized ascorbate oxidase beads as an excellent recognition element for copper(I1) ions. Carbonic Anhydrase as the Recognition Element

Carbonic anhydrase purified from erythrocytes (EC 4.2. 1.1) is known for its remarkably high turnover numbers. The enzyme is a dominant factor in equilibrium between carbon dioxide and bicarbonate in blood as shown in Equation 3, which also expresses esterase activity as seen in Equation 4. CO, + H,O p-nitrophenyl acetate + H,O

-+

H,CO,

(31

3

p-nitrophenol + acetate

(4)

Application of bovine carbonic anhydrase immobilized on the same sort of porous glass beads (CPG) for the specific determination of zinc(I1) and cobalt(I1) ions in combination with flow-calorimetric monitoring turned out to be feasible for the first time (Satoh et al., 1986, 1989b). Injection of 0.5 ml ofp-nitrophenyl acetate into the carrier streams (Tris-HC1 buffer, 0.1 M, pH 8.0) in the calorimetric system resulted in an exothermic response. Since values of changes in enthalpy for ester hydrolysis are normally zero, we considered that the exothermic change resulted from the protonizing heat of acetic acid enzymatically formed in the tris buffer. Exposing the immobilized enzyme beads in a column (packed volume, 0.3 ml) to 5 ml of 10 mM 2,6-pyridine dicarboxylate solution (pH 5.0) as the chelating agent caused regeneration of the apoenzymes. Zinc(I1) ions in the range of 25 to 250 pM could be calorimetrically measured using 0.5 ml injections. Generally, zinc(I1) ions complexed in the active site of zinc enzymes (e.g., hydrolases) are reversibly exchanged with cobalt(I1) ions, and the cobalt-coordinated enzymes still retain their catalytic activity. We also succeeded in assaying submillimolar levels of cobalt(I1) ions (0.05 to 0.2 mM) using cobalt-substituted enzymes (Satoh, 1989b). In this case, less volume of the chelating agent (2.5 ml) was sufficient for regeneration of the enzymes. Galactose Oxidase as the Recognition Element

Galactose oxidase (EC 1.1.3.9) catalyzes oxidation of D-galactose as follows: D-Galactose + 0, 3 D-galacto-hexodialdose + H,O,

(5)

IKUO SATOH

470

This enzyme isolated from Dactilium dendroidesreadily converted to the apoenzyme under water soluble conditions while it takes a much longer time for reactivation of the apoenzyme. Utilization of the immobilized preparation in combination with the flow-injection technique for monitoring of enzyme activity resolved the problem. Copper(I1) ions were calorimetrically determined in millimolar levels (Satoh, 1990e). Amperometric monitoring with an oxygen electrode and a hydrogen peroxide electrode made the assay more sensitive and rapid (Satoh et al., 1990a). There is room for improvement for a more sensitive response.

B. Apoenzyme Membrane as the Sensing Element Table 1 summarizes successful biosensing of heavy metal ions based on the apoenzyme reactivation method using bead-type immobilized enzymes. Application of the apoenzyme column to flow-injection biosensing demonstrated their long-term stability. Practical use has already been found in clinical and food analyses. In order to obtain a more rapid assay, a contact type of apoenzyme sensor has been developed (Satoh et al., 1992, 1993). Ascorbate oxidase immobilized onto a porous polymer membrane (partially aminated polyacrylonitrile,50 pm thickness, Asahi Chemical Industry Co. Ltd.) was directly attached to a flow-through polarographic oxygen electrode and used as the recognition element for copper(I1)

Table 1. Flow-Injection Biosensing of Heavy Metal Ions Based on the Apoenzyme Reactivation Methods Metal

Zn(I1) Zn(I1) Zn(I1) Zn(I1) Cu(I1) Cu(I1) Cu(I1) CU(I1) Cu(I1) Cu(I1) Co(I1) Co(I1) Co(I1)

Recognition Element

ALP ALP ALP BCA ASOD ASOD ASOD GalOD GalOD GalOD ALP ALP BCA

Monitoring Method

Ranae lmMl

Calonmetry Potentiometry Spectrophotometry Calorimetry Amperometry Calorimetry spectrophotometry Amperometry Amperometry Calorimetry Calorimetry Spectrophotometry Calorimetry

/T /I /P /T 10 /T /P

0.010-1.0 0.010-1.0

0.00014.010 0.0254.25

0.000~.002 0.0014.05

0.00014.010

10

0.1-10.0

/H /T

0.01-10.0 5.0-20.0 O.O&I.O 0.0014.2 0.0054.2

/T /P /T

Nofes: Enzyme:ALP, alkalinephosphatase; ASOD, Ascorbate oxidase; IBCA, Bovine carbonic anhydrase; GalOD,

Galactose oxidase. Method T (thermistor), I (pH-ISFET),P (photomultiplier), 0 (polarographic oxygen electrode), H (hydrogen peroxide electrode).

Biosensing of Heavy Metals

471

ions (Satoh et al., 1992a). Use of the system mounted with the apoenzymemembrane sensor resulted in a more rapid assay (26 min). Furthermore, microdetermination of dual heavy metal ions was tried with hybrid enzymes immobilized onto the same kind of membrane (Satoh, 1993). Alkaline phosphatase and ascorbate oxidase were coimmobilized onto the polymer membrane and tightly attached to a polarographic oxygen electrode. Ascorbate oxidase functioned not only as the recognition element for copper(I1) ions but also as an indicator enzyme for amperometric monitoring of alkaline phosphatase activity on the same membrane. For sensing zinc(I1) ions, the following coupled reactions were amperometrically monitored: L-Ascorbyl-2 phosphate + H,O

-+ L-ascorbate + orthophosphate

(6)

2 L-Ascorbate + 0, + 2 dehydroascorbate + 2 H,O

(7) The apoenzyme membrane was regenerated by pumping cofactor-complexing agents for removing each cofactor, namely copper(I1) and zinc(I1) ion from the catalytic site of alkaline phosphatase or ascorbate oxidase. Thus, zinc(I1) ions in 2 to 200 pM concentrations and copper(I1) ions in 2 to 100 pM concentrations were determined through activation of each of the immobilized apoenzymes on the same supporting membrane.

VI. CONCLUSION Flow-injection microassay of heavy metal ions such as cobalt(II), copper(II), and zinc(I1) ions based on apoenzyme reactivation methods was found feasible with immobilized metalloenzymes. Regardless of the shape of the immobilized metalloenzymes, that is, beads or membrane, this method of unique monitoring is now widely applied in several analytical areas. Further developmental studies towards establishing the generality and versatility of these analytical techniques involving immobilized apoenzymes in flow systems are currently in progress.

REFERENCES Danielsson, B. (1990). Calorimetric biosensors. J. Biotechnol. 15, 187-200. Danielsson, B. & Mosbach, K. (1988). Enzyme thermistors. In: Methods in Enzymology (Mosbach, K., Ed.), Vol. 137, pp. 181-197. Academic Press, New York. Mattiasson, B., Nilsson, H., & Olsson, B. (1979). An apoenzyme electrode. J. Appl. Biochem. I , 377-3 84. Risinger, L., Ogren. L., & Johansson, G. (1983). Determination ofzinc(I1) ions with a reactor containing immobilized carboxypeptidase A in a flow system. Anal. Chim. Acta 154,251-257. Satoh, I. (1989a). Biosensing using calorimetric devices. In: Chemical Sensor Technology (Seiyama, T., Ed.), Vol. 2, pp. 26%282. Kodansha Ltd., Tokyo, Japan. Satoh, I. (1989b). Continuous biosensing of heavy metal ions with use of immobilized enzyme-reactors as recognition elements. In: Proceedings of the MRS International Meeting on Advanced Materi-

472

IKUO SATOH

als, Vol. 14 (Biosensors, Karube, I., Ed.), pp. 4550. Material Research Society, Pittsburgh, Pennsylvania, USA. Satoh, I. (19904. Apoenzyme reactivation microassay for zinc(1I) ions with flow-through transducers. In: Proceedings of the 3rd International Meeting on Chemical Sensors, pp. P/lO&P/l07. The Organizing Committee of the 3rd International Meeting on Chemical Sensors, Cleveland, OH, USA. Satoh, I. (1990e). Calorimetric biosensing of heavy metal ions with the reactors containing the immobilized apoenzymes. AM. N.Y. Acad. Sci. 613,401404. Satoh, I. (1991d). An apoenzyme thermistor microanalysis for zinc(I1) ions with use of an immobilized alkaline phosphatase reactor in a flow system. Biosensors and Bioelectronics6(4), 375-379. Satoh, I. (1991a). Flow-injection calorimetry of heavy metal ions using apoenzyme-reactors. Netsusokutei (Calorimetry and Thermal Analysis) (in Japanese) 18(2), 89-96. Satoh, I. (1991e). Flow-injection microdetermination of heavy metal ions using a column packed with immobilized apoenzyme beads. J. Flow Injection Anal. 8(2), 111-126. Satoh, I. (1993). Amperometric biosensing of heavy metal ions using a hybrid type of apoenzyme membrane in flow streams. Sensors and Actuators, in press. Satoh, I. (1992b). Use of immobilized alkaline phosphatase as an analytical tool for flow-injection biosensing of zinc(I1) and cobalt(I1) ions. AM. N.Y. Acad. Sci. 672,240-244. Satoh, I., Abe, R., & Nambu, T. (1988). Bioelectrochemical sensing of copper(I1) ions using an immobilized apoenzyme column. Denki Kagaku 56(12), 10451049. Satoh, I. & Aoki, Y. (1990d). Biosensing of zinc(I1) ions using an apoenzyme reactor and an ISFET detector inflow streams. DenkiKagaku58(12), 1114-1118. Satoh, I., Ikeda, K., & Watanabe, N. (1986). Microanalysis of zinc@) ion by using an apoenzyme thermistor. In: Proceedings of the 6th Sensor Symposium (Takahashi, K., Ed.), pp. 203206. Institute of Electrical Engineers of Japan, Tokyo, Japan. Satoh, I., Itoh, H., & Anzai, H. (1992a). Flow-injection amperometric biosensing of copper (11) ions using a contact-type of an apoenzyme sensor. In: Proceedings of the 2nd World Congress on Biosensors, pp. 1 8 H 9 0 . Elsevier Advanced Technology, Oxford. UK. Satoh, I., Kasahara, T., & Goi, N. (199Oa). Amperometric biosensing of copper(I1) ions with use of an immobilized apoenzyme reactor. Sensors and Actuators BI, 499-503. Satoh, I., Kimura, S., & Nambu, T. (1987). Biosensing of copper(I1) ions with an apoenzymethermistor containing immobilized metalloenzymes in flow system. In: Digest of Technical Papers, The 4th International Conference on Solid-state Sensors and Actuators (Transducers '87, Matsuo, T., Ed.), pp. 789-790. Institute of Electrical Engineers of Japan, Tokyo, Japan. Satoh, I. & Masumura, T. (1990b). Flow-injection biosensing of zinc(I1) ions with use ofan immobilized alkaline phosphatase reactor. In: Technical Digest of the 9th Sensor Symposium (Sasaki, A., Ed.), pp. 197-200. Institute of Electrical Engineers of Japan, Tokyo, Japan. Satoh, I. & Nambu, T. (1991b). Flow-injection photometric biosensing of copper(I1) ions with use of an immobilized ascorbate oxidase column. In: Technical Digest of the 10th Sensor Symposium (Nakamura, T., Ed.), pp. 77-80. Institute of Electrical Engineers of Japan, Tokyo, Japan. Satoh, I. & Yamada, Y. (1991~).Flow-injection biosensing of cobalt(I1) ions with use of an immobilized alkaline phosphatase reactor in a flow system. In: Digest of Technical Papers of the 6th International Conferenceon Solid-state Sensorsand Actuators (Transducers '91, Chang, S.-C., Ed.), pp. 699-702. The Institute of Electrical and Electronics Engineers, Inc. Piscataway, NJ, USA. Townshend, A. & Vaugham, A. (1970). Application of enzyme-catalysed reactions in trace analysis - V Determination of zinc and calcium by their activation of the apoenzyme of calf-intestinal alkaline phosphatase. Talanta 17,289-298. Vallee, B.L. (1980). Zinc and other active metals as probes of local conformation and htnction of enzymes. Carlsberg Res. Commun., 15,423-441. Wagner, F.W. (1988). Preparation of metal-free enzymes. In: Methods in Enzymology (Riordan, J.F. & Vallee, B.L., Eds.), Vol. 158 A, pp. 21-32. Academic Press, San Diego.

DESIGN OF HIGH-ANNEALlNG-TEMPERATURE PCR PRIMERS AND THEIR USE IN THE DEVELOPMENT OF A VERSATILE LOW-COPY-N UMBER AMPLI F ICATION PROTOCOL

Michael W. Meckl enburg

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. MATERIALS AND METHODS . . . . . . . . . . . . . . . . . . . . . . . . A. Oligodeoxynucleotide Synthesis and Purification . . . . . . . . . . . . . B. Purification of HCMV DNA . . . . . . . . . . . . . . . . . . . . . . . . C. Preparation and Use of PCR Master Mix . . . . . . . . . . . . . . . . . D. The Amplification Procedure . . . . . . . . . . . . . . . . . . . . . . . . E. Analysis of the PCR Amplified Material . . . . . . . . . . . . . . . . . .

Advances in Molecular and Cell Biology Volume 15B, pages 473-490. Copyright 8 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

473

474 474 476 477 477 477 478 479

MICHAEL W. MECKLENBURG

4 74

111. RESULTS AND DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . A. Primer Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Amplification Protocol Design . . . . . . . . . . . . . . . . . . . . . . C. Detection of HCMV . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

,480 .480 .482

.484 ,486

ABSTRACT High-annealing-temperature (HAT) primers for polymerase chain reaction (PCR) were designed and tested. These HAT primers were used to develop an amplification procedure that employs hypervariable extension times in combination with two temperature amplification cycles, master mixes and hot start to achieve highly reproducible, low-copy-number PCR amplification. The employment of HATprimers drastically reduced cycling times and essentially eliminated nonspecific amplification products even in the presence of vast excesses of nonspecific DNA sequences. Hypervariable extension times provided a simple, noninvasive approach for dynamically modifying the reaction environment during amplification. In addition, this procedure makes extremely efficient use of enzyme, reducing the amount of Taq or Replitherm polymerase to 0.4 and 0.12 U per 50 p1 reaction, respectively. The use of PCR master mixes increased the reproducibility and portability of the assay. And finally, a modified form of hot sturt was employed to reduce primer oligomer formation. HAT primers were designed for the amplification of human cytomegalovirus (HCMV). These primers were used to develop a highly specific, low-copynumber PCR assay that employed a combined annealing/extension temperature of 70 "C or 72 "C. Nonspecific bands, other than primer oligomer bands, were not detected even in the presence of a vast excess of human genomic DNA. These primers can be employed to calibrate the temperature on different thermocylcers as well as to take into account environmental factors that influence amplification. This would be especially advantageous when porting protocols to different machines and simplifying the use of PCR in field studies as well as for identifying factors that influence the local stability at the 3' end of primers.

1. INTRODUCTION Advancements in DNA probe technology have led to the development of simple, nonisotopic detection systems for use in basic research and clinical diagnostics (Wetmur, 1991; Fernandes and Coffman, 1992; Kricka, 1992; Kessler, 1993). Although numerous ideas have been put forward to improve the sensitivity of nucleic acid based detection assays, none has had a more dramatic effect than PCR (Saiki et al., 1985; Innis et al., 1990). In essence, PCR takes advantage of the inherent genetic capability that nature has endowed upon nucleic acids, namely its ability to replicate itself. This ability, upon which all life depends, is both obvious in its simplicity and staggering in its implications. Since its introduction, PCR has

Design and Use of HAT PCR Primers

4 75

transformed the way DNA analysis is carried out in both research and clinical laboratories. In this procedure, the target DNA sequences are amplified by sequential rounds of polymerization using two oligodeoxynucleotide (ODN) primers complimentary to opposite strands of the target region. The procedure can amplify a single target sequence up to lo9 molecules in a few hours. PCR is now routinely employed to manipulate and detect specific DNA sequences. In many cases, the sensitivity of the PCR technique has surpassed that of even the most sensitive ELISAs and biological assays (Arnhein and Erlich, 1992; Rapley et al., 1992; Lew et al., 1992; Rolfs et al., 1993). PCR demonstrated the fundamental principles involved in nucleic acid based amplification and it was not long before variations on this basic theme were developed (Kwok et al., 1989; Wu and Wallace, 1989; Duck et al., 1990; Guatelli et al., 1990; Barany, 1992; Walker et al., 1992; Segev, 1993). The potential of these new nucleic acid based amplification techniques has only begun to be assessed, but they will undoubtedly lead to the development of additional low-copy-number detection systems (Gingeras et al., 1991; Bush et al., 1992; Kalin et al., 1992; Carpenter et al., 1993; Devlin et al., 1993). PCR has also been used as a generalized reporter system (Sano et al., 1992; Mecklenburg, 1995a). The realization that nucleic acid based amplification procedures can be used as highly sensitive universal reporter systems will have dramatic effects both in basic research and clinical diagnostics. Nucleic acid amplification procedures have provided us with techniques that allow the detection ofjust a few copies of a specific nucleic acid sequence. Employment of these techniques has the potential to significantly increase the sensitivity of immunoassays, as well as for nucleic acid probe detection. Nucleic acid based amplification techniques are dependent upon the cooperative, reversible nature of DNA duplex formation. The ability to accurately predict the overall stability of duplexed ODN from base sequence data is an essential prerequisite in the development of a knowledge based strategy for the construction of high-annealing-temperature PCR primers. Most precise methods for calculating helix stability are based upon empirically determined nearest neighbor thermodynamic parameters using di- or trinucleotides (Breslauer et al., 1986: Freier et al., 1986). These methods are surprisingly accurate at predicting the T, of DNA and RNA duplexes. However, not all the sequences that function efficiently as probes make good PCR primers. This is, at least in part, due to the fact that PCR primers are not only required to hybridize but are also required to function as efficient substrates for primer extension. It has been shown that the initiation of polymerization is highly dependent upon the stable association of the 3' end of the primer with the template (Rychlik and Rhoads, 1989). It is, therefore, not surprising that primers designed using standard probe methodology do not always function efficiently in PCR. Thermal denaturation data derived from di- and trinucleotide studies reflect the average dynamic equilibrium over the entire length of the DNA

476

MICHAEL W. MECKLENBURG

duplex and has no mechanism for taking into account the local stability at the 3' ends of DNA duplexes. This issue has led to the development of a number of computer programs for the design of PCR primers (Lowe et al., 1990; Hillier and Green, 1991; Lucas et al., 1991; Montpetit et al., 1992). These programs analyze parameters such as hair-pin formation, overall GC content, primer-primer annealing, etc. Primer annealing temperatures are calculated by, in one way or another, skewing the standard nearest neighbor analysis predictions so as to take the stability at the 3' ends of the primers into account.These methods can therefore only serve as guidelines for the construction of PCR primers. Refinement of these methods requires expanding our understanding of the local stability of primers in duplexes. Ideally, one would like to develop a nearest neighbor analysis that employs a measurement parameter that is sensitive to the stability of the terminal 3' nucleotides in DNA duplexes. This in turn requires the development of a technique that is capable of quantitating the local stability at the 3' ends of primers hybridized to their templates. One approach would be to perform a thermal amplification profile for a particular primer pair and determine the temperature at which amplificationno longer occurs. The presence of specific PCR amplification product would be used as the indicative measurement parameter. A T,, could be derived in much the same way T, is derived for the thermal denaturation of primers (Mecklenburg, 1995b).The empirical TPCRdata derived from systematically exchanging the 3' terminal nucleotides in the primerltemplate would provide the data required for the development of a PCR-based nearest neighbor analysis. The accurate determination of TPCRwould require the development of a highly reproducible amplification procedure and thermocycling conditions. Here, I present the integration of HAT PCR primers into a highly standardized amplification protocol capable of amplifying low-copy-number sequences with a high degree of reproducibility. HCMV was used as the model system in these studies. HCMV is a member of the herpes group and is a ubiquitous human pathogen that causes a wide variety of ailments (Stagno et al., 1982). HCMV has a genome of linear double-stranded DNA of about 235 kb. The detection of the virus using PCR is of clinical interest (Pillay and Griffiths, 1992; The et al., 1992).

II. MATERIALS AND METHODS The Taq thermostable polymerase and the human genomic DNA were obtained from Promega. Replitherm, a thermostable polymerase, was obtained from Epicenter. The T7 DNA polymerase and BSA (RIA grade) were purchased from US Biochemical Corp. Agarose was supplied by Sigma. The nucleotide triphosphates digoxygenin-labeleddUTP, alkaline phosphatase labeled antidigoxygeninantibodies and salmon sperm DNAwere purchased from Boehringer Mannheim. Low-peroxide Tween 20 and NP-40 (protein grade) were bought from Calbiochem. The

Design and Use of HAT PCR Primers

477

AMPPD [di-sodium 3-(4-methoxyspiro-(1,2-dioxetane-3-2'-tricylo[3.3.1.13.7]decan)4-yl]phenyl phosphate) was obtained from Tropix, USA. The single-sided X-ray film was bought from Kodak. The Zetaprobe hybridization membrane was purchased from Bio-Rad. All other reagents were purchased from Merck and were ofpro analysis grade. Phenol (Merck, p.a.) was redistilled as follows: 10 pl of dH,O was added to 100 g of solid phenol and melted at 65 "C. Once melted, 0.5 g tin powder, 100 mg NaHCO, and 10 boiling chips were added and the distillation started. After the temperature had reached 170 "C, the phenol was collected in a dark glass bottle containing 0.1 M Tris buffer, pH 8, and stored at -20 "C. A. Oligodeoxynucleotide Synthesis and Purification

The ODNs were synthesizedon an Applied Biosystems 380A synthesizer by the author at the Biomedical Center Unit, Lurid, Sweden according to the instructions supplied by the manufacturer. ODNs were purified by ion exchange fast protein liquid chromatography (FPLC) on a Mono Q column (Pharmacia LKB Biotechnology, Inc.) using a 0.05 to 1 M potassium phosphate, pH 6.5 gradient containing 20% acetonitrile.The samples were concentrated on a Speed-Vacand desalted using an ultrafiltration unit with a molecular weight cut-off of 5,000 daltons (Ultrafree, Millipore). The concentration was determined by measuring the OD,,,, and the volume was adjusted to give a final concentration of 0.5 mg/ml.

B. Purification of HCMV DNA The HCMV was grown on human foreskin fibroblasts according to standard procedures (DeMarchi and Kaplan, 1976; Stinski, 1978). The HCMV genomic DNA was isolated as described (Stinski et al., 1979; DeMarchi, 1981; Sambrook et al., 1989). The virus preparation was extracted three times with an equal volume of phenokhloroforrn (1: 1, vol/vol) and once with chloroform-isoamyl alcohol (24: 1, vol/vol). The DNA was precipitated by the addition 0.5 volumes of 7.5 M ammonium acetate and 2 volumes of 96% ethanol. After centrihgation, the pellets were washed with 70% ethanol, dried, and resuspended in DNA buffer (1 mM EDTA, 10 mM Tris, pH 8). The DNA concentration was determined by measuring the OD,,,. C. Preparation and Use of PCR Master Mix

In order to reduce the risk of carry-over contamination a number of precautions were taken. Stock solutionswere prepared in sterile tissue culture facilitiesthat had not been exposed to amplified material. The solutions were prepared using individually wrapped UV irradiated plastic disposables,baked glassware (300 OC, 4 h) or autoclaved materials, which have been exposed to UV irradiation packaged in UV transparent material (15 W, 30 cm, 60 min). Only pipette tips with aerosol

478

MICHAEL W. MECKLENBURG

barriers were used (Continental Labs, California, USA).The buffer components were prepared from bottles specifically for PCR use only. Stock solutions of Tris (1 M, pH 8.4), KC1 (1 M), MgCl, (1 M), BSA (10 mg/ml), dNTPs (10 mM of each deoxynucleotide), and primers (0.5 mg/ml) were prepared. The 10 mM dNTP solution was prepared in 10 mM Tris, pH 8 using the sodium salt of the nucleotide triphosphates, except in the case of dGTP in that the lithium salt was used and the pH was adjusted to 7 with LiOH. The solutionswere aliquoted into microfuge tubes in appropriate volumes and stored at -20 "C. Each tube was used only once and then discarded. Different master mix preparations were used for the Taq and Replitherm polymerases. One milliliter of Taq master mix was prepared by combining the following: 20 p1 Tris, 50 p1 KCl, 4 pl MgCl,, 20 p1 BSA,50 pl dNTP, and 772 p1 distilled water. One milliliter of the Replitherm master mix was prepared as follows: 100 pl lox buffer from Epicenter (15 mM MgCl,, 500 mM KC1,0.05% Tween 20,0.05% NP-40,0.01% gelatin, and 100 mM Tris, pH 8.3), 10 pl Tris, 2.5 pl MgCl,, 20 p1 BSA, 50 pl dNTP, and 733.5 pl distilled water. It is essential that the lox buffer from Epicenter be used for the preparation ofthe Replitherm master mix. Typically, 50 ml of master mix was prepared, aliquoted into 916 pl portions (equivalent to 1 ml or 20 x 50 pl reactions) and stored at -20 "C. The primers ( 2 pl of each primer/$ 16 pl portion) were added subsequently in order to increase the flexibility and portability of the master mixes. Each primer pair received a unique number that was used to identify the master mixes once primers have been added. The master mixes were stable for at least 9 months when stored at -20 "C. The mixes could be freeze-thawed at least three times without any detectable loss of sensitivity.

D. The Amplification Procedure The amplification was performed using the in the tube temperature sensing thermocycler from Cambio, Cambridge, UK.The calibration was performed using a digital microthermometer (Tradoterm, Sweden). A drop of paraffin oil was added to each well to ensure proper thermal contact. In the standard protocol, 46 pl of the master mix was transferred into a 1.2 ml microfbge tube (Sarstedt). The 2 pl DNA sample was added, followed by two drops of baked paraffin oil (300 "C, 4 hr). The tubes were placed in the thermocycler and the denaturation step initiated. After denaturation, 2 pl of the diluted Taq polymerase (0.2 U/pI) or Replitherm (0.06 U/pl) were added to each tube. The reaction tubes remained in the thermocycler during the addition of polymerase (Ward et al., 1989). Two temperature PCR was employed in these studies. The standard procedure employed 60 cycles. The anneaIing/extension temperature was 70 "C and the cycling denaturation temperature was 92 "C. The annealing/extension step was carried out for 15 s for the first 25 cycles, 60 s for the next 30 cycles and 420 s for the last 5 cycles. The denaturation step was carried out for 15 s throughout the

Design and Use of HAT PCR Primers

479

amplification procedure, except for the initial denaturation step that was carried out for 5 min at 95 OC.

E. Analysis of the PCR Amplified Material Analysis of the amplified material was performed in separate, self-contained facilities that were UV irradiated on a routine basis. The equipment used in this facility was dedicated for this purpose. A method of testing for contamination was developed. Asmall piece of W-treated ( 15 min on a Transilluminator model TM36, W P , California, USA) filter paper (2 x 3 mm) was used to wipe the surface to be tested. The paper was placed in a microhge tube and a standard PCR amplification was run. The detection limit of the contamination test was 200 copies. The test was run once a week at 10 predetermined locations and 10 randomly chosen locations. The procedure was also used to identify contaminated individuals. The PCR amplification products were analyzed on 3% agarose gels. The gels were prepared as follows: 100 mL of TAE buffer (1 mMEDTA, 40 mM Tris-acetate, pH 8) was thoroughly degassed and 3 g of agarose were added. The aluminum foil covered beaker was swirled immediately prior to being microwave irradiated for 3 min at maximum intensity (800 W). The gel was stirred while cooling to 55 "C and 5 pL of an EtBr solution (1 0 mg/mL,) was added. The gel was poured and allowed to cure at 4 "C for 30 min. The PCR-amplified material (9 p1) was mixed with 1 pl of the dye buffer (0.0 1% bromphenol blue, 30% glycerol, 50 mM EDTA, pH 8) on a virgin piece of parafilm. The sample was loaded onto the gel and electrophoresis carried out for 2 h at 90 V. The size marker was MspI cut pBR322 (New England Biolabs). The DNA bands were visualized on a UV table and photographed with a Polaroid camera using Polaroid 665 film. The Southern transfer onto the Zeta Probe membranes was carried out according to the manufacturers' instructions. The gel was soaked for 15 min in transfer solution (0.4 N NaOH, 1.5 M NaC1) and blotted onto to the nylon membrane. The membrane was rinsed with neutralizing solution (1 M NaC1, 0.5 M Tris, pH 7.2), dried at room temperature for 60 min, and finally W irradiated face down on a Transilluminator UV light table for 90 s (Khadjian, 1987; Chou et al., 1992). The end labeling of the OD" with digoxygenin-tagged dUTP was performed using a terminal deoxynucleotide transferase labeling kit (Boehringer Mannheim). The blocking, hybridization, and immunological procedures were performed as described (Boehringer Mannheim). The membranes were blocked for 1 h at 55 "C in hybridization solution (4 x SSPE, 1% sodium lauryl sulfate, 1% (wh) powdered nonfat dry milk, and 100 yg/ml sheared salmon sperm DNA). The solution was replaced with fresh hybridization solution containing 20 ng/ml digoxygenin-labeled probe and incubated at 55 "C for 2 h with agitation. The membranes were rinsed twice with hybridization solution at room temperature. Two high stringency washes were performed at 55 "C using 0.2 x SSPE containing 0.1% sodium lauryl sulfate for 20 min (Sambrook et al., 1989). The membranes were rinsed twice with

MICHAEL W. MECKLENBURG

480

antibody buffer (0.1% Tween 20,0.15 M NaC1, 0.1 M Tris, pH 7.5), blocked for 30 min with antibody buffer containing 0.5% (w/v) powdered nonfat dry milk. The solution was replaced with fresh antibody buffer containing alkaline phosphatase labeled antidigoxygenin antibodies (1 50 mU/mL) and incubated for 1 h at room temperature. The membranes were washed 3 x 10 min in antibody buffer and then rinsed twice with assay buffer (50 mM MgCI,, 0.1 M NaC1, 0.1 M Tris, pH 9.5). The membrane was then wetted with assay buffer containing 0.1 m g / d AMPPD, wrapped in plastic wrap, and exposed to X-ray film for 2 h (Bronstein et al., 1990).

111. RESULTS AND DISCUSSION PCR is very sensitive to the stability of the DNA duplex at the 3' ends of primers. This makes PCR an ideal model system to study the subtle factors that influence local denaturation at the 3' ends of primers hybridized to target templates. This requires the development of techniques that can be used to calculate the ability of This in turn requires a primer pair to amplify as a function of temperature, T,. the development of a highly reproducible amplification procedure and thermocycling conditions. The purpose of this investigation was to develop an amplification protocol that can be used as a tool for accurately and reproducibly determining the ability of primer pairs to amplify as a function of the annealing temperature. The procedure employs hypervariable extension times in combination with HAT primers, two temperature amplification cycles, master mixes and hot start to achieve low-copynumber sensitivity. Two HAT primer pairs were designed and used to develop the amplification protocol for the detection of HCMV described here. A. Primer Design

The primer pairs were chosen to amplify regions ranging in size from 200 to 500 basepairs, having a GC content ranging from 60% to 65% and a theoretical melting temperature of about 78 "C (Suggs et al., 1981). In addition, a number of other standard parameters were also checked, such as hairpin formation, 3' end homologies, GC content balancing and a general control for homologous sequences in the EMBL nucleic acid database (Williams, 1989; Rychlik et al., 1990; Lowe et al., 1990). The minimal fragment size was chosen to be 200 basepairs for two reasons. First, the analysis of fragments below this size requires the use of polyacrylamide gels or very high agarose concentrations, both of which increase the labor and expense of the analysis. Second, many primer pairs produce a detectable primer oligomer band(s) when low-copy-number amplification protocols are employed (Li et al., 1990; Chou et al., 1992). These oligomer bands are typically 50 to 100 basepairs in length. Thus, by maintaining a minimum fragment size of 200 basepairs, nonspecific primer dimer bands can easily be differentiated from specific amplifi-

Design and Use of HAT PCR Primers

48 1

cation products. The maximal fragment size was chosen to allow the employment of short extension times and to minimize the risk that minor degradation of the template would affect the sensitivity of the assay. The GC content of the primers was chosen to lie between 60% and 65% in order to minimize primer length. A theoretical annealing temperature of 78 "C was chosen because it was the maximal temperature at which an acceptable number of primers in the size range 22 to 26 nucleotides could be selected in the HCMV regions targeted for analysis. These constraintsprovided a way of maintaining the overall uniformity between primers, which simplified performing and analyzing their thermal amplification profiles. The use of HAT primers was considered to be of central importance for several reasons. First, increasing the annealing temperature reduces the maximal temperature differential during the amplification cycle, that in turn reduced cycling times. Second, primer pairs that are capable of amplifying at 65 "C or higher can use a two-temperature cycling scheme in that the annealing and extension steps are combined (Kim and Smithies, 1988).This combined annealing/extension step shall hereafter be referred to as the annealing step. In order to maximize the efficiency of this scheme, the annealing temperature should be as close to 75 "C as possible (the temperature at that thermostable polymerases have maximal enzymatic activity). Third, the employment of HAT primers generally improved sensitivity, increased reproducibility, and reduced nonspecific amplification products (Chou et al., 1992; Mecklenburg, 1995b). Two distinct regions of the HCMV genome were targeted (Table 1). Each region has been shown to code for mRNAs that are expressed during viral infection (Stenberg et al., 1984; Hutchinson and Tocci, 1986). Expressed regions of the HCMV genome were chosen since they are generally more genetically stable. In addition, placement of the 3' end of a primer at the third position of the tRNA recognition sequence, the wobble codon, was avoided for similar reasons. The two primer pairs, 9/5 and 13/7chosen for this studyuse an in-the-tubeannealingtemperature of 70 "C and 72 "C, respectively.

Table 1. HCMV Primer and Probe Sequences Prirner/Probe Sequence 5'-3' Name

MMC9 MMC5 probe: MMC6 MMC13 MMC7 pobe: h4MC8

AGGTTCGAGTGGACATGGTGCGGC AGCGGCGCCCTTGCTCACATCATG GGGAGGATGTTTGCAGAATGCCTTAGATATC ACAAGGCGTTGTCAAGCGTGCGGC CGCCGCAGCTGTGGCAGTTAACGT CGTGGGTGGTGCGAGAGTACACGATGGGTG

Target Primary Length(nt) Source

278

this paper

224

thispaper

482

MICHAEL W. MECKLENBURG

In order to obtain maximal sensitivity using the amplification protocol described here, a thermal annealing temperature profile must be empirically determined for each new primer pair. This profile is constructed by performing amplifications at 5 "C intervals until the maximal annealing temperature is found. A detailed thermal profile is then performed at 2 OC intervals in the appropriate region. The optimal annealing temperature is usually chosen to be 4 "C below the temperature at which the PCR signal disappears completely. This approach greatly simplifies the process of porting amplification procedures to different environments and/or thermocyclers. New systems are calibrated simply by determining the temperature profile for a single primer pair. Once calibrated, other primer pairs then merely need be adjusted accordingly. One could envisage the development of a series of primer pairs that could be used to compare and calibrate thermocyclers over a range of temperatures. This would be ideal for calibrating portable systems and would be especially advantageous in field studies. B. Amplification Protocol Design

All low-copy-number amplification protocols must address the fact that the conditions in the reaction tube are constantly changing during the amplification. A balance must be maintained between the synthetic capacity of the system and the amount of synthesis required during any given cycle. Excess synthetic capacity will lead to the development of nonspecific amplification products, whereas insufficient synthetic capacity will result in a decrease in sensitivity. In order to determine the amount of synthetic capacity required during any given cycle, a number of factors such as the size of the fragment to be extended, the concentration of primerhemplate complex, the relative activity of the polymerase at a particular annealing temperature, the effect of thermal inactivation of the polymerase during amplification, and the length of the extension step must be taken into account. The synthetic capacity and the amount of synthesis required during a given cycle must be balanced. During the initial rounds of amplification, primers and Taq polymerase are in excess, while only a few copies of target DNA are present. The combination of excess synthetic capacity and primers drastically increases the chance that nonspecific primer oligomers and/or other nonspecific amplification products will form. Once formed, these eficiently amplified products compete with the target fragment for primers and enzyme. In low-copy-number reactions the formation of primer oligomers is the single most important factor affecting low-copy-number amplification reactions (Li et al., 1990; Mullis, 1991; Chou et al., 1992). Primer oligomer formation must be suppressed until a critical concentration of the target sequence has been reached. Low-copy-number amplification protocols must incorporate a mechanism that maintains maximal sensitivity while minimizing the risk for primer oligomer forination. This requires adjusting the synthetic capacity of the reaction mixture during amplification.

Design and Use of HAT PCR Primers

483

A number of approaches have been developed to resolve this problem such as multiple additions of Taq polymerase (Powell et al., 1987; Kim and Smithies, 1988), booster PCR (Ruano et al., 1989), reamplification (Higuchi et al., 1988; Li et al., 1990), nested PCR (Mullis and Faloona, 1987; Haqqi et al., 1988; Kemp et al., 1989; Albert and Fenyo, 1990) and heminested PCR (Li et al., 1990; Witsell et al., 1990). These approaches involve either the addition of components during the amplification reaction or performing a second amplification reaction. This increases the labor involved as well as the overall complexity of the amplification procedure. In addition, the manipulation of amplified material may in some cases increase the risk of carryover contamination. The amplification procedure described in this paper employs hypervariable extension times in order to address this problem. The protocol modulates the synthetic capacity by varying the extension time. This provides a noninvasive, programmable protocol that is simple to implement and modify. While other amplification protocols have employed variable extension times (Demmler et al., 1988; Bradbury et al., 1990), none have consequently employed hypervariable extension times in the way described here. In order for this scheme to achieve maximal effect, the largest possible annealing time differential must be employed. This requires determining the smallest time interval necessary to achieve a stable in-the-tube temperature at every position in the thermocycler using standard cycling temperatures. External thermocouple devices were employed to empirically determine this value. The minimal time was determined to be 8 s (+/- 1 "C). In order to ensure a high degree of reproducibility the minimum time was chosen to be 15 s. The thermocycler was set to start counting the time when the in-the-tube temperature was within 1 O C of the set temperature. The importance of testing across-the-block temperature variation cannot be overstated (Resendez-Perez and Barrera-Saldana, 1990). Over 100 different time extension protocols were tested with a number of different HAT primer pairs (two of which are described in this paper). The scheme that resulted in the most consistent

Table 2. Amplification Cycling Scheme # of Cycles

1

Time (s)

T (''c)

300

95

HOLD at 70 "C, add Taq polymerase 25 30 5

15 15 15 60 15 420

92 70172 92 70172 92 70172

484

MICHAEL W. MECKLENBURG

amplification of low-copy-numbersamples is shown in Table 2. The annealing time was varied from 15 to 420 s. Primer oligomer formation during the addition of polymerase has been shown to markedly influence the sensitivity and reproducibility of low-copy-number amplification procedures (Li et al., 1990; Mullis, 1991; Chou et al., 1992). In order to achieve low-copy-number sensitivity and a high degree of reproducibility, it is absolutely necessary to use some sort of hot start procedure. The protocol described here employs a manual hot start procedure that involves inserting a hold step at 70 "C after the denaturation step. It is during this hold step that polymerase is added to each tube. It is important that the tubes remain physically seated in the thermocycler block during this procedure. When this procedure is employed with primers, that have annealing temperatures of at least 70 "C, primer oligomer formation that interferes with the amplification of the specific band has not been detected. The use of HAT primers also increases the overall efficiency of the protocol by allowing the incorporation of a combined annealing/extension step. This not only decreases the cycling time but also increases the eaciency of the amplification procedure. This is due to the fact that a three temperature cycle requiring primers (primers with annealing temperatures below 65 "C) must be extended at least partially if they are to remain bound when the temperature is increased to the extension step. In the initial cycles, when the polymerase saturates the primerhemplate complexes all the annealed primers are easily extended (adequately to make them stable during the extension step), even though polymerase has significantly lower activity. However, as the polymerase concentration gradually becomes limiting, the efficiency of the amplification gradually decreases because the number of annealed primers that are not extended sufficiently to make them stable at the extensionstep increases.This effectbecomeseven more pronouncedat lower annealing temperatures since polymerase activity is strongly temperature dependent. High combined annealinglextensiontemperature increases the overall efficiency of the amplification procedure since the primerhemplate complexesremain available for extension during a longer period of time. Moreover, the procedure makes very efficient use of enzymeby graduallyincreasingthe extensiontimes in order to fulfillthe ever increasing demands the system places upon the synthetic machinery. C. Detection of HCMV

A general amplification procedure has been developed based upon using master mixes and the amplification protocol described above. The sensitivity, specificity, and reproducibility of two primer pairs targeted against HCMV were analyzed (Figure IA). The 60 cycle amplification procedure took 3.5 h to complete and gave good yields of product of the expected size. Both primer pairs had low-copy-number sensitivity (Figure lA, lanes 2, 8). On occasion, the intensity of the PCR fragment band can decrease slightly in samples containing 1000 copies or more (see Figure 1, lane 10). The drop-out rate for these primers was tested using samples

Design and Use of HAT PCR Primers

485

Figure 1. (A) An ethidium stained agarose gel showing the DNA products obtained after amplification of HCMV with MMC9/MMC5 (lanes 1-6) and MMCl3/MMC7 (lanes 7-12). No DNA control (lanes 1, 7), 10 copies HCMV (lanes 2, 8), 100 copies HCMV (lanes 3, 9), 1000 copies HCMV (lanes 4, 1O), 1 pg Human DNA (lanes 5 , 11), and 10 copies HCMV + 1 pg Human DNA (lanes 6,12). After blotting the membrane was divided. (6)The MMC9/MMC5 amplificate was probed with MMC6 and (C) the MMCl3/MMC7 amplificate was probed with MMC8 as described in Materials and Methods.

that contained 10 and 1000 copies. No dropouts were detected in the 100 samples that were run at each target concentration (data not shown). In this system primer oligomer bands were always detected in controls that contained no target DNA. The presence of this band was used as a functional control for the master mixes and the polymerase. The presence of nonspecific human DNA did not affect the sensitivityor the specificityofthe amplificationprocedure (Figure 1A, lanes 6,12).

486

MICHAEL W. MECKLENBURC

No amplification products were detected in the presence of human DNA sequences (Figure 1A, lanes 5, 11). Both the Taq and Replitherm polymerases gave equivalent results when used as described in Material and Methods. It is important to note that when a homemade lox buffer was substituted for the Epicenter lox buffer, the concentration of enzyme had to be increased from 0.12 to 0.24 Ui50 p1 in order to obtain equivalent results. At present, no explanation for this can be given. When purified template was employed, low-copy-number sensitivity was obtained using a 45-cycle procedure. In this procedure only ten 15 s-cycles were used. However, the 60-cycle procedure was shown to give more consistent results when amplifying clinical material (Mecklenburg, 1995b). In order to ensure that the amplification products were authentic, Southern analysis was carried out (Figures 1B and C). After blotting the gel, the membrane was cut into two pieces and probed with the appropriate digoxygenin-labeled ODN (see Table 1). The results confirm that the amplification products from both primer pairs were authentic. No spurious bands were detected. A detailed analysis of the effect of annealing temperature on PCR dependency of these primers will be published elsewhere (Mecklenburg, 1995b).

IV. CONCLUSIONS PCR is very sensitive to the stability at the 3' ends of primers. This makes PCR an ideal model system to study the subtle factors that influence local denaturation at the 3' ends ofprimers hybridized to target templates. This requires the development of techniques that can be used to calculate the ability of a primer pair to amplify as a function of temperature, TPCR.This in turn requires the development of a highly reproducible amplification procedure and thermocycling conditions. The protocol described here provides such a tool. Thus, the purpose of this investigation was to develop a highly reproducible amplification protocol as a tool to probe the stability at the 3' ends of DNA duplexes. The procedure employs hypervariable extension times in combination with HAT primers, two temperature amplification cycles, master mixes, and hot start to achieve low-copy-number sensitivity. Hypervariable extension times provide a simple, versatile approach for dynamically modifying the reaction environment. In addition, this noninvasive approach is simple to implement. HAT PCR primers have the added advantage of reducing cycling times and, in general, eliminate nonspecific amplification bands even in the presence of excess nonspecific DNA. In addition, the increase in amplification efficiency that results from employing this protocol reduces the amount of thermostable polymerase required in the amplification procedure. The HCMV was chosen as the model system in these studies because detection of the virus using PCR is of clinical interest. Two HAT primer pairs were designed and used to develop the amplification protocol for the detection of HCMV described

Design and Use of HAT PCR Primers

487

here. The primer pairs described here were shown to have low-copy-number sensitivityand to be specific for HCMV. No spurious bands, other than primer dimer bands, were detected even in the presence of a vast excess of human genomic DNA. A highly reproducible PCR amplification protocol has been developed with low-copy-number sensitivity. In order to obtain optimal sensitivity, a thermal annealing temperature analysis must be performed for each new primer pair. In addition to increasing the sensitivity of the assay, this approach greatly simplifies the process of porting amplification procedures to different environments andor thermocyclers.The new system is calibratedsimply by determining the temperature profile for a single primer pair. The other primer pairs then merely need be adjusted accordingly. One could envisage the development of a series of primer pairs that could be used to calibrate thermocyclers, an amplification based thermometer. This would provide a built-in system for the calibration of portable systems that would be especiallyadvantageous in field studies. Moreover, this design strategy enhances the reproducibility and the portability of the amplification protocol. The development of a nearest neighbor analysis employing a measurement parameter sensitive to the stability of the terminal 3’ nucleotides in DNA duplexes would provide a very useful tool for studying DNA duplex stability. This would compliment more standard nearest neighbor analyses and provide a more accurate picture of local DNA duplex stability. Results from these studies would be of great interest to those working with other nucleic acid based amplification procedures, large scale sequencing projects, as well as those involved in studying replication and repair mechanisms.

REFERENCES Albert, J. & Fenyo, E.M. (1990). Simple, sensitive, and specific detection of human immunodeficiency virus type 1 in clinical specimens by polymerase chain reaction with nested primers. J. Clin. Microbiol. 28, 15661564. Amheim, N. & Erlich, H. (1992). Polymerase chain-reaction strategy. Ann. Rev. Biochem. 61, 131-156. Barany, F. (1991). Genetic disease detection and DNA amplification using cloned thermostable ligase. Proc. Natl. Acad. Sci. USA 88, 18s-193. Boehringer Mannheim (1989). Dig DNA labeling and detection nonradioactive: applications manual. Biochemica, Beohringer Mannbein, Germany. Bradbury, M.W., isola, L.M., & Gondon. J.W. (1990). Enzymatic amplification of a Y chromosome repeat in a single blastomere allows identification of the sex preimplantation mouse embryos. Proc. Natl. Acad. Sci. USA 87,4053-4057. Breslauer, K.J., Frank, R., Blocker, H.. & Marky, L.A. (1986). Predicting DNA duplex stability from the base sequence. Proc. Natl. Acad. Sci. USA 83,37463750. Bronstein, I., Voyta, J.C., Lazzari, K.G., Murphy, O., Edwards, B., & Kricka, L.J. (1990). Rapid and sensitive detection ofDNAin Southern blots with chemiluminescence. Biotechniques 8,3 10-3 14. Bush, C.E., Donovan, R.M., Peterson, W.R., Jennings, M.B., Bolton, V, Sherman, D.G., Vnader-Brink, K.M., Beninsig, L.A., & Godsey, J.H. (1992). Detection of human immunodeficiency virus type 1 RNA in plasma samples from high-risk pediatric patients by using self-sustaining sequence replication reaction. J. Clin. Microbiol. 30, 28 1-286.

488

MICHAEL W. MECKLENBURG

Carpenter, W.R., Schutzbank, T.E., Tevere, V.J., Tocyloski,K.R., Dattagupta, N. andYeung,K.K. (1993). A transcriptionally amplified DNA probe assay with ligatable probes and immunochemical detection. Clin. Chem. 39, 193&1938. Chou, Q., Russel, M., Birch, D.E., Raymond, J., &Bloch. W. (1992). Preventionofpre-PCRmis-priming and primer dimerization improves low-copy-number amplifications. Nucleic Acids Res. 20, 1717-1723. DeMarchi, J.M. & Kaplan, S.A. (1976). Replication of human cytomegalovims DNA: lack ofdependence on cell DNA synthesis.J. Virol. 18, 10631070. DeMarchi, J.M. (1981). Human cytomegalovims DNA: restriction enzyme cleavage maps and map locations for immediate-early, early, and late RNAs. Virology 114,2>38. Demmler, G.J., Buffone, G.J., Schimbor. C.M., & May, R.A. (1988). Detection of cytomegalovirusin urine from newborns by using polymerse chain reaction DNA amplification.J. Infect. Dis. 158, 1177-1 184. Devlin, R., Sfudhohe, R.M., Dandliker, W.B., Fahy, E., Blumeyer, K., & Ghosh, S.S. (1993). Homogeneous detection of nucleic acids by transient-state polarized fluorescence. Clin. Chem. 39, 193el943. Duck, P., Alvarado-Urbina, G., Burdick. B., & Collier, B. (1990). Probe amplifier system based on chimeric cycling oligonucleotides. Biotechniques 9, 142-1 49. Femandes, J.J. & Cofhan, N.B. (1992). DNA technology. J. Am. Osteopath. Assoc. 92,777-783. Freier, S.M., Keirzek, R., Jaeger, J.A., Sugimoto, N., Caruthers, M.H., Neilsen, T., & Turner, D.H. (1986). Improved free-energy parameters for predicts of RNA duplex stability. Proc. Natl. Acad. Sci. (USA). Gingeras, T.R., Prodanovich, P., Latimer, T., Guatelli, J.C., Richman, D.D., & Barringer, K.J. (1991). Use of self-sustained replication amplification reaction to analyze and detect mutations in zidovudine-resistant human immunodeficiency virus. J. Infect. Dis. 164, 1066-1074. Guatelli, J.C., Whitfield, K.M., Kwok, D.Y., Barringer, K.J., Richman, D.D., & Gingeras, T.R. (1990). Isothermal, in vitro amplification of nucleic acids by a multienzyme reaction modeled after retroviral replication. Proc. Nat. Acad. Sci. USA 87, 1874-1878. Haqqi, T.M., Sarkar, G., David, C.S., & Sommer, S.S. (1988). Specific amplification with PCR of a refractoly segment of genomic DNA. Nucleic Acids Res. 24, 11844-1 1844. Higuchi, R., von Beroldingen, C.H., Sensabaugh, G.F., & Erlich, H.A. (1988). DNA typing from single hairs. Nature 332,543-546. Hillier, L. & Green, P. (1992). OSP: a computer program for choosing PCR and DNA sequencing primers. PCR Methods Appl. 1, 124-128. Hutchinson, N.I. & Tocci, M.J. (1986). Characterization of a major early gene form the human cytomegaloviruslong inverted repeat; predicted amino acid sequenceof a 30-kDa protein encoded by the 1.2-kb mRNA. Virology 155, 172-182. Innis, M.A., Gelfand, D.H., Sninsky, J.J., &White, T.J. (1990). PCR Protocols In: AGuide to Methods and Applications. Academic Press, New York. Ishikawa, E., Hashida, S., Kohno, T.. & Hirota, K. (1990). Ultrasensitive enzyme immunoassay. Clin. Chim. Acta 194,51-72. Kalin, I., Shepard, S., & Candrian, U. (1992). Evaluation of the ligase chain reaction (LCR) for the detection of point mutations. Mutat. Res. 283, 11%123. Kemp, D.J., Smith, D.B., Foote, S.J., Samaras, N., & Peterson, M.G. (1989). Colorimetricdetection of specific DNA segments amplified by polymerase chain reaction. Proc. Natl. Acad. Sci. USA 86, 242S2427. Kessler, C. (1993). Nonradioactive Labeling and Detection of Biomolecules. Springer-Verlag, Berlin. Khandjian, E.W. (1987). Optimizedhybridization of DNA blotted and fixed to nitrocelluloseand nylon membranes. BioiTechnology 5, 165-169. Kim, H.S. & Smithies, 0. (1988). Recombinant fragment assay for gene targeting based on the polyemase chain reaction. Nucleic Acids Res. 16,211-216.

Design and Use of HAT PCR Primers

489

Kricka, L. (1992). Nonisotopic DNA Probe Techniques. Academic Press, New York. Kwok, D.Y., Davis, G.R., Whitfield, K.M., Chappelle, H.L., DiMichelle, L.J., & Gingeras, T.R. (1989). Transcription-based amplification system and detection of amplified human immunodeficiency virus type I with a bead-based sandwich hybridization format. Proc. Natl. Acad. Sci. USA 86, 1173-1177.

Lew, A.M., Brandon, R.B., Panaccio, M., & Morrow, C.J. (1992). The polymerase chain reaction and other amplification techniques in immunological research and diagnosis. Immnology 75,3-9. Li, H., Cui, X., & Arnheim, N. (1990). Direct electrophoretic detection of theallelic state of single DNA molecules in human sperm by using the polymerase chain reaction. Proc. Natl. Acad. Sci. USA 87,4580-4584.

Lowe, T., Sharefkin, J., Yang, S.Q., & Dieffenbach, C.W. (1990). A computer program for selection of oligonucleotide primers for polymerase chain reactions. Nucleic Acids Res. 18, 1757-1 761. Lucas, K., Busch, M., Mossinger, S., & Thompson, J.A. (1991). An improved microcomputer program for finding gene- or gene family-specific oligonucleotides suitable as primers for polymerase chain reaction or as probes. Comput. Appl. Biosci. 7, 525-529. Mecklenburg, M. (1995a). Application of PCR as a reporter system. In: Preparation, Analyses, and Applications of Synthetic Oligonucleotides: A Laboratory Manual (Guameros, G., De LaVega, EM., Eds.), in press, Springer Verlag, Berlin. Mecklenburg, M. (1995b). Manuscript in preparation. Montpetit, M.L.. Cassol, S., Salas, T.,& O’Shaughnessey, M.V. (1992). OLIGOSCAN: a computer program to assist in the design of PCR primers homologous to multiple DNA sequences. J. Virol. Methods 36, 119-128. Mullis, K.B. & Faloona, F. (1987). Specific synthesis of DNA in vitro via a polymerase-catalyzed chain reaction. Meth. Enzymol. 155, 335-350. Pillay, D. & Griffiths, D.P. (1992). Diagnosis of cytomegalovirus infection: a review. Genitourin Med. 68, 183-188.

Powell, L.M., Wallis, S.C., Pease, R.J., Edwards, Y.H., Knott, T.J., & Scott, J. (1987). Anovel form of tissue-specific RNA processing produces apolipoprotein-B48 in intestine. Cell 50,83 1-840. Rapley, R., Theophilus, B.D.M., Bevan, 1,s..&Walker, M.R. (1992). Fundamentalsofpolymerase chain reactiofuture in clinical diagnostics. Med. Lab. Sci. 49, 119-128. Resendez-Perez, D., & Barrera-Saldana, H.A. (1990). Thermocylcer temperature variation invalidates PCR results. Biotechniques 9,286-294. Rolfs, A,, Schuller, I., Finckh, U., & Weber-Rolfs, I. (1993). PCR Clinical Diagnostics and Research. Springer-Verlag, Berlin, Germany. Ruano, G., Fenton, W., & Kidd, K.K. (1989). Biphasic amplification of very dilute DNA samples via ‘Booster’ PCR. Nucleic Acids Res. 17, 5407-5416. Rychlik, W. & Rhoads, R.E. (1989). A computer program for choosing optimal oligonucleotides for filter hybridization, sequencing and in vim amplification of DNA. Nucleic Acids Res. 17, 8543-855 1 .

Saiki, R.K., Scarf, S., Faloona, K.B., Mullis, K.B., Horn, G.T., Erlich, H.A., & Amhein, N. (1985). Enzymatic amplification of beta-globin genomic sequences and restriction site analyses for diagnosis of sickle cell anemia. Science 230, 1350-1354. Rychlik, W., Spencer, W.J., & Rhoads, R.E. (1990). Optimization ofthe annealing temperature for DNA amplification in vitro. Nucleic Acids Res. 18,6409-6412. Sambrook, J., Fritsch, E.F., & Maniatis, T. (1989). Molecular Cloning: A Laboratory Manual, 2nd edition. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Sano, T., Smith, C.L., & Cantor, C.R. (1992). Immuno-PCR: very sensitive antigen detection by means of specific antibody-DNA conjugates. Science 258, 12C122. Segev, D. (1993). Amplification of nucleic acid sequences by the repair chain reaction. In: Nonradioactive Labeling and Detection of Biomolecules (Kricka, C., Ed.), Springer-Verlag, Berlin.

490

MICHAEL W. MECKLENBURG

Stagno, S., Pass, R.F., Dworsky, M.E., Henderson, R.E., Moore, E.G., Walton, P.D., & Alford, C.A.

(1982). Congenital cytomegalovirus infection: the relative importance of primary and recurrent maternal infections. N. Engl. J. Med. 306,945-949. Stenberg, R.M., Thomsen, D.R., & Stinski, M.F. (1984). Structural analyses of the major immediate early gene of human cytomegalovirus. J. Virol. 49, 190-199. Stinski, M.F. (1978). Sequence of protein synthesis in cells infected by human cytomegalovirus: early and late virus-induced polypeptides. J. Virol. 26,686701. Stinski, M.F., Mocarski, E.S., Thomsen, D.R. (1979). DNA of human cytomegalovirus: size heterogeneity and defectiveness resulting fiom serial undiluted passage. J. Virol. 3 1.23 1-239. Suggs, S.V., Hirose, T., Miyake, T., Kawashima, E.H., Johnson, M.J., Itakura, K., & Wallace, R.B. (1981). Use of synthetic oligodeoxyribonucleotides for the isolation of specific cloned DNA sequences. In: Developmental Biology Using Purified Genes (Brown, D.D. Fox, C.F., Eds), pp. 683-693. Academic Press, New York. The, T.H., van der Ploeg, M., vander Berg, A.P.. Vlieger, A.M., van der Giessen, M., & van Son, W.J. (1992). Direct detection of cytomegalovirus in peripheral blood leukocytesa review of the antigenemia assay and polymerase chain reaction. Transplantation 54, 193-1 98. Walker, G.T., Little, M.C., Nadeau, J.G.. & Shank, D.D. (1992). Isothermal in vitro amplification of DNA by a restriction enzymeiDNA polymerase system. Proc. Natl. Acad. Sci. USA 89,392-396. Ward, E.S., Gussow, D., Griffiths, A.D., JonesP.T., & Winter, G. (1989). Bindingactivitiesofarepertoire of single immunoglobulin variable domains secreted fiom Escherichia coli [see comments]. Nature 341,54&546. Wetmur, J.G. (1991). DNA probes: application of the principles of nucleic acid hybridization. Crit. Rev. Biochem. Mol. Biol. 26,227-259. Williams, J.F. ( 1989). Optimization strategies for the polymerase chain reaction. BioTechniques 7, 762-768. Witsell, A.L. & Schook, L.B. (1990). Clonal analyses of gene expression by PCR. BioTechniques 9, 318-322. Wu, D.Y. & Wallace, R.B. (1989). The ligation amplification reaction (LAR)-amplification ofspecific sequences using sequential rounds of template-dependent ligation. Genomics 4,560-569.

ON-LINE MONITORING OF INDUSTRIAL FERMENTATIONS USING A SPLIT-FLOW MODIFIED THERMAL BIOSENSOR

M. Rank and B. Danielsson

I. 11. 111. IV.

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491 INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492 INSTRUMENTS AND SAMPLING . . . . . . . . . . . . . . . . . . . . . . 493 ANALYSIS AND ENZYME COLUMNS . . . . . . . . . . . . . . . . . . . 493 RESULTS AND DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . 495

ABSTRACT A biosensor was modified in several ways in order to adapt it for use in industrial scale automated broth culture monitoring. All instrumentation was assembled inside a steel cabinet and controlled by a PC situated at a safe distance from the fermentors. Penicillin V, glucose, and ethanol were monitored on-line in 0.5 and 2.5 m3 bioreactors

Advances in Molecuiar and Cell Biology Volume 15B, pages 491497. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN0-7623-0114-7

49 1

492

M. RANK and B. DANIELSSON

using an enzyme thermistor modified for split-flow analysis. Penicillin V and L-lactate were also monitored in 160 m3bioreactors. The work was performed at Novo Nordisk A/S in the fermentation pilot plant in Bagsvaerd or in the production plant in Kalundborg, Denmark. The injected samples were split into two equal streams immediately before simultaneously entering the enzyme column and the identical reference column (with immobilized BSA instead of enzyme). This results in a temperature response (6T) without the nonspecific heat arising from mixing and solvation effects from the fermentation broth. Only the enzymatic hydrolysis is reflected in the 6T, and all calculated values correlated very well with off-line HPLC determinations. Instead of the previously used j3-lactamase, three penicillin productions in 160 m3 bioreactors were monitored on-line using penicillin V acylase. This enzyme was purified from a broth supernatant. The L-lactate concentration was monitored from the start in another Penicillum fermentation. Glucose was monitored from start in a Saccharomyces fermentation and after five hours the analysis was switched to ethanol monitoring.

I. INTRODUCTION An elegant combination of theory and craftsmanship some 20 years ago resulted in

the birth of the enzyme thermistor (Mosbach and Danielsson, 1974).The instrument described in this paper is the outcome of several years of improvement and fine tuning. The enzyme thermistor, a calorimetric biosensor, has to date been employed in 50 different applications in the fields of biochemistry and biotechnology (Danielsson, 1990). Since almost all enzymatic reactions are exothermic, the only limitation is the availability of a purified and fairly cheap enzyme specific for the substrate of interest (Danielsson and Mosbach, 1986).Abiosensor that can measure fermentation broth metabolites in combination with an automated and computerized flow injection analysis would prove a very valuable tool for the industry (Brooks et al., 1991). The use of the split-flow technique is a convenient method for subtracting the nonspecific heat that is associated with complex samples such as fermentation broths (Mattiasson et al., 1976).A trouble-free sampling device is necessary if a whole fermentation is to be monitored without fouling or changes due to altered broth viscosity (Mattiasson et al., 1981). Due to the duration of an average fermentation, only a fully automatic computerized system can handle the amount of data and control the fully automatic FIA analysis. Rather than adapt the instruments to 100% humidity, vibrations and temperature variations between 10 and 35 "C that may be encountered in the fermentor hall, the cool dry environment of a separate cabinet is used to provide a more practical and cheaper solution. Novo Nordisk A/S is an international pharmaceutical company whose major production is based on fermentation. On-line monitoring of new microorganisms and of expensive large-scale productions have never been more important, especially with the rapid development of new products through recombinant gene technology. Knowledge of the concentrations of substrates, products, inhibitors,

On-line Monitoring of Industrial Fermentations

493

and other metabolites is of vital importance to carefully control industrial fermentations. The model system of choice for the on-line monitoring is based on the easy access to Penicillum fermentation projects at Novo Nordisk.

II. INSTRUMENTS A N D SAMPLING At first, on-line sampling was performed by using a monopump to pump broth through an external loop connected to a modified tangential flow filtrationunit from Millipore (Gram et al., 1990). On-line sampling of the fermentation broth was later performed with a hydrophilized polypropylene filtration probe from Advanced Biotechnology Corporation (Puchheim Germany). The filtrating part of the ABC probe was inserted through the bioreactor wall using a fitting normally employed in pH- electrodes. About 1 to 3 ml/min of a clear 0.2 pm filtered sample was continuously aspirated to the outside of the bioreactor using a peristaltic pump. The sample stream was then led to the analyzing FIA-system placed inside a steel cabinet flushed with cool dry air (see Figure 1). A split-flow bottom plate was constructed and installed at the bottom of the calorimeter in order to split the buffer stream immediatelybefore entering the two columns. The initial experimentsin the pilot plant were recorded with an integrator inside the cabinet. Later, the signals from the amplifier were transferred to a Datalogger. All information was then transferred via a local modem connection to a software program from Labtech Notebook, Wilmington,USA, installed in a PC located 60 m away in a control room.

111. ANALYSIS A N D ENZYME COLUMNS The eluting phosphate buffer contained 2 mM azide to prevent microbial contamination. The azide was exchanged with benzoic acid when columns with catalase were used. In order to prevent nucleophilic breakdown ofpenicillin V in phosphate buffer, all standards were made in 10-times diluted phosphate buffer that was kept cold in an insulated box filled with ice (Bundgaard and Hansen, 1981). The enzyme matrix was controlled pore glass and all enzyme immobilizations were performed on glutaraldehyde-activated silanized glass beads (Weetall, 1976). All enzymes were dialyzed before coupling. In earlier studies, penicillin V was monitored with immobilized p-lactamase. The reference column was made in an identical way except bovine serum albumin instead of enzyme was bound. Later, penicillin V determinations utilized a purified Penicillin V acylase. This enzyme was purified from a broth supernatant supplied by Novo Nordisk A / S (Hussey et al., 1983). The enzyme activity was determined spectrophotometrically by measuring the amount of 6-APA formed using p-dimethyl aminobenzaldehyde as the color reagent. The assay was a (6-aminopenicillanicacid) modified procedure (Balasingham et al., 1972). On-line enzyme thermistor values from penicillin fermentations were compared with manually withdrawn samples analyzed by off-line HPLC. The broth concen-

M. RANK and B. DANIELSSON

494

&

STANDARDS TWO

WASH SOLUTION

ELUENT

('I

Figure 7. Schematic representation of the set-up. A continuous sample stream was aspirated from the bioreactor using a filtration probe and a peristaltic pump. The filtered broth sample was pumped into a six-position pneumatic sample selector. The same multichannel peristaltic pump also aspirated the carrier buffer through the sample valve and further into the thermostated calorimeter with the two columns. The split-flow system divides the sample immediately before it enters the enzyme column and the inactive reference column. The signals were recorded with an amplifier and an interface enabled computer monitoring. The whole analysis was followed and controlled from the PC. Samples were injected every 10 min in the order: low concentration standard, broth sample, and high concentration standard. The broth concentrations were calculated by linear regression from an average of the three latest values of each standard (From Rank et al., 1992).

On-line Monitoring of Industrial Fermentations

495

trations of penicillin V, penicilloic acid, p-hydroxy penicillin V, and 6-APA were determined. The assay was performed at the analytical laboratories of Novo Nordisk A/S using a modified method (Hussey et al., 1983). Glucose was monitored by glucose oxidase from Aspergihs niger coimmobilized with 40,000 units of catalase from bovine liver. Ethanol was monitored with alcohol oxidase from Pichiapastoris. For comparison broth samples were analyzed by gas chromatography at Novo Nordisk NS. Lactate was monitored with lactate oxidase from Pediococcus sp. Once every day on-line samples were also analyzed using a UV method kit from Boehringer Mannheim Biochemica.

IV. RESULTS AND DISCUSSION So far, eight complete fermentations have been monitored on-line. The first tests at the University of Lund proved valuable for the development of a working instrument, but the important information concerning enzyme specificity and reproducibility was obtained later in the fermentor hall. The ABC-sampling probe provides a clear microfiltrate throughout the fermentations with PenicilIum or Sacharomyces. The delay from sample withdrawal to the registered peak was determined to be five minutes. In the first three fermentation experiments, filtered samples were collected on-line with a modified tangential flow filtration unit from Millipore together with a monopump. Immobilized p-lactamase was then used as the detecting biocatalyst. The linear range for on-line monitoring of penicillin V with p-lactamase was 0.1 to 500 mM with sample volumes of 20 to 500 pl. The concentration of the two p-lactam substrates penicillin and p-hydroxypenicillin V were determined by HPLC. When the two values were added and plotted against the on-line biosensor values, a deviation of 10% was observed (Rank et al., 1992). It was later discovered that the sample pump had a grinding effect on the fungi mycelia and intracellular j3-lactams such as 6-APA were released into the broth. This may explain the 10% higher values. Penicillin V acylase, however, hydrolyzed the side chain of penicillin, which is a four times less exothermic reaction than that of P-lactamase. Penicillin V acylase was purified and immobilized before use for on-line monitoring of penicillin production in a 160 m3 bioreactor. The linear range was 0.5 to 150 mM using sample volumes between 20 and 500 pl. The sum of the concentrations of the three side chain penicillin V acylase substrates in the fermentation broth, penicillin V, penicilloic acid, and p-hydroxy penicillin V, were determined by off-line HPLC, and when compared with the on-line biosensor values they correlated very well. Glucose was monitored from the start in a Saccharomyces cerevisae fermentation. After seven hours the column was exchanged and ethanol was monitored. The time from the exchange of columns and standards to the first on-line ethanol value was 60 minutes (Rank, 1993). Early experiments with alcohol oxidase (Pichia pastoris) revealed a slightly higher exothermic response for methanol and a poor

M. RANK and B. DANIELSSON

496 100

0 90 80

JO 60 !>I2 40 30

20 10

I

20

0

40

bO

80

I00

rci I,me

Ll

oriiiiic

CT

+

0550"

(Unchruiqcr)

figure 2. On-line monitoring of L-lactate during penicillin production in a 160 m3 bioreactor using an enzyme column with lactate oxidase coimmobilized with catalase. Values from fermentation broth samples were calculated every 30 min and samples were collected manually and analyzed using a spectrophotometric assay kit. The accuracy of the kit assay is not satisfactory and is only used as an indication. The lactate concentration slowly declines until midway into the fermentation when it falls below 0.1 5 mM and can no longer be measured with a minimal error constant.

affinity for I-butanol (Danielsson, 1991). A broad substrate specificity was experienced with other alcohol oxidases and this has to be considered when searching for the most suitable enzyme. Results using an alcohol oxidase from Candida boidinii (Boehringer Mannheim) showed that ethanol instead of methanol is the best substrate for this enzyme (Guilbaultet al., 1983).L-lactate was monitored from start in a Penicillurn fermentation. The lactate slowly declined until halfway into the fermentation when the concentration fell below 0.15 mM and could no longer be monitored (see Figure 2). The limitations of this system are mainly the insufficient availabilityof pure and specific enzymes. Although many enzymes involved in the important citric acid cycle and amino acid synthesis pathways are available, they are very expensive.

ACKNOWLEDGMENTS This project was supported by a grant from the Nordic program on bioprocess engineering under the auspices of NI (The Nordic Fund for Technology and Industrial Development).

On-line Monitoring of Industrial Fermentations

497

The authors especially thank Michael O’Reilly of Novo Nordisk and Birgitta Rees-Jonsson of the Biosensor Group for their help and cooperation.

REFERENCES Balasingham, K., Warburton. D., Dunnill, P., & Lilly, D.M. (1972). The isolation and kinetics of penicillin amidase from E. coli. Biochem. Biophys. Acta 276,25&256 Brooks, S.L., Higgins, I.J., Newman, J.D., & Turner, A.P.F. (1991). Biosensors for process control. Enzyme. Microb. Technol. 13,946955. Bundgaard, H. & Hansen, J. (I98 I). Nucleophilic phosphate-catalyzed degradation of penicillins: demonstration of a penicilloyl phosphate intermediate and transformation of ampicillin to a piperazinedione Int. J. Pharm. 9,27>283. Danielsson, B. (1990). Calorimetric biosensors. J. Biotech. 15, 187-200. Danielsson, B. (I99 I). Enzyme thermistor devices. In: Biosensors Principles and Applications (Blum. L.J. & Coulet, P.R., Eds.), pp. 83105. Marcel Dekker Inc., New York. Danielsson, B. & Mosbach, M. (1986). Theory and applications of calorimetric sensors. In: Biosensors: Fundamentals and Applications (Turner, A.P.F.. Karube, I., & Wilson, G.S., Eds.), pp. 575595 Oxford University Press, Oxford. Decristoforo, G. & Danielsson, B. (1984). Flow injection analysis with enzyme thermistor detector for automated determination of p-lactams. Anal. Chem. 56,263268. Gram, J., Nikolajsen, K., Holm, K., & De Bang, M. (1990). Automated sampling and glucose analysis for a pilot-plant penicillin V production. Abstract presentation ECB 5, p. 3 11 Copenhagen. Guilbault, G.G., Danielsson, B., Mandenius, C.-F., & Mosbach, K. (1983). A comparison of enzyme electrode and thermistor probes for assay of alcohols using alcohol oxidase. Anal. Chem. 55, 1582-1 585. Hussey, R.L., Mascher, W.G., & L a g , A.L. (1983). Determination of p-hydroxy penicillin V p-hydroxyphenoxy acetic acid, phenoxyacetic acid and penicillin V in production fermentation broth. J. Chrom. 268, 12&124. Mattiasson B., Danielsson. B., & Mosbach, K. (1976). A split-flow enzyme thermistor. Anal. Lett. 9. 8674389. Mattiasson, B., Danielson. B., Winquist, F., Nilsson, H., & Mosbach, K. (1981). Enzyme thermistor analysis of penicillin G in standard solutions and in fermentation broth. Appl. Environ. Microbiol. 41,90>908. Mosbach, K. & Danielsson, B. (1974). An enzyme thermistor. Biochim. Biophys. Acta 364, 140-145. Rank, M., Gram, J., & Danielsson, B. (1992). Implementation of a thermal biosensor in a process environment: on-line monitoring of penicillin V in production-scale fermentations. Biosens. Bioelectr. 7, 9, !A. Rank, M., Gram, J., & Danielsson, B. (1993). Industrial on-line monitoring ofpenicillin V, glucose and ethanol using a split-flow modified thermal biosensor. Anal. Chim. Acta 281,521-526. Weetall, H.H. (1976). Covalent coupling methods for inorganic support materials. Meth. Enzymol. (Mosbach, K., Ed.), vol. 44, 134-148. Academic Press, New York.

This Page Intentionally Left Blank

MINIATURlZE D THERMAL BIOSENSORS

U. Hedberg, B. Xie, and B. Danielsson

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CONSTRUCTION 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CONSTRUCTION 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CONSTRUCTION 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I. 11. 111. IV.

499 500 500 501 503 505

ABSTRACT A summary of our efforts to develop miniaturized biosensors based on the enzyme thermistor is presented. T h e e constructions are described. The work focuses on the measurement of glucose in whole blood. The first biosensor was used to analyze concentrated and tenfold diluted blood samples. The glucose concentration obtained using this construction correlated well with the reference method.

Advances in Molecular and Cell Biology Volume 15B, pages 499-505. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

499

500

U. HEDBERG, B. XI€, and B. DANIELSSON

Using the second biosensor, the response time could be reduced from 3 minutes down to 30 seconds using a sample volume of 1 pl. The relative standard deviation for 100 blood samples (3.8 mM) was 3.7%. Micromachining technology was used to construct the third thermal biosensor on a silicon chip. Enzymes were immobilized directly onto the 30 parallel flow channels. Injection of 200 samples containing 10 mM hydrogen peroxide gave a relative standard deviation of 3%.

1. INTRODUCTION The enzyme thermistor is a thermometric biosensor that has been developed in our department. This biosensor detects the heat evolved during enzyme catalysis. The instrument has been used for many applications over the years (Danielsson, 1990; Rank, 1992). We have now made several miniaturized enzyme thermistor constructions. Miniaturization allows these sensors to be used in portable instrumentation for applications such as measurement of various metabolites in body fluids. Here we describe our attempts at developing a miniaturized instrument. Blood glucose determinations using two miniaturized constructions are described. An account of the technique of micromachining to construct a thermal biosensor on a small silicon chip is also described here.

II. CONSTRUCTION 1 This device is 54 mm long and 24 mm in diameter. The fluid temperature is equilibrated when the buffer solution passes through the steel tubing wound inside an aluminum cylinder surrounding the column (Figure 1).After passing the column,

Adiabatic shield Enzyme column

\

Parallel coil

Figure 7. Schematic figure of construction 1.

Miniaturized Thermal Biosensors

501

Table 1. Blood Glucose Concentrations (Means of 10 Determinations) Measured with Construction 1 and with the Boehringer Reflolux Instrument Group

Refrolux

1

5.1 4.5 4.6 5.6

2 3 4 5

4.6

Miniaturized Enzyme Thermistor 5.0 4.5 4.2 5.7 4.5

the buffer flows through a piece of tubing placed around a copper cylinder next to the column (adiabatic shield). The stainless steel column (1.5 x 15 mm) containing the immobilized enzyme is sealed with a polyethylene filter in both ends. The construction is insulated with air, aluminum, and Plexiglas. Glucose Determination in Diluted Whole Blood

Glucose oxidase and catalase were coimmobilized on superporousagarose beads (106-180 pm). Buffer containing sodium fluoride and EDTA was continuously pumped through the column via a six-way injection valve at a flow rate of 50 pVmin. The sample volume was 20 pl. Blood was collected from the fingertips of human volunteers on five different occasions and the glucose concentration was measured with the Reflolux S glucose analyzer from Boehringer Mannheim. The injected samples were divided into five different groups. After diluting the samples tenfold, the blood glucose concentrations were measured. The response time was about 3 minutes. The system gave a linear response up to 1.6 mM glucose. With a tenfold dilution, blood glucose concentrationsas high as 16 mM were recorded. The results of the glucose measurementsare shown in Table 1.

111. CONSTRUCTION 2 This construction was made in order to improve the response time (Xie, 1993). A schematic drawing of the construction is shown in Figure 2. The diameter of the column was reduced from 1.5 mm to 0.6 mm and the length was 15 mm. Thermistors were mounted on gold capillary tubings in close proximity to the column. The column was placed inside an inner plastic jacket and then covered with an outer jacket. Stainless steel tubing was mounted on a small aluminum block that acted as a heat sink. This arrangement increased the stability of the baseline.

U. HEDBERG, B. XIE, and B. DANIELSSON

502

External

Micro-

Thermistor Outlet tubing

I

I

/

P1exigl ass adaptor

Internal jacket

Gold capillary

5 mm

H

Figure 2. Schematic figure of construction 2 (Xie, 1993).

Glucose Determination i n Whole Blood

Using a sample volume of 1 pl and a flow rate of 50 pl/min, the system had a linear range of up to 20 mM glucose. The response time was about 30 seconds. Human blood collected from the vein was put in heparinized tubes containing sodium fluoride. After two hours the glucose concentration was analyzed using Reflolux S from Boehringer Mannheim and a miniaturized biosensor. The relative standard deviation was 3.7% when over 100blood samples (3.8 mM glucose) were injected as shown in Figure 3.

0.0

0

20

40

60

80

100

120

Number of blood samples Figure3. Temperature responses obtained for over 100 repeated injections of a blood sample containing 3.8 mM glucose (Xie, 1993).

Miniaturized ThermaI Biosensors

503

20

04 0

I

i

10

20

Blood glucose (mM) with the sensor Figure 4. Correlation between construction 2 and Reflolux S in the concentration range from 3.6 to 19.3 mM glucose (Xie, 1993).

Small volumes of a 200 mM glucose solution were added to the blood in order to obtain glucose concentrationsranging from 3.6 to 19.3 mM. The results obtained with the miniaturized biosensor and Reflolux S were compared over a concentration range as shown in Figure 4. Acorrelation coefficient of 0.980 was obtained for the 37 blood samples.

IV. CONSTRUCTION 3 The biosensor shown in Figure 5 was designed and fabricated on a silicon wafer (14 x 6 x 0.4 mm) with minimal heat capacity. The total volume of the reactor cell (5 x 1 x 0.014 mm) was 0.02 pl and consisted of 30 parallel channels. The cell was Glass cover

Inlet

Thermistor

Channels

Gold tubing

Silicon chip

Outlet

Connecting film

Figure 5. Schematic figure of the thermal biosensor fabricated onto a silicon chip (Xie, 1992).

U. HEDBERG, B. XIE, and 8. DANIELSSON

504

30

c? I

0 v 7

t-

10

Q

0

1

Penicillin with the enzyme thermistor EI

Penicillin with the silicon based device

v

HP0

9

10

2

/

with the silicon based device

I

I

I

I

4

6

8

10

Sample Conc.

(mM)

Figure 6. Sensitivity comparison between the silicon-based thermal biosensor and the conventional enzyme thermistor (Xie, 1992).

constructed by anisotropic etching of the silicon wafer surface. At the inlet and outlet of the cell ultramicrobead thermistors were placed on gold capillary tubings (Xie, 1992). Enzymes (catalase and penicillinase) were immobilized directly on the silanized (3-aminopropyl-triethoxysilane)silicon chip using glutaraldehyde.The sensor was

1 "

I

0

1

100

200

Number of batches Figure 7. Reproducibility obtained with immobilized catalase injecting samples of 10 m M hydrogen peroxide (Xie, 1992).

Miniaturized Thermal Biosensors

505

placed in an aluminum box insulated with polyurethane foam. In order to compare the sensitivity of the silicon chip with that of the conventional enzyme thermistor, 20 pl samples of different penicillin concentration were injected (Figure 6 ) . The flow rate was 20 pl/min for the silicon chip construction and 500 pl/min for the enzyme thermistor. The reproducibility of the sensor was tested using immobilized catalase by continuously injecting 10 mM hydrogen peroxide (Figure 7). After 200 injections, the loss of enzyme activity was seen as a decrease in the temperature response. For the first 200 injections the standard deviation was 3%.

V. CONCLUSIONS During our initial studies, the blood was diluted tenfold and 20 pl were injected. At least 100 samples were injected into a column containing glucose oxidase and catalase immobilized on superporous agarose. The measured glucose concentration correlated well with the reference method (Reflolux S). The response time was reduced to 30 seconds using a simpler construction.The column size, the tubing diameter, and the distance between the injection valve and the column were minimized. Instead of injecting 20 pl tenfold diluted blood, 1 p1 samples of undiluted blood were injected. Eliminating the dilution step simplifies sample handling and reduces the overall response time. These results indicate that thermal biosensors can be miniaturized without any significant loss in sensitivity.In the case of the silicon chip based thermal biosensor, the sensitivity was identical to that obtained with the conventional enzyme thermistor.

ACKNOWLEDGMENTS This work was supported in part by grants from the National Swedish Board for Technical Development (NUTEK) and Novo Nordisk N S .

REFERENCES Danielsson, B. (1990). Calorimetric biosensors. J. Biotechnol. 15, 187-190. Rank, M., Gram, J., & Danielsson, B. (1992). Implementation of a thermal biosensor in a process environment: on-line monitoring of penicillin V in production scale fermentors. Biosensors Bioelectronics 7,63 1 4 3 5 . Xie, B., Danielsson, B., Norberg, P., Winquist, F., & Lundstrom, I. (1992). Development of a thermal microbiosensor fabricated on a silicon chip. Sensors Actuators B 6, 127-130. Xie, B., Hedberg, U., Mecklenburg, M., & Danielsson, B. (1993). Fast determination ofwhole blood glucose with a calorimetric micro-biosensor. Sensors Actuators B, 15-16, 141-144ss.

This Page Intentionally Left Blank

PART V

AFFINITY TECHNIQUES FOR SEPARATION AND BIORECOGN ITION

This Page Intentionally Left Blank

AFFINITY TECHNIQUES FOR SEPARATION AND BIORECOGNITION: INTRODUCTORY REMARKS

Per-Olof Larsson

Isolation of@protein from a cell homogenate is certainly not a trivial task. Often the protein is present in small amounts and mixed with thousands of other protein species that have very similar physical properties, making it difficult to find a discriminating separation method. Traditional purification techniques (precipitation, ion exchange chromatography, etc.) therefore require many steps before a sufficientlypurified protein is obtained. The introduction of affinity chromatography in the late sixties and seventies changed the situation dramatically. Affinity chromatography relies on the specific interaction between the protein and a complementary ligand attached to the affinity adsorbent and therefore offers unique opportunities for selecting one desired protein from a complex mixture. During the last 25 years numerous examples of impressive purification achievements with

Advances in Molecular and Cell Biology Volume 15B, pages 50%511. Copyright 0 1996 by JAI Press Inc. AIL rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

509

51 0

PER-OLOF LARSSON

affinity chromatography have been reported, and the technique is now one of the most used. In the first chapter of this section, Lowe gives an historic overview of the field starting as far back as 1910 with Starkenstein’s affinity purification of a-amylase. Lowe continues with modem rrequirements of the affinity chromatography technique considering upstream engineering, adsorbent design, and process development. Some emphasis is put on the development of new biomimetic dyes, which act as efficient ligandsthey may be tailored to high specificity and are comparatively much more stable than derivatives of natural ligands, e.g., nucleotides. In the next chapter Berna, Moraes, Barbotin, Thomas, and Vijayalakshmi give experimental details about the rapid purification of a recombinant enzyme to homogeneity, cyclodextrin glycosyl transferase. The purification is based on metal ion affinity chromatographyusing two tandemly coupled columns.The first column [Zn(II)]has a negative affinity (removingrelated proteins but not the desired one), whereas the second column [Cu(II)] has a positive affinity for the desired enzyme. Tjerneld describes an alternative affinity technique for purification, namely aqueous two-phase affinity separation. The technique is now well established and is based on the fact that two properly prepared aqueous polymer solutions are easy to mix but will, upon standing, spontaneously separate into two phases. If one of the polymers is derivatized with an affinity ligand, an affinity purification system is obtained. Tjerneld describes here how the concept of aqueous two-phase affinity separations can be extended by using polymers whose phase-forming properties are temperature dependent. For example, an enzyme that has been partitioned to an affinity polymer phase in an initial step may then be released in a second step by merely raising the temperature of the affinity polymer phase. In this way a pure enzyme is obtained, free from contaminating polymers. Schwarz and Wilchek reveal the mysteries of thiophilic adsorption, an affinity technique originally proposed by Porath. The technique has special importance when isolating immunoglobulins, e.g., monoclonal antibodies.Very strong binding occurs between certain sulfur-containing ligands and immunoglobulins,indicating a specific binding mechanism. Schwarz and Wilchek present an array of evidence that collectively suggest the chemical nature of the binding site and where it is situated on immunoglobulins. In the next chapter affinity chromatography is used for fundamental studies instead of purification and isolation of proteins. The fact that true affinity chromatography is based on the defined molecular interaction between two (sometimes three) molecules soon attracted researchers to use the technique as a means of characterizing and quantifying interactions between proteins and their ligands, e.g., cofactors, substrates, and inhibitors. Easily measured quantities such as retention times, break-through curves, and peak widths could be mathematically transformed into Kdconstantsand even rate constants. Here Chaiken, Myszka, and Morton give a short overview of analytical affinity chromatography. They continue to describe new applications where, in particular, interactions between HIV proteins and

Affinity Techniques for Separation and Biorecognition

51 1

immune system components are studied with the aid of analytical affinity chromatography. The results are compared with results obtained with a biosensorAhe BIAcore machine. This biosensor directly measures the interaction between an immobilized ligand and a protein via the resulting increase of refractive index close to the biosensor surface. Fauchere describes peptides as multiple purpose tools. Peptides, natural or modified, may be used in a number of applications, e.g., as drugs, inhibitors, and affinity ligands to name a few. Fauchere gives a thorough overview of applications, synthetic methods, design strategies for obtaining a secondary structure that will give the peptide the desired biorecognition properties, and finally, the special problems encountered in drug design. In the next chapter Larsson describes superporous agarose. Significantly, this new material contains two kinds of pores, diffusion pores and flow pores. When the material is used in chromatographic beds it gives high chromatographic efficiency. The new material can be derivatized with ion exchange groups, affinity ligands, etc. Kozulic and Heimgartner describe the properties of several hydrophilic and amphiphatic gels related to polyacrylamide. Gel electrophoresis with these new gels gives superior results compared with standard polyacrylamide. Some of the gels may be used for hydrophobic interaction electrophoresis. The introduction gives us a rather full background to the project, where we can learn that research by-products may be very important. In the final chapter of this section, Kasche takes a broad view of immobilized systems. He notes the similarities of the processes occurring in affinity chromatography, in biosensors, and in immobilized biocatalysts, namely the diffusion distances and the ligate (or biocatalyst) density. These considerations provide an integrated approach to the analytical description of these systems.

This Page Intentionally Left Blank

AFFl NITY CHROMATOGRAPHY AND RELATED TECHNIQUES: PERSPECTIVES AND TRENDS

Christopher R. Lowe

I. 11. 111. IV.

v.

VI .

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 14 MODERN REQUIREMENTS FOR AFFINITY CHROMATOGRAPHY . . . 5 16 5 17 UPSTREAM ENGINEERING . . . . . . . . . . . . . . . . . . . . . . . . . ADSORBENT DESIGN AND CONSTRUCTION . . . . . . . . . . . . . . . 5 17 5 19 PROCESS DEVELOPMENT . . . . . . . . . . . . . . . . . . . . . . . . . . CONCLUSIONS AND FUTURE PROSPECTS . . . . . . . . . . . . . . . . 520

ABSTRACT Few of the recent advances in biotechnology would be commercially viable without effective methods of protein purification. Affinity techniques exploit specific biorecognition phenomena and are ideally suited to the purification of high value pharmaceutical proteins. This review focuses on the technique of affinity chromatography,

Advances in Molecular and Cell Biology VoIume 153, pages 513-522. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. SBN: 0-7623-0114-7

51 3

CHRISTOPHER R. LOWE

514

provides a historical perspective, and describes the current status of the technique and its future prospects. It is anticipated that new developments in affinity chromatography will lead to new separation media better able to withstand the rigorous conditions required for sanitization and cleaning in situ of industrial scale processes but still retaining the high resolution, capacity, and yield of classical affinity media.

1. INTRODUCTION Protein purification has until recently remained one of the more empirical aspects of modern biotechnology. Traditional techniques based on precipitation with pH, temperature, salts, organic solvents, or high molecular weight polymers are being replaced by sophisticated chromatographic techniques based on biological specificity (Cuatrecasas et al., 1968; Cuatrecasas and Anfinsen, 1971; Lowe and Dean, 1974). The technique of affinity chromatography exploits the unique capacity of biological macromolecules to being complementary substrates, hormones, repressors, coenzymes, allosteric effectors, oligonucleotides, and other ligands, specifically and reversibly. The technique has now been exploited to resolve and purify enzymes, genetically and chemically modified proteins, cells and viruses, supramolecular structures and organelles, and to investigate and explore a wide variety of binding sites, topographies, and kinetic mechanisms (Lowe, 1977). The potentially facile, rapid, and virtually limitless application of affinity chromatography has prompted an almost exponential growth in use of the technique over the last two decades. The concept of separating macromolecules by means of biospecific interactions with immobilized substrates is not new: Starkenstein (19 10) reported the isolation of a-amylase by adsorption onto insoluble starch, while Willstatter et al. (1923) appreciably enriched lipase by adsorption onto powdered stearic acid. Likewise, Campbell et al. (195 1) isolated rabbit anti (bovine serum albumin) antibodies using a specific immunoadsorbent comprising bovine serum albumin coupled to diazotized p-aminoebenzylcellulose, and Lerman (1953) isolated mushroom tyrosinase on variousp-azophenol-substituted cellulose columns. Subsequently, Arsenis and McCormick (1964, 1966) purified liver flavokinase and other flavin mononucleotide-dependentenzymes on flavin-substituted celluloses. However, despite the fact that the antecedents of affinity chromatography clearly reach back to the beginning of the twentieth century, it was only in 1968 that the immense power of biospecificity as a means of purification was appreciated and the term “affinity chromatography” was coined (Cuatrecasas et al., 1968). Since then, thousands of proteins and other biomolecules have been isolated by this technique using almost every conceivable class of biochemical as the immobilized ligand. A recent survey found that affinity chromatography was the second most widely used purification technique after ion exchange chromatography and was used in 60% of the purification schemes investigated (Bonnerjea et al., 1986).This astounding technological

51 5

Affinity Techniques

success story can be attributed to several factors:the exquisite selectivityof affinity chromatography, the commercial availability of suitable matrix materials, the development of facile methods for activating these supports to attach appropriate ligands, and the market-pull exerted by the nascent biological industries. The key developments responsible for the successhl introduction of affinity chromatography as the method of choice are shown in Figure 1. For example, Cuatrecasas et al. Year 1900

a-amylase purified on starch

1910

1920

=-Lipase

purification on powdered stearic acid

1930

1940

1950

1960

1970

1980

--

immunoaffinity Chromatography Purification of tyrosinase

-

-

CNBr Activation and Poi saccharides Modern Concept of Affinyty Chromatography (Blospeclflcedsoorbents) Group specific Adsorbents (C4enzymes,lectlns,nuclelcaclds etc) Reactive Dyes High Performance Liquid Affinity Chromatography 'Biomimetic' Dyes

+Purification 'tags'

1990

p e novo Ligand design

2000

Figure 1. Historical perspective and key milestones in the development of affinity chromatography.

CHRISTOPHER R. LOWE

51 6

(1968) introduced agarose as a matrix and the insertion of a spacer molecule between the ligand and the matrix to relieve steric hindrance (Cuatrecasas and Anfinsen, 1971). Other developments including introduction of the cyanogen bromide procedure for activation of polysaccharide matrices (Axen et al., 1967), the concept of group-specific or general ligand adsorbents (Lowe and Dean, 1971; Mosbach et al., 1972), identification of the problem of nonspecific adsorption (O’Carra et al., 1973), the use of reactive dyes (Lowe, 1984), and the introduction of the related technique of high performance liquid affinity chromatography (Ohlson et al., 1978) have all contributed significantlyto the establishment of the approach. Despite its acknowledged advantages, the use of affinity chromatographyis still largely limited to research laboratories and is only recently beginning to make an impact in industry for the purification ofhigh-value biopharmaceuticals.The reason for this lies in the high cost and lability of traditional ligands, the presence of troublesome nonspecific adsorption, and fouling problems and the severe difficulties associated with sterilization and cleaning in situ. This review describes new developments in affinity chromatography and related techniques that should offset some of the previous shortcomings of the technique and establish it finally as the routine tool for large-scale protein purification (Clonis, 1987).

II. MODERN REQUIREMENTS FOR AFFINITY CHROMATOCRAPHY Substantial worldwide markets exist for highly purified pharmaceutical proteins, enzymes, vaccines, antibodies, and hormones. Clinical validation of such products is an absolute requirement by the regulatory authorities and all steps in the production must be in compliance with current Good Manufacturing Practices (GMP). Steps must be taken to ensure that the final therapeutic product is hlly active and unadulterated with potentially lethal contaminants. For example, biological molecules isolated from natural sources,or most commonly,expressed from recombinant DNA systems, must carry documentation to show that they have acceptably low levels of biologically active contaminants such as DNA (
51 7

Affinity Techniques

by the initial concentration of the target protein in the mother liquor (Knight, 1989) and the yields from multistage outmoded purification protocols. Consequently, the development of new, robust purification procedures that possess high resolution, high recovery, and high throughput is the most critical area demanding innovative upstream engineering,adsorbent design, and constructionand process development (Luong et al., 1987; Lowe, et al., 1990; Sassenfeld, 1990).

111. UPSTREAM ENGINEERING The advent of sophisticated recombinant DNA techniques has led to the idea that such approaches may well be applied to the source of the target protein in such a way as to facilitate the subsequent downstream processing steps involved in the purification of the protein. In one such technique, DNA encoding additional polypeptide or protein tags, is fused either to the 5' or 3' end of the gene of interest (Sassenfeld, 1990; Ford et al., 1991). Expression of these gene fusions results in the hybrid protein comprising the target protein and a purification tag that may be removed by enzymic or chemical cleavage. Purification tags have been devised to facilitate separation by affinity, ion exchange, hydrophobic, covalent, immunoaffinity, and metal chelate chromatography (Ford et al., 1991). Choice of the most appropriate tag will depend on individual requirements, scale, and selection of the expression system. However, removal of the purification tag still presents a significant hurdle to further exploitation,because, althoughmany methods have been used successfully in the research environment, seemingly intractable problems like denaturation,heterogeneity,and low yield still arise and are exacerbated on scale-up (Sassenfeld, 1990). Nevertheless, if such problems could be resolved purification tag technology may save prodigious amounts of time; purification effort is directed and rational and does not involve random screening and optimization.Nevertheless, even if such upstream events could be optimized, there is still a pressing requirement for durable, high capacity affinity adsorbents to form the key component of any downstream processing protocol.

IV. ADSORBENT DESIGN AND CONSTRUCTION Originally, selective adsorbents were constructed with natural biological ligands (Lowe and Dean, 1974); the exquisite specificity of enzymes, polynucleotides, enzyme substrates, and antibodies offered a beguiling prospect. Experience has shown that most mono- and group-specific biological ligands are fragile, tend to be difficult to immobilize with retention of activity, and lead to expensive adsorbents (McLoughlin and Lowe, 1988; Lowe et al., 1990). Consequently, despite the obvious attractions of the technique, the large-scale exploitation of affinity chromatography has been hampered by two problems: the high cost of current affinity media and the difficultiesof making such media operate stably with high throughput in a multicycle sterile environment. Paradoxically, the high selectivity of affinity

518

-

CHRISTOPHER R. LOWE

Affinity Chromatography

9 Biornirnetic Pseudospecific

Biospecific

I

I

Monospecific

Group Specific

A

I

Antibodies Substrates Receptors Products Coenzymes Nucleotides

I

I

Dyes Designer Dyes De novo ligands

Metal Chelate Hydrophobic thiol

Lectins Heparin

Oligo-

Polynucleotides

Figure 2. Classification of affinity chromatography adsorbents on the basis of the specificity of the immobilized ligand.

chromatography is also its most profound weakness because suitable of-the-shelf adsorbents are often unavailable. Ideal media should incorporate elements of both selective and non-selective absorbents, be stable, and have general applicability (Figure 2). Specially synthesized ligands offer the best hope of finding general purpose, inexpensive and stable ligands. The reactive textile dyes are a group of synthetic ligands that have been responsible for purifying an astounding range of individual proteins and are now exploited in over 75% of all affinity separations (Figure 3) (Lowe, 1984; Clonis, 1987). However, despite the fact that thousands of research papers have been published, textile dyes have not yet proved to be the panacea originally envisaged. This is because textile dyes are relatively impure bulk chemicals and bonding processes Gelatin Metalchelate

IgG

\

\

Other Protein A

Other triazin dyes

C1 Reactive Blue 2

Figure 3. Comparison of usage of affinity ligands for large scale affinity chromatography. Data from Clonis (1 987).

Affinity Techniques

519

between dye and matrix poorly developed such that there is a near universal perception that all synthetic ligand products are toxic and leak into the product stream. Substantialresearch effort on the nature of dye-protein interactions(Small et al., 1982; Lowe and Pearson, 1984; Mcloughlin and Lowe, 1988; Lowe et al., 1990) and the use of purified dyes and improved coupling procedures has virtually eliminated ligand leakage and significantly increased column life-time capacity, batch-to-batch reproducibility, and sterilizability (Stewart et al., 1992). Furthermore, by exploiting a combination of empirical observation and computer-assisted molecular modeling, novel dye ligands could be designed and synthesized, which retained all the cost and stability advantages of textile dyes but yet more nearly mimicked the binding behavior and specificityof natural biological ligands (Lowe et al., 1986;Burton et al., 1988, 1990; Lowe et al., 1992). More recent studies have shown that ligands may be designed de novo by computer-aided molecular design. The ligands were designed to mimic the side chains of a number of arginyl dipeptides for the selectivepurificationof trypsin-like proteases (Burton and Lowe, 1992). The ligand with the highest affinity for the enzyme, an analogueof a Phe-Arg dipeptide, when immobilizedto sepharose CL4B via a hexamethylene spacer arm, purified pancreatic rallikrein 110-fold in one step from a crude pancreatic acetone extract. This study clearly demonstrates the value in exploiting computer modeling for the design of novel biomimetic ligands for affinity chromatography. Stable and sterilizable support matrices are also an essential ingredient for the large-scale use of affinity chromatography for the purification of pharmaceutical proteins. Porous supports for affinity chromatography have traditionally comprised beaded polysaccharides,synthetic polymers, and inorganic oxides (Knight, 1989). Very few of the commercially available affinity media satisfy the criteria for affinity and adsorbents (Groman and Wilchek, 1987) and display clear deficiencies with respect to degradation of the ligand and support under conditions used for sanitization and sterilization(Knight, 1989).Newer materials are now beginning to make an impact (Narayanan and Crane, 1990); a new generation of ultrastable, nontoxic perfluoropolymer supports was developed (Stewart et al., 1989, 1990).These new affinity adsorbents display exceedingly low rates of leakage of immobilized ligands under conditions of exposure to chaotropes (5 M sodium thiocyanate), acid (1 M HCl), and base (1 M NaOH) (Stewart et al., 1992). Shortly, these materials are expected to be available in a porous, beaded, format.

V. PROCESS DEVELOPMENT To date, affinity adsorbents have nearly always been operated under batchwise conditions with packed beads that become liable to fouling and plugging (Luong et al., 1987). Consequently,pretreatment to remove solid debris is a prerequisite to the effective use of affinity chromatography. Unfortunately, the low throughput creates a very low productivity system with processing rates as low as 10 kg h-', compared to >lo00 kg h-' for conventional precipitation procedures. The technique

CHRISTOPHER R. LOWE

520

has, thus far, been applied only to high-value, low-volume pharmaceuticalproducts such as urokinase, interferon, factor VIII, and antithrombin 111. Fluidized bed arrangements may obviate the problems associated with fouling and entry streams containing particulates and may permit affinity separations to be moved more upstream from their current position as a polishing step. Recent work has demonstrated the value of exploitingperfluorocarbon affinity emulsions in fluidized beds for the semi-continuous recovery of human serum albumin from blood plasma (McCreath et al., 1992). The affinity emulsion was generated by homogenization of perfluorodecalin with a dyed polymeric fluorosurfactant (Stewart et al., 1989, 1990). Under fluidized bed conditions, the emulsions behave as normal chromatographicmaterials and are stableunder operational conditionswith no coalescence being observed for periods greater than one year. These novel high density liquid affinity supportspresent excitingopportunitiesto develop a range of unit operations for the continuous purification of proteins (McCreath et al., 1992). The productivity of affinity separations may also be improved by increasing the flow rates as in high performance liquid affinity chromatography (Ohlson et al., 1978; Clonis et al., 1986) or be exploiting related high capacity techniques such as affinity cross-flow filtration (Mattiasson and Ling, 1986), affinity two-phase partition (Johansson, 1985), and affinity precipitation (Larsson and Mosbach, 1979; Pearson et al., 1986). Indeed, in the latter case, productivity compares favorably with classical precipitation techniques, while retaining all the advantages of selectivity created by the exploitation of specific biological interactions.

VI. CONCLUSIONS AND FUTURE PROSPECTS The unique specificity and reversibility of biological interactions have opened new horizons for the development of purification strategies. Affinity chromatography and related techniques display qualities of scale, resolution, recovery, yield, and capacity and are likely to displace existing process technologies. However, the requirements for protein purification are also becoming increasingly stringent at an equally alarming rate. The approval of any pharmaceutical by the FDA relies on a convincing demonstration by the manufacturer of the safety and efficacy of the product (Anicetti et al., 1989). Sophisticated electrophoretic,chromatographic,and mass spectrometric techniques are now available to assess product purity, although

Table 1. Common Impurities in Pharmaceutical Proteins Produced by Recombinant DNA Technology Endotoxin Host cell and media proteins DNA Infectious agents

Product isofotms / variants Aggregated forms Proteolytic products Leachates from separationmedia

521

Affinity Techniques

in general, preparative procedureshave yet to match such analyticalprecision. Table 1 lists some typical impuritiesto be found in protein pharmaceuticals.Clearly, new high resolution preparative chromatographictechniques will have to be developed to resolve deleterious contaminants such as pyrogens, host cell and media proteins, infectious agents, and the plethora of potential product variants that are known to exist in recombinant proteins (Anicetti et al., 1989).Novel adsorbents for resolving protein and glycoprotein isoforms and the variety of oxidation, deamidation, aggregated, proteolysed, and mutant forms of the protein product will be required. Needless to say, such problems will pose new challenges to the bioprocess engineer well into the next century.

REFERENCES Anicetti, V.R., Keyt, B.A., & Hancock, W.S. (1989). Purity analysisofproteinpharmaceuticalsproduced by recombinant DNA technology, Trends Biotechnol. 7,342-349. Arsenis, C. & McCormick, D.B. (1964). Purification of liver flavokinase by column chromatography on flaviwellulose compounds. J. Biol. Chem. 241,33&334. Arsenis, C. & McCormick, D.B. (1966). Purification of flavin mononucleotide-dependent enzymes by column chromatography on flavin phosphate cellulose compounds. J. Biol. Chem. 239,3093-3097. Axen, R., Porath, J., & Emback, S. (1967).Chemical couplingofpeptides and proteins to polysaccarides by means of cyanogen halides. Nature 214. 1302-1304. Bonnejea, J., Oh, S., Hoare, M.. & Dunnill, P. (1986). Protein purification: The right step at the right time. Bio/Technology 4,954-958. Burton, N.P. & Lowe, C.R. (1992). Design of novel affinity adsorbentsfor the purification of trypsinlike proteases. J. Mol. Recognit. 5 , 5 5 6 8 . Burton, S.J., Stead, C.V., & Lowe, C.R. (1988). Design and applications of biomimetic anthraquinone dyes. 111. Anthraquinone immobilized C.I. reactive Blue 2 analogues and their interaction with horse liver alcohol dehydrogenase and other adenine nucleotide binding proteins. J. Chromatogr. 508, 109-125. Campbell, D.H., Luescher, E.L., & Lerman, L.S. (1951). Immunologic adsorbents 1. Isolation of antibody by means of a cellulose protein antigen. Proc. Natl. Acad. Sci. USA 37.575-578. Clonis, Y.D. (1987). Large scale affinity chromatography Bio/Technology 5 , 1290-1293. Clonis, Y.D., Jones, K., & Lowe, C.R. (1986). Process scale high performance liquid affinity chromatography. J. Chromatogr. 363,31-36. Cuatrecasas, P. & Anfinsen, C.B. (1971). Affinity chromatography. Ann. Rev. Biochem. 40,25%278. Cuatrecasas, P., Wilchek. M., & Anfinsen, C.B. (1968). Selective enzyme purification by affinity chromatography. Proc. Nat. Acad. Sci. USA 61,636-643. Ford, C.F., Suominen, I., & Glatz, C.E. (1991). Fusion tails for the recovery and purification of recombinant proteins. Prot. Exp. Pur. 2,95107. Groman, E.V. & Wilchek, M. (1987). Recent developmentsin affinity chromatographysupports.Trends Biotechnol. 5,220-224. Johansson, G. (1985). Aqueous two-phase systems in protein purification. J. Biotechnol. 3, 11-18. Knight, P. (1989). Chromatography: 1989. Report. BiolTechnology 7,24>249. Larsson, P.-0. & Mosbach, K. (1979). Afinity precipitationof enzymes. FEBS Lett. 98,333-338. Lerman, L.S. (1953). A biochemically specificmethod for enzyme isolation.Proc.Natl. Acad. Sci. USA 39,232-236. Lowe, C.R. (1977). Affinity chromatography: the current status. Int. J. Biochem. 8,177-181. Lowe, C.R. (1984). Applications of reactive dyes in biotechnology. In: Enzyme and Fermentation Biotechnology(Wiseman, A., Ed.), Vol. 9, pp. 78-161. Ellis Honvood, Chichester.

522

CHRISTOPHER R. LOWE

Lowe, C.R., Burton, S.J., Burton, N.P., Alderton, W.K., Pitts, J.M., & Thomas, J.A. (1992). Designer dyes: biomimetic ligands for the purification of pharmaceutical proteins by affinity chromatography. Trends Biotechnol. 1 0 , 4 4 2 4 8 . Lowe, C.R., Burton, S.J., Burton, N.P., Steart, D.J., Purvis, D.R., Pitfield, I., & Eapen, S. (1990). New developments in affinity chromatography. J. Mol. Recognit. 3. 117-122. Lowe, C.R., Burton, S.J., Pearson, J.C., Clonis, Y.D., & Stead, C.V. (1986). Design and application of biomimetic dyes in biotechnology. J. Chromatogr. 376, 121-130. Lowe, C.R. & Dean, P.D.G. (1971). Affinity chromatography of enzymes on insolubilized cofactors. FEBS Lett. 14.313316. Lowe, C.R. & Dean, P.D.G. (1974). Affinity chromatography. John Wiley, London. Lowe, C.R. & Pearson, J.C. (1984). Affinity chromatography on immobilized dyes. Methods Enzymol. l04,97-111. Luong, J.H.T., Nguyen, A.L., & Male, K.B. (1987). Recent developments in downstream processing based on affinity interactions. Trends Bitechnol. 5.28 1-286. Mattiasson, B. & Ling, T.G.I. (1986). In: Membrane Separation in Biotechnology (McGregor, W.C., Ed.). Marcel Dekker. McCreath, G.E., Chase, H.A., Purvis, D.R.. & Lowe. C.R. (1992). Novel afinity separations basedon perfluorocarbon emulsions. Use of a perfluorocarbon affinity emulsion for the purification of human serum from blood plasma in a fluidised bed. J. Chromatogr. 597, 18P-196. McLoughlin, S.B. & Lowe, C.R. (1988). Applications of triazinyl dyes in protein purification. Rev. Prog. Coloration 18, 247-259. Mosbach, K., Guildford, H., Ohlsson, R., & Scott, M. (1972). General ligands in affinity chromatography: Co-factor substrate elution of enzymes to the immobilized nucleotides adenosine 5'-mOnOphosphate and nicotinamide adenine dinucleotide. Biochem. J. 127, 625-63 1. Narayanan, S.R. & Crane, L.J. (1990). Affinity chromatography supports: a look at performance requirements. Trends Biotechnol. 8, 12-1 6. O'Carra, P., Bany, S., & Griffin, T. (1973). Spacer arms in affinity chromatography: the need for a more rigorous approach. Biochem. SOC. Trans. 1,28P-290. Ohlson, S., Hansson, L., Larsson, P-O., & Mosbach. K. (1978). High performance liquid affinity chromatography (HPLAC) and its application to the separation of enzymes and antigens. FEBS Lett. 9 3 , 5 9 . Pearson, J.C., Burton, S.J., & Lowe, C.R. (1986). Affinity precipitation of lactate dehydrogenase with a triazine dye derivative. Anal. Biochern. 15,382-389. Sassenfeld, H.M. (1990). Engineering proteins for purification. Trends Biotechnol. 8,88-93. Small, D.A.P., Lowe, C.R., Atkinson, A., & Bruton, C.J. (1982). Affinity labelling of enzymes with triazine dyes. Isolation of a peptide in the catalytic domain of horse liver alcohol dehydrogenase using Procion Blue MX-R as a structural probe. Eur. J. Biochem. 128, 11P-123. Starkenstein, E.V. (19 10). ijber Fermtenwir-Kung und deren Beeinflussung durch Neutralsalze. Biochem. Z. 24, 14. Stewart, D.J., Hughes, P., & Lowe, C.R. (1989). Afinity chromatography on triazine dyes immobilised on novel perfluorocarbon suppons. J. Biotechnol. 11,253266. Stewart, D.J., Purvis, D.R., & Lowe, C.R. (1990). Affinity chromatography on novel perfluorocarbon supports. Immobilization of C.I. Reactive Blue 2 on a poly(viny1 alcohol)-coated perfluoropolymer support and its application in affinity chromatography. J. Chromatogr. 5 10, 177-187. Stewart, D.J., Purvis, D.R., Pitts, J.M., & Lowe, C.R. (1992). Development of an enzyme-linked immunoadsorbent assay for C.I. reactive Blue 2 and its application to a comparison of the stability and performance ofa perfluorocarbon support with other immobilised C.I. Reactive Blue 2 affinity adsorbents. J. Chromatogr. 623, 1-14. Willsriitter, R., Waldschmidt-Leitz, E., & Memman, F. (1923). Bestimmung der Pankreatischen Fettspaltung. Erste Abhandlang Uber Pankreasenzyme. Z. Physiol. Chem. 125.93.

0NE-STEP AF F I NITY PURI F I CAT10N OF A RECOMBINANT CYCLODEXTRIN CLYCOSYL TRANSFERASE BY (CU(II), Zn(ir) TANDEM COLUMN) IMMOBILIZED METAL ION AFFINITY CHROMATOGRAPHY

P. Berna, F.F. Moraes, J.N. Barbotin, D. Thomas, and M.A. Vijayalakshmi

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. MATERIALS AND METHODS . . . . . . . . . . . . . . . . . . . . . . . . A. Chemicals.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. EnzymeSource.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Matrix Denvatization . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Chromatographic Procedure . . . . . . . . . . . . . . . . . . . . . . . .

Advances in Molecular and Cell Biology Volume lSB, pages 523-537. Copyright 0 1996 by JAI Press h e . All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

523

524 524 525 525 525 526 526

524

BERNA, MORAES, BARBOTIN, THOMAS, and VIJAYALAKSHMI

E. Biospecific Affinity Chromatography . . . . . . . . . . . . . . . . . . . F. Pseudobiospecific Affinity Chromatography . . . . . . . . . . . . . . . G. ProteinAssay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H. Activity Assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. Electrophoresis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. RESULTS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. IMAC on Cu(n)-IDA Sepharose 6B . . . . . . . . . . . . . . . . . . . . B. IMAC on Zn(n)-IDA Sepharose 6B . . . . . . . . . . . . . . . . . . . . C . IMAC on Tandem Cu(rr)/Zn(rr)IDA-Sepharose 6B . . . . . . . . . . . . D. Biospecific Affinity Chromatography on P CD-Sepharose 6B . . . . . . E. Electrophoresis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. CGTase capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. Enzyme Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

,526

. 526 .526 .527 .527 .527 .527 ,530 . 530 .531 .531 ,532 .532 .533

ABSTRACT The one step purification to homogeneity using immobilized metal ion affinity chromatography (IMAC) of a recombinant cyclodextrin glycosyl transferase (rCGTase) from alkalophillic Bacillus cloned in Escherichia coli is described. Tandem columns with copper and zinc were used. Negative affinity on Zn(l1) and positive affinity on Cu(I1) were observed. We compared this pseudobiospecific affinity chromatography technique with the biospecific affinity chromatography using P cyclodextrin (CD) as the immobilized ligand in terms of purification factor and activity recovery. We obtained the same purification factor in both cases but with 79% activity recovery with the P CD ligand against 89% enzymatic activity recovery for IMAC. No further treatment of the enzyme was necessary for pseudobiospecific affinity chromatography because the elution conditions used did not inhibit the enzyme, as opposed to the biospecific approach. The presence of 10 mg/ml of p CD or 150 mM of P-D-glucose in the adsorption buffer did not disturb rCGTase adsorption on Cu(I1) columns. These facts suggested that at least one histidine outside of the active site is accessible on the surface of the rCGTase molecule for binding on Cu(1I).

1. INTRODUCTION Cyclodextrin glycosyl transferase (CGTase) is an enzyme which produces cyclodextrins from starch. This enzyme has an industrial importance since various applications of the CD have been found (Bender, 1986; Szejtli, 1990). CGTase purification was usually achieved by multiple step processes involving non-specific precipitation with acetone or ammonium sulfate (French, 1962),nonspecific sorption on ion exchange and size exclusion chromatography (Horikoshi et al., 1973; Kitahata et al., 1974) or biospecific sorption on native (Norberg and French, 1950) or modified starch (Gottvaldova et al., 1988). Biospecific affinity chromatography

One-Step Affinity Purification of a Recombinant CGTase by IMAC

525

using a CD (Lazslo et al., 1981) or j3 CD (Makela et al., 1988), products of the enzyme, is the most utilized purification process at present on an analytical scale. As described by Vijayalakshmi (1989), pseudobiospecific ligand affinity chromatography has many advantagesover biospecific ligand affinity chromatography. Immobilized metal ion affinity chromatography (IMAC) is one of the most potent pseudobiospecific affinity systems. M A C was introduced by Porath (Porath et al., 1975) who used the affinity of proteins for metal ion immobilized on a chromatographic matrix via a chelating agent. Further work by Sulkowski with several model proteins established the ground rules for protein recognition by immobilized metal ion [CU(II),Ni(rI), Zn(II), Co(Ir)] on iminodiacetic acid chelating matrices (Sulkowski, 1985; 1989). IMAC is based on the accessible histidine content of protein that can form coordinate bonds with the immobilized metal ion under appropriate conditions. Therefore, since its creation, IMAC has been used not only for the purification of histidine-containing proteins (Krishnan and Vijayalakshmi, 1985; Anderson et al., 1987)but also for histidine residues surface topography of those proteins (Hemdan et al., 1989; Zhao et al., 1991). The recombinant CGTase used has 12 histidine residues in its primary sequence (Schmid et al., 1988). However, the degree of accessibilityof the histidine residues is not known. Hence, we decided to exploit IMAC as an alternative to the biospecific ligand affinity chromatography for CGTase purification and to study its accessible histidine content.

II. MATERIALS A N D METHODS A. Chemicals

Sepharose 6B and electrophoresis calibration kits for molecular weight determination were from Pharmacia fine chemicals AE3 (Uppsala, Sweden). 1,CButanediol diglycidyl ether, epichlorhydrine, and zinc sulfate were obtained fiom Merck (Darmstadt, Germany).Bovine serum albumin (Fraction V), copper sulfate, soluble starch, and iminodiacetic acid (IDA) were products of Sigma Chemical Co. (St. Louis, MO, USA). The p CDs were from Chinoin (Budapest, Hungary). All other chemicals used were of reagent grade. B. Enzyme Source

rcCGTase from alkalophillicBacillus (strain 1.1) cloned in Escherichiu coli was a gift from the Consortium f3r elektrochemische Industrie GmbH (Munich, Germany). This filtrate of E. coli culture broth had a protein content of 1.4 mg/ml and a specific activity of 4 1 U / d .

526

BERNA, MORAES, BARBOTIN, THOMAS, and VIJAYALAKSHMI

C. Matrix Derivatization

The p CD-Sepharose 6B was prepared according to Lazslo (Lazslo et al., 1981) using 1,Cbutanediol diglycidyl ether (bisoxirane)and p CD instead of a CD as the ligand. The IDA-Sepharose 6B was prepared as described by Porath and Olin (1983). CU(II)capacity for this gel was about 60 pmol Cu(Ir)/mlof gel bed. D. Chromatographic Procedure

All the chromatographic procedures used a linear flow of 25 c d h on a colllmn of 1.13 x 9 cm (9 ml) and were carried out at 4 "C with a low pressure liquid chromatography system (Minipuls I1 Gilson peristaltic pump, Uvicord SII LKB U.V. detector, wavelength set at 278 nm, Redirac 2 112 LKB fraction collector)with degassed buffer. Aprotein sample, 1 ml of filtered culture broth containing approximately 1.4 mg of protein and 41 units of activity was applied on the column. The column was washed with the adsorption buffer until all unbound material was eluted. The adsorbed protein was then displaced by the desorption buffer. 2.5 ml fractions were collected and analyzed for activity and protein content.

E. Biospecific Affinity Chromatography Tris-HC150 mM, pH 8 , 5 mM CaC1, used as the adsorption buffer. The gel was washed with ten column volumes of adsorption buffer before chromatography.The elution was carried out by adding 10 mg/ml of p CD in the adsorption buffer.

F. PseudobiospecificAffinity Chromatography The IDA-Sepharose 6B column was loaded with a solution of CuSC, or ZnSO, (50 mM) in distilled water until the saturation of the gel, and equilibrated with the adsorption buffer (Tris-HC1 50 mM, pH 7, 1 M NaC1) prior to chromatography. Competitive elution with a Tris-HC1 buffer, 50 mM, pH 7, containing 0.5 M NaCl and 25 mM imidazole, was used to elute the bound proteins. For tandem chromatography a Zn(I1)-IDA column was placed before the CU(II)-IDA column. The sample was loaded on the Zn(Ir)--IDA column and washing done in the tandem mode. Then the columns were disconnected and only the CU(II)-IDAcolumn was eluted. After each chromatography the metal ions on the gels were displaced by 50 mM EDTA and the gels thoroughly washed before a new metal load was applied. G. Protein Assay

Protein content was estimated by the micro-Bradford assay (Bradford, 1976). The values were compared to a standard curve obtained with bovine serum albumin.

One-Step Affinity Purification of a Recombinant CGTase by IMAC

527

H. Activity Assay Cyclization activity was assayed with soluble starch as substrate by measuring the production of p CD using a modification of the method of Vikmon (1982). This method followed the activity by the decrease in absorption (550 nm) of phenolphthalein when complexing with p CD. The reaction mixture (1 ml total) containing the enzyme suitably diluted with distilled water to a volume of 0.5 ml, and 0.5 ml of 1% starch solution in 10 mM Tris-HC1 buffer (PH8) containing 5 mM CaCl, was incubated at 50 "C for 20 minutes. The reaction mixture was immersed in a boiling water bath for five minutes to stop the reaction. To 0.5 ml of a dilution of the above reaction mixture 2.5 rnl of a phenolphthalein working solution was added to develop the color. The reactants were mixed. Absorption at 550 nm was measured and compared to a calibration curve made with p CD in water (0-400 pM). The phenolphthalein working solution contained 60 pM phenolphthalein and 1.9% ethanol in sodium carbonate buffer (120 mM pH 10.5). One activity unit is equivalent to the formation of 1 pmol of p CD per minute under the chosen conditions. I. Electrophoresis

Sodium dodecyl sulfate poly acrylamide gel electrophoresis (SDS-PAGE) on 10% acrylamide gel was performed according to Laemmli (Laemmli, 1970) on a Biorad Protean 11. Coomassie blue was used for staining.

111. RESULTS A. IMAC on C u w I D A Sepharose 6B

Immobilized CU(II)-IDAis known to bind proteins that have accessible histidine on their surface, with a minimum requirement of a single accessible histidine (Sulkowski, 1989).After injection of the culture filtrate on the Cu(I1)-IDAcolumn, all the rCGTase was considered bound to the gel because no activity was found during the wash (Figure 1). The elution by Tris-HCl buffer, 50 mM, pH 7,0.5 M NaCl containing 25 mM imidazole, gave a 100% activity recovery of the enzyme with a purification factor of 1.29 (Table 1). The chromatography of rCGTase in the presence of 10 mg/ml of p CD, or different concentrations of p-D glucose up to 150 mM did not disturb the binding of the enzyme on the CU(II)column (data not shown). It was recommended by Sulkowski (1985) that when competitive elusion with imidazole is used, one should presaturate the Cu(I1)-IDA-sepharose6B column with 1 mM imidazole solution. However, the presaturation of the gel with imidazole before chromatography of the rCGTase did not change anything concerning adsorption, but the amount of imidazole necessary to elute the bound proteins was reduced resulting in a time gain on the elusion operation.

528

BERNA, MORAES, BARBOTIN, THOMAS, and VIJAYALAKSHMI 140

I20

5

100

4

no 3 60 2

40

1

20 0

0 60

40

20

Elution volume (ml)

figure 1. Immobilized metal ion affinity chromatography of crude rCGTase from Alkalophilic Bacillus cloned in E. coli on a Cu(ii)-IDA-Sepharose66 column. Column 1.1 3 x 9 cm; adsorption buffer, Tris-HCI 50 mM, pH 7 , 1 M NaCI; desorption buffer, Tris-HCI 50 mM, pH 7, 0.5 M NaCI, 25 m M imidazole; protein applied, 1.4 mg, 40 U in 1 ml buffer. Protein determination by micro-Bradford assay. A, protein (pghnl); 0, activity (U/ml).

Table 1. Purification by Affinity Chromatography of Recombinant Cyclodextrin Glycosyl Transferase on Different Pseudobiospecific and Biospecific Ligands

Ligand

Cu(I1)

Zn(I1) Tandem Zn(II)+ Cu(I1)

P CD

Fraction

Sample Wash Eluted Sample Wash Eluted Sample Wash Eluted* Sample Wash Eluted

Total Total Protein (mg) Activiry (v)

1.36 0.15 1.05 1.36 1.10 0.20 1.20 0.11 0.76 1.38 0.43 0.78

40.0 0 40.0 40.0 35.0 0 40.0 0 35.8 40.0 0 31.7

SpeciJic Activity

Total Activity Recovev

(U/mg)

PW

29.4 0 38.1 29.4 31.8 0 33.4 0 46.8 29.0 0 40.6

100 0 100 100 87 0 100 0 89 100

Note: 'This represents the elution from the Cu(I1) column only, see text for details.

0 79

Purijkation Factor 1.oo 0

1.29 1.oo 1.08 0 1.oo 0 1.40 1.oo 0 1.40

One-Step Affinity Purification of a Recombinant CGTase by IMAC

529

From the SDS-PAGE electrophoregram (Figure 2, Lane 5 ) we could see that some of the contaminants from the culture filtrate were bound also to the CU(II)-IDA column and therefore should also possess histidine residues on their surface. It is known that the later elements in the first transition series of the periodic classification have different histidine topography requirements for protein adsorption. Sulkowski reported that zinc required the presence on the protein of at least two

Figure 2. SDS-polyacrylamide vertical gel electrophoresis pattern of the crude and purified rCGTase on 10% acrylamide gel. Coomassie blue staining. Lane 1 to 4, eluted fraction from p CD biospecific affinity chromatography at different concentrations; Lane 5, eluted fraction from immobilized metal ion affinity chromatography on Cu(ii)-IDA; Lane 6, crude extract, filtrate from E. coli culture broth; Lane 7, low molecular weight markers (From the top: 94 kD, 67 kD, 43 kD, 30 kD, 20.1 kD, 14.4 kD); Lane 8, eluted fraction from tandem immobilized metal ion affinity chromatography on Zn(ll)/Cu(ll)-IDA.

530

BERNA, MORAES, BARBOTIN, THOMAS, and VIJAYALAKSHMI

accessible histidines in cluster for binding, whereas CU(II)needed only one accessible histidine (Sulkowski, 1989).Therefore, the presence of two (His-His) doublets in the primary sequence of the rCGTase (Schmid et al., 1988) led us to try chromatography on Zn(I1).

B. IMAC on Zn(il)-IDA Sepharose 6 B We injected the filtrate of the E. coli culture broth on Zn(I1)-IDASepharose 6B expecting to discriminate between proteins with and without histidine clusters. No binding of rCGTase was observed under the conditions chosen because 87% of the activity was recovered during the wash (Figure 3). Some of the impurities bound to the column gave a purification factor of 1.08 (Table 1), which was even lower than that obtained with the CU(II)-IDAcolumn. Whereas we could partially purify the rCGTase on Zn(I1)-IDAby negative affinity, it was of interest to arrange the two columns [Zn(rI)-IDAand Cu(I1)-IDA]in tandem. The Zn column was arranged first in sequence to exploit the negative afinity.

C. IMAC on Tandem Cu(rr)/Zn(ll) IDA-Sepharose 6 B Figure 4 shows the elusion pattern in the tandem mode. As expected, rCGTase passed through the Zn(II)-IDAcolumn and bound to the CU(II)-IDAcolumn. After 6

5

4

3 2 1

0

20

Elution volume

40

0 60

(ml)

Figure 3. Negative immobilized metal ion affinity chromatographyof crude rCGTase from Alkalophilic Bacillus cloned in E. coli on a ZnWDA-Sepharose 66 column. Column 1.1 3 x 9 cm; adsorption buffer, Tris-HCI 50mM, pH 7, 1 M NaCI; desorption buffer, Tris-HCI 50 mM, pH 7, 0.5 M NaCI, 25 mM imidazole; Protein applied, 1.4 mg, 40 U in 1 ml buffer. Protein determination by micro-Bradford assay. A, protein (pghl);0, activity (U/ml).

One-Step Affinity Purification of a Recombinant CGTase by /MAC r

I40 7

531 6

120

5

100

4

80

3 60 2

40

1

20 0

0

20

40

60

0

80

Elution volume (ml) Figure 4. Tandem immobilized metal ion affinity chromatography of crude rCGTase from Alkalophilic Bacillus cloned in E. colion Zncit)/Curti)-IDA-Sepharose68 columns. Columns 1.13 x 9 cm; adsorption buffer, Tris-HCI 50 mM, pH 7 , l M NaCI; desorption buffer for the Cucii)-IDA-Sepharose6B column, Tris-HCI 50 mM, pH 7, 0.5 M NaCI, 25 mM imidazole; Protein applied: 1.4 mg, 40 U in 1 ml buffer. Protein determination by micro-Bradford assay. A, protein (pglrnl);0, activity (U/ml).

disconnection of the two columns, the rCGTase was eluted from the CU(II)-IDA column under the same conditions as previously described. The purification factor was 1.4 and the activity recovery was about 89% (Table 1). D. Biospecific Affinity Chromatography on p CD-Sepharose 6B

Biospecific affinity chromatography with immobilized p CD was run in order to compare the results with the above data. All the rCGTase bound to the column under the conditions used. Ten mg of p CDIml of buffer eluted the enzyme (Figure 5). Under these conditions only 79% activity recovery was found. The purification factor was 1.4 (Table l), a value identical to that observed with the tandem chromatography on Zn(II)-IDAplus CU(II)-IDA. Other attempts to elute the enzyme without cyclodextrins failed. In fact, no leaking of enzyme was observed during a wash with distilled water at room temperature in contrast to that reported by Makela et al. (1988), and removing the CaC1, from the adsorption buffer did not elute the rCGTase.

E. EI ectrophoresis The SDS-PAGE electrophoregram (Figure 2) showed that the eluted fractions from the p CD-column or from the Cu(Ir)-IDAcolumnof the tandem procedure had

532

BERNA, MORAES, BARBOTIN, THOMAS, and VIJAYALAKSHMI

the same purity and were homogeneous. Adensitometryreading ofthe crude extract pattern on the gel showed that the rCGTase band represented 70% of the total protein. This was not unexpected since the CGTase is a recombinant protein excreted in the medium. The estimated molecular weight of the enzyme was about 75,000 Da, which is consistent with the molecular weight calculated from the primary sequence (75,400 Da).

F. CGTase capacity Culture medium was passed through the affinity columns until CGTase activity was found at the outlet. The enzyme was adsorbed from the fluid with a binding capacity of 33 mg CGTase/ml of gel bed for the p CD-column and with 13 mg of CGTase/ml of gel bed for the CU(II)-IDAcolumn. G. Enzyme Stability

It is interesting to note that the rCGTase showed a loss of only 10% of activity after 3 years of storage at -20 "C.

Elution volume

(ml)

Figure 5. Biospecific affinity chromatography of crude rCGTase from Alkalophilic Bacilluscloned in E. colion immobilized p cyclodextrin-Sepharose66. Column 1.1 3 x 9 cm; adsorption buffer, Tris-HCI 50 mM, pH 8, 5 mM CaC12; desorption buffer, Tris-HCI 50 mM, pH 8, 5 mM CaC12, 10 mg/ml of p CD. Protein applied, 1.4 mg, 40 U in 1 ml buffer. Protein determination by micro-Bradford assay. A, protein (pg/ml); 0, activity (U/ml).

One-Step Affinity Purification of a Recombinant CGTase by M A C

533

IV. DISCUSSION Tandem pseudobiospecificand biospecific affinity chromatography gave the same purification factor of 1.4. This purification factor is small but consistent with the fact that rCGTase represents 70% of the protein in the starting material. Both purified rCGTases were homogeneous on SDS-PAGE; therefore, IMAC is as efficient as biospecific chromatography for the purification of rCGTase. Activity recovery was about 90% in the tandem pseudobiospecific chromatography and 80% in the biospecific one. However, 90% activity recovery for biospecific chromatography was previously reported but this involved an additional dialysis step (Laszlo et al., 1981) in order to remove the eluting agent P CD from the CGTase fractions because P CD is an inhibitor of the enzyme (French, 1962). The same yield was also observed after dialysis when either urea or ethanol was used as eluent (Makela et al., 1988). The lower yield observedby us may be at least partially explained by the omission of the post chromatographic dialysis step. Therefore, the pseudobiospecific affinity chromatography is as good as biospecific chromatography regarding activity recovery, with the important advantage of avoiding a dialysis step because the eluting agent used, namely imidazole, did not show any inhibitory effect. Several authors reported that soluble CU(II)and Zn(I1) were strong inhibitors of CGTase (French, 1962; Akimura et al., 1991). During IMAC on CU(II),we did not see a deleterious effect on enzyme activity because 100% of the activity was recovered. Furthermore, any final problem resulting from metal leakage can be easily taken care of by addition of a short, metal ion-free chelating bed after the affinitycolumns. The free matrix removes any eventual metal ion from the solution and avoids the inhibition. We reported a CGTase capacity that found in of 33 m g / d of gel bed for biospecific chromatography, which was equivalent to previous studies (Laszlo et al., 1981; Makela et al., 1988).Nevertheless, Villette et al. (199 1) reported a drastic decrease in capacity (0.3 mg/ml of gel bed) and a marginal activity recovery (only 10%) during their attempt to scale up biospecific affinity chromatography. For IMAC the capacity on a CU(II)column was 13 mg/ml of gel bed, 2.5-times less than biospecific chromatography on a laboratory scale but 10-times more than that reported by Villette et al. (199 1) on epichlorhydrin reticulated copolymer of beta cyclodextrin. Villette et al. (1991) also reported a total absence of CGTase binding to the affinity sorbent with P CD as ligand when the total neutral carbohydrate content of the injected extract was above 35 mM. In that case, they had to prepurify the crude extract before loading on the column. Absence of CGTase binding was certainly due to the maslung of the P CD binding site in CGTase. In the case of chromatography of rCGTase on Cu(I1)-IDA-Sepharose6B, 10 mg/ml of P CD or different concentrations of P-D-glucose up to 150 mM did not compete with the potential sites for IDA-CU(II)binding. Hence, the CGTase binding on the CU(II)column was

534

BERNA, MORAES, BARBOTIN, THOMAS, and VIJAYALAKSHMI

not affected by these carbohydrates. This implies that the accessible histidine residues involved in immobilized IDA-CU(II)recognition are not at or near the active site. Control columns with free Sepharose6B and Sepharose6B-IDA(data not shown) did not show nonspecific binding of rCGTase on those gels. Therefore, the immobilized metal ion was responsible for enzyme binding. If we refer to Sulkowski’s rules (histidine accessibility requirement for binding on Cup), N~(II),Zn(II), or CO(II)-IDA), the binding of rCGTase on the Cu(n)-IDAcolumnsuggests that at least one histidine amongthe twelve present on the molecule is accessibleon the surface. It seemed reasonable to use the data obtained on CGTase from other species, as we compared six available primary sequences (Table 2) and found that five histidine residues were always conserved in each of the six studied CGTases (His 98, 140, 176, 177, and 327, Bacillus circulans numeration) (Binder et al., 1986; Takano et al., 1986; Kimura et al., 1987; Kaneko et al., 1988; Schmid et al., 1988; Nitschke et al., 1990). Studies of CGTase histidine residue modification by diethyl pyrocarbonate (DEPC) (Villette et al., 1992) showed the modification of seven histidine residues that are accessible on the surface. Two of them (1.5 for Bender) (Bender, 1991) were faster reacting and seemed to be involved in the active site catalytic mechanism because their modification suppressed enzymaticactivity. In our case, we saw that 10 mg/ml of p CD or glucose at 150 mM were inefficient to elute the enzyme bound to CU(II)columns. This observation again demonstrated that the binding between rCGTase and the CU(II)which occurred via histidine did not involve the active site. Hence, histidines suggested by Bender (1991) to be involved in the active site, namely His-327 and His-I76 or 177, which are conserved in our rCGTase, are not the ones that bind to the immobilized CU(II).It would be interesting to study the catalytic behavior of rCGTase immobilized on Cu(II) columns, since the active site seems not to be involved in the metal binding site. The nonbinding of rCGTase on Zn(m) gave additional information. It is known that protein with the accessible sequence (His-(X)2,3-His)in an a helix configuration or (-His His-) by folding binds to Zn(n) (Sulkowski, 1989). In our case, the (His-His) sequence is present in the primary structure of the enzyme, this sequence is conserved in the six different CGTases, and from Bender (199 1) we know that they should be implicated in the active site and hence present on the surface.Further, Hofmann et al., (1989) described the three-dimensional structure of one CGTase, showing this histidine doublet on the edge of the molecule suggesting its accessibility. So the (His-His) sequence in our enzyme should be available for binding, but the rCGTase did not show any retention on Zn(I1). Sulkowski observed the same phenomena with human myoglobin, which presented His-His in a p turn of the protein, but here again, no binding was observed on Zn@)-IDA columns (Sulkowski, personal communication). One explanation could be that structural requirements on the His-His doublet are not hlfilled for binding on Zn(II)-IDA. This implies that a histidine doublet does not bind to

One-Step Affinity Purification of a Recombinant CGTase by IMAC

535

Table 2. Comparison of the Histidine Residue Position in 6 Different CGTases Alkalophillic bacillus strain 1011 CGTase Alkalophillic bacillus 38-2 CGTase Alkalophillic bacillus strain 1- 1 CGTase cloned in E. coli Bacillus circulans strain 8 CGTase Bacillus macerans CGTase Klebsiella pneumoniae M5 CGTase Alkalophillic bacillus strain 1011 CGTase Alkalophillic bacillus 38-2 CGTase Alkalophillic bacillus strain 1-1 CGTase cloned in E. coli Bacillus circulans strain 8 CGTase Bacillus macerans CGTase Klebsiella pneumoniae M5 CGTase

93 43

83

78

98

126 128 140 176 177 202

98

126 128 140 176 177 203

91

119

133 169 170

98

126

140 176 177 202

98

140 176 177 202

91 104 119

135 169 170 186

233

270

327 333

233

270

327 333

410

226 242 263 264 327 233 244

327

25 1

328 332

227

502 630 667 669 390

Note: *References are in the text; conserved histidines in the 6 CGTases are in hold letters.

Zn(II)-IDAand needs at least to be separated by an amino acid residue, as suggested by Sulkowski (1989). Finally, IMAC was once again demonstrated to be a competitive technique that can score over biospecific affinity chromatographybecause it gives the same purity and activity recovery with no additional steps required either to prepare the crude extract free from neutral carbohydrates or to eliminate the specific eluant (e.g., p CD). These are major advantages in view of a future industrial application. IMAC has the additional advantage of giving information on the histidine content of the protein. Furthermore, the noninvolvement of the active site in the binding mechanism predicts an easy method for CGTase immobilization without loss of activity.

ACKNOWLEDGMENT The authors thank the Consortium fiir elektrochemische Industrie GmbH (Miinich, Germany) for providing the filtrate culture broth of recombinant E. coli. This work was supported in part by a grant from the E.E.C.

536

BERNA, MORAES, BARBOTIN, THOMAS, and VIJAYALAKSHMI

REFERENCES Akimura, K., Yagi. T., & Yamamoto, S. (1991). Purification and properties of Bacillus coagulans cyclomaltodextrin glucanotransferase. J. Ferment. Bioeng. 7 1(5), 322-328. Anderson, L., Sulkowski. E., & Porath. J. (1987). Facile resolution of a fetoproteins and s e w albumins by immobilized metal ion affinity chromatography. Cancer Res. 47, 3624-3626. Bender, H. (1986). Production, characterization and application of cyclodextrins. Adv. Biotechnol. Process. 6, 31-71. Bender, H. (1991). On the role of histidine residues in cyclodextrin glycosyltransferase: chemical modification with ðyl pyrocarbonate. Carbohydr. Res. 209, 145453. Binder, F., Huber, O., &Bock, A. (1986). Cyclodextrin- glycosyltransferasefrom Klebsiellapneumoniae M5al: cloning, nucleotide sequence and expression. Gene 47,26%277. Bradford, M. ( 1976).Arapidand sensitivemethod for the quantitationof microgram quantitiesof protein utilizing the principle of protein-dye binding. Anal. Biochem. 72,248-254. French, D. (1 962). Cyclodextrin transglycosylase (Bacillus rnacerans Amylase) In: Methods in Enzymology (Colowick, S.P., & Kaplan, N.O., Eds.), Vol. V, pp. 48-155. Academic Press, New York. Gottvaldova, M., Hrabova, H., Sillinger, V., & Kucera, J. (1988). Biospecific sorption of cyclodextrin glucosyltransferase on physically modified starch. J. Chromatogr. 427,33 1-334. Hemdan, E.S., Zhao, Y.J., Sulkowski, E., & Porath, J. (1989). Surfacetopography of histidine residues: A facile probe by immobilized metal ion affinity Chromatography. Proc. Natl. Acad. Sci. USA. 86,1811-1815. Hofmann, B.E., Bender, H., & Schulz, G.E. (1989). Three dimensional structure of cyclodextrin glycosyltransferasefrom Bacillus circulans at 3.4 angstrbmresolution. J. Mol. Biol. 209,793-800. Horikoshi, K., Ando, T.K., Yoshida,N.,Nakamura,N., & Kunitachi, N. (1973). Japanpatent 2.453.860. Kaneko, T., Hamamoto, T., & Horikoshi, K. (1988). Molecular cloning and nucleotide sequence of the cyclomaltodextrin glucanotransferase gene from the alkalophilic Bacillus sp. strain N o 38-2. J. Gen. Microbiol. 134,97405. Kimura, K., Kataoka, S.. Ishii, Y., Takano, T., & Yamane, K. (1987). Nucleotide sequence of the j3 cyclodedextrin glucanotransferase gene of alkalophillicBacillus sp. strain 1011 and similarity of its amino acid sequence to those of a amylases. J. Bacteriol. 169(9),4 3 9 W 0 2 . Kitahata, S., Tsuyama, N., & Okada, S. (1974). Purification and some properties of cyclodextrin glycosyltransferase from a strain of Bacillus species. Agric. Biol. Chem. 38,387-389. Krishnan, S. & Vijayalakshmi, M.A. (1985). Purificationof an acid protease and a serine carboxypeptidase from Aspergillus niger using metal chelate affinity chromatography. J. Chromatogr. 329, 165170. Laemmli, U.K. (1970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227,680-685. Laszlo, E., Banky, B., Seres, G., & Szejtli, J. (1981). Purification of cyclodextrin-glycosyltransferase enzyme by affinity chromatography. Starch 33,281-283. Makela, M., Mattsson, P., Schinina, M.E., & Korpela T. (1988). Purificationand properties of cyclomaltodextrin glucanotransferase from an alkalophilic Bacillus. Biotechnol. Appl. Biochem. 10, 414427. Nitschke, L., Heeger, K., Bender, H., & Schulz, G.E. (1990). Molecular cloning, nucleotide sequence and expression in Escherichia coli of the p cyclodextrin glycosyltransferasegene from Bacillus circulans strain No 8. Appl. Microbiol. Biotechnol. 33,542-546. Norberg, E. & French, D. (1950). Studies on the Schardinger dextrins 111, redistributionsreactions of B. maceram amylase. J. Am. Chem. SOC.72,1202-1206. Porath, J., Carlsson, I., Olsson, I., & Belfrage, G. (1975). Metal chelate affinity chromatography, a new approach to protein fractionation.Nature 258,598-599. Porath, J. & O h , B. (1983). Immobilized metal ion affinity chromatography of biomaterials. Serum proteins affinities for gel-immobilized iron and nickel ions. Biochem. 22, 1621-1630.

One-Step Affinity Purification of a Recombinant CCTase by /MAC

537

Schmid, G., Engelbrecht, A., & Schmid, D. (1988). Cloning and nucleotide sequence of a cyclodextrin glycosyltransferase gene fiom the alkalophilic Bacillus 1-1. In: Proceedings of the Fourth International Symposium on Cyclodextrins (Huber, O., & Szejtli, J. Eds.), pp. 71-76. Kluwer Academic Press, Dordrecht, Netherlands. Sulkowski, E. (1985). Purification ofproteins by IMAC. Trends Biotechnol. 3(1), 1-7. Sulkowski, E. (1989). The saga of IMAC and MIT. Bioessays 10, 17&175. Szejtli, J. (1990). The cyclodextrins and their applications in biotechnology. Carbohydr. Polym. 12(4), 375-392. Takano, T., Fukuda, M., Monma, M., Kobayashi, S., Kainuma, K., & Yamane, K. (1986). Molecular cloning, DNA nucleotide sequencing and expression in Bacillus subtilis cells of the Bacillus maceruns cyclodextrin glucanotransferase gene. J. Bacteriol. 166(3), 11 1S1122. Vijayalakshmi, M.A. (1989). Pseudobiospecific ligand affinity chromatography. Trends Biotechnol. 713,71-76. V i h o n M. (1982). Rapid and simple spectrophotometric method for determination of microamounts of cyclodextrins. In: Proceedings of the First International Symposium on Cyclodextrins (Szejtli J., Ed.), pp. 69-74. Reidel, D., Dordrechf Netherlands. Villette, D.P., Looten, J., & Bouquelet, S.J.-L. (1991). Fast purification of cyclodextrin-glucosyltransferase from Bacillus circulans E 192 by affinity chromatography using an epichlorhydnn-reticulated copolymer of beta-cyclodextrin. Chromatographia 32(7/8), 341-344. Villette, J.R., Sicard, P.J., & Bouquelet, S.J.-L. (1992). Cyclomaltodextrin glucanotransferase from Bacillus circulans E 192: 111, Chemical modification by diethyl pyrocarbonate: evidence for an induced fit at the active site resulting from the binding of an acceptor. Biotechnol. Appl. Biochem. 15, 69-79. Zhao, Y.J., Sulkowski, E., & Porath, J. (1991). Surface topography of histidine residues in lysozymes. Europ. J. Biochem. 202, 1 1 1 5 1 1 19.

This Page Intentionally Left Blank

AFFl NITY PURIFICATION OF ENZYMES USING TEMPERATURE-INDUCED PHASE SEPARATION

Folke Tjerneld

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 539 I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 540 11. TEMPERATURE-INDUCED PHASE SEPARATION APPLIED TO PROTEIN PURIFICATION . . . . . . . . . . . . . . . . . . . 54 1 111. AFFINITY PARTITIONING . . . . . . . . . . . . . . . . . . . . . . . . . . 543 Iv. AFFINITY PARTITIONING COMBINED WITH TEMPERATURE-INDUCED PHASE SEPARATION . . . . . . . . . . . . . 544 V. PURIFICATION OF ENZYME FROM YEAST EXTRACT . . . . . . . . . 545 VI. CONCLUDING REMARKS . . . . . . . . . . . . . . . . . . . . . . . . . . 546

ABSTRACT Temperature-induced phase separation can be made with polymers which have a cloudpoint, that is, which phase separate in water after a temperature increase. An

Advances in Molecular and Cell Biology Volume 15B, pages 53-46. Copyright 0 1996 by JAI Press h e . All rights of reproduction in any form reserved. ISBN 0-7623-0114-7

539

540

FOLKE TJERNELD

ethylene oxide/propylene oxide random copolymer (UCON) was used for protein purification. Target proteins were first partitioned to the UCON phase of a UCON/dextran aqueous two-phase system. The UCON phase was removed and isolated in a separate container. The temperature was increased above the cloud point of UCON, which leads to the formation of a new two-phase system with an upper water phase and a lower UCONiwater phase. The proteins were recovered in the water phase free ofpolymer. The UCON/water phase was free ofprotein and easily recycled. Temperature-induced phase separation was used with the affinity ligand Procion Yellow HE3G coupled to UCON. An initial affinity partitioning was made in a phase system containing UCON, UCON-ligand, and dextran. The upper UCON phase containing the UCON-ligand-enzyme complex was removed. A small amount of salt was added to the UCON phase. After temperature increase above the cloud point of UCON (40 "C), the enzyme was recovered in the water phase and the UCON-ligand in the UCON phase. Temperature-induced phase separation is a very promising purification technique where the ligand carrying polymer is easily and effectively removed from the isolated protein by a mild temperature increase.

1.

INTRODUCTION

A number of polymers exist which in water solution have a lower critical solution temperature (LCST). These polymers display inverse solubility in water, that is, as temperature increases they become increasingly insoluble. At temperatures below the LCST the solution is a one-phase system. Above the critical temperature the solution separates into a polymer-rich phase and a water-rich phase. This temperature is called the cloud point of the system. The most well-studied polymer with this property is polyethylene glycol (PEG). At temperatures above the cloud point of a PEG solution, a liquid PEG-rich lower phase is formed that is in equilibrium with a water-rich upper phase (Saeki et al., 1976).Theoretical explanations for the LCST of PEG have been given by Kjellander and Florin (198 l), and Karlstrom (1985). Other examples of polymers with LCST in water are ethyl(hydroxyethy1) cellulose (EHEC) (Carlsson, 1988) and polyvinyl caprolactam (Galaev and Mattiasson, 1992). In contrast to PEG, when heated above the cloud point these two polymers form a precipitate rather than a liquid polymer-rich phase. The cloud point of PEG is influenced by the molecular weight of the polymer and by addition of salts to the solution (Kjellander and Florin, 1981; Karlstrom, 1985). PEG of MW 20,000 has a cloud point of 112 "C, which is too high for most biomolecules. A method for lowering the cloud point of PEG is to introduce propylene oxide groups into the polymer. With a random copolymer composed of 50% ethylene oxide (EO) and 50% propylene oxide (PO) groups, with MW of4000, a cloud poinf of 50 "C is obtained (see Figu're 1) (Harris et al., 1991; Alred et al., 1993). This polymer is commercially available as UCON 50-HB-5 100. The phase behavior of random EOPO copolymers is similar to PEG. Above the cloud point a phase separation into a liquid copolymer-rich lower phase and a water-rich upper

541

45 0.

10

20

30

40

50

60

70

Conc. of UCON 50-HB-5100 In w/w%

Figure 1. Temperature-concentration phase diagram for the binary system UCON 50-HB-51 OO/water. Molecular weight of the polymer is 4000. All concentrations above the solid line are in the two-phase region. Tie-line at 53 "C (----). As indicated by (-) a 10% UCON solution at 53 "C will separate in one phase with 1YO UCON, 99% water, and one phase with 37% UCON, 63% water. From Johanssonet al. (1993).

phase is obtained. The copolymer-rich phase contains about 60% water at 4 "C above the cloud point (Figure 1) (Johansson et al., 1993).

II. TEMPERATURE-INDUCED PHASE SEPARATION APPLIED T O PROTEIN PURIFICATION Temperature-induced phase formation with an EOPO copolymer (UCON) was studied for protein purification (Harris et al., 1991). As a first step an aqueous two-phase system of UCON/dextran was used (Figure 2). In the phase system the UCON polymer is enriched in the top phase and dextran in the bottom phase. This phase system resembles the conventional PEG/dextran aqueous two-phase system, which has widespread use for separation of biomolecules and cell particles (Walter et al., 1985; Albertsson, 1986). UCON is here used instead of PEG as top-phase polymer. In order to lower the cloud point of UCON, 0.2 M Na$O, was added to the system. The target protein is partitioned preferentially to the UCON phase whereas bulk proteins and cell debris are partitioned to the dextran phase. The

542

FOLKE TJERNELD

.

h

3 3

5 8

-

?

m

?

0

m

z

8 3

"

0

5

10

15

20

25

Dextran T500 (% w/w)

Figure 2. Phase diagram for the system UCON 50-HB-5100 (MW 4000), Dextran T500 (MW 500,000) and water at 22 "C. Tie-lines connecting compositions of top and From Harris et al. (1991). bottom phases (u).

UCON phase is withdrawn and isolated in a separate vessel. The temperature is raised to 40 "C, which is above the cloud point of UCON, resulting in a phase separation with a lower UCON-rich phase and an upper water-rich phase. In this new two-phase system the protein partitions exclusively to the upper water phase. The protein has been totally separated from the polymer. The polymer can easily be recycled and used for a second extraction (Harris et al., 1991). Table 1 shows Table 7. Partition of 3-PhosphoglycerateKinase and Hexokinase* Enzyme 3-PGK Hexokinase Note:

K(22 "C) 0.90 0.76

K(40 "C) >I00 >loo

*Primary phase system: 6% UCON 50-HB-5100,3% Dextran T500,90units 3-phosphoglyceratekinase, 16 units hexokinase, and 0.01 M sodium phosphate buffer, pH 7.0. K and G values at 40 "C are for the partition between the water and UCON phases formed by the increase in temperature. From Hams et al. (1991).

Temperature-Induced Phase Separation

543

the partition of two enzymes at 22 "C in the UCON/dextran system and at 40 "C in the waterAJCON system formed by the increase in temperature.

111. AFFINITY PARTITIONING AUCON conjugatewith the affinity ligand Procion Yellow HE-3G was synthesized and used for enzyme purification (Alred et al., 1992). The enzyme was first partitioned in a two-phase system composed of UCON, UCON-ligand, and dextran. The enzyme was extracted into the UCON-ligand carrying upper phase. This is analogous to earlier work using PEG with bound affinity ligands in aqueous two-phase systems (Johansson, 1984).

M

s

-1.5

, . . , . I

u.0

0.1

0.2

0.3

0.4

0.5

0.6

%UCON-Procion Yellow H E 3 G Figure 3. Affinity partitioning of glucose-6-phosphatedehydrogenase. The partition coefficient (K)as a function of the concentration of UCON-Procion Yellow HE-3G. (H) is the line obtained in the system 5.1% UCON 50-HB-5100, 7.0% Dextran T500 and 0.01 M sodium phosphate buffer, pH 7.0. (a)is the line obtained in the system 7.0% PEG 4000, 7.0% Dextran T500 and 0.01 M sodium phosphate buffer, pH 7.0. Temperature: 22 "C. From Alred et al. (1 992).

FOLKE TJERNELD

544

The logarithm of the partition coefficient (K) for glucose-6-phosphatedehydrogenase is shown in Figure 3 as a h c t i o n of the amount of UCON-ligand included in UCON/dextran and PEG/dextran two-phase systems. The ligand-enzyme complex partitions to the top (UCON or PEG) phases of the two systems. The figure shows that by using only 0.2% UCON-Procion Yellow HE-3G in the system the enzyme is effectively extracted into the top phase. Log K is changed from -1.2 to + 1.1 when 0.4% UCON-ligand is included in the UCON/dextran system (Alred et al., 1992).

IV. AFFINITY PARTITIONING COMBINED W I T H TEMPERATURE-I NDUCED PHASE SEPARATION Glucose-6-phosphate dehydrogenase was partitioned in a UCON/dextran phase system with 0.2% UCON-Procion Yellow HE-3G (Table 2) (Alred et al., 1992).In this phase system G6PDH was partitioned to the top UCON-rich phase (K = 4.5). After phase separation had occurred at 22 "C, the UCON-containing phase was removed and isolated in a separate container. Sodium sulfate was added to this UCON-phase to a concentration of 0.2 M. The solution was then placed in a water bath at 40 "C for 15 minutes. The increase in temperature caused formation of a new phase system with an upper phase composed of water, buffer, and salt and a lower phase composed of UCON (approximately 40%) and water. By the salt addition and temperature increase the enzyme was dissociated from the affinity ligand. In the new phase system, glucose-6-phosphate dehydrogenase was totally partitioned to the upper water-salt phase. The enzyme was recovered (88%) in the water phase. The UCON-Procion Yellow partitioned to the lower, UCON-rich phase (KL = 0.06) and could be recovered along with UCON for recycling (77%). There was no enzyme present in the lower UCON and UCON-Procion Yellow phase at 40 "C (Alred et al., 1992).

Table 2. Affinity Partition of Glucose-6-PhosphateDehydrogenasewith Recovery of UCON 50-HB-5100-Procion Yellow HE-3G'

K ~ ( 4 0"C)

KL(40 oC)b

Water Phase

% UCON-Pry Recovered in UCON Phase

>loo

0.06

88.0%

77.1?o'

% G6PDH Recovered in

KE(22 oC)a 4.51 Notes:

K, is the partition coefficient for G6PDH. K, is the partition coefficient for UCON-Procion Yellow HE-3G. 'System is 5.1% UCON 50-HB-5100,7% Dextran T500,0.2% UCON-Procion Yellow HE-3G. and 0.01M sodium phosphate buffer, pH 7.0. The amount of glucose-6-phosphatedehydrogenase was 34 units. K values at 40 "C are for the partition between the water and UCON phase formed by the increase in temperature. From Alred et al. (1992).

a

Temperature-lnduced Phase Separation

545

Table 3. Purification of Glucose-6-PhosphateDehydrogenasefrom Yeast Extract' K (22 QC)

Sample

G6PDH Protein UCON-PrY

12.4 0.32

24.6

K (40 "C)

>I00 >loo 0.32

Purification FactoP (40 "C)

% Recovered ar

4.2

78.gb

40 'C

-

-

-

84.6'

Notes: aThespecific activity of G6PDH in the PEG precipitated homogenate was 0.475 units mg-' protein, equivalent to a purification factor of I. bRecovered in water phase at 40 'C. CRecoveredin UCON phase at 40 "C. 'System is 6.3% UCON 50-HEi-5100.9% DextranT40,0.2% UCON-Procion Yellow HE-3G. 0.02 M sodium phosphate buffer, pH 7.0,, and 5.7% yeast extract. K values at 40 "C are for the partition between the water and UCON phase formed by the increase in temperature. K is the partition coefficient, CJC,. From Alred et al. (1992).

V. PURIFICATION OF ENZYME FROM YEAST EXTRACT Glucose-6-phosphate dehydrogenase was purified from yeast extract in a UCON/dextran aqueous two-phase system using 0.2% UCON-Procion Yellow HE3G (Table 3) (Alred et al., 1992). The purification scheme is shown in Figure 4. In the initial phase system, the enzyme was extracted by the UCON-ligand to the top phase (K = 12). The bulk proteins were partitioned to the bottom phase (K = 0.32). The upper phase was isolated in a separate container. Sodium sulfate and sodium chloride were added, both at 0.2 M, and the temperature was raised to 40 "C. In the new two-phase system formed at 40 OC, the enzyme was recovered in the water phase with a yield of 79% and a purification factor of4.2. The partition coefficient for the enzyme in the water/UCON phase system was >loo. UCONProcion Yellow was recovered in the lower UCON phase with a yield of 85%. No protein could be detected in this UCON phase.

d

RecycleUCON 5OHBd100 and UCON-Pry HE3C

1

Figure 4. Enzyme purification scheme using affinity extraction of target enzyme with UCON-Procion Yellow HE-3G. Temperature-induced phase separation is used for eizyme recovery and recycling of UCON 50-HB-5100 and UCON-ligand. From (Ilred et al. (1 992).

546

FOLKE TJERNELD

VI. CONCLUDING REMARKS A new protein purification technique is demonstrated in these examples where the protein after only two purification steps is obtained in a waterbuffer solution with a high degree of purification and free of contaminating polymers (see Figure 4). The polymer that carries the affinity ligand can be separated from the protein by raising the temperature above the cloud point of the polymer. Thus, both protein recovery and recycling of affinity ligand are much facilitated by temperature-induced phase separation. An ethylene oxide-propylene oxide random copolymer with a cloud point of 18 "C was recently synthesized and used for protein purification (Alred et al., 1994).With polymers that have low cloud point temperature-induced phase separation can be made at room temperature.

ACKNOWLEDGMENTS The author gratefully acknowledges the financial support of the Swedish Research Council for Engineering Sciences (TFR).

REFERENCES Albertsson, P.-.k(1986). Partition of Cell Particles and Macromolecules, 3rd ed., Wiley, New York. Alred P.A., Tjerneld F., Kozlowski A,, & Harris J.M. (1992). Synthesis of dye conjugates of ethylene oxide-propylene oxide copolymers and application in temperature-induced phase partitioning. Bioseparation 2,363373. Alred, P.A., Tjerneld, F., & Modlin. R.F. (1993). Partitioning of ecdysteroids using temperature-induced phase separation. J. Chromatogr. 628,205-214. Alred, P.A., Kozlowski, A., Harris, J.M., & Tjerneld, F. (1994). Application of temperature-induced phase partitioning at ambient temperature for enzyme purification. J. Chromatogr. 659.28P-298. Carlsson, A. (1988). Interaction between ethyl(hydroxyethy1)cellulose and sodium dodecyl sulphate in aqueous solution. Colloid Polym. Sci. 266, 1031-1036. Galaev, I.Yu. & Mattiasson, B. (1992). Affinity thermoprecipitation of trypsin using soybean hypsin inhibitor conjugated with a thermo-reactive polymer. poly(N-vinyl caprolactam). Biotechnol. Techniques 6,35>358. Harris P.A., Karlstrom, G., & Tjerneld, F. (199 I). Enzyme purification using temperature-induced phase formation. Bioseparation 2,237-246. Johansson, G. (1984). In: Methods inEnzymology (Jakoby, W.B., Ed.), Vol. 104, pp. 356364. Academic Press, New York. Johansson, H.-O., Karlstrom, G., & Tjemeld, F. (1993). Experimental and theoretical study of phase separation in aqueous solutions of clouding polymers and carboxylic acids. Macromolecules 26, 447W83. Karlstrom, G. (1985). A new model for upper and lower critical solution temperatures in poly(ethy1ene oxide) solutions. J. Phys. Chem. 89,4962-4964. Kjellander R. & Florin, E. (1981). Water structure and changes in thermal stability of the system poly(ethylene oxide)-water. J. Chem. SOC.,Faraday Trans. I , 77, 205S2077. Saeki, S., Kuwahara, N., Nakata, M., & Kaneko, M. (1976). Upper and lower critical solution temperatures in poly(ethy1ene glycol) solutions. Polymer 17,685489. Walter, H., Brooks, D.E., & Fisher, D. (Eds.) (1985). In: Partitioning in Aqueous Two-Phase Systems: Theory, Methods, Uses, and Applications to Biotechnology, Academic Press, New York.

POSSIBLE SITES ON ANTIBODIES INVOLVED IN THIOPHILIC ADSORPTION

Alexander Schwarz and Meir Wilchek

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. RESULTS AND DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . .

547

548 548

ABSTRACT A possible binding site of antibodies involved in thiophilic binding is deduced from experimental and theoretical work. A possible general mechanism of thiophilic interaction chromatography is suggested.

Advances in Molecular and Cell Biology Volume 15B, pages 547-551. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-01 14-7

547

548

ALEXANDER SCHWARZ and MElR WILCHEK

1. INTRODUCTION Mononclonal and polyclonal antibodies have many applications in biotechnology such as immunoaffinity chromatography, immunodiagnostics, drug targeting, biosensors, and many more applications. For all these purposes homogeneous antibody preparations are needed. Conventional purification methods like protein A, protein G, hydroxyapatite, and ion exchange are either expensive or time consuming. Thiophilic adsorption chromatography (TAC), a chromatographic method introduced by J. Porath et al. in 1985, shows selective binding of immunoglobulins in the presence of structure-forming salts. Different sulfur-containing ligands were found to promote this type of adsorption. The types of ligand suitable for TAC consist oftwo parts. One part can be characterized as a hydrophilicelectron acceptor like a sulfone or pyridine, whereas the other part is an electron donor like a sulfur, nitrogen, or oxygen molecule. The best electron donor seems to be su1fi.u whereas the electron acceptor can be varied, provided it is of hydrophilic nature (Porath et al., 1985; Nopper et al., 1989; Oscarsson et al., 1990). Porath et al. (1987) undertook a study trying to correlate overall structural features like net surface charge, isoelectric point, molecular mass, carbohydrate content, and disulfide linkage with thiophilic adsorption and found no correlation. The results obtained in our studies may shed some light on a possible mechanism and on the location of binding sites on the immunoglobulin molecules.

11. RESULTS AND DISCUSSION Thiophilic adsorption chromatography is highly selective for immunoglobulins with affinity constants ranging in the lod to M range as shown in Figure 1 for some thiophilic smctures. The ability of such simple, chemically well defined ligands to recognize specific surface properties of proteins is not understood. It is believed to be a charge transfer mechanism because studies with aromatic peptides found them to bind strongly to the thiophilic matrix. In our studies, we found that if BSA is dinitrophenylated this derivative binds strongly to thiophilic matrices whereas BSAdoes not interact with the matrix. This experiment supportsthe notion that exposed aromatic moieties are responsible for the interaction. First, we med to locate the approximate binding site or sites on the immunoglobulin molecule. We loaded a monoclonal mouse IgG,, antibody against biotin on the 3s column and injected BSA. The BSA was not bound to the column, whereas in a similar experiment with biotinylated BSA, the protein was bound. Approximately 1.5 molecules of biotinylated BSA were bound per anti-biotin antibody, indicating that the variable part of the antibody is free for antigen binding and does not participate in the binding of the antibody to the thiophilic matrix. Protein A also does not bind to the thiophilic matrix under optimized conditions for antibody binding. Again, loading a mouse IgG,, antibody to the column and

Thiophilic Adsorption Sites on Proteins

Structure

SUPPORT SOLID

iixsa

SOLID

SUPPORT

SUPPORT

549

Name Kd [MI

PYS

'.o

Thia

4.1

lo-'

-2s

1.5 10"

3s

8.7

4s

2.4

I0I

SUPPORT

Figure 7. Structure, name, and Kd of some thiophilic structures.

then injecting protein A resulted in its binding to the column. Approximately 0.8 molecules of protein A were bound per IgG,, indicating that the binding of the antibody is not promoted through its F, region. Furthermore, purified Fab and F(ab), fragments also bound to the thiophilic matrix, albeit under slightly higher Na2S0, concentrations than needed for the binding of whole antibodies. Thus, excluding the variable part from the first experiment and the CH2 and CH3 part of the F, of an antibody from the second and third experiment outlined above, these experiments indicated that the possible binding site or sites are located at the CHI or H region. As all IgGs from a variety of species bind to the thiophilic matrix, we proceeded to investigate common sequences of these antibodies. We believed that there must be a common sequence responsible for the binding. Bearing in mind that a possible mechanism is charge transfer, we looked for exposed aromatic sequences in the

550

ALEXANDER SCHWARZ and MElR WILCHEK

antibody molecule. Furthermore, consideringthe selectivity of TAC for antibodies, we postulated that one exposed aromatic amino acid is not enough for the observed selectivity, whereas at least two exposed residues might explain it. From our above outlined experiments, we concentrated our search for the CH1 and H regions of the antibodies. Comparing the antibody sequences of the CH1 and H regions of all antibodies stored in the Swiss Protein data bank, we found one exposed aromatic sequence in the CH1 region in position 150-15 1 of the heavy chain, which exists in all but IgM antibodies. The sequence G/DF/YFP is located directly in a loop and is therefore highly exposed. There are a number of other proteins that bind with equal strength to thiophilic matrices. These proteins are listed in Table 1. We compared the known sequences of these proteins with our initial result from our sequence comparison and found several exposed aromatic dipeptides that are located in a loop structure. We chose soybean trypsin inhibitor (SBTI) for hrther evaluation of our hypothesis. SBTI can be conveniently cleaved by CNBr into two fragments of similar size, and both fragments contain an exposed aromatic dipeptide sequence. According to our hypothesis, both fragments should bind and, in fact, we found both fragments bound to the thiophilic matrix. Principally, many proteins can be adsorbed to thiophilic matrices depending on the salt content of the solution. Using a gradient system many proteins can be purified on thiophilic matrices (Lihme et al., 1991). There is a whole spectrum of interactions ranging from proteins like cytochrome c, RNase A, and myoglobin that do not interact with the thiophilic matrix to very strong interactions like those found with antibodies. Many proteins interact at intermediate strength with the thiophilic column. We postulate that the mechanism by which thiophilicadsorption chromatography works is a measure of the exposure of aromatic amino acids on the surface or in

Table 1. Proteins which Strongly Interact with Thiophilic Matrices and Their Proposed Binding Sites Protein

IgG SBTI Lysozyme Sweet potato amylase Lens culinaris lectin Carboxypeptidase A Alkaline phosphatase Insulin Hemopexin

Sequence

DGWFR

EWFWD

DlGYFF'E GTYYI SRWWCN PFPWYD DTFYN GVWFA DDYFD GFFYTP RDYFMP

Thiophilic Adsorption Sites on Proteins

551

accessible cavities of proteins. The aromatic sites on the protein interact via their n-electron system with the electron acceptor site like the sulfone or the heterocyclus on the ligand, while the electron donor interacts with the now electron acceptor site on the protein. The chaotropic salt, which is essential for this interaction, helps to remove surface-bound water molecules from these aromatic sites rendering them accessible to the thiophilic interaction. The interaction can be reversed by deletion of the salt when the water returns to the protein surface and the thiophilic interaction is interrupted.Another possibility is to add an electron donor to the elution solution, also interrupting the thiophilic interaction. For the strong interaction seen in antibodies, we postulate that they have a two aromaticamino acid sequence exposed to the aqueous surroundingand are therefore able to strongly interact with thiophilic matrices. We explain the intermediate interactionby a mechanism whereby these proteins have only one exposed aromatic amino acid or two aromatic amino acids that are interrupted by a nonaromatic amino acid. There are several proteins that do not bind to the matrix, and this can be explained by the above outlined mechanism. They do not possess any accessible aromatic amino acids and are therefore not able to interact with the matrix.

ACKNOWLEDGMENTS A. S. thanks the Minerva Foundation for a postdoctoral fellowship and Steven Becker for linguistic advice.

REFERENCES Lihme, A. & Heegaard, P.M.H. (1991). Thiophilic adsorption chromatography, the separation of serum proteins, Anal. Biochem. 192,6449. Nopper, B., Kohen, F., & Wilchek, M. (1989). A thiophilic adsorbent for the one-step high performance liquid chromatography purification of monoclonal antibodies. Anal. Biochem. 1 8 0 , 6 6 7 1. Oscarsson, S. & Porath, J. (1990). Protein chromatography with pyridine and alkylthioether-based agarose adsorbents. J. Chromatogr. 499,235-247. Porath, J. & Hutchens, T.W. (1987). Thiophilic adsorption: Anew kind ofmolecular interaction revealed by chromatography. Intern. J. Quantum Chem., Quantum Biol. Symp. 14,297-315. new method for protein Porath, J., Maisano, F., & Belew, M. (1985). Thiophilic adsorptiofractionation. FEBS Lett. 185,306310.

This Page Intentionally Left Blank

NEW OPPORTUNITIES FOR USING IMMOBILIZED LIGANDS TO CHARACTERIZE MACROMOLECULAR RECOGNITION AND DESIGN RECOGNITION MOLECULES

Irwin Chaiken, David Myszka, and Thomas Morton

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554 I. INTRODUCTION: PROTEIN HEAVEN AND THE UNIFYING THEME OF PROTEIN RECOGNITION . . . . . . . . . . . . . 554 11. OFUGINSOFANALYTICALAFFINITYCHROMATOGRAPHY . . . . . . 555 111. SOME RECENT APPLICATIONS . . . . . . . . . . . . . . . . . . . . . . . 556 A. HIV Envelope gp120 - T Cell CD4 . . . . . . . . . . . . . . . . . . . . 557 B. HIV gag p24 Self-Association . . . . . . . . . . . . . . . . . . . . . . . 557 IV. ADVENT OF THE BIOSENSOR. . . . . . . . . . . . . . . . . . . . . . . . 561 A. New Capabilities in Recognition Analysis Using Immobilized Ligands . 561 B. Comparative Studies with sCD4-gp120 . . . . . . . . . . . . . . . . . . 564 C. Human Interleukin 5 Interaction with its Receptor . . . . . . . . . . . . 566 V. PROSPECTS FOR BIOLOGY AND BIOTECHNOLOGY . . . . . . . . . . 566 567 NOTEINPROOF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Advances in Molecular and Cell Biology Volume 15B, pages 553468. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

553

554

IRWIN CHAIKEN, DAVID MYSZKA, and THOMAS MORTON

ABSTRACT Characterization of protein interactions and interaction sites can provide a means both to learn about molecular recognition and assembly processes in biology and to identify and evaluate new recognition molecules of practical use in biotechnology. We have had a long-standing interest in using immobilized ligands as analytical tools for characterizing recognition mechanisms of proteins and other biological macromolecules. This interest started with analytical affinity chromatography (AAC). Recently, a newcomer among analytical technologies using immobilized ligands has appeared, namely the surface plasmon resonance (SPR) biosensor. The SPR biosensor, as AAC, enables detection and measurement of noncovalent interaction of a soluble macromolecule with a solid phase containing covalently attached ligand. Importantly, the biosensor offers several unique features including access to kinetics (hence deeper mechanistic understanding), ability to analyze molecules in mixtures (hence access to more biologically relevant conditions), and real-time observation of the interaction process (hence ability to observe interacting molecules as they form or are added). Recent results with HIV proteins including p24 self assembly and CD4-a120 interactions, as well as with interleukin 5 and its receptor reflect some of the growing uses of both AAC and the SPR biosensor as macromolecular recognition tools. Overall, the advent of the SPR biosensor and the likely follow-up development of other automated devices promise to stimulate evolution of the analytical use of immobilized ligands that started with AAC into a broad-based analytical solid phase science for the field of biomolecular recognition.

1. INTRODUCTION: PROTEIN HEAVEN AND THE UNIFYING THEME OF PROTEIN RECOGNITION One of the challenges-and pleasures-of current biotechnology drug discovery is being able to find solutions to practical problems in medicine against the backdrop of “protein heaven”, the explosion in biological sciences of newly discovered proteins and newly revealed principles of protein structureand function. A unifying feature which provides a useful paradigm to assemble this wealth of new knowledge in biology and to conceive technological solutions is molecular recognition. Protein interactions are at the root of virtually all biological processes including chromosomal organization and gene expression, assembly of cell organelles and metabolic pathways, the immune response, cell signaling and trafficking, and development. We are learning a lot about the identity of biological macromolecules-synthetases and proteases; adhesion receptors and counterreceptors; nucleosome assembly proteins and transcriptional activators; antibodies and antibody receptors; cytokines, cytokine receptors and signal transducers; and on and on. Normal interactions of such proteins and other macromolecules are what keep a biological system going, whereas inappropriate recognition events, due to too much of a normal recognition molecule or its loss or mutation to a form with altered interaction properties, lead to disease. Thus, mimicking protein recognition and

Macromolecular Recognition and Design Recognition Molecules

555

antagonizing it are practical goals common to much of therapeutics discovery. As a consequence, we seek methods to detect proteins with specific interaction properties, to characterize protein interactions at the molecular and structural level and to design mimics and antagonists. Some time ago, we became interested in immobilized ligands to characterize macromolecular interactions. Our work started with ligands on chromatographic supports and evolved into analytical affinity chromatography.The recent discovery and development of the surface plasmon resonance biosensor has led us to explore this new technology, using BIAcoreTMof Pharmacia Biosensor, for macromolecular recognition studies. The intent of this paper is to describe our recent experiences with both technologies and ideas on their use to discover and characterize interactions of biological macromolecules and to identify new recognition molecules as potential mimetics and antagonists.

II. ORIGINS OF ANALYTICAL AFFINITY CHROMATOGRAPHY The idea that affinity chromatography could be used as an analytical method arose largely from its success as a preparative method. The laboratory of Porath had discovered chromatographic supports and activationprocedures amenable to covalent attachment of ligands with selective affinity for proteins (Axen et al., 1967; Porath and Kristianson, 1975), while Anfinsen’s laboratory had demonstrated the use of such an affinity support to purify staphylococcal nuclease (Cuatrecasas et al., 1968). The power of preparative affinity chromatography, seen repeatedly for a wide range of proteins and other biomolecules (Jakoby and Wilchek, 1974; Dean et al., 1985), argues that a high level of fidelity must occur in the interactions that take place between proteins and ligands immobilized on affinity supports. Figure 1Aand 1B show several key features ofaffinity chromatographic interaction which, based on the above noted fidelity of protein interactions with immobilized ligands, no doubt operate in preparative affinity chromatography for proteins. These key features are: (1) accessibility of the immobilized ligand to protein active sites; (2) selectivity of ligands to discriminate between proteins with different binding sites; and (3) reversibility of proteirr--affinity support interaction allowing elution under gentle conditions. Accessibility, selectivity, and reversibility also provide the underlying basis for analytical affinity chromatography.Based on these properties, it was predicted that chromatographic elution volume of a protein on an affinity support would be a faithful reflection of its biologically relevant interaction affinity and that the conditions of binding to affinity support would allow nonchaotropic elution to measure such elution volumes. We therefore devised a strategy of isocratic elution under binding conditions, along with competitive elution, to measure interaction properties of eluting proteins through the measurement of their elution volumes (Figure 1C). Isocratic elution of macromolecules on the affinity support yields

IRWIN CHAIKEN, DAVID MYSZKA, and THOMAS MORTON

556

A. BINDING: ACCESSIBILITY. SELECTIVITY

B. ELUTION: REVERSlBlLlTY

C. ANALYTICAL AFFINITY CHROMATOGRAPHY

4

isocratic

elution

competitive

b

elution

-

Figure 7. Scheme depicting the key features of accessibility and specificity (A) and reversibility (B), which are common to preparative and analytical affinity chromatography. (C) Scheme depicting isocratic elution and competitive elution that are the major experimental strategies of analytical affinity chromatography.

elution volumes that allow equilibrium binding afinity to be determined as KMR. Competitive elution, isocratic elution in the presence of soluble ligand competing with matrix ligand, allows simultaneous measurementof both the matrix interaction (KMp) and interaction in solution (KLIp).Descriptions of this methodology have been presented (Dunn and Chaiken, 1974;Chaiken, 1979; Swaisgood and Chaiken, 1987; Chaiken et al., 1992). Similar approaches have been devised in other laboratories (Lowe et al., 1974;Nichol et al., 1974;Kasai and Ishii, 1975;Brodelius and Mosbach, 1976).

111. SOME RECENT APPLICATIONS Understanding and controlling protein recognition are dominant themes in such present-day therapeutics discovery efforts as the characterization of the human

Macromolecular Recognition and Design Recognition Molecules

557

immunodeficiency virus (HIV) and the search for therapeutics approaches for HIV-mediated acquired immune deficiency syndrome (AIDS). The protein components of HIV are known and their interactions are integral to such events as viral invasion of T cells (through envelope protein gp120 recognition of CD4), viral propogation (includingreverse transcriptaseheterodimer formation and interaction with RNA and DNA), and viral maturation (including p24 self-association in nucleocapsid assembly and protease dimerization and recognition of gag precursor protein substrates). All of these processes represent targets for therapeutics development. We have established analytical affinity chromatographic assays for characterizing some of these protein-protein recognition processes and potentially for identifying antagonists or modulators of these interactions. A. HIV Envelope gpl20 - T Cell CD4

We have devised a frontal elution AAC method using immobilized gp120 to measure interactions of envelope protein with soluble CD4 (sCD4) as well as with domain constructs of sCD4 and their sequence mutants. Figure 2 shows schematic views of the g p 1 2 M D 4 interaction (as it would occur biologically as well as on an affinity chromatographic support) and elution data obtained. The dissociation values determined are denoted as Kw’s for P (sCD4 or VlV2) interaction with matrix bound gp120 (M). The affinities measured here are quite strong with dissociation constants in the nM range. The chromatographic affinities are similar to those determined by antibody-based assays (Arthos et al., 1989; Moebius et al., 1992). The close-to-sCD4-magnitude binding affinity of V1V2 is consistent with crystal structure and mutagenesis studies of VlV2, which show that the amino terminal two-domain molecule is quite rigid and contains the major residues responsible for gp120 recognition (Arthos et al., 1989; Ryu et al., 1990). Several single site mutants with changes in putative binding site residues Phe-43 and Ala-55 show undetectable binding to immobilized gp120 (D. Myszka, data not shown). Beyond the determinationof interaction properties of sCD4 and its mutants, elution of CDCrelated molecules also could be a convenient, sensitive, and direct means to determine the efficacies of small molecules as mimetics of sCD4 (or of gp120 for that matter) by their ability to antagonize the extent of sCD4 (or VlV2) chromatographic retardation on immobilized gp 120. B. HIV gag p24 Self-Association

For nucleocapsid assembly studies, we have immobilized p24 as its monomer and found that soluble p24 can interact with the affinity support (Rose et al., 1992). The interactions envisioned and examples of elution data are shown in Figure 3. Here, we have been able to perform both zonal and frontal elutions of p24, with the affinities determined by both methods being closely similar to one another. As denoted in Figure 3A, the chromatographic method leaves open the possibility to identify small molecule agents that would antagonize or otherwise alter p24

558

IRWIN CHAIKEN, DAVID MYSZKA, and THOMAS MORTON

A. HIV docking

on Tcell

bB

B. Solid phase affinity analysis

(continued

Figure 2. Frontal elution analytical affinity chromatographic measurements of interaction of soluble CD4 (sCD4) with human immunodeficiency virus envelope protein gpl20 (strain BHlO). A: Schematic diagram of HIV-1 docking onto CDCbearing T cell through CD4-gpl20 recognition. B: Scheme of solid phase affinity system with immobilized gp120 and soluble CD4. C: Frontal elution profiles for sCD4 at a concentration range of 22 to 440 nM. Vo is the elution volume for a nonretarded protein of similar size to sCD4 (ovalbumin). BH1Ogpl20 was immobilized to Affiprep -1 0 through amines (Pierce). Experimental conditions were: chromatography buffer 20 mM sodium phosphate, pH 7.4, 150 mM sodium chloride and 0.01 % Tween 20; flow rate 0.05 ml/min; temperature 20 "C; detection by absorbance at 21 0 nm; the gp120 surface was regenerated between elutions using 0.1 M phosphoric acid in 2 ml. D: Linearized replot of AAC data using the procedure described by Rose et al. (1 992). These data led to a calculated KM = 20 nM.

interaction and hence might antagonize either normal nucleocapsid formation during viral maturation or nucleocapsid disassembly during cell infection. Viral uncoating inhibitors (stabilizers of nucleocapsid) have been identified for the nucleocapsid protein of rhinovirus (Smith et al., 1986; Diana et al., 1989) and proposed to function in a similar way with HIVgag 24 (Rossman, 1988).

Macromolecular Recognition and Design Recognition Molecules

559

C. Frontal Elution Profiles 440

nM

0.14 vo

0.12

E

0 v

-

0.10

h(

3C

0.08

$

0.06

m

u)

n

a

0.04 0.02 0

0

20

40

80

60

100

120

140

Time minutes

D. Linearized Replot of AAC data 6000

-

5000 -

-

4000

-

3000

-

2000

-

7

2

I

1

0

?

.

L 7

K M ~ 20 = nM

1000 -

0

1

2

3

4

sCD4 concentration [M x 1W7]

Figure 2. (Continued)

5

220 nM

IRWIN CHAIKEN, DAVID MYSZKA, and THOMAS MORTON

560

A) Scheme Immobilized p24 (M)

Soluble p24 (P)

p

B) Frontal Elution K~~ = 3.7 x

C) Zonal Elution K~~ = 3.0 x 10-5 M

105M

I

0

10 Time (min)

+ EFFECTORS

20

0

V.

1.78 rnl

I

5

10

Time (min)

figure 3. Analysis of HIV-1 p24 selfassociationand screening for effectors of selfassociation by elution of soluble p24 (P) on immobilized p24 (M).A: scheme showing the dimerization interactions of p24 in solution (defined by the dissociation constant Kpp) and on the solid phase (defined by dissociation constant Kwp). Parts B and C show typical analytical chromatographic elution profiles for the p24 system described in A. The Km values shown were determined from elution data obtained over a range of p24 concentrations. B: frontal chromatography profile for p24 at a concentration of 80 ug/ml (continuous line); the minimum of the first derivative of the elution profile (dashed line) defines the experimental elution volume Ve. Experimental conditions were: buffer, PBS - pH 7.0; flow rate, 0.25 ml/min; temperature, 20 "C; detection by absorbance at 280 nm. C: zonal chromatography profile for p24 at the same experimental conditions as in B. VOis the nonretarded elution volume. (Figure adapted from Rose et al., 1992).

Macrornolecular Recognition and Design Recognition Molecules

561

IV. ADVENT OF THE BIOSENSOR A. N e w Capabilities in Recognition Analysis Using I m m o b i l i z e d Ligands

We investigated the use of the SPR biosensor as a tool to characterize protein interactions and compared the results obtained with data using AAC. We have focused our work so far on the system from Pharmacia called BIAcore. Its configurationinvolves the interaction of solublemacromolecules in a flow cell with ligands immobilized on the dextran coating of a gold chip. The interaction is measured directly as an increase in refractive index, a change directly related to an increase of molecular mass on the chip. In principle, the SPR and AAC methods for interaction analysis have several common features including interaction of soluble macromolecule with immobiCommon Features of AAC and BIA as Tools in Molecular Recognition o Solid phase immobilized ligand o Flow through of macromolecule o Means to measure binding event directly

Some Unique Features and Opportunities with BlAcore o enalvte

-

mobile macromolecule can be mixed with other molecules

Hence, greater access to "biological conditions"

o Klnetlcs

-

kinetic parameters, not just equilibrium, can be measured

Hence, improved understanding of mechanisms of macromolecular recognition

o p e aI time - recognition event can be visualized as it occurs Hence, rapid and direct evaluation of binding ability of biomolecules as they are added or form

Figure 4. Some comparative features of analytical affinity chromatography and BlAcore SPR biosensor for characterization of macromolecular interactions using immobiIized Iigands.

562

IRWIN CHAIKEN, DAVID MYSZKA, and THOMAS MORTON A. Sensorgrams 1200

1000

-

BOO

-c 111

3 Q u)

c

0

600

n u)

d

400

mo

0

I

I

I

I

I

I

I

1

I

mo

300

400

500

600

700

800

goo

1000

I

o

100

I

I

I

I

Time (sac)

(continued

Figure 5. BlAcore analysis of sCD4 binding to immobilized gpl20. A: Sensorgram overlays for sCD4 at a concentration range of 50 to 1100 nM. Initially, 2000 response units (RU) of BH1Ogpl20 was immobilized to the sensorchipsurface through standard amino-coupling chemistry (Pharmacia).Experimentalconditions were: buffer - 20 m M HEPES, pH 7.4, 150 mM sodium chloride, 0.01% Tween 20; flow rate, 5 pl/min; temperature, 20 “C; the gpl20 surface was regenerated between runs by a 10-pl wash with 0.1 M phosphoric acid. 6: The on rate determination plot is an analysis of rate constants expressed in slope values (derived from a plot [not shown] of binding rates versus relative response) versus the sCD4 protein concentration as described by Karlsson et at. (19911. From the slope of the curve, kon = 85,000 M-’s-’. C: the off-rate determination plot represents the dissociation of bound sCD4 from gpl20 in continuous buffer flow. Data shown from the 1100 nM injection (see part A) was analyzed as described by Karlsson et al. (19911. The slope of the curve from 0 to 200 seconds gave a koty = 0.0005 s-’ . Hence, the apparent K,j = 6nM.

Macromolecular Recognition and Design Recognition Molecules

8. On rate determination 0.10-

-e

0.08

-

r

Y

0)

-

0.06-

3

9

8

0.04-

0 cn

0.02

-

O.M)?. 0

ken= 85,000 M'b-' 1

200

.

I

.

400

I

.

600

1

.

I

,

1

1000 1200

800

sCD4 concenlration [nMJ C. OW rate determination In(R1IRn) = k,(tr

1,)

0.15

0.05

0.00 0

100

200

Time (sec)

Figure 5. (Continued)

300

563

564

IRWIN CHAIKEN, DAVID MYSZKA, and THOMAS MORTON

lized ligand, interaction during flow, and direct measurement of interaction process (as opposed to measuring the interaction indirectly through activity changes or spectroscopic transitions). These similarities are listed in Figure 4. In contrast, BIAcore interaction analysis carries with it some important advantages. These also are highlighted in Figure 4. Perhaps most noteworthy is the capacity to determine kinetic rate constants for association and dissociation. Obtaining such information about macromolecular interactions has not been routine in biochemistry and biophysics. The capacity to characterize kinetic constants can improve mechanistic understanding of protein structure and finction. In addition, comparing on and off rates for potential agonists and antagonists in drug discovery can allow choices to be made amongst therapeutic leads based on the expected lifetimes of their complexes with macromolecular targets. B. Comparative Studies with sCD4-gpl20

In the context of studies on the recognition properties of sCD4 for the HIV envelope protein gp120, we have established a BIAcore assay with chip-immobilized gp120 and compared the results obtained with those from AAC. A representative set of sensorgrams obtained for sCD4 is shown in Figure 5. The value for equilibrium dissociation constants was obtained as the ratio of off and on rates determined from the descending and ascending parts of the sensorgrams. Interestingly, the BIAcore value obtained for sCD4,6 nM,is close to, but somewhat smaller than the dissociation constant determined by AAC. The reasons for this small difference are not fully understood at present. They may reflect differences in the nature of gp120 immobilized on chromatographic beads versus the sensor chip, for example differences in density or local environment.Nonetheless,in general, these and other data obtained so far suggest that BIAcore-determined equilibriumaffinity constants are close reflections of the analogous values obtained by AAC and other methods. Similar comparisons of BIAcore-derived rate constants with those from other methods are more difficult to make at present due to the relative lack of such data by non-BIA methods.

figure6. BlAcore analysisof hlL5 interaction with soluble IL5 receptor alpha subunit. A: Schematic diagram of IL5 interacting with membrane bound alpha and beta subunits of the IL5 receptor in eosinophils. B: Configuration of the BlAcore binding assay. The soluble portion of the IL5 receptor alpha subunit (slL5R) was immobilized on the BlAcore sensor chip and the binding of hlL5 was monitored. The interaction has also been studied using immobilized I15 and free receptor (data not shown). (continued)

Macromolecular Recognition and Design Recognition Molecules

B. Solid phase affinity assay

IL5 binding to Eosinophil

A.

565

m IUR IUR

C. BlAcore sensorgrams for IL5 binding to immobilized IL5 receptor.

1 200

--

150

--

(I)

L

E

3

g

$ 100 --

a

50

-koft = 0.005 5-l KD = 2.5 nM

0

-0

100

200

300

400

500

600

700

800

Time (sec)

Figure 6. (Continued)C: Overlay plot of sensorgrams for binding of various concentrations of IL5 to immobilized soluble hlL5R. Soluble receptor was immobilized using amine coupling. Human IL5 (1-29 nM) in HEPES buffered saline, pH 7.4 (HBS), was injected at 100 sec. At 460 sec, the buffer was changed to HBS alone. Analysis of the binding and release phases of the curves (Karlsson et al., 1991) gave values for bn and k0H, respectively, from which Kd was calculated.

566

IRWIN CHAIKEN, DAVID MYSZKA, and THOMAS MORTON

C. Human lnterleukin 5 Interaction with its Receptor

We have initiated studies of recombinant forms of hIL5 and soluble IL5R alpha subunit using BIAcore. Human IL5 is a covalent dimer (McKenzie and Sanderson, 1992); the receptor alpha subunit forms a heterodimer with a receptor beta subunit when associated in membranes as full length (membrane spanning region containing) constructs (Tavernier et al., 1991). In the IL5 case, chips with each of the protein interactors have been prepared, and both are functional in binding the corresponding protein partner. Some of the initial data obtained are shown in Figure 6 for hIL5 binding to immobilized receptor alpha subunit. Interestingly, whereas the equilibrium dissociation constant for hILS/sILSR alpha is similar in magnitude to that for sCDWgp120, the rate constants are different by at least an order of magnitude. This differential lends support to the view that the rate constants being obtained with BIAcore are molecule-related and not instrument-related and thus apparently reliable descriptors of particular proteins.

V. PROSPECTS FOR BIOLOGY AND BIOTECHNOLOGY The advent of the SPR biosensor for analysis of protein interactions has reemphasized the basic usefulness of immobilized ligands to probe molecular recognition in biology. When AAC was first conceived, it was appealing in part because it allowed analysis across a wide affinity range (from mh4 to nM) and wide size range (from small molecules to macromolecules) and with sufficient sensitivity (limited only by detection of eluted protein and thus down to picamoles or even below) to make it accessible for most biomolecules. AAC also required only simple instrumentation, basically a column and a postcolumn detector and collector. And, while the interaction being analyzed is on a solid phase and therefore different from interactions more usually studied in dilute aqueous solution, the interaction can be viewed as a strength (more relevance to biology where interactions often occur on solid surfaces).Nonetheless,the optical biosensor concept, and BIAcore in particular, promise to popularize the analytical use of immobilized ligands because of the rapidity and apparent ease in setting up the interaction system. The computational side of BIAcore is still in relative infancy; much advancement is likely to come in extracting useful information from sensorgrams. Further effort is needed to validate the binding parameters generated by BIAcore, or at the least to compare these data with those from other methods including methods fully in solution. In spite of the caveats, however, the hope of course is that interaction analysis will be accessible and attractive enough that the explosion of discovery of proteins and other biomolecules will be matched by characterization of their recognition properties and consequent advances in our basic understanding of the principles of recognition.

Macromolecular Recognition and Design Recognition Molecules

567

NOTE IN P R O O F Since the presentation of this paper at the Mosbach Symposium in 1992, we and others have extended investigation of the use of optical biosensors to characterize macromolecular interactions. Some key papers from our own work recommended for further reading are Morton et al. (1993, 1995a, 1995b), Li et al. (1 996).

REFERENCES Arthos, J., Deen, K.C., Chaikin, M.A., Fornwald, J.A., Sathe, G., Sattenau, Q.J., Clapham, P.R., Weiss, R.A., McDougal, J.S., Pietropaolo, C., Axel, R., Truneh, A,, Maddon, P. J., & Sweet, R.W. ( 1989). Identification of the residues in human CD4 critical for the binding of HIV. Cell 57,46%48 I . Axen, R., Porath, J., & Ernbach, S. (1967). Chemical coupling of peptides and proteins to polysaccharides by means of cyanogen halides. Nature 214, 1302-1304. Brodelius, P. & Mosbach, K. (1976). Determination of binding constants for binary dehydrogenase-coenzyme complexes by (bio)affinity chromatography on an immobilized AMP-analogue. Anal. Biochem. 72,62%636. Chaiken, I.M. (1979). Quantitative uses of affinity chromatography. Anal. Biochem. 97, 1-10. Chaiken, I., Rose, S., & Karlsson, R. (1992). Analysis ofmacromolecular interactions using immobilized ligands. Anal. Biochem. 201,197-210. Cuatrecasas, P., Wilchek, M., & Anfinsen, C.B. (1968). Selective enzyme purification by affinity chromatography. Proc. Natl. Acad. Sci. USA 61,636443. Dean, P.D.G., Johnson, W.S., & Middle. F.A. (Eds.) (1984). Affinity Chromatography-A Practical Approach. IRL. Press, Oxford. Diana, G.D., Pevear, D.C.,Otto, M.J.,McKinlay, M.A.,Rossmann,M.G., Smith, T., &Badger, J. (1989). Inhibitors of viral uncoating. Pharmac. Ther. 42,289-305. Dunn, B.M. & Chaiken, I.M. (1974). Quantitative affinity chromatography. Determination of binding constants by elution with competitive inhibitors. Proc. Natl. Acad. Sci. USA 71,2382-2385. Jakoby, W.B. & Wilchek. M. (Eds.) (1974). Methods in Enzymology Vol. 34. (Affinity T e c h n i q u e s Enzyme Purification Part B). Academic Press, New York. Karlsson, R., Michaelsson, A., & Mattsson, L. (1991). Kinetic analysis of monoclonal antibodqcantigen interactions with a new biosensor basedanalytical system. J. Immunol. Methods 145,229-240. Kasai, K. & Ishii, S. (1975). Quantitative analysis of affinity chromatography of trypsin. A new technique for investigation of proteiwligand interactions. J. Biochem. 77, 261-264. Li, J., Cook, R., Dede, K., & Chaiken, I. (1996). Single chain human interleukin 5 and its asymmetric mutagenesis for mapping receptor binding sites. J. Biol. Chem. 271, 1817-1820. Lowe, C.R., Harvey, M.J., & Dean, P.D.G. (1974). Affinity chromatography on immobilized adenosine 5'-monophosphate VI. Some kinetic parameters involved in the binding ofgroup specific enzymes. Eur. J. Biochem. 42, 1 4 . McKenzie, A.N.J., & Sanderson, C.J. (1992). Interleukin 5. In: Interleukins: Molecular Biology and Immunology (Chem. Immunol. Vol. 51) (Kishimoto, T., Ed.), pp. 135152. Karger, Basel. Moebius, U., Clayton, L.K., Abraham, S., Harrison, S.C., & Reinherz, E.L. (1992). The human immunodeficiency virus gp120 binding site on CD4: delineation by quantitative equilibrium and kinetic binding studies of mutants in conjunction with a high-resolution CD4 atomic structure. J. Exp. Med. 176,507-5 17. Morton, T.A., Bennett, D.B., Appelbaum, E.R., Cusimano, D.M., Johanson, K.O., Matico, R.E., Young, P.R., Doyle, M., & Chaiken, I.M. (1994). Analysis of the interaction between human interleukin 5 and the soluble domain of its receptor using a surface plasmon resonance biosensor. J. Mol. Recognition 7,47-55.

568

IRWIN CHAIKEN, DAVID MYSZKA, and THOMAS MORTON

Morton, T.A., Myszka, D.G., & Chaiken, I.M. (1995a). Interpreting complex binding kinetics from optical biosensors: a comparison of analysis by linearization, the integrated rate equation and numerical integration. Anal. Biochem. 227, 176.185. Morton, T., Li, J., Cook, R., & Chaiken, I. (1995b). Mutagensis in the carboxyl terminal region ofhuman interleukin 5 reveals a recognition patch for the receptor a Chain. Proc. Natl. Acad. Sci. USA 92, 10879-1 0883. Nichol, L.W., Ogston, A.G., Winzor, D.J., & Sawyer, W.H. (1974). Evaluation of equilibrium constants by affinity chromatography. Biochem. J. 143,435-443. Porath, J. & Kristianson, T. (1975). Biospecific affinity chromatography and related techniques. In The Proteins Vol. 1 (Neurath, H. & Hill, R.L., Eds.), pp. 95-178. Academic Press, New York. Rose, S., Hensley, P., O’shannessy, D., J., Culp, J., Debouck, C., &Chaiken, 1. (1992). Characterization of HIV- 1 p24 self-association using analytical affinity chromatography. ProteinsStructure, Function, Genetics 13. 112-119. Rossmann, M. G. (1988). Antiviral agents targeted to interact with viral capsid proteins and a possible application to human immunodeficiency virus. Proc. Natl. Acad. Sci. USA 85,462-627. Ryu, S.-E., Kwong, P.D.. Truneh. A,, Porter, T.G., Arthos, J., Rosenberg, M., Dai, X., Xuong, N.. Axel, R., Sweet, R.W., & Hendrickson, W.A. (1990). Crystal structure of an HIV-binding recombinant fragment of human CD4. Nature 348,419426. Smith, T.J., Kremer, M.J.,Luo, M., Vriend, G., Arnold, E., Kamer, G., Rossmann, M.G.. McKinlay, M.A., Diana, G.D., & Otto, M.J. (1986). The site of attachment in human rhinovirus 14 for antiral agents that inhibit uncoating. Science 233, 1286-1293. Swaisgood, H.E. & Chaiken, LM. (1 987). Analytical affinity chromatography and characterization of biomolecular interactions. In: Analytical Affinity Chromatography (Chaiken, 1.M.. Ed.), pp. 65-1 15. Academic Press, Boca Raton. Tavemier, J., Devos, R., Cornelis, S., Tuypens, Van der Heyden, J., Fiers, W., & Plaetinck, G. (1991). A human high affinity interleukin 5 receptor (IL5R) is composed of an IL5-specific a chain and a p chain shared with the receptor for GM-CSF. Cell 66, 1175-1 184.

PEPTIDES: MULTIPLE PURPOSE TOOLS

Jean-Luc Fauchh-e

I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 569 570 11. SYNTHETIC PEPTIDE LIBRARIES . . . . . . . . . . . . . . . . . . . . . 111. THE DESIGN OF THE SECONDARYSTRUCTURE OF PEPTIDES . . . . 571 IV. PEPTIDE DRUG DESIGN . . . . . . . . . . . . . . . . . . . . . . . . . . . 574 V. MULTIPLE USES OF PEPTIDES . . . . . . . . . . . . . . . . . . . . . . . 576

1. INTRODUCTION The intermediate size-lying between that of proteins and of the amino acid building bloch+confers on peptides a specificity of their own in several ways. First, as a rule, the peptide behaves as a relatively flexible chain with little defined secondary structure in water and high adaptability to partners such as receptor, antibody, or catalytic site of a protease. However, naturally cross-linked peptides such as oxytocin and vasopressin exist, and insulin, a 5 1-residue triply bridged peptide, behaves as a protein. Secondly, after being liberated from a precursor protein by the action of a processing enzyme, the peptide has its own identity as a Advances in Molecular and Cell Biology Volume 15B, pages 569583. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 47623-0114-7

569

JEAN-LUC FAUCH~RE

5 70

functionally active species, either as a hormone or a neurotransmitter, or an antibiotic.Thirdly,as a chemically well defined small molecule, the peptide is easily synthesized and modified by the chemist, and peptide engineering has become as useful as its protein counterpart. In this short survey, the following related topics will be stressed: (1) the new strategiesand applications made availableby the establishmentof peptide libraries; (2) the design of the secondary structure of peptides; (3) some new methods and recent achievements in peptide drug design; (4) a number of other applications of chemically modified peptides demonstrating again the versatility of peptides as multipurpose tools in pharmacology and biotechnology.

II. SYNTHETIC PEPTIDE LIBRARIES The introduction of peptide synthesis on solid phase by Merrifield (1963) dramatically increased the speed of synthesis and became an indispensable tool for the rapid production and screening of pharmacologically active analogs. We face now an explosion of new powerful methods for the multiple parallel synthesis of peptides which are likely to gain comparable importance. The representative approaches of this new technology and their evaluation according to five parameters (as well as to the opinion of the reviewer!) are shown in Table 1. The multipin peptide synthesis (Geysen et al., 1987), in which a peptide chain is growing on each of the 96 pins arranged on a microtitration plate, was the first to be proposed. Multipin peptide synthesis does not require special equipment; it increases the synthesis speed by two orders of magnitude and can be combined with an ELISA test of the peptide on the pin. The one-bead-one-peptide approach (Lam et al., 1991) is much more powerful in terms ofnumber ofpeptides (up to 34 million hexapeptides); however, it requires a mix and divide procedure after each coupling

Table 7. Peptide Synthesis Libraries Method and Authors

Multipin peptide synthesis: systematic screening for bioactive peptides (Geysen et al., 1987) One bead-one peptide approach: a new type of synthetic peptide library for identifying ligand-binding activity (Lam et al., 1991) Modified tea-bag method: rapid solid-phase synthesis of large numbers of peptides (Houghten et al., 1991) Light-directed spatially addressable parallel chemical synthesis (Fodor et al., 1991)

Speed

Number

Quantity

+

+

+

Form Identification

-pin (free)

+++ (+)

+++

+++

+

-bead

+

+

+

+++

Free (-resin)

+

+++

+

+

-chip

+++

Peptides: Multiple Purpose Tools

571

step and a sequencing of the peptide on the bead after the ELISA test. The modified tea-bag method (Houghtenet al., 1991) also requires the mix and divideprocedure but can be conducted so as to permit peptide identificationwithout sequencing (see later). Finally, the light-directed spatially addressable parallel peptide synthesis (Fodor et al.,1991) is the most sophisticatedprocedure, combiningphotolithographywith solidphase peptide synthesis that can produce and identify as many as lo4peptides on a one square centimeter chip. However, this procedure requires high-technologyequipment, not yet easily accessible to a nonspecializedlaboratory. To fully define a hexapeptide epitope according to the strategy outlined by Houghten et al. (1991), a library of about 34 million hexamers of structure O,O,XXXX is first established. The process of dividing, coupling, and recombining the resins ensures equimolecularity within the X X X X peptide resin, whereas the residues 0, and 0, are identified and contained in 324 (18’) individual peptide resins. In the next step, synthesis is resumed in positions 0, and 0, using the residues of the ELISApositive mixture. The second ELISA locates residues 0, and 0,. Finally, a new synthesis followed by ELISA identifies the two last residues 0, and 0,. This procedure, at first sight cumbersome, is in fact very effective, since it allows one to screen large combinatorial peptide libraries and to possibly discover, for example, new leads for the development of peptide drugs. Robotic devices for automatically dividing and recombining the resin in multiple peptide synthesizers are being developed (Zuckermannet al., 1992;Boutin and Fauchiere, 1996a;Boutin et al., 1996). Genetic approaches for the production of epitope libraries, in which each individual peptide is displayed on the surface of a bacteriophage clone, are also progressing at high speed (see, e.g., Smith and Scott, 1992).

Ill. THE DESIGN OF THE SECONDARY STRUCTURE OF PEPTIDES More than just reproducing the primary sequence of a peptide, the chemist has learned over the years how to predict and modulate its three-dimensional structure in several ways (Table 2). Turns (mainly p- and y-turns), which extend over three or four amino acids and are stabilized by hydrogen bonds in proteins (Rose et al., 1985), can be reproduced in peptides either by introduction of pairs of prolinelike constrained residues [(1,2,3,4)-tetrahydroisoquinoline-3-carboxylicacid (Tic), octahydroindole-2-carboxylicacid (Oic) in positions 7 and 8 of bradykinin; Kyle et. al., 19921, or by the synthesis of covalently stabilized turn mimetics (p-turns: Kahn et al., 1988; y-turns: Huffman et al., 1988). Both these important secondary structural elements can be seen either as the beginning of an a-helix or P-pleated sheet or as the first step towards peptide cyclization. Pseudopeptides result from modifications of the peptide backbone, in contrast to those of the side chains more easily obtained via amino acid substitutions.Possible peptide bond surrogates (for a review, see Spatola, 1983)include CO-N(CH,), CH,-NH, CH(0H)-CH,, CS-NH,

Table 2. The Design of Peptide Secondary Structure (Peptide Engineering) 3D-Structure

cn v N

Design Method

Turns

Pro(0-Alkyl),Tic,Oic mimics

Pseudopeptide (altered geometry of peptide bond) Cyclic peptides

backbone modifications XHOHXH, -NH-C(CH3)+(r cyclization with exocyclic Tyr bicyclic [&lo, -1decapeptide alternating, D, L-residues

Amphipathic lefthanded P-helix Amphipathic a-helix choice of a-forming residues helical wheel analysis 3,,-helix Ca, a’-disubstituted amino acids Amphipatbic helix, choice of a-forming residues, leucine zipper leucine in helical wheel Helix bundles template-assembled synthetic proteins Helix bundles choice of residues in active site and in a-helical domains, molecular modeling

Examples

bradykinin p-turn tuftsin pepstatin, renin HIV protease alamethicin

Biological Meaning

potent antagonists

Reference

enkephalins GnRH

peptailbol antibiotics potent agonists potent antagonists

Kyle et al., 1991 Kahn et al., 1988 Sueiras et al., 1990 Huff, 1991 Schmitt and Jung, 1985 Wilkes and Schiller, 1992 Rivier et al., 1990

gramicidin A dimers

transmembrane ion channels

Wallace, 1990

P-endorphin glucagon

potent analogs

Kaiser, 1987, DeGrado et al., 1989 Toniolo and Benedetti, I991 Landschultz et al., 1988

transition state analogs

alamethicin peptaibol antibiotics dimers:interdigitating leucine side DNA-binding peptides chains AChR fragments pentameric ion channels in lipid bilayers chymohelizyme- I

enzyme mimics

Mutter et al., 1992 Montal et al., 1990 Hahn et al., 1990

Peptides: Multiple Purpose Tools

5 73

CH,-S, CO-CH,, CH=CH, and NH-CO. Other modifications involve the a-carbon atom as in 2-amino-isobutyricacid (Aib), inD-amino acids or in a,@-dehydroamino acids. Although the local consequences of a given backbone change can be described, the overall effect on the peptide structure and on biological activity cannot generally be predicted. A common finding is the resulting increased resistance towards proteolytic degradation at the substitution site and not rarely in remote positions (review: Fauchere and Thurieau, 1992). Compared to side chain substitutions,backbone modifications more drastically alter the secondary structure. For example, when replacing the peptide bond by its reduced form CH,-NH [a very popular substitution that can be integrated easily in the usual repetitive process of solid phase synthesis (Sasaki and Coy, 1986)], one loses the planarity and rigidity of the peptide bond, the trans substitution on the bond, the hydrogen bond acceptor character of the CO group, and the amide neutrality. Successful examples of the use of peptide bond surrogates have been reported for renin inhibitors (pepstatin analogs) in which they mimic the transition state of aspartyl protease subtrates (Sueiras-Diaz et al., 1990). Examples of the C,-involving pseudopeptides are those in which Aib (2-amino-isobutyricacid) is known to stabilize @-turnsand to induce 3,,-helical conformation (a specific helix type encountered in several peptide antibiotics (Toniolo et al., 1991) or used to maximize a-helicity (Cervini et al., 1992). Cyclization has also become a routine procedure to decrease the conformational degrees of freedom in peptides and, in case of end-to-end cyclization, to protect the peptide against degradation by exopeptidases. The secondary structure of cyclic peptides is also more easily studied by spectroscopic methods and molecular modeling than their linear counterparts. The a-helix, P-pleated sheet and the 3 ,o-helixare very common structural motifs in proteins, which can be built into even medium-size peptides. The choice of the residues according to the refined Chou and Fasman rules (Fasman, 1989) and to the helical wheel analysis (Schiffer and Edmundson, 1967) enables the chemist to design the amphipathic helix, one of the most widely used chiral building blocks in large bioactive peptides. Kaiser (1987) was guided by these principles when he designed several middle-size peptides (e.g., calcitonin, glucagon, @-endorphin) with minimal sequence homology to the parent peptide but with the conformational features responsible for biological activity. Helix amphipathy is the key element for dimerization of both alternating D-L and all-L-residuehelices (DeGrado et al., 1989; Wallace, 1990).Helix amphipathy is also the case in the leucine zipper, a structure common to a class of DNA-binding proteins, in which the interdigitating leucine side chains on one side of the helix facilitate dimerization (Landschultz et al., 1988). The contribution of other structural elements to the formation of helix bundles has been investigated and used in a minimalist approach to protein design (DeGrado et al., 1989; cf. also review on a-helical coiled coils and bundles: Cohen and Parry, 1990). The four-helix bundle motif is encountered in natural channels, and reproduced in synthetic ion channels

5 74

JEAN-LUC FAUCH~RE

(synporins: Montal et al., 1990) and enzymes (synzymes: Hahn et al., 1990). Progress has also been reported towards the total synthesisof proteins, for example, HIV protease, a homodimer of two 99-residue chains (Kent et al., 1992) and of its all-D enantiomer, the latter showing reciprocal chiral specificity on peptide substrates (Milton et al., 1992). These impressive synthetic achievements are generally confirmed by strong spectroscopic evidenceand very often derived from the need to reproduce naturally occurring structures and modulate biological activity.

IV. PEPTIDE DRUG DESIGN Polypeptide hormones, neuropeptides, and protease inhibitors have many of the properties of an ideal drug. However, their short half-life and poor bioavailability have greatly impaired their use as drugs. Progress in recent years towards the preparation of pseudopeptides and peptide mimetics has improved the chances of obtaining useful drugs structurally related to the parent peptide. Elements for the rational design of peptide drugs have been reviewed (Fauchere, 1986; Fauchere, 1989). Taking into account a few recent developments, the main existing strategies are outlined in this section. Starting with the natural sequence as the lead, the designer of peptide agonists will generally go through a number of steps in order to find the pharmacophore (minimal sequence) and to restrict the conformational freedom of the analog. Peptide shortening and an alanine scan may help to locate the mandatory sequence elements. A Chou and Fasman analysis (Fasman, 1989)and a Aib scan (introduction of 2-amino-isobutyric acid successively in each position) will give clues regarding the secondary structure requirements. An early assessment of the scissile bonds in physiologically relevant proteolytic media will also locate the first bonds to be stabilized. Molecular modeling, although useful from the very beginning, will become more powefil as the number of analogs to be compared increases and as their size and conformational freedom is reduced. Classically, partial or total cyclization and the introduction of turns and peptide bond surrogates will be attempted, obviously resulting in pseudopeptide analogs in which the original peptide may no longer be recognized. A classical outcome of these strategies has been the discovery of shortened, potent, and selective agonists of the natural peptide. Well-known examples are pentagastrin (Morley et al., 1963, goserelin (Dutta et al., 1978), octreotide (Pless et al., 1986), and ebiratide (Wiemer et al., 1988). The design ofpeptide antagonistsmay follow a similar strategy, starting with the same agonist sequence and trying to modify it so as to lose bioactivity but retain affinity. Although the rationale is that agonist and antagonist will compete for the same site, there is no rule by which to convert a peptide agonist into an antagonist. Amazingly, substitution or deletion of a single residue in a precise position is often the key discovery to obtain antagonistic behavior: Ile-9 in corticotropin (ACTH,

Peptides: Multiple Purpose Tools

575

Kumar, 1975), Ala-8 in angiotensin I1 (Hall et al., 1974), D-Phe-7 in substance P (Yamaguchi et a1.,1979), D-Phe-7 in bradykinin (Vavrek and Stewart, 1985), des-His-2 in gonadotropin-releasing hormone (GnRH) (also called luteinizing hormone-releasing hormone (LHFW, Vale et al., 1972).Optimizationthen follows as for an agonist, mixed effects having to be eliminated. The rational design of an orally active dipeptoid antagonist of the cholecystokinin C-terminal octapeptide (CCKS) with central nervous system activity (Home11et al., 1992)and of an orally active dipeptide substance P antagonist FK888 (Fujii et al., 1992) have been reported recently. Another strategy, based on screening, has been shown to be very efficient for the discovery of peptide antagonists. The efficient and possibly automated radioreceptor binding assays available nowadays for most peptide hormones or neuropeptides permit a large series of compounds, such as the classified chemicals of a pharmaceutical company, to be screened within a short time. Alternatively, the source of binding material can be a bacterial or fungal broth from which the binding components can be isolated and characterized. These often scorned methods have already led to impressive results such as the discovery of several nonpeptide neurokinin antagonists (review: Watling, 1992). The list of peptide mimetics that antagonize the action of natural peptide agonists is rapidly increasing (Table 3). One major advantage of this strategy is that it identifiesnonpeptide structures,thus doing away with the drawbacks of the peptide as drug. It has to be stressed that these screening strategies only lead to the discovery of antagonists. An agonist apparently must be much better adjusted to the recognition

Table 3. Nonpeptide Ligands (Antagonists)of Peptide Hormone and Neuropeptide Receptors ~~~

~

Natural Peptide

Cholecystokinin ODN' Bradykinin Angiotensin 11 Neurokinins

Oxytocin Vasopressin GnRH (LHRH) Endothelin Note:

Nonpeptide Mimetic

asperlicin, L364718 tifluadom diazepam MV 8612 DuP 753 WL 19 (PD 121981) CP 96345 SR 489689 RP 67580 L 365209 OPC 21268 ketoconazole anthraquinones

Reference

Chang et al., 1985 Bock et al., 1990 Miyata et al., 1987 Calixto et al., 1988 Duncia et al., 1992 Blankley et al., 1991 Snider et al., 1991 Emonds-Alt et al., 1992 Garret et al., 1991 Pettibone, 1989 Yamamura et al., 199 1 De et al., 1989 Oohata et al., 1990

ODN = octadecaneuropeptide, trypsin fragment of diazepam-binding inhibitor from rat brain.

576

JEAN-LUC FAUCHERE

site in order to trigger the receptor, whereas for an antagonist only high binding affinity is required. New leads for the development of antagonists of bioactive peptide hormones and neuropeptides may be expected from screening of the synthetic libraries described above. However, the discovery of a new lead does not save the chemical optimization of potency, selectivity, and stability of the peptide structure. Inclusion of nonproteinogenic amino acids in the epitopes may however provide better starting points. The design of inhibitorsof proteases, of which peptides are the natural substrates, is also a matter of intense current research. Following the discovery of captopril (Cushman and Ondetti, 1980), two new generations of ACE inhibitors have been successively developed, represented by the peptidelike enalapril (Patchett et al., 1980) and derivatives (e.g., perindopril: Laubie et al., 1984), and the nonpeptide structures such as cilazapril (Attwood et al., 1984).Renin inhibitors have also been derived from peptide structures on the basis of transition state analogs of the substrate of aspartyl proteases and with the objective of obtaining oral activity, a goal that has been at least partially achieved (Greenlee, 1990; Kleinert et al., 1992). Intense efforts to discover potent inhibitors of HIV protease, another aspartyl protease, are underway that use pseudopeptides containing a transition state mimic of the scissile bond as lead compounds (Huff, 1991). Finally, specific inhibitors of a-thrombin derived from the C-terminal end of hirudin (Krstenansky et al., 1990), combined with an active site-directed tripeptide (Maraganore et al., 1990), are under clinical evaluation as anticoagulant agents. A number of peptide drugs are currently available (for a list see Fauchere and Thurieau, 1992, their Table 3) and a larger number is expected on the market in the near future (Fauchere and Thurieau, 1992;their Table 4). In most ofthese structures, stabilizationagainst proteolysis has been the major objective of the designer, as can be seen from the structure modification of the natural peptide sequence including end-group modification, partial cyclization, introduction of nonproteinogenic amino acids or D-residues or modifications ofthe backbone. Besides these chemical means, improvements of galenic devices and formulations(see, e.g., GnRH administration: Perren et al., 1986) will further widen the acceptance and clinical use of therapeutic peptides.

V. MULTIPLE USES OF PEPTIDES In the preceding section, the merits of the peptides as drugs were stressed, either in the unmodified form (salmon calcitonin) or as pseudopeptides (bradykinin antagonist Hoe 140; for the full structure ofthis decapeptide analog see Hock et al., 199l), or as peptide mimetics (nonpeptide neurokinin antagonists;for references see Table 3). These applications,to which the research and development ofprotease inhibitors such as angiotensin-converting enzyme, renin, or HIV-protease inhibitors must be added, certainly represent major achievements in the peptide field. However,

Peptides: Multiple Purpose Tools

577

Table 4. Peptides: Multiple Purpose Tools Application

Protease inhibitor Drug targeting Bireceptor ligand Prodrug Sweetener Tumor imaging Sequence-specific DNA-cleaving Ion channel

Example

s-calcitonin bradykinin substance P ACE HIV, renin a-melanotropin KDEL signal galantide tripeptide arm aspartame somatostatin Hin-recombinase fragment gramicidin A AChR mimic chymohelizyme

Structure

Reference

Azria, 1989 unmodified 32-residue Hock et al., 1991 pseudopeptide antagonist nonpeptide antagonists Watling, 1992 Lawton et al, I992 stabilized di-, tripeptides Rich, 1990; Huff, 1991 transition state analogs Varga, 1985 daunomycin conjugate Brinkmann et al., 1991 toxin peptide conjugate peptide-peptide conjugate Bartfai et al., 1992 Trouet et al., 1982 daunorubicinalbumin conjugate DuBois, 1991 dipeptide ester "In-octreotide conjugate Pless et al., 1992 EDTA-Fe-peptide complex Sluka et al., 1987

'

alternating D,L-residues four-helix bundle protein triad in four-helix bundle

Wallace, 1990 Grove et al., 1992 Hahn et al.. 1990

peptide engineering has produced a number of other useful pharmacological tools, some of which are presented in Table 4. Drug targeting using the biological address contained in polypeptide hormones was suggested long ago. Conjugates of a-MSH with daunomycin had a ten-times higher cytotoxicity index than daunomycin in murine melanoma cell cultures, and the cells could be protected from the toxicity of the conjugate by a large excess of a-melanotropin (Varga, 1985). Despite the imprecisely defined structure of the conjugate (probable daunomycida-MSH ratio 3:l) and the drop of its receptor affinityby two orders of magnitude compared to a-MSH, significantcell specificity was achieved in these peptide-drug conjugates. Tumor and metastasis imaging is another analogous application: a conjugate of octreotide with an "Indium-chelating agent recognized the somatostatin receptors on endocrine-related tumors, thus providing an efficient identification of the tumor and hopefilly a therapeutic benefit (Pless et al., 1992). There are indications that large proteins can be targeted to discrete intracellular locationsby small peptides such as Lys-Asp-Glu-Leu(KDEL) for the endoplasmic reticulum (Brinkmann et al., 1991) or Pro-Lys-Lys-Lys-ArgLys-Val (PKKKRKV) for the nucleus (Kalderon et al., 1984). It can be expected that other motifs will be discovered that, when attached by chemical modification or by genetic engineering, will guide the protein to an organelleor even to a receptor family, as is the case, for example, for the naturally occurring fragment Arg-GlyAsp (RGD) in the integrins (Maeda et al., 1989).Another interesting development 1; the synthesis of bireceptor-recognizingpeptides, which combine in their chimeric

578

JEAN-LUC FAUCHERE

structurethe membrane-bindingpart (address)with the receptor-triggeringsegment (message) of polypeptide hormones, in one of the possible permutations (Bartfai et al., 1992).These manipulations may change the receptor specificity,for example, from p- to &opiate sites in dermorphins and deltenkephalins (Sagan et al., 1991) and increase the affinity by linking the message to a better address (generally via a synthetic p-tum) or even produce bireceptor-recognizing peptides when the message of two hormones acting on the same tissue are connected. Hybrid peptides of galanine and substance P (galantide,Langel et al., 1992),of galanine and bradykinin (Wiesenfeld et al., 1992), and of vasointestinal peptide and neurotensin (Gozes et al., 1991) were constructed in this way. The concept is analogous to the covalent dimerization of the same hormone fragment, such as ACTH,,-2, (the address of corticotropin),in which cooperative phenomena are probably the cause of the high increase in affinity (Fauchkre et al., 1985). Chimeric bivalent ligands are likely to become the source of potent and subtype-specific antagonists in the early stage of peptide drug design where the mode of action of the natural agonist has to be elucidated and before nonpeptide antagonists can be found. Binary drugs in which only one ofthe partners is apeptide (fragment of substance P) and the second partner is a compound that produces the same final effect (analgesia by adenosine agonists) were also described (Jacobson et al., 1987). Finally, the peptide Ala-Leu-Ala-Leu was used as a serum-stable but lysosome-hydrolyzable link between daunorubicin and serum albumin, as required for lysomotropic drug-carrier conjugates (Trouet et al., 1982). A DNA-binding, 52-residue peptide was converted into a sequence-specific, DNA-cleaving peptide by chemical attachment of an iron chelator to its N-terminus (Sluka et al., 1987), thus opening the way to oncogene inactivating agents. As predicted early on and later confirmed by X-ray analysis (Wallace and Ravikumar, 1988), D,L-alternating peptides have the propensity to fold so as to form a channel with a polar cavity (Urry et al., 1983). Gramicidin A is a natural, alternating D,L-peptide that forms head-to-head dimers in bilayers. Gramicidin A is the prototype for the development of D,L-peptides with ion channel properties, since these compounds are now accessible by synthesis. The bundle of four a-helices is another scaffold that can be used for the construction of ion channels of variable specificity(Montal et al., 1990). The same helix bundle motif (to which the characteristic amino acid triad of serine proteases (Ser, His, Asp) was incorporated) was also involved in chymohelizyme, a synthetic enzyme with chymotrypsinlike catalytic activity (Hahn et al., 1990, Stewart et al., 1992). These examples demonstrate that both the predicted structure and functionality can be achieved in these complex bundles of polypeptide chains. Among the multiple uses of peptides, not shown in Table 3, one can think of synthetic peptide vaccines (see, e.g., Tam, 1988) the development of which is limited more by economic factors than by chemical feasibility.

Peptides: Multiple Purpose Tools

5 79

The amazing versatility of peptides as bioactive agents as well as molecular devices exceeds by far the examples given in this survey and promises a multitude of exciting new applications.

REFERENCES Attwood, M.R., Francis, R.J., Hassall, C.H., Krohn, A., Lawton, G., Natoff, I.L., Nixon, J.S., Redshaw, S., & Thomas, W.A. (1984). New potent inhibitors of angiotensin-converting enzyme. FEBS Lett. 165,20 1-206. Azria, M. (1989). In: The Calcitonins (Azria, M. Ed.), pp. 1-152. Karger, Basel. Bartfai, T., Fisone, G., & Langel, U. (1992). Galanin and galaninantagonists:molecular and biochemical perspectives. TIPS 13.312-317. Blankley, C.J., Hodges, J.C., Klutchko, S.R., Himmelsbach, R.J., Chucholowski, A., Connolly, C.J., Neersgard, S.J., Van Nieuwenhze, M.S., Sebastian, A., Quinn, J., Essenburg, A.D., & Cohen, D.M. (199 1). Synthesis and structure-activity relationships of a novel series of non-peptide angiotensin I1 receptor binding inhibitors specific for the AT, Subtype. J. Med. Chem. 34,3248-3260. Bock, M.G., Dipardo, R.M., Evans, B.E., Rittle, K.E., Whitter, W.L., Veber, D.F.,Freidinger, R.M., Chang, R.S., Chen, T.B., & Lotti, V.J. (1990). Cholecystokinin-A receptor ligands based on the k-opioid agonist tifluadom. J. Med. Chem. 33,45&455. Boutin, J.A. & Fauchere, J.L. (1996a). Second generation robotic synthesizer for peptide, pseudopeptide and non-peptide libraries. In: Proc. 13th Int. Symp. on Laboratory Automation and Robotics '95, Boston, 197-2 10, Zymark Corp., Hopkinton, MA. Boutin, J.A. & Fauchere, A.L. (1996b). Combinatorial peptide synthesis: Statistical evaluation of the peptide distribution. TIPS 17, 8-12. Boutin, J.A., Hennig, P., Lambert, P.H., Bertin, S., Petit, L., Mahie& J.P., Serkiz, B., Volland, J.P., & Fauchere, J.L. (1996). Combinatorial peptide libraries: robotic synthesis and analysis by NMR, MS, MS/MS and HPEC techniques. Anal. Biochem. 234, 126-141. Brinkmann, U., Pai, L.H., Fitzgerald, D.J., Willingham, M., & Pastan, I. (1991). B3(Fv)-PE38KDEL, a single-chain immunotoxin that causes complete regression of a human carcinoma in mice. Proc. Natl. Acad. Sci. USA 88,861643620. Calixto, J.B., Pizzolatti, M.G.. & Yunes, R.A. (I 988). The competitive antagonistic effect of compounds from Mandevilla velutina on kinin-induced contractions of rat uterus and guinea-pig ileum in vitm. Br. J. Pharmacol. 94. 1 1 3 3 1142. Cervini, L., Galyean, R., Donaldson, C.J., Yamamoto, G.. Koerber, S.C., Vale, W., & Rivier, J.E. (1992). In: Peptides, Chemistry and Biology (Smith, J.E. & Rivier, J.E. Eds.), pp. 437438. Escom, Leiden. Chang, R.S., Lotti, V.G., Monaghan, R.L., Bimbaum, J., Stapley, E.O., Goetz, M.A., Albers-Schonberg, G., Patchett, A.A., Liesch, J.M., Hensens, O.D., & Springer, J.P. (1985). A potent nonpeptide cholecystokinin antagonist selective for peripheral tissues isolated from Aspergillus alliaceus. Science 230, 177-179. Cohen, C. & Parry, D.A. (1990). a-helical coiled coils and bundles: How to design an a-helical protein. Proteins 7, 1-15. Cushman, D.W. & Ondetti, M.A. (1980). Inhibitors of angiotensin-converting enzyme. Prog. Med. Chem. 17,42-104. De, B., Plattner, J.J., Bush, E.N., Jae, H.S., Dim, G., Johnson, E.S., & Perun, T.J. (1989). LH-RH antagonists: Design and synthesis of a novel series of peptidomimetics. J. Med. Chem. 32, 2038-2041. DeGrado, W.F., Wasserman, Z.R., & Lear, J.D. (1989). Protein design, a minimalist approach. Science 243.622428.

5 80

JEAN-L

uc FAUCH ERE

Duncia, J.V., Carini, D.J., Chiu. A.T., Johnson, A.L., Price, W.A., Wong, P.C., Wexler, R.R., & Timmermans, B.M.W.M. (1992). The discovery of DuP 753, a potent, orally active nonpeptide angiotensin I1 receptor antagonist. Med. Res. Rev. 12, 149-191. DuBois, G.E. (1991). In: Sweeteners, Discovery, Molecular Design and Chemoreception (Walters. D.E., Orthoefer, F.T., & DuBois, G.E., Eds.), ACS Symposium 450, pp. 261-276. Dutta, A.S., Fun; B.J., Giles, M.B., & Valcaccia, B. (1978). Synthesis and biological activity of highly active a-am analogues of luliberin. I. Med. Chem. 21, 1018-1024. Emonds-Ah, X., Vilain, P., Goulaouic, P., Proietto, V., Van Broek, D.. Advenier, C., Naline, E.. Neliat, G., Le Fur, G., Breliere, J.C. (1992). A potent and selective non-peptide antagonist of the neurokinin A ( N U ) receptor. Life Sci. 50, 101-106. Fasman, G.D. (1989). Protein conformational prediction. TIBS 14,295-299. Fauchere, J.L., Rossier, M., Capponi, A., & Vallotton, M.B. (1985). Potentiation of the antagonistic effect of ACTH, on steroidogenesis by synthesis of covalent dimeric conjugates. FEBS Lett. 183,28>286. Fauchere, J.L. (1986). Elements for the rational design of peptide drugs. Adv. Drug Res. 15-29-69. Fauchere, J.L. (1989). Towards the rational design of peptide drugs. Actual. Chim. Ther. 16, 5 5 7 2 . Fauchere, J.L. & Thurieau, C. (1992). Evaluation of the stability of peptides and pseudopeptides as a tool in peptide drug design. Adv. Drug Res. 23, 127-159. Fodor, S., Read, J., Pirmng, M., Stryer. L., Lu, A,, & Solas, D. (1991). Light-directed, spatially adressable parallel chemical synthesis. Science 25 1. 767-773. Fujii, T., Murai, M., Morimoto, H., Maeda, Y., Yamaoka, M., Hagiwara, D., Miyake, H.. Ikari, N., & Matsuo, M. (1992). Pharmacological profile o f a high afinity dipeptide NKI receptor antagonist, FK888. Br. J. Pharmacol. 107,785-789. Garret, C., Carmette, A,, Fardin, V., Moussaoui, S., Peyronel, J.F., Blanchard, J.C.. & Laduron, P.M. (1991). Pharmacological properties of a potent and selective nonpeptide substance P antagonist. Proc. Natl. Acad. Sci. USA 88, 10208-10212. Geysen H.M., Rodda, S.J., Mason, T.J., Tribbick, G., & Schoofs, P.G. (1987). Strategies for epitope analysis using peptide synthesis. J. Immunol. Meth. 102,259-274. Gozes, Y., Brenneman, D.E.,Fridkin,M., Asofsky. R., &Gozes, I. (1991). AVIPantagonistdistinguishes VIP receptors on spinal cord cells and lymphocytes. Brain Res. 540, 3 19-321. Greenlee, W.J. (1990). Renin inhibitors. Med. Res. Rev. 10, 173-236. Grove, A,, Tomich, J.M., Iwamoto, T., Marrer, S., Montal, M.S.. & Montal, M. (1992). In: Peptides, Chemistry and Biology (Smith, J.A. & Rivier. J.E., Eds.). ESCOM, Leiden, pp. 329-331. Hahn, K.W., Klis, W.A., & Stewart, J.M. (1990). Design and synthesis of a peptide having chymotrypsin-like esterase activity. Science 248, 1544-1 547. Hall, M.M., Khosla, M.C., Khairallah, P.A., & Bumpus, EM. (1974). Angiotensin analogs: The influence of sarcosine substituted in position 1. J. Pharmacol. Exp. Ther. 188,222-228. Hock, F.J., Wirth, K., Albus. U., Linz, W., Gerhards, H.J., Wiemer, G., Henke, S., Breipohl, G., Konig, W., Knolle, J., & Scholkens, B.A. (1991). Hoe 140, a new potent and long acting bradykinin-antagonist: in vitm studies. Br. J. Pharmacol. 102, 769-777. Honvell, D.C., Pritchard, M.C., Richardson, R.S., & Roberts, E. (1992). In: Peptides, Chemistry and Biology (Smith, J.A. & Rivier, J.E., Eds.), pp. 779-780. ESCOM, Leiden. Houghten, R., Pinilla, C., Blondelle, S., Appel, J., Dooley, C., & Cuervo, J. (1991). Generation and use of synthetic peptide combinatorial libraries for basic research and drug discovery. Nature 354, 84-86. Huff,J.R. (1991). HIV protease: A novel chemotherapeutic target for AIDS. J. Med. Chem. 34, 2305-2314. Huffinan, W.F., Callahan, J.F., Eggleston, D.S., Newlandeer, K.A., Takata, D.T., Codd, E.E., Walker, R.F., Schiller, P.W., Lemiew, C., Wire, W.S., & Burks, T.F. (1988). In: Peptides, Chemistry and Biology (Marshall, G.R., Ed.), pp. 105-108. Escom, Leiden.

Peptides: Multiple Purpose Tools

581

Jacobson, K.A., Lipkowski, A.W., Moody, T.W., Padgett, W., Pijl, E., Kirk, K.L., & Daly, J.W. (1987). Binary drugs: Conjugates of purines and a peptide that bind to both adenosine and substance P receptors. J. Med. Chem. 30, 1529-1532. Kahn, M., Wilke, S., Chen, B., Fujita, K., Lee, Y.H., & Johnson, M.E. (1988). Thedesignandsynthesis of mimetics of peptide p-turns. J. Mol. Recognit. 1,75-79. Kaiser, E.T. (1987). Design principles in the construction of biologically active peptides. TIBS 12, 305-309. Kalderon, D., Roberts, B.L., Richardson, W.D., & Smith, A.E. (1984). Ashort amino acid sequence able to specify nuclear location. Cell 39,499-509. Kent, S.B.H., Alewood, D., Alewood, P., Baca, M., Jones, A., & Schnolzer, M. 1992. In: Solid Phase Synthesis (1992) (Epton, R., Ed.), pp. 1-22. Intercept Ltd., Andover. Kleinert, H.D., Rosenberg, S.H., Baker, W.R., Stein, H.H., Klinghofer, V., Barlow, J., Spina, K., Polakowski, J., Kovar, P., Cohen, J., & Denissen, J. (1992). Discovery of a peptide-based renin inhibitor with oral bioavailability and efficacy. Science 257, 1940-1943. Krstenansky, J.L., Broersma, R.J., Owen, T.J., Payne. M.H. Yates, M.T., & Mao. S.J.T. (1990). Development of MDL 28050, a small stable antithrombin agent based on a fi~nctionaldomain of the leech protein hirudin. Thromb. Haemost. 63,20%214. Kumar, S. (1975). Ile9-ACTH,,,, a competitive antagonist of ACTHl-24-induced cyclic AMP and corticosterone production. Biochem. Biophys. Res. Commun. 66, 1063-1068. Kyle, D.J., Green, L.M., Blake, P.R., Smithwick, D., & Summers, M.F. (1992). A novel p-turn mimic usefbl for mapping the unknown topology of peptide receptors. Peptide Res. 5,206-209. Lam, K., Salmon, S., Hersh, E., Hruby, V., Kazmiersky, W., & Knapp, R. (1991). Anew type ofsynthetic peptide library for identifying ligand-binding activity. Nature 354, 82-84. Landschultz, W.H., Johnson, P.F., & McKnight, S.L. (1988). The leucine zipper: A hypothetical structure common to a new class of DNA binding proteins. Science 240, 1759-1764. Langel, U., Land, T., & Bartfai, T. (1992). Design of chimeric peptide ligands to galanin receptors and substance P receptors. Int. J. Peptide Protein Res. 39,5 16-522. Laubie, M, Schiavi, P., Vincent, M., & Schmitt, H. (1984). Inhibition ofangiotensin I-converting enzyme with S 9490: Biochemical effects, interspecies differencesand role of sodium diet in hemodynamic effects. J. Cardiovasc. Pharmacol. 6, 1076-1082. Lawton, G., Paciorek, P.M., & Waterfall, J.F. (1992). The design andbiological profile ofACE inhibitors. Adv. Drug Res. 23, 161-220. Maeda, T., Oyama, R., Ichiara, K., Kimizuka, F., Kato, I., Titani, & Sekiguchi, K. (1989). A novel cell adhesive protein engineered by insertion of the Arg-Gly-AspSer tetrapeptide. 3. Biol. Chem. 264, 15165-15 168. Maraganore, J.M., Bourdon, P., Jablonski, J., Ramachandran. K.L., & Fenton. J.W. (1990). Design and characterization ofhirulogs: Anovel class ofbivalent peptide inhibitors ofthrombin. Biochemistry 29,7095-7101. Memfield, R.B. (1963). Solid phase peptide synthesis I. The synthesis of a tetrapeptide. J. Am. Chem. SOC.85,214%2154. Milton, R.C.Del, Milton, S.C.F., & Kent, S.B.H. (1992). Total synthesis of a D-Enzyme: The enantiomers of HIV-1 protease show demonstration of reciprocal chiral substrate specificity. Science 256, 1445-1448. Miyata, M., Mocchetti, I., Ferrarese, C., Guidotti, A., & Costa, E. (1987). Protracted treatment with diazepam increases the turnover of putative endogenous ligands for the benzodiazepine/P-carboline recognition site. Proc. Natl. Acad. Sci. USA 84, 1444-1448. Montal, M., Montal, M.S., & Tomich, J.M. (1990). Synporinssynthetic proteins that emulate the pore structure of biological ionic channels. Proc. Natl. Acad. Sci. USA 87,6929-6933. Morley, J.S., Tracy, H.J., & Gregory, R.A. (1965). Structure-function relationships in the active C-terminal tetrapeptide sequence of gastrin. Nature 207, 1356-1 359.

582

JEAN-L UC FA UCH ERE

Mutter, M., Carey, R.I., Domer, B., Emest, I., Flogel, R., Giezendanner, U., Rivier, J.E., Servis, C., Sigel, C., Steiner, V., Tuscherer, G., Vuilleumier, S., & Wyss, D. (1992). In: Peptides, Chemistry and Biology (Smith, J.A. & Rivier, J.E., Eds.), pp. 326328. ESCOM, Leiden. Oohata, N., Nishikawa, M., & Kiyoto, S. (1991). European Patent EP 0405421A. Patchett, A.A. et al. (1980). A new class of angiotensin-converting enzyme inhibitors. Nature 288, 280-284. Perren, T.J., Clayton, R.N., Blackledge, G., Bayley, L.C., Holder, G., Lynch, S.S., Arkell, D., Cottam, G., Famr, D., & Young, C.H. (1986). Pharmacokinetic and endocrinological parameters of a slow-release depot preparation of the GnRH analogue ICI 118630 (Zoladex) in patients with prostatic cancer. Cancer Chemother. Pharmacol. 1 8 , 3 9 4 3 . Pettibone, D.J., Clineschmidt, B.V., Anderson, P.S., Freidinger, R.M., Lundell, G.F., Koupal, L.R., Schwartz, C.D., Williamson, J.M., Goetz, M.A., Hensens, Liesch. O.D., & Springer, J.P. (1989). A structurally unique, potent and selective oxytocin antagonist derived from Strepfomyces silvensis. Endocrinology 125, 217-222. Pless, J., Albert, L., Bruns, C., Marbach, P., Stolz, B., Bakker, W.H., Krenning, E.P., & Lamberts, S.W. (1992). In: Peptides, Chemistry and Biology (Smith, J.A. & Rivier, J.E., Eds.), pp. 106108. ESCOM, Leiden. Pless, J., Bauer, W., Briner, U., Doepfner, W., Marbach, P., Maurer, R., Petcher, T.J., Reubi, J.C., & Vonderscher, J. (1986). Chemistry and pharmacology of SMS 201 995, a long-acting octapeptide analogue of somatostatin. Scand. J. Gastroenterol. 21 (Suppl. 119), 54-64. Rich, D.H. (1990). In: Comprehensive Medicinal Chemistry (Sammes, P.G., Ed.), Vol. 2, pp. 391441. Pergamon Press, Oxford. Rivier, J.E., Rivier, C., Vale, W., Koerber, S.C., Corrigan, A,, Porter, J., Gierasch, L.M.. & Hagler. A. (1990). In: Peptides, Chemistry, Structure and Biology (Rivier, J.E. & Marshall, G.R.. Eds.), pp. 3 S 3 7 . Escom, Leiden. Rose, G.D., Gierasch, L.M., & Smith, J.A. (1985). Turns in peptides and proteins. Adv. Protein Chem. 37, 1-109. Sagan, S., Amiche, M., Delfour, A,, Camus, A,, Mor, A,, & Nicolas, P. (1989). Differential contribution of the C-terminal regions of dermorphin and detmenkephalin to opioid sites selection and binding potency. Biochem. Biophys. Res. Commun. 163,726732. Sasaki, Y. & Coy, D.H. (1986). Solid phase synthesis of peptides containing the CH,NH peptide bond isostere. Peptides 8, 119-121. Schiffer, M. & Edmundson, A.B. (1967). Use of helical wheels to represent the structures of proteins and to identify segments with helical potential. Biophys. J. 7, 121-135. Schmitt, H. & Jung, G. (1985). Total synthesis of the a-Helical eicosapeptide antibiotic alamethicin. Liebigs Ann. Chemie 321-344. Sluka, J.P., Horvath. S.J., Bruist, M.F., Simon, M.I., & Dervan, P.B. (1987). Synthesisofsequence-specific DNA-cleaving peptide. Science 238, 1129-1 132. Smith, G.P. & Scott, J.K. (1992). In: Peptides, Chemistry and Biology (Smith, J.A. & Rivier, J.E., Eds.), pp. 48-88, ESCOM, Leiden. Snider, R.M., Constantine. J.W., Lowe 111, J.A.. Longo, K.P., Lebel, W.S., Woody, H.A., Drozda, S.E, Desai, M.C., Vinick, F.J., Spencer, R.W., & Hess, H.J. (1991). APotent nonpeptide antagonist of the substance P (NKl) receptor. Science 25 1 , 4 3 M 3 7 . Spatola, A.F. (1983). Peptide backbone modifications: A structure-activity analysis of peptides containing amide bonds surrogates. In: Chemistry and Biochemistry of Amino Acids, Peptides and Proteins (B. Weinstein, Ed.), Vol. 7, pp. 267-357. Dekker, New York. Stewart, J.M., Cann, J.R., Corey, M.J., Hahn, K.W.,&Klis, W.A. (1992). In: Solidphase Synthesis 1992 (Epton, R., Ed.), pp. 23-27. Intercept Ltd., Andover. Sueiras-Diaz, J., Jones, D.M., Svensson, L., Deinum, J., & Szelke, M. (1991). In: Peptides 1990 (Giralt, E. & Andreu, D., Eds.), pp. 378-379. Escom, Leiden.

Peptides: Multiple Purpose Tools

583

Tam, J.P. (1988). Synthetic peptide vaccine design: Synthesis and properties of a high-density multiple antigenic peptide system. Proc. Natl. Acad. Sci. USA 85,540%5413. Toniolo, C. & Benedetti, E. (1991). The polypeptide 310-helix.TIBS 16,356353. Trouet, A., Masquelier, M., Baurain, R., & Deperez-De-Campaneere, D. (1982). A covalent linkage between daunorubicin and proteins that is stable in serum and reversible by lysosomal hydrolases, as required for a lysosomotropic drug-carrier conjugate. Proc. Natl. Acad. Sci. USA 79,626629. Uny, D.W., Trapane, T.L., & Prasad, K.U. (1983). Is the gramicidin A a transmembrane channel single-stranded or double-stranded? A simple unequivocal determination. Science 2 I, I O W I 067. Vale, W., Grant, G., Rivier, J., Monahan, M., Amoss. M., Blackwell, R., Burgus, G., & Guillemin, R. (1972). Synthetic polypeptide antagonists of the hypothalamic luteinizing hormone releasing factor. Science 176,933-934. Varga, J.M., (1985). Hormone-drug conjugates. Methods Enzymol. 112,25!&269. Vavrek, R.J. & Stewart, J.M. (1985). Competitive Antagonists of Bradykinin. Peptides 6, 161-164. Wallace, B.A. (1990). Gramicidinchannelsandpores. Ann. Rev. Biophys. Biophys. Chem. 19,127-157. Wallace, B.A. & Ravikumar, K. (1988). The gramicidin pore: Crystal sbucture of a cesium complex. Science 241, 182-187. Watling, K.J. (1992). Nonpeptide antagonists herald new era in tachykinin research. TiPS 13,266269, Wiemer, G., Gerhards, H.J., Hock, F.J.. Usinger, P., von Rechenberg, W.. & Geiger. R. (1988). Neurochemical effects of the synthetic ACTH- analog Hoe 427 (ebiratide) in rat brain. Peptides 9, 1081-1087. Wiesenfeld-Hallin, Z., Xu, X.J., Langel, O., Bedecs, K., Hokfelt, T., & Bartfai, T. (1992). Galanin-mediated control ofpain: Enhanced role after nerve injury. Proc. Natl. Acad. Sci. USA89,3334-3337. Wilkes, B.C. & Schiller, P.W. (1992). In: Peptides, Chemistry and Biology (Smith, J.A. & Rivier, J.E., Eds.), pp. 255-256. ESCOM, Leiden. Yamaguchi, I., Rackur, G., Leban, J., Bjorkroth, U., Rosell, S., & Folkers, K. (1979). Synthesis and biological activity of analogs of substance, modified for conformational information by D-amino acids. Acta Chem. Scand. B33,63-68. Yamamura, Y., Ogawa, H., Chihara, T., Kondo, K., Onogawa, T., Nakamura, S., Mori, T., Tominaga, M., & Yabuuchi, Y. (1991). OPC-21268, an orally effective nonpeptide vasopressin V1 receptor antagonist. Science 252, 572-574. Zuckermann, R.N., Kerr, J.M., Siani, M.A., Banville, S.C., & Santi, D.V. (1992). Identification of highest-affinity ligands by affinity selection from equimolar peptide mixtures generated by robotic synthesis. Proc. Natl. Acad. Sci. USA 89.45054509.

This Page Intentionally Left Blank

SUPERPOROUS AGAROSE-A NEW MATERIAL FOR CHROMATOGRAPHY

Per-Olof Larsson

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 585 I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 586 588 11. THEORETICAL CONSIDERATIONS . . . . . . . . . . . . . . . . . . . . . 588 A. Choice of Superpore Diameter . . . . . . . . . . . . . . . . . . . . . . . B. Expectations of Superporous Particles . . . . . . . . . . . . . . . . . . . 588 111. PREPARATIONOFSUPERPOROUSAGAROSEPARTICLES . . . . . . . 589 IV. DIRECT OBSERVATION OF PORE FLOW . . . . . . . . . . . . . . . . . .590 V. CHROMATOGRAPHIC EXPERIMENTS . . . . . . . . . . . . . . . . . . . 590 592 VI. CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ABSTRACT Superporousparticles contain two sets ofpores, normal difision pores and very large flow pores. If such particles are used chromatographically,part of the flow will pass through each particle. This leads to a rapid equilibrationbetween the stationary and the mobile phases, even if the superporous particle has a large outer diameter.

Advances in Molecular and Cell Biology Volume 15B,pages 58S592. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

585

586

PER-OLOF LARSSON

Superporous particles are thus chromatographically efficient and at the same time give a low flow resistance, a combination of advantageous properties not found with normal chromatographic supports.

1. INTRODUCTION Chromatographic methods are important in most protein purification schemes, particularly ion exchange chromatography, affinity chromatography, and size exclusion chromatography.For several decades most chromatography packing materials intended for the separation of biomolecules have been polysaccharide based, and for good reason. These materials, for example, cross-linked dextran and agarose, are commercially available in suitable porosities, are generally inert towards proteins, and are stable towards hydrolysis (Jansson and Kristiansson, 1990). Furthermore, they withstand strongly alkaline solutions, which is a cheap and very efficient all-purpose sanitizing agent in industrial operations.Polysaccharide packings are also easy to derivatize with groups that will improve their selectivity, for example, ion exchange groups and bioaffinity ligands. The main draw-back of the polysaccharide supports are their limited mechanical strength. For standard packings (particle diameter 0.1-0.2 mm) the mechanical stability is hardly a problem. Such packings should be run at a low flow rate, which gives little mechanical strain, to allow for the slow diffusion in the large particles. If HPLC-type separations are attempted, however, to achieve improved chromatographic efficiency and thereby improved throughput, much smaller particles are required. The resulting high flow resistance will cause agarose particle beds to collapse. Mechanically strong particles, especially of silica or polystyrene, have therefore become choice materials for HPLC separations. Silica and polystyrene in their native form are hardly suitable for protein separations, but a number of chemistries have been developed to confer on them suitable surface properties (Unger, 1979; Larsson et al., 1983; Chicz et al., 1986; Jansson and Kristiansen, 1990). Still, the new materials are chemically not as resistant nor as robust as the polysaccharide materials. In this communication I describe new support material that combines desirable properties of traditional agarose supports and HPLC-type materials (Larsson, 1992).The new support is prepared from agarose and has a fairly large particle size. Significantly, the support material contains two sets of pores, normal diffision pores characteristic of all agarose materials and very wide, so-called superpores or flow pores. A chromatographed substance will be transported byflow in these superpores to the interior of each individual particle, leaving only short distances to be covered by slow diffusion processes (Figure 1). In this way the new particles will be chromatographically as efficient as several times smaller homogeneous particles without suffering from the drawback of high back pressure. Figure 2

Superporous Agarose

587

arose with normal diffusion pores (about 300 8, diameter)

Superpore (flow pore) diameter = 1/10 of particle diameter

Figure 1. The flow pattern with superporous particles. Chromatographed substances are transported by flow to the interior of the superporous particle leaving only short distances to be covered by slow diffusion.

compares schematically the performance of three columns packed with small standard particles (A), large standard particles (B), and large superporous particles (C). The principle of pore flow and its profound influence on chromatographic efficiency (perfision chromatography) has been amply demonstrated (Afeyan et al., 1990), primarily for polystyrene-based supports. A range of polystyrene-based materials is now also commercially available. In contrast to the materials described here, the marketed separation materials have a very small particle diameter (below 20 pm).

Small, standard Large, standard Large, superporous particles particles particles - High back-pressure - Low back-pressure - Low back-pressure - Short diffusion - Long diffusion - Short diffusion - High efficiency - Low efficiency - High efficiency

Figure 2. Comparison of small and large standard particles with large superporous particles.

PER-OLOF LARSSON

5 88

The development of superporous agarose in our laboratory was initiated several years ago as an extension of the work with macroporous gelatin supports for animal cell cultures (Nilsson et al., 1986). These supports possess very large pores and cavities able to accommodate animal cells, thereby increasing the effective surface of the particles. It was then suggested by Professor Jan-Christer Jansson that this type of porous particle could be made of agarose and that manufacturing conditions should be developed to allow the formation of interconnected pores through which pore flow might occur. Such a pore flow was thought to give superior chromatographic performance, an anticipation now proven valid.

I I. THEORETICAL CONS1DERATlONS A. Choice of Superpore Diameter

Figure 1 illustratesa chromatographicparticle with very large flow pores. If such a particle is used in a chromatographic bed (Figure 2), it is easily understood that part of the chromatographic flow will actually be diverted through the particle, a situation that could have great implications for mass transport. A critical parameter for a satisfactory pore flow is obviously the diameter of the flow pores. A simple calculation may give approximate guidance when determining the proper diameter. Consider a chromatographic bed packed with superporous particles (Figure 2). The channels between the particles may, as a first approximation,be considered to be hydraulically similar to the flow pores inside the particle. The diameter of the channels between the particles is dependent on particle shape and packing pattern and could be approximated to be one-third of the particle diameter (Coulson and Richardson, 1991). Thus, when the flow pores inside the particles have the same diameter, for example one-third ofthe particle diameter, the linear flow rate should be the same inside and outside the particles-an ideal situation. In practice, a flow pore diameter of one-tenth of the particle diameter and larger should give a satisfactory flow distribution (Gustavsson and Larsson, 1996).

B. Expectations of Superporous Particles Figure 2a shows a standard separation particle with normal difision pores and Figure 2c shows a superporous particle also containing flow pores. A question of overriding interest is, of course, how big an improvement can be expected of a support material with properly designed flow pores. A qualitative assessment may be presented as follows. In many realistic separation situations the performance of a chromatographic support is limited by the diffision through the matrix. In such cases the performance of a superporous particle could be similar to that of a standard particle with a diameter equal to the distance between the flow pores (but it couldnever be better!). For example, a bed packed with 150 pm diameter superporous particles having 30

589

Superporous Agarose

pm flow pores and a 30 pm distance between the superpores could, under certain conditions, be expected to behave as a bed packed with 30 pm standard particles. Superporous particles could thus be expected to behave comparably to standard particles one-fifth the size. The bonus of having a bed packed with 150 pm superporous particles instead of 30 pm standard particles is that the generated pressure drop is reduced by a factor of 25, as the pressure drop is inversely proportional to the square of the particle diameter (Coulson and Richardson, 1991). Such a reduction in pressure could be a most significant aspect, especially in large-scale applications.

111. PREPARATION OF SUPERPOROUS AGAROSE PARTICLES The preparation of superporousagarose particles was describedby Larsson (1992). Figure 3 provides a schematicoverview.Warm aqueous agarose solution was stirred with an organic solvent containing a surfactant stabilizing oil-in-water emulsion. The formed emulsion (emulsion 1) was then transferred to a second vessel with a stirred, warm organic solvent containing a water-in-oil stabilizing surfactant. Droplets of emulsion 1 were then formed. At this point the organic solvent was cooled to room temperature causing the emulsion 1 droplets to solidify. The superporous particles were isolated on a sieve and washed with water, ethanolwater, and water. The particles were wet classified using graded metal screens. The diameter of the flow pores were controlled by adjusting the composition of emulsion 1 and the stirring speed. The particle size was controlled mainly by the stirring speed in the second vessel.

0.03 m m

SUPERPOROUS AGAROSE PARIICLE

Figure 3. Preparation of superporous agarose particles.

PER-OLOF LARSSON

590

Microscope Pipette

Teflon tubing 0.25 mm inner diameter

To vacuum

Figure 4. Direct observation of pore flow. The bead had a diameter of 0.30 mm. The pipette positioned with a micromanipulator delivered water, dye solutions or suspensions of 0.5 pm latex particles.

IV. DIRECT OBSERVATION OF PORE FLOW In the early phase of development of superporous agarose beads, the presence of functioning flow pores was very much in doubt. Convincing proof of pore flow was looked for and found using the experimental apparatus depicted in Figure 4. The figure shows a 0.35 mm superporous bead that was sucked into an HPLC-type teflon tubing with 0.25 mm internal diameter. The bead was observed under the microscope. By using a micromanipulator, a thin glass capillary was placed on selected places on the bead where it delivered water that was readily sucked into the pores by the applied vacuum. To further visualize the pore flow principle, the pipette was allowed to deliver dye solutions and solutions containing suspended latex particles (0.5 pm) that were easily transported through the gel bead.

V. CHROMATOGRAPHIC EXPERIMENTS A number of size-exclusion chromatography experiments were carried out to characterize the properties of the superporous beads. Chromatographic data (retention times and peak widths) were converted into HETP values. In all cases reference runs were carried out with homogeneous agarose beads of the same diameter. Size-exclusion chromatography experiments were chosen as they offered a very clean representation of pore flow effects. In corresponding ion exchange experiments, for example,the results might have been obscured or enhanced by adsorption kinetics. The results obtained were generally in line with what could be expected. Thus, very high molecular weight compounds gave essentially the same HETP value regardless of the flow rate through the column (data not shown). Completely excluded substances do not need to diffuse into the support material and any band broadening should therefore not occur at increased flow rates. Salts and proteins on the other hand will diffuse into the particle matrix and the HETP value should be dependent on the flow rate. Figure 5 shows the results with the salt sodium azide and the protein bovine serum albumin. Figure 5a (sodium

591

Superporous Agarose

1 IStand particles

I

1

0

1

2

3

0

1

2

Linear flow (cm/min.) Figure 5. Chromatographic comparison of standard agarose particles and superporous particles at different flow rates. The particles had a diameter of 0.40 mm. The plate height (HETP) calculations were based on retention time, peak width at half height and column length (Snyder and Kirkland, 1979). (A): data for sodium azide (low molecular weight) and (B): data for bovine serum albumin (high molecular weight).

azide) compares the HETP values for homogeneous and superporousparticles as a function of the flow rate. The difference between the two types of particles is only moderate at low flow rates. Interestingly, the so-called reduced plate (Snyder and Kirkland, 1979) for the superporous particle is only about one at the lowest flow rates, a value half the size theoretically possible. This clearly indicates that the superporous particles behave as much smaller particles than their outer geometry would suggest. At high flow rates the superporous particles show their superiority. Figure 5a shows that they retain rather low HETP values, while the homogeneous particles now yield high values, indicating that the long diffusion distances within the homogeneous particles become limiting, and hence, diffusion cannot keep up with the flow outside the particles. The most convincing performance with superporous gels was, as expected, achieved when chromatographingproteins (Figure 5b). A protein such as bovine serum albumin (MW = 68,000) diffuses more slowly (diffusion coefficient = 6 x 1U7cm2s-') than salts, and because its diameter is comparable to the diameter of the diffusion pores (it is partly excluded from the gel), its effective diffusion coefficient will be even lower. The diffusion-controlled domain will therefore be entered already at low flow rates. Under such conditions the superpores should be especially valuable since they diminish the diffusion distances. Figure 5b shows this very clearly. The plate height for the superporousparticles increased only moderately at increased flow rate, whereas it very soon reached very high values for the homogeneous material. At flow velocities above 0.5 cm m i d , a direct breakthrough

PER-OLOF LARSSON

592

was observed with the homogeneous particles making HETP-calculations meaningless.

VI. CONCLUSIONS The new type of agarose particles described here is characterized by two sets of pores: (1) normal diffusion pores characteristic of all agarose materials. And (2) very wide, so-called superpores that allow part of the chromatographic flow to pass through each individual particle. The pore flow gives improved mass transfer, especially in situations where diffusion is the limiting factor for the overall performance. The benefits of superporosity were therefore most clearly demonstrated in cases involving: (1) large particles. (2) high flow rates and (3) slow-diffusing substances such as proteins. Superporous agarose beads have also been derivatized with ion exchange groups and (bio)affinity ligands. The resulting adsorbents have proven to be efficient tools in protein purification (Gustavsson and Larsson, 1996).

ACKNOWLEDGMENT Financial support from The Swedish National Board for Technical Development (NUTEK) is gratefully acknowledged. I also wish to express m y sincere gratitude t o Klaus Mosbach for his suggestion of the present area of research, for his continued support, and not least for a number of good and shared laughs.

REFERENCES Afeyan, N.B., Gordon, N.F., Mazsaroff, I., Varady, L., Fulton, S.P., Yang, Y.B., & Regnier, F.E. (1990). Flow-through particles for the high-performance liquid chromatographic separation of biomolecules: Perfision chromatography. J. Chromatogr. 5 19, 1-29. Chicz, R.M., Shi, Z., & Regnier, F. (1986). Preparation and evaluation of inorganic anion-exchange sorbents not based on silica. J. Chromatogr. 359, 121-130. Coulson, J.M., Richardson, J.F., Backhurst, J.R., & Harker, J.H. (1991). Chemical Engineering, Volume 2,4th edn. Pergamon Press, Oxford. Gustavsson, P.-E. & Larsson, P.-0. (l996b). Superporous agarose, a new material for chromatography. J. Chromatogr. A., in press. Janson, J.-C. & Kristiansen, T. (1990). Packings in affinity chromatography. Chromatogr. Sci. 47, 747-78 1. Larsson, P.-0. (1992). Superporous polysaccharidegels. Patent application SE 9200827 (Sweden). Larsson, P.-O., Glad, M., Hansson, L., MAnsson, M.-O., Ohlson, S., & Mosbach, K. (1983). High performance liquid affinity chromatography. Adv. Chromatogr., 2 1 , 4 4 5 . Nilsson, K., Buzsaky, F., & Mosbach, K. (1986). Growth of anchorage-dependent cells on macroporous microcarriers. Bio/technology 4, 989-990. Snyder, L.R. & Kirkland, J.J. (1979). Introductionto Modem Liquid Chromatography, 2nd edn, Wiley, New York. Unger, K.K. (1979). Porous silica. J. Chromatogr. Library, Vol. 16, Elsevier, New York.

HYDROPHILIC A N D AMPHIPHATIC MONOMERS A N D USE OF THEIR GELS AS SEPARATION MEDIA

Branko KozuI ic and Urs Heimgartner

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593 I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594 11. RESULTS AND DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . 595 A. Poly(NAT) Gels. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595 B. Hydrophilic Gels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598 C. Characte;ization of Hydrophilic Gels . . . . . . . . . . . . . . . . . . . 600 D. Amphiphatic Gels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 601

ABSTRACT In this contribution we describe some results of the work carried out by Professor Klaus Mosbach’s group at the Swiss Federal Institute of Technology and later at Elchrom. Synthesis of new acrylic monomers was required for the protein imprinting project and, after the finding that N-acryloyl-tris(hydroxymethy1)aminomethane gels

Advances in Molecular and Cell Biology Volume 15B, pages 593-604. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

593

594

BRANKO KOZULIC and URS HEIMCARTNER

offer advantages for electrophoresis, several series of monomers were synthesized and used to prepare separation media. Particularly useful were hydrophilic monomers based on amino sugar alcohols and corresponding amphiphatic monomers that enable separation of proteins by hydrophobic interaction electrophoresis.

1. INTRODUCTION In 1984 when we joined Professor Mosbach’s group at the Swiss Federal Institute of Technology (ETH), which included B. Kozulic as a post-doc and U. Heimgartner as a Ph.D. student, it was important to define research projects that would fall within the interest area of Professor Mosbach but different from those already underway in his group at the University of Lund. Given the extraordinary broadness of Professor Mosbach’s scientific interests, we needed several brain-storming sessions to choose two projects, defined as applied glycoprotein biochemistry and imprinting of proteins. As is often the case, some groundwork was already done and the group had experience working with glycoproteins (Barbaric et al., 1984; Kozulic et al., 1984) and imprinting of proteins (Glad et al., 1985). In the glycoprotein project the initial emphasis was on improvement of methods for the detection of glycoproteins. The results of these studies constitute the major part of U. Heimgartner’s Ph.D. thesis. Thus, we found that new polyacrylic polyhydrazides represent an excellent reagent for the detection of glycoproteins bound to a solid support (Heimgartner et al., 1989). Since antibodies are glycoproteins, a slightly modified reaction sequence was utilized for detection of antigens (Heimgartner et al., 1990a). The same basic chemistry, that is the reaction of aldehydes generated in the sugar part of a glycoprotein with a hydrazide reagent, was successfully applied in the study of spatial proximity of sugar chains in immunoglobulins from different species (Heimgartner et al., 1990b). A cleavable dihydrazide cross-linker was used in this work. In some of the above studies we used Endo-H to release sugar chains from the glycoproteins. Since the enzyme was rather expensive we considered purifying it by affinity chromatography of course. Initially we attempted to prepare a competitive inhibitor by chemical modifications of its substrate (invertase oligosaccharides and glycopeptides), but it turned out that it is much simpler to use the intact substrate in a specially designed substrate-affinity chromatography (Greber et al., 1989). In the published work on imprinting of proteins (Glad et al., 1985), the matrix was based on porous silica, and silane monomers were used for imprinting. It was reasonable to assume that other matrices, and different monomers, may also be suitable. It appeared that vinyl-type monomers, in particular acrylamide-type monomers, may be advantageous because it is known that acrylamide polymerizes readily at low temperatures in a water solution whose pH is close to neutrality. Such mild conditions are essential for maintenance of the native conformation of the protein being imprinted. However, there was one serious obstacle to the use of

New Monomers for Preparation of Separation Media

595

acrylamide. Polymerization of acrylamide in the presence of a protein results in entrapment of that protein within the matrix and that procedure is actually used as an immobilization method (Johansson and Mosbach, 1974). The entrapment is attributed to small pores of the polyacrylamide gel relative to the size of the protein. For successfulimprinting it is essential to remove the protein after matrix formation and, therefore, it was necessary to find a suitable monomer that will give a highly porous matrix while being hydrophilic and able to polymerize under mild conditions. A known acrylic monomer (Jedlinski and Paprotny, 1966), N-acryloyltris(hydroxymethy1)aminomethane (NAT), appeared suitable because it is hydrophilic, has an amide vinyl group, and was reported to form highly porous beads suitable for ion exchange chromatography (Girot and Boschetti, 1981).After synthesis of the monomer and mastering suspension polymerization, numerous experimentswere carried out with NAT in combination with other monomers. Once the NAT monomer was available in the lab we also decided to test its suitability for preparation of gels for electrophoresis. That work was clearly a by-product of the imprinting project. At that time we were not aware that what started initially as a small side project would profoundly influence our future activities not only at ETH but also at Elchrom, the company in which the authors are cofounders. In this contribution we shall describe some results of these activities.

II. RESULTS AND DISCUSSION A. Poly(NAT) Gels

Electrophoresis and chromatography are the two most important methods for separation of biomolecules. The separation is achieved by forcing the molecules to migrate through a matrix. Even though both methods employ a matrix, the requirements imposed on a matrix for electrophoresisare different from those imposed on a matrix for chromatography. The currently used matrices for chromatography are made of different materials including inorganic, organic, and natural materials such as silica, cross-linked synthetic polymers, and polysaccharides. In contrast to this richness of matrices for chromatography,the choice of matrices for electrophoresis is surprisingly poor, as already noted by Righetti (1989). Thus, agarose and polyacrylamide gels are still almost exclusive matrices for electrophoresis. Our results with poly(NAT) gels demonstrated that this matrix is an alternative to polyacrylamide and agarose gels for analysis of proteins (Kozulic et al., 1987) and nucleic acids (Kozulic et al., 1988). Poly(NAT) gels are less restrictive than polyacrylamidebut more restrictive than agarose gels, enabling separation of large proteins and medium size DNA that are not optimally resolved in polyacrylamide or agarose gels (Kozulic et al., 1987; Kozulic et al., 1988; Kozulic and Mosbach, 1994).After publication of the cited work, the separation of double-stranded DNA has been optimized in the gels run in the submerged electrophoresismode so that current precast 6%, 9%, and 12%poly(NAT) gels give excellent resolution of DNA

596

BRANKO KOZULIC and URS HEIMGARTNER

Figure 1. Submerged gel electrophoresis of ds DNA restriction fragments in a 6% poly(NAT)gel in the improved submarine unit (Kozulic and Heirngartner, 1991). Lanes 1 and 2 contain IambddHae II digest; lane 3 contains 0.5 pg of 123 bp ladder (Pharrnacia); lanes 4 and 5 contain IambddMva I; lane 6 contains 1 kbp DNA ladder (BRL); lanes 7 and 8 contain IambddAva II. The gel was run in TAE buffer at room temperature at 7 V/crn for 3 hours.

in the 50 to 5000 bp range. Figure 1 shows a typical separation in a 6% poly(NAT) gel. The gels are particularly suitable for analysis of PCR fragments. An example is shown in Figure 2. During optimization of poly(NAT) gels it was realized that standard submarine units, although suitable for most applications,do not allow the best gel performance and, therefore, a new apparatus was designed (Kozulic and Heimgartner, 1991). In the initial work (Kozulic et al., 1987), proteins were run in poly(NAT) gels mostly under nondenaturing conditions in a continuous buffer system. SDS electrophoresis in a discontinuous buffer system confirmed that poly(NAT) gels are less restrictive than polyacrylamide gels, but the protein bands were more difise (unpublished observations). This band diffusenesswas found to be related to slower destacking in the separating gel when the buffer system included glycine as the

New Monomers for Preparation o f Separation Media

597

Figure 2. Submerged gel electrophoresis of amplified DNA in a 12% poly(NAT) gel. Lanes 1, 3, and 8 contain 0.5 pg of 100 bp ladder (Pharmacia); lanes 2 and 7 contain 0.5 pg of pBR 322/Hae Ill digest; lane 4 contains the amplification product of a homozygote mutant individual (95 bp) having the most common mutation in the cystic fibrosis gene (deletion F508); lane 5 contains the amplification products of a heterozygote individual with the same mutation; and lane 6 contains the amplification product of a homozygote normal individual (98 bp). The gel was run at 10 V/cm at 20 "C for 4 hours.

trailing ion, chloride as the leading ion, and Tris as the common ion (Laemdi system). By changing the buffer system it became possible to destack relatively small proteins (about 40 kDa) in a 7% poly(NAT) gel, which resulted in very sharp bands (Figure 3). In addition, poly(NAT) gels can be stained with silver without noticeable background (Figure 3) by a modification of a published procedure (Gottlieband Chavko, 1987). Taken together, these results demonstrate that poly(NAT) gels can be advantageously used as an alternative matrix to polyacrylamide gels. One additional advantage is low toxicity of the NAT monomer (it is not lethal to rats at 2 g/kg administered orally) compared to the fact that acrylamide is a neurotoxin in humans (Bailey et al., 1986), as well as highly toxic (LD,, in mice 0.150 gkg).

598

BRANKO KOZULIC and URS HEIMCARTNER

Figure 3. SDS electrophoresis of proteins in 7% poly(NAT) gels. A 1-mm thick gel was run in Bio-Rad Mini Protean I I apparatus at 35 mA for 1 h and 10 min. The gel was silver stained by a modification of the procedure described by Gottlieb and Chavko (1987). The samples include markers (Bio-Rad, lanes 1, 5, and 9) with molecular weights 21 6,116,94,67, and 45 kDa, three dilutions of wheat germ extract (lanes 2 4 , three dilutions of rabbit reticulocyte lysate (lanes 6 4 , and protein markers with molecular weight 94, 67, and 45 kDa (lane 10).

€3.

Hydrophilic Gels

The most important features of poly(NAT) gels include pronounced hydrophilicity and a higher effective porosity compared with polyacrylamide gels. The higher porosity of poly(NAT) gels was advantageously used for separation of large biomolecules as discussedabove. Amonomerwas sought that produces even a more porous gel because such a gel would be beneficial in many applications including isoelectric focusing, multiphasic zone electrophoresis, and electrophoresis of proteins, lipoproteins, proteoglycans, and nucleic acids. As a working hypothesis it was assumed that NAT yields gels of higher porosity because its molecular weight is higher than that of acrylamide. Thus, an NAT solution has a molar concentration lower than the acrylamide solution of the same percentage. After polymerization, the lower molar concentration of the NAT solution presumably results in fewer polymer chains per unit volume leading to gels of increased porosity. The finding that a poly(NAT) gradient gel exhibited effective porosity approximatelythree-fold higher than porosity of the corresponding polyacrylamide gel (Komlic et al., 1987), in accordance with the 2.5-fold lower molarity, lend support to the above assumption. If that simple assumption is correct then even more porous gels will be formed from monomers of higher molecular weight. However, there is probably an upper limit to the molecular weight of a monomer due to required solubility and minimal concentration of double bonds necessary for efficient polymerization.For example,

N e w Monomers for Preparation of Separation Media

599

a 0.5 M solution of acrylamide (MW 71) is about 3.5%, and this concentration is close to the lower limit for gel formation. A monomer with MW of 1000 would need to be dissolved at 50% in order to achieve the same molarity. It is hard to predict whether such a gel containing 50% dry material would be more or less sieving to macromolecules than a 3.5% polyacrylamide gel. On the other hand, it is clear that at lower monomer concentration polymerization efficiency will be lower. A low polymerization efficiency generally results in gels of poor stability and, therefore, the lack of mechanical strength may be a drawback with gels produced from monomers of very high molecular weight. The optimal properties are expected to be inherent to the monomers of medium size, also due to a balanced ratio between the size of the polymer backbone and the size of the side chains present in every repeating unit. However, our current knowledge of gels does not provide a basis for definition of the medium size in terms of molecular weight. In addition to a higher molecular weight, the new monomers should fulfill at least two firther requirements. First, they should be hydrophilic in order to give homogenous aqueous gels. Second, the double bond of these monomers should efficiently polymerize under mild conditions used for the preparation of gels for electrophoresis. From the above assumptions and considerations, it appeared that gels with desirable properties may be formed from monomers derived from amino sugar alcohols as described by the formula shown in Figure 4. Sugar alcohol monomers are hydrophilic because they contain at least three hydroxyl groups. Further, due to an adjacent amide group the double bond in the monomers is more reactive than a typical double bond. Two of the monomers

R1

R2

R3

I

I

I

HC - N - C - C = CH2

I (HCOH1 "

II 0

I CH,OH

where R, is H, CH20H or (CHOH),CH20H, m being 1 or 2 ; R, is H or CH3;

R3 is H or CHI: and

n is an integer of 1-4; Figure 4.

600

BRANKO KOZULIC and URS HEIMGARTNER

represented by the above formula, N-acryloyl-1-amino- 1-deoxy-D-glucitol and N-methacryloyl-1-amino- 1-deoxy-D-glucitol, as well as their linear polymers, were described (Whistler et al., 1961; Klein and Herzog, 1987). However, in the two references no data were reported concerning polymerization of either of the two monomers in the presence of a cross-linker to form an aqueous gel. Moreover, there was no indication as to whether such an aqueous gel may represent a matrix suitable for separation of biomolecules. Many monomers represented by the above formula were synthesized by a modified process as described in detail in the patent literature. See amendment No. 1. Here we shall describe some properties of the gels comprising these monomers. The gels were made in the form of beads as well as in the form of a continuous bed. The beads were packed into columns and used for chromatographic separations, whereas the gels in the form of rods and sheets were used in electrophoresis. The gels used in electrophoresis were studied using the theory based on the extended Ogston model of gel electrophoresis.

C. Characterization of Hydrophilic Gels Electrophoretic migration of macromolecules in polyacrylamide gels is most often described in terms of the extended Ogston model (Rodbard and Chrambach, 1970;Tietz, 1988).Accordingly,the measured mobility, p, can be related to the free mobility, po,of a migrating molecule with radius R, as well as to the gel percentage, T, total length of the gel fibers, l’, and the fiber radius, r: log p = log p, - xl’(r + R ) ~ Tx

10-l~

or log p = log p, - Y T where the retardation coefficient, K,, is defined as

Y = xl’(r + R ) x~ The extended Ogston model has been extensively used for analysis of electrophoretic migration of various macromolecules, mainly to estimate molecular weight and radius of an unknown molecule. However, since the retardation coefficient is correlated also to the length of gel fibers and their radius, this model can be used for characterization of different gels. Characterization was carried out by preparing a series of gels of different percentage, that is, running a set of proteins in all gels under identical denaturing (SDS) conditions followed by construction of Ferguson plots and R-plots (Rodbard and Chrambach, 1970; Tietz, 1988).According to the extended Ogston model, in R-plot the intersection of the straight line on abscissa corresponds to -r, which is the radius of the gel fiber. The intersection on the ordinate relates to the square root of the fiber volume (VF, in ml per gram polymer dry weight), and slope of the straight line relates to the square root of the

New Monomers for Preparation of Separation Media

601

total fiber length (1’, in c d g ) . Table 1 summarizes the values estimated for five gels comprising new monomers. Since the major goal of this evaluation was to compare different gels, all values are given relative to polyacrylamide gel. It is apparent that all gels have a fiber radius larger than a polyacrylamide gel. Further, fiber volumes of the new gels are also higher but the fiber lengths are smaller than in polyacrylamide gel. There are clear differences among gels made of different monomers. For example, N-acryloyl-N-methyl-1-amino- 1-deoxy-Dgalactitol gel has 1’ and Vf values more similar to that of poly(NAT) gel than to other monomers. In addition, fiber volume of N-acryloyl-1-amino- 1-deoxy-Dgalactitol is almost twice as high as the fiber volume of the corresponding N-methyl derivative. Most significant are differences in the total fiber length among the new gels. The values vary from 0.29 to 0.97 relative to polyacrylamide. It is noteworthy that pronounced differences exist between gels made of monomers that are enantiomers such as glucitol and galactitol monomers, which differ only in configuration at C4 in the sugar part. At present it is not clear how this minor change in monomer structurecan cause such profound changes in gel properties. Even though the above gels differ, they are all suitable for separation of biomolecules. By slab gel electrophoresis we obtained very sharp protein and DNA bands and the monomers are particularly suitable for capillary gel electrophoresis (not shown). D. Amphiphatic Gels

From the structural formula shown in Figure 4, it is evident that all compounds represented by it are very hydrophilic. If the substituent on nitrogen (RJ became more hydrophobic, it was not clear whether the monomers would form stable aqueous gels and how biomolecules would migrate in such gels under the influence of an electric field. A practical question was whether it is possible to form methacrylate derivatives of the sterically hindered secondary amino group without acylation of hydroxyl groups. A synthetic route was found and many monomers

Table 1. New Monomers for Preparation of Separation Media Monomer

r

Acrylamide NAT N-acryloyl-1-amino- 1 -deoxy-D-glucitol N-acryloyl-2-amino-2-deoxy-D-glucitol N-acryloyl-1-amino-1 -deoxy-D-galactitol N-acryloyI-N-methyl- 1 -amino- 1-deoxy-D-glucitol N-acryloyl-N-methyl1-amino-1-deoxy-D-galactitol

1 1.1

Note;

1.8 1.7 2.3 2.9 1.3

VC

P

1

1

1.2 2.2 2.5 2.6 2.0 1.4

0.56 0.86 0.53 0.29 0.97

0.84

The fiber radius r, fiber volume V,, and fiber length I‘ of gels comprising different monomen, as estimated from the R-plots. The values for acrylamide gel were taken as 1.

602

BRANKO KOZULIC and URS HEIMGARTNER

with different substituentson nitrogen were prepared as described in the published patent literature. A series of monomers was prepared having the same hydrophilic part (galactitol) and increasingly hydrophobic nitrogen substituents. The substituents included methyl, ethyl, propyl, butyl, hexyl, and other hydrocarbon radicals. Such monomers are amphiphatic compounds because they contain a hydrophilic and a hydrophobic part. Many polymers with unique properties were obtained by polymerization of these monomers, but here we focus on some electrophoresis results with gels comprising amphiphatic monomers. Gels were prepared in which the cross-linker was Np-methylene-bis-acrylamide and the monomer was N-acryloyl-N-methyl-1-amino- 1-deoxy-D-galactitol, N-acryloyl-N- eth y 1- 1-amino - 1-deoxy -D -g alac t it o 1, N-acryloyl-N-propyl-1amino- 1-deoxy-D-galactitol or N-acryloyl-N-butyl-1-amino- 1-deoxy-D-galactitol. Thus, the only difference between the monomers was substitutionon the amide nitrogen, ranging from methyl to butyl. Accordingly, hydrophobicity of the gels increased in very small increments. Migration rate of bromophenol blue (3',3",5',5''-tetrabromophenol sulfonphthalein) was similar in the gels made of methyl and ethyl monomers. In the first gel the dye migrated slightly behind the 123-base-pair (bp) DNA fragment (from 123 DNA ladder, BRL) and in the second gel slightly ahead of the 123 bp fragment. In the gel with propyl groups, the dye migrated approximately as the 246 bp fragment. More importantly, in the gel with butyl groups bromphenol blue migrated approximately as the 1000 bp fragment. In addition, at the beginning of electrophoresis the dye zone was concentrated as it entered the gel and a slight change in color (to pale blue) of bromphenol blue was observed in this gel. Such effects were not noticed in the other three gels. These findings indicate that electrophoretic migration of bromophenol blue is decreased due to hydrophobic binding to the butyl groups in the gel. It is important to note that the electrophoretic migration of DNA fragments from 123 to 6000 bp was comparable in all gels, indicating a similar effective porosity of these four gels. Many gels with hydrophobic residues were used for electrophoresisof proteins. The gels were usually composed of acrylamide, an amphiphatic monomer, and Np-methylene-bis-acrylamide. The relative ratios of the three components were chosen to give essentially transparent gels. When bovine serum albumin (BSA), as a model protein, was electrophoresed under native conditions in a gel comprising N-hexyl-N-acryloyl-1-amino- 1-deoxy-D-glucitol, the protein was not retarded although bromophenol blue was. Likewise, in gels with octyl groups no retardation of BSA was observed. However, when the electrophoresisbuffer contained SDS, BSA was retarded and the retardation was dependent on SDS concentration and amount of the hydrophobic groups in the matrix. Interestingly, the retardation was stronger at higher SDS concentrations.Without going into possible mechanisms of protein retardation in the presence of SDS, the results indicate that the new gels offer the possibility of separating proteins on the basis of their hydrophobicity. Separations of proteins according to their size and charge are well established, and

New Monomers for Preparation o f Separation Media

603

separation based on hydrophobicity may add the third dimension to electrophoretic separation of proteins. Monomers with other substituents on the nitrogen were also prepared. One of 1-deoxy-D-galactitol). This is an extremely them was N-acryloyl-amino-N,iV,-bis( hydrophilic monomer and its polymers contain 10 hydroxyl groups per repeating unit. The molecular weight ofthat monomer is 399. Since stable aqueous gels were prepared with Bis as the cross-linker, when the monomer concentration was around lo%, it is evident that the upper limit for monomer molecular weight has not yet been reached. See amendment No. 2.

ACKNOWLEDGMENTS We thank M . Machler, Medizinische Genetik, Universitat Zurich, Switzerland, for providing us with the cystic fibrosis PCR fragments.

REFERENCES Bailey, E., Farmer, P.B., Byrd, I., Lamb, J.H., & Peal, J.A. (1986). Monitoring exposure to acrylamide by the determination of S-(2-caroxyethyl)cysteine in hydrolyzed hemoglobin by gas chromatography-mass spectrometry. Anal. Biochem. 157,241-248. Barbaric, S., Kozulic, B., Ries, B., & Mildner, P. (1984). Physicochemical and kinetic properties of acid phosphatase from Succharomyces cerevisiue. J. Biol. Chem. 259,878-883. Glad, M., Norrlow, O., Sellergren. B., Siegbahn, N., & Mosbach, K. (1985). Use of silane monomers for molecular imprinting and enzyme entrapment in polysiloxane-coated porous silica. J. Chromatogr. 347, 11-23. Girot, P. & Boschetti. E. (1981). Physico-chemical and chromatographic properties of new ion exchangers. J. Chromatogr. 213,38!&396. Gottlieb, M. & Chavko, M. (1987). Silver staining of native and denatured eucaryotic DNA in agarose gels. Anal. Biochem. 165, 33-37. Greber, U., Kozulic. B., & Mosbach, K. (1989). Purification of endo-N-acetyl-beta-D-glucosaminidase H by substrate-affinity chromatography. Carbohydr. Res. 189,289-299. Heimgartner, U., Kozulic B.. & Mosbach, K. (1989). Polyacrylic polyhydrazides as reagents for detection of glycoproteins. Anal. Biochem. 181. 182-189. Heimgartner, U., Kozulic B.. & Mosbach, K. (1990a). Polyacrylic polyhydrazides as novel reagents for detection of antibodies in immunoblotting assays. J . Immunol. Methods, 132,239-245. Heimgartner, U., Kozulic B., & Mosbach, K. (1990b). Reversible and irreversible cross-linking of immunoglobulin heavy chains through their carbohydrate residues. Biochem. J. 267. 58S591. Jedlinski, Z. & Paprotny, J . (1966). Synthesis and polymerization of N-alkylolacrylamides. Rocmiki Chem. 40, 1487-1493. Johansson, A.C. & Mosbach, K. (1974). Acrylic copolymers as matrices for immobilization of enzymes 1. covalent binding or entrapping of various enzymes to bead-formed acrylic copolymers. Biochim. Biophys. Acta 370,33!&347. Klein, J. & Henog, D. (1987). Poly(viny1saccharide)s 2. Synthesis of some poly(viny1saccharide)s of the amide type and investigation of their solution properties. Macromol. Chem. 188, 1217-1232. Kozulic, B., Barbaric, S., Ries, B., & Mildner, P. (1984). Study of the carbohydrate part of yeast acid phosphatase. Biochem. Biophys. Res. Commun. 122, 1083-1090. Kozulic, M., Kozulic, B., & Mosbach, K. (1987). Poly-N-acryloyl-tris gels as anticonvective media for electrophoresis and isoelectric focusing. Anal. Biochem. 163,506-5 12.

604

BRANKO KOZULIC and URS HEIMGARTNER

Kozulic, B., Mosbach, K., & Pietrzak, M. (1988). Electrophoresis of DNA restriction fragments in poly-N-acryloyl-tris gels. Anal. Biochem. 170,478-484. Kozulic, B. & Heimgartner, U. (1991). An apparatus for submerged gel electrophoresis. Anal. Biochem. 198,256-262. Righetti, P.G. (1989). Of matrices and men. J. Biochem. Biophys. Methods. 19, 1-20. Rodbard, D. & Chrambach, A. (1970). Unified theory of gel electrophoresis and gel filtration. Proc. Natl. Acad. Sci. USA 65,910-977. Tietz, D. (1988). Evaluation of mobility data obtained from gelelectrophoresis: Strategies in the computing of particle and gel properties on the basis of the extended Ogston model. Adv. Electrophoresis 2, 109-1 69. Whistler, R.L., Panzer, H.P., & Roberts, H.J. (1961). 1-Acrylamido-1-deoxy-D-glucitol,l-deoxy-lmethacrylamido-D-glucitol and their polymerization. J. Org. Chem. 26, 158W588.

A N INTEGRATED APPROACH IN THE ANALYTICAL DESCRIPTION OF AFFINITY CHROMATOGRAPHY, BIOSENSORS, IMMOBILIZED BIOCATALYSTS, A N D SIMILAR SYSTEMS

VoI ker Kasche

606 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 606 1. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. SIMILARITIESIN THE CHARACTERIZATION OF AFFINITY ADSORBENTS AND IMMOBILIZED BIOCATALYSTS . . . . 608 111. CONSEQUENCES FOR THE DESIGN OF PREPARATIVE SYSTEMS . . 609 A. Mass Transfer Limited Systems . . . . . . . . . . . . . . . . . . . . . . 609 B. Biocatalyst or Ligate Molecule Density . . . . . . . . . . . . . . . . . . 61 1

Advances in Molecular and Cell Biology Volume 15B, pages 605-617. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN:0-7623-0114-7

605

606

VOLKER KASCHE

W. SIGNAL GENERATION AND SIZE IN BIOSENSORS AND ANALYTICAL CHROMATOGRAPHY . . . . . . . . . . . . . . . . . . . . 612 A. Biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .612 B. Analytical Chromatography . . . . . . . . . . . . . . . . . . . . . . . . ,614 V. CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ,614

ABSTRACT Affinity chromatography adsorbents, biosensors, and artificial and natural systems with immobilized biocatalysts have phenomenological similarities and can be analyzed using an integrated system approach. This is possible because the molecules in the bulk phase must d i f i s e through a diffusion layer into the part of the system where biotransformations or associations occur with immobilized biocatalysts or ligate molecules. Two properties of the system limit its behavior. The magnitude of the dijjsion distance determines both concentration gradients and the rate of mass transfer. This influences the selectivity and the time to reach the end point of the process. The biocatalyst or ligate density limits the rate of biotransformations or the capacity in chromatographic systems. These system properties can be used to develop an integrated analytical description of these systems.

1. INTRODUCTION Natural and artificial heterogeneous systems that are used forpreparative purposes as biotransformations, selective isolation of molecules, regulation of biotransformations, or analytical purposes can be described using the system shown in Figure 1. One of the first investigators who observed that the behavior of these systems cannot be described by the methods developed for homogeneous systems was Otto Warburg (Warburg, 1923). He found that the oxygen consumption rate in tissue slices depended on the thickness of the slices. In subsequent years systems were analyzed extensively by cell physiologic techniques (Rashevsky, 1940). Similar systems exist in soil where McLaren found that the pH optimum of adsorbed enzymes differed from the pH in free solution due to a pH gradient between the soil surface and the bulk solution caused by the electrostatic double layer formed by stationary charges on the surface (McLaren and Packer, 1970). Artificial systems as heterogeneous catalysts in chemical engineering are also described in Figure 1, which were first characterized in the 1930s (Damkohler, 1937; Thiele, 1939; Zel’dovich, 1939). In the 1950s,enzymes and proteins were immobilized for analytical (chromatography, later biosensors (Keston, 1961)) and preparative (biotransformations or isolation of antigens or antibodies) purposes by investigators including Campbell, Luescher and Lerman, Grubhofer and Schleith, Katchalski, Manecke, and Mitz. This is well documented in one of the early reviews in this field (Silman and

*

An Integrated Description of Systems with Immobilized Molecule

607

Concentration

Distance from cente&f the immobilized phase Phase with immobilized molecules or cells (biofilm, tissue)

Diffusion Bulk phase layer with (mass transfer flowing Only by solution diffusion)

Figure 1. Natural and artificial systems where the rate of mass transfer from the bulk phase through the diffusion layer and in the volume element, where biotransformations or interactions occur, influences the overall rate of biotransformation or association of solute molecules with the immobilized molecules. The result is the formation of substrate (ligand) complex and product concentration gradients.

Katchalski, 1966). That these systems can be analyzed using the tools developed to describe the natural and artificial systems mentioned above was realized in the 1960s.At this time Klaus Mosbach started his work in this field. He has contributed much in the development of immobilized molecules for preparative (immobilized biocatalysts for biotransformations, preparative affinity chromatography) and analytical (biosensors) uses. This is well documented in the several volumes of Methods in Enzymology, especially those he edited (Mosbach, 1976; Mosbach, 1987). In this context some remarks on the phenomenological similarity of these systems may be timely. The question is whether this similarity may be used to derive an integrated analytical description of these systems. This description can be based on properties of the system given in Figure 1, which influence the rate and selectivity in their application to preparative and analytical processes. One such factor is the dzfusion distance from the bulk phase to the volume element where biotransformations or interactions occur. It equals the characteristic length of this element (radiusR for sphericalparticles) and the thickness ofthe diffusion layer 6 . The other property is the density of the immobilized molecules that determines the maximum rate of biotransformation or the capacity in chromatographic systems.

VOLKER KASCHE

608

II. SIMILARITIES IN THE CHARACTERIZATION O F AFFINITY ADSORBENTS A N D IMMOBILIZED BIOCATALYSTS The overall rate of an enzyme catalyzed reaction is generally reduced when the enzyme is immobilized. The same applies to the rate of complex formation between a free ligand and a ligate molecule that is immobilized. The reduced reaction rate is caused by the mass transfer from the bulk solution to the phase where the association or biotransformation reaction occurs that may cause the formation of substrate (ligand) and product (ligate) concentration gradients (Figure 1). The establishment of gradients increases the time required to reach the desired end point of the process, expressed as the fraction of substrate converted to product or ligand that has been adsorbed (Figure 2). The kinetics of the immobilized systems can be characterized using the ratio of the initial rates of the reaction with the immobilized and free system. These stationary effectiveness factors can be used to characterize systems for analytical purposes. For preparative purposes, however, it is essential to approach the theoretical yield of the processes given by the end points in Figure 2. The ratio of the times required to reach the desired end point with the free to immobilized system (the operational effectiveness factor) can be used to characterize the latter when it is used for preparative purposes (Kasche et al., 1979). The stationary and operational effectiveness factors have been calculated for immobilized biocatalysts (see Kasche, 1983). For this purpose a dimensionless quantity that is a function of the properties of the system is introduced. The Thiele modulus is defined for systems with immobilized biocatalysts as follows:

17

(Thiele modulus)2=

Maximum reaction rate - R2vmax Maximum mass transfer rate D,,K,,,

BIOTRANSFORMATION

O/O

Product formed

TIME

CHROMATOGRAPHY Oh

;

p

(1)

o

Ligand adsorbed

i

n

y

r

w

TIME

Figure2. Time dependenceto reach the desired end point of a reaction with free (-) and immobilized (---) enzyme (biotransformation)or ligate molecule (chromatography).

An Integrated Description of Systems with Immobilized Molecule

609

where R is the particle radius, a measure of the distance that a substrate must diffise in the porous structurewith immobilized enzyme; V,, a measure of the biocatalyst density; De,the diffusion coefficient of the substrate in the immobilized phase; and K, the Michaelis-Menten constant. A similar quantity may be used to characterize immobilized ligates used for preparative chromatography (Thiele modulus for chromatography)2= R2kassnL Defi ~

where k,,, is the bimolecular rate constant for the reaction between ligand and immobilized ligate and nL the density of immobilized ligate. In Equations 1 and 2 the rate of mass transfer through the diffusion layer is not considered. A measure for this is the dimensionless Sherwood number Sh given by the following relation Sh = 2R I6

(3)

which increases with the relative velocity between the bulk phase and the particles. To consider the difision distance through the diffision layer in the expressions for the Thiele modulus Equations 1 and 2 must be multiplied with the factor [(2 + Sh)’/Sh’]. The effectiveness factors decrease with the Thiele modulus. The behavior of preparative systems (rate, capacity) can be improved by increasing the rate of mass transfer in mass transfer limited systems or by increasing the biocatalyst or ligate density in systems where they are the limiting factors.

111. CONSEQUENCES FOR THE DESIGN OF PREPARATIVE SYSTEMS A. Mass Transfer Limited Systems

In preparative systems with immobilized biocatalysts for biotransformations or immobilized ligates for chromatography, a high space time yield in substrate converted or ligand adsorbed is desirable. To obtain this, a high biocatalyst or ligate density, giving a system where the rate of mass transfer is the rate limiting step, is advantageous.Thus, the Thiele modulus is large and the effectiveness factors low. The latter can then only be increased when the rate of mass transfer is increased. Then the diffusion length and the concentration gradients are also decreased. The substrate and product concentration gradients in systems with immobilized biocatalysts or ligate molecules, shown in Figure 1, reduce the rates and increase the product inhibition compared with the homogeneous system. When hydrolases are used to hydrolyze substrates, H‘ is a frequent product. One example is the hydrolysis of Penicillin G, one of the first systems where hydrolysis with an immobilized enzyme was successfully developed for an industrial application (Carleysmith and Lilly, 1979). Then pH gradients may be formed that decrease the

610

VOLKER KASCHE

rate firther and can, for pH-sensitive enzymes like penicillin amidase, increase the rate of enzyme denaturation. The formation of concentration gradients may also reduce the stereospecificity of the enzyme or lead to increased by-product formation due to increased product concentration (Kasche et al., 1991). To reduce these disadvantages of the immobilized system the concentration gradients should be reduced. Concentrationgradient reduction can be accomplished by increasing the rate of mass transfer in the diffusive layer. In batch reactors this implies the use of higher stirring speeds. At practical particle densities (= lo%), however, this leads to an increased rate in particle abrasion due to particle-particle collisions. To avoid abrasion, fixed bed reactors can be used where the rate of external mass transfer is increased with the interstitial flow rate. This is shown in Figure 3 for the hydrolysis of Penicillin G by immobilized penicillin amidase in packed bed recycle reactors. To measure the pH gradient between the particle and the bulk solution during the reaction fluorescein, whose fluorescence intensity is pH dependent, was also immobilized in the particles (Renken, 1993). At the highest flow rates the pH gradient is not negligible, even in the presence of buffer. To decrease the gradients further by increased rate of mass transfer to the particles at constant biocatalyst density, either the particle size or the average diffusion distance from the bulk solution to the immobilized biocatalyst

0

An Integrated Description of Systems with Immobilized Molecule

611

must be decreased or the pore size in the porous structures increased. In packed beds, the solution in the contact area between particles may be assumed to be stagnant. This reduces the fraction of the surface area that may be used for mass transfer from the bulk liquid phase. This effect should increase with decreasing particle size, as found for packed bed reactors and chromatographic systems. These results indicate that at particle sizes below 50 pm the gradients cannot be reduced significantly by further reduction in particle size (Hinberg et al., 1974; Kasche et al., 1992). Then the rate of mass transfer can only be increased by using macroporous perfusible supports for the immobilization of the biocatalysts or ligates. Such supports, either spherical, such as membranes or as porous discs have already been introduced for preparative chromatography (Abou-Rebyeh et al., 1991; Champluvierand Kula, 1991; Fulton et al., 1992). The results show that such supports increase the rate of preparative chromatographicisolation of biopolymers when compared with the traditional supports. Preliminary data for biocatalysts immobilized in such perfusible macroporous systems indicate that concentration gradients are much smaller than in the systems that are predominantly used now. B. Biocatalyst or Ligate Molecule Density

The highest density that can be obtained is reached when the available surface is completely covered with immobilized molecules. For most immobilization methods it has been shown that the biological function (catalytic or binding) is retained when proteins are immobilized. Whether all immobilized enzymes or ligate molecules can execute this function simultaneously,however, depends on the size of the interacting molecules (Figure 4). From this figure it follows that the biological

A

B

Immobilized Enzyme (0)-small substrate

Immobilized Antibody (Y) - Antigen

Increasing Biocatalyst Ligate density

0

4

Figure 4. Influence of immobilized biocatalyst or ligate (antibody) density and substrate or ligand (antigen) size on the fraction of immobilized molecules whose biological function can be used simultaneously. When the average distance between immobilized molecules approaches the size of the substrate or ligand molecule this fraction decreases below 1.

VOLKER KASCHE

61 2

function depends on the size of the interacting molecules and the density of the immobilized biocatalyst or ligate. When the average distance between the immobilized molecules approaches the size of the substrate or ligand molecule, the biological function of two adjacent biocatalysts or ligates cannot be used simultaneously. This is frequently observed with immobilized antibodies at concentrations > 2 mg/ml wet support (= 10 pM). Then only a fraction of the immobilized antibodies can bind a large antigen protein simultaneously, as the binding of the latter leads to steric hindrance in adjacent binding sites (Nachman, 1992). This is also demonstrated by the following observation. At concentrations up to 5 mg/ml wet support (z 60 mM) all immobilized penicillin amidase (MW = 90 kD) molecules retained their biological function to hydrolyze penicillin G. Only about 10% of the immobilized enzyme could, however, bind monoclonal antibodies against penicillin amidase. It follows from this observation that there exists an optimal biocatalyst or ligand density that depends on the size of the interacting molecules.

IV. SIGNAL GENERATION A N D SIZE IN BIOSENSORS A N D ANALYTICAL CHROMATOGRAPHY A. Biosensors

Based on the type of signal generation, biosensors can be classified as dynamic, integral, or afinity biosensors. In dynamic biosensors the signal observed is proportional to the concentration gradient between the bulk phase and the phase with immobilized enzymes (Figure 2). To this class of biosensors belong enzyme electrodes, ENFETS, and optrodes where an 0, or pH-gradient is generated. In integral biosensors the signal observed is proportional to the amount of analyte that has been converted to product in the biosensor. The signal can be measured as the amount of product formed or as the temperature change caused by the reaction as occurs in enzyme thermistors that have been developed extensively by Klaus Mosbach. These two biosensor types can be described using the analytical methods developed for systems with immobilized biocatalysts. In afinity biosensors the signal is proportional to the amount of analyte that has formed a complex with ligates immobilized in the biosensor. To this class of biosensors belong those where biospecific interactions are used. Phenomenologically their description is similar to adsorption chromatography. The time to obtain the steady state signal depends on the diffusion distances (R + 6) in Equations 1 through 3 and the amount of analyte (integral biosensors). It is normally in the range 1 to 10 minutes. The time can be reduced by decreasing the diffusion distance or R. In dynamic biosensors this leads to a reduction in concentration gradients and signal size that can be partly counteracted by increasing the biocatalyst density V,,. Thus in this analytical system, the reduction of R is disadvantageous compared with the preparative systems where

An Integrated Description of Systems with Immobilized Molecule

61 3

the opposite applies (Renken, 1993). To increase the measuring frequency with biosensors, presteady state signals can be measured. This requires the use of biosensors in flow injection systems like FIA that have been used for enzyme thermistors for a long time and are increasingly used for the other biosensor types. The signal size is then smaller than the steady state signal. The measuring frequency is then determinedby the dispersion in the biosensor element. This can be evaluated as the dispersion in analytical chromatography. Biosensors in flow systems can be used to monitor biotechnological processes. For this purpose it is essential to determine the end point of the process (Figure 2 ) . One example here is the hydrolysis of p-lactams as penicillin G. This is described by Penicillin G CI Phenylacetic acid + 6-Aminopenicillanic acid When a dynamic or integral (thermistor)biosensor with a hydrolase that catalyzes the above hydrolysis is used to monitor this reaction, the following influences the biosensor signal near the end point of the reaction. In the sample the product concentration is high and the enzyme may catalyze the reverse reaction or the

Plate height, cm

0

.

l o o 'j

0

0

+ '

1o-'q

10"

0 0

A

A

i

10 3 10"

Supports with monoclonal antibodies IV F 19, 111 E 1 A IV F 19 Sepharose + IV F 19 Eupergit 250 8 IV F 19 Knauer A 111 E 1 Eupergit 250

0

0

A

STI-Sepharose STI-Lichrosphere

3

10"

10-1

100

10'

Interstitialflow rate, cm/sec

figure 5. Plate height as a function of interstitial flow rate in isochratic analytical affinity chromatography using different supports and ligand-ligate systems. STI (soybean trypsin inhibitor) immobilized to different supports was used for the isochratic elution of a-chymotrypsin (MW = 25 kD) (Kasche et at., 1981); the monoclonal antibodies against penicillin amidase immobilized to the different supports were used for the isochratic elution of penicillin amidase (MW = 88 kD). More than 90% of the activity of the enzymes in the samples was recovered in the eluates. Radius of the spherical support particles (Lichroshpere, 10 pm; Sepharose, 40 pm; Eupergit 250, 100 pm). The Knauer support is a macroporousperfusabledisc [radius, 40 rnm; height, 4 mm (Abou-Rebyeh et al., 1991)l.

VOLKER KASCHE

614

kinetically controlled synthesis of the analyte penicillin (Kasche, 1986; Kasche et al., 1992).This reduces the signal near the end point and it is impossibleto calibrate the biosensor without the use of another analytical instrument. Thus, the use of hydrolases in dynamic and integral biosensors to monitor the end points of enzyme hydrolysis reactions is limited. B. Analytical Chromatography

The resolution and measuring rate in analytical chromatography is determined by the plate height that is described by the van Deemter equation (Horvath and Lin, 1976). It is a function of the interstitial flow velocity and the diffusion distance between the flowing liquid and the adsorbent particles. As for the preparative systems, the reduction of particle dimensions below P 50 pm has a marginal influence on plate heights compared with the expected dependence on R2 (Figure 5). An alternative here is to reduce the diffusion distances by using perfisable particles, membranes, or macroporous discs (Abou-Rebyeh et al., 1991 ; Champluvier and Kula, 1991; Fulton et al., 1992). The data given in Figure 5 show that the macroporous discs have the smallest plate heights observed in analytical aenity chromatography of biopolymers.

V. CONCLUSIONS For preparative systems with immobilized biocatalysts or ligate molecules a high space time yield is desirable. In analytical systems the object is to obtain a high rate in signal generation at minimal signal dispersion. This high rate coincides with a high space time yield. The biocatalyst or ligate density per unit support volume is given by the following expression: (n 2 P P P (4) where n is the biocatalyst or ligate density per unit support area (Figure 4), p the porosity of the support, and rp the average pore radius in the support. Analytical expressions for the space time yield for immobilized biocatalysts as a function of V,, and particle dimensions have been derived previously (Weisz, 1962; Kasche, 1986). When the biocatalyst density is given by Equation 4 and the particle size R is used to characterize the diffusion distance, the following expression for the maximum space time yield vobs in biotransformations or in the adsorption step in chromatography can be derived for systems in steady state (n/rp)k 'obs =

.

b'

(5)

( n / r J 6R2(1- E)k

where Cbis the bulk substrate or ligand concentration, E the fraction void space in a packed bed, k a second order rate constant characteristicfor the system (= k,,, for

An Integrated Description of Systems with Immobilized Molecule

61 5

chromatography; = k,,, I K , for one substrate biotransformations at K,,, > C, with negligible product inhibition). The last term in the denominator is the square of the Thiele modulus given in Equations 1 and 2 or the Damkohler number (Damkohler, 1937; Thiele, 1939). Equation 5 is thus a relation that can be used to analyze how the biocatalyst (ligate) density and the diffuon distance influence the space time yields and dynamic behavior of analytical and preparative systems with immobilized biocatalysts or ligates. Equation 5 can be used to estimate the density of the latter that is required to obtain mass transfer controlled system where the second term in the denominator is 210 (Figure 6). A higher density will therefore not increase the space time yield but will give rise to larger concentration gradients that are favorable for dynamic biosensors but unfavorable for product-inhibited biotransformations. For the latter systems the second term in the denominator should be 110. With Equation 5 the required biocatalyst or ligate density can be estimated.The equation can be changed by increasing or decreasingthe pore radius. Changing the pore radius, however, does not change the maximal space time yield,

A Spacetime yield (“obs)

Reaction

mass transfer

rate limiting Ill

Optimal operating point for preparative systems

dynamic biosensors

Biocatalyst or ligate density, (nh p)

Figure 6. The space time yield as a function of biocatalyst or ligate density given by Equation 5. At low biocatalystdensities the system i s reaction controlled and the space time yield increases with increasing biocatalyst density until mass transfer along the diffusion distance becomes the rate limiting step. Then the space time yield is independent of the biocatalyst density. The operating point of preparative systems is given when the maximal space time yield is obtained. At higher biocatalyst densities the concentration gradients increase. This is unfavorable when product inhibition cannot be neglected or in racemate resolution. For dynamic biosensors the signal size depends on the concentration gradients. In this case a high biocatalyst density is required. 1-111 the influence of pore radius and diffusion distance on the space time yield while all other system properties are kept constant. I + II the pore radius is increased by a factor of 10; I -+ Ill the diffusion distance is reduced by a factor of 2.

VOLKER KASCHE

616

which can only be increased by decreasing the particle size that is a measure of the dzfision distance. This is, in packed beds, only possible down to about 50 pm due to particle-particle interactions. An alternative here is to use perfusable particles, membranes, or discs where the pore length, a measure for the diffusion length, can be reduced to 1 to 10 pm without a large increase in back pressure as observed with HPLC supports.With these supports the pore radius is about an order of magnitude larger than in most supportscurrentlyin use (< 0.1 pm). This reduces the biocatalyst density in the support but has no influence on the space time yield in the mass transfer limited system provided this can be realized. In these supports a reduction in the diffusion distance by a factor of p increases the maximum space time yield by a factor of p2 (Equation 5 and Figure 6). The results presented here (Figures 3 and 5 ) together with the analysis of Equation 5 indicates that perfbsable particle membranes or discs may have significant potential applications in analytical and preparative systems with immobilized biocatalysts or ligates. Some results that demonstrate this have been presented here and in other studies.

REFERENCES Abou-Rebyeh, H., Korber, F., Schubert-Rehberg, K., Reusch, J., & Josic, Dj. (1991). Carrier membrane as a stationary phase for affinity chromatography and kinetic studies of membrane-bound enzymes. J. Chromatog. 566,341-350. Carleysmith, S.W. & Lilly, M.D. (1979). Deacylation of benzylpenicillin by immobilised penicillin acylase in a continuous four-stage stirred-tank reactor. Biotechnol. Bioeng. 21, 1057-1073. Champluvier, B. & Kula, M.R. (1991). Microfiltration membranes as pseudo-affinity adsorbents: modification and comparison with gel beads. J. Chromatog. 539,3 15325. Damkohler, G. (1937). Influence of d i f k i o n , fluid flow, and heat transport on the yield in chemical reactors. Der Chemie-Ingenieur, 3,35%485. Reprinted (1988). Int. Chem. Eng. 28, 132-198. Fulton, S.P., Shahidi, A.J.. Gordon, N.F., &. Afeyan, N.B. (1992). Large-scale processing & highthroughput perfusion chromatography. Bio/Technology 10,635639. Hinberg, I., Korus. R., & O’Driscoll, K.F. (1974). Gel-entrapped enzymes: Kinetic studies of immobilized P-galactosidase. Biotechnol. Bioeng. 16,94%963. Horvath, C. & Lin, H.-Y. (1976). Band spreading in liquid chromatography, general plate height equation and a method for the evaluation of the individual plate height contributions. J. Chromatog. 149, 4S70. Kasche, V. (1983). Correlation of theoretical and experimental data for immobilized biocatalysts. Enzyme Microbiol. Technol. 5,2-13. Kasche, V., Schwegler, H., & Kapune, A. (1979). The operational effectiveness factor of immobilized enzyme systems. Enzyme Microbiol. Technol. 1 , 4 1 4 6 . Kasche, V., Buchholz, K., & Galunsky, B. (1981). Resolution in high performance liquid affinity chromatography. Dependence on eluate diffusion into the stationary phase. J. Chromatog. 216, I6%174. Kasche, V., Michaelis, G., & Wiesemann, T. (1991). Stereo- and sequence specificity of serine proteases in peptide synthesis. Biomed. Biophys. Acta 50,38-43. Kasche, V., Renken, E., Schietke, G., Ulrich, R., Biicke, R., Gnewuch, H., & Jansen, K. (1992). Signal generation and evaluation in fluorescence based on-line fiber optic biosensors. In: Biosensors: Fundamentals, Technology and Applications (Scheller, F. & Schmid, R., Eds.), pp. 265274. VCH Verlagsgesellschaft, Weilheim.

An integrated Description of Systems with Immobilized Molecule

61 7

Kasche, V. (1986). Mechanism and yields in enzyme catalyzed equilibrium and kinetically controlled synthesis of p-lactam antibiotics, peptides and other condensation products. Enzyme Microbiol. Technol. 8 , 4 1 6 . McLaren, A.D. &Packer, L. (1970). Some aspects of enzyme reactions in heterogeneous systems. Adv. Enzymol. 33,245-303. Keston, A.S. (1961). Glucose indicators for body fluids. US Patent 2 981 606,25 April 1961. Mosbach, K. (Ed.), (1976). Immobilized Enzymes. Methods Enzymol. 44. Mosbach, K. (Ed.), (1987). Immobilized enzymes and cells. Methods Enzymol. 135-137. Nachman, M. (1992). Kinetic aspects ofmembrane-based immunoaffinity chromatography. J. Chromatog. 597, 167-172. Rashevsky, N. (( 1940). Mathematical Biophysics. The University of Chicago Press, Chicago. Renken, E. (1993). Signal generation and size in dynamic and afinity biosensors. Thesis, Technical University Hamburg-Harburg. Silman, I.M. & Katchalski, E. (1966). Water insoluble derivatives of enzymes, antigens and antibodies. AM. Rev. Biochem. 35,873-908. Thiele, E.W. (1939). Relations between catalytic activity and size of particle. Ind. Eng. Chem. 31, 9 16-920. Warburg, 0. (1923). Experiments on surviving carcinoma tissue. Methods. Biochem. Z. 142,3 17-333. Weisz, P.B. (1962). Enzymatic reaction sequences and cytological dimensions. Nature 195,772-774. Zel’dovich,Ya. B. (1939). The theoryofreactions onpowdersandporous substances.ActaPhysicochim. U.R.S.S. 10,583-592.

This Page Intentionally Left Blank

PART VI

MOLECULAR RECOGNITION

This Page Intentionally Left Blank

MOLECULAR RECOGNITION: A N INTRODUCTION

Ian A. Nicholls

Molecular Recognition provides the basis for Life as we perceive it. The study of Nature’s own recognition systems and of Science’s attempts to mimic her amazing specificities and subtleties will undoubtedly give rise to a better understanding of Life and its underlying chemical processes. The results from such studies are therefore of fundamental importance to the development of all aspects of the chemical and biological sciences. Professor Klaus Mosbach and his colleagues have contributed a comprehensive review detailing the origins of molecular imprinting: from the conceptual developments of Pauling, which pointed the way to imprinting, through the seminal works of several groups, especially using silica matrices, through to the development of organic based imprinted polymer systems, in which his group has been a dominant force. The present status of molecular imprinting and its current and developing areas of application, e.g. for the development of antibody combining site mimics, chiral chromatographicstationary phases, sensor elements,synthesismediators and artificial enzymes, are discussed.

Advances in Molecular and Cell Biology Volume 15B, pages 621422. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

62 1

622

IAN A. NICHOLLS

The use of reversible covalent interactions to produce ligand selective recognition sites in molecularly imprinted polymer systems has been extensively utilised by the groups of Wulff and Shea. Wulff, in an overview concentrating on the development of this important aspect of molecular imprinting technology, has drawn on these studies to illustrate the utility of molecular imprinting for constructing recognition sites in attempts to produce enzyme mimics. In this review Wulff highlights the involvement of main chain chirality and functional group placement, critical factors for subsequent recognition. The lecture presented by Shea at the Mosbach Symposium further illuminated the development of molecular imprinting as a technique for producing recognition sites for use in various applications. This gathering of the World’s leading researchers in the area of molecular imprinting , within the framework of the Mosbach Symposium, in a sense constitutes the first international mini-symposium focussing upon the technique. The ease of preparation and the vast range of potential ligands amenable to the technique make molecular imprinted polymers ideal systems for the study of fundamental physical properties relevant to molecular recognition.The adaption of thermodynamic factorisations of the physical terms underlying recognition phenomena and a basis for studying these factors using molecularly imprinted polymers is presented by Nicholls. Thermodynamicfactors,e.g. the contributionofrotor freezing and of functional group intrinsic binding energies, are of fundamental importance, improved understanding of these and related terms is of significant consequence to the design of imprinted polymer systems and to their use in further evaluating recognition phenomena. The significance of molecular complementarity in recognition is beautifully illustrated in many biological systems. Katchalski-Katzir and co-authors have described an algorithm for evaluating the extent of geometric fit between a ligand and areceptor and have evaluated the utility of the approach through its application to a range of recognition systems. Employing two hemoglobin dimer variants, the tRNA synthetase-tyrosinyl adenylate complex and the inhibition of aspartic proteinase and trypsin; the authors have demonstrated the use of this approach for the study of protein-protein and protein-small ligand interactions. This elegant approach highlights the versatility of computational methodology for the rationalization of molecular recognition phenomena and the study of molecular biology. The value of this type of approach for the study of interactions in biological systems was further underscored by the lecture presented by Rees at the Mosbach Symposium. Collectively, the papers presented in the molecular recognition section of the Mosbach Symposium on Biochemical Technologyreflect the diversity of scientific endeavour which is brought together in this vital field of study.

THE ROLE OF GEOMETRIC FIT BETWEEN PROTEIN MOLECULES A N D THEIR LIGANDS IN DETERMINING BIOLOGICAL SPECIFICITY’

Ep hraim KatchaIs ki-Katzir, Isaac Shar iv, Miriam Eisenstein, Asher A. Friesem, Claude Aflalo, and llya A. Vakser

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. PARTICIPATION OF PROTEINS IN BIOLOGICALLY SPECIFIC INTERACTIONS . . . . . . . . . . . . . . . . . . . . . . . . . . 111. THE MEANING OF “BIOLOGICAL SPECIFICITY” . . . . . . . . . . . . .

624 624 625 625

‘A lecture based on this article was delivered by Prof. Ephraim Katchalski-Katzir at the Mosbach Symposium, held in Lund, Sweden, Dec. 24,1992.

Advances in Molecular and Cell Biology Volume 15B,pages 623437. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

623

624

EPHRAIM KATCHALSKI-KATZIR ET AL.

Jv. THE IMPORTANCE OF MOLECULAR SURFACE RECOGNITION IN SPECIFIC PROTEIN-LIGAND INTERACTIONS . . .626 V. GEOMETRIC ALGORITHM FOR IDENTIFICATIONOF MOLECULAR SURFACE COMPLEMENTARITY . . . . . . . . . . . . . , 6 2 7 VI. IMPLEMENTATION OF THE ALGORITHM DEVELOPED AND ITS APPLICATION TO SEVERAL KNOWN COMPLEXES . . . . . ,632 V. CONCLUDING REMARKS . . . . . . . . . . . . . . . . . . . . . . . . . .635

ABSTRACT The three-dimensional (3D) structure of most protein complexes reveals a close geometric match between those parts of the respective surfaces of the protein and the ligand that are in contact. In many cases the 3D structure of the components in the complex closely resembles that of the molecules in their free, native state. Geometric matching thus appears to play an important role in determining the structure of a complex. A geometric recognition algorithm was developed to identify molecular surface complementarity. It is based on a purely geometric approach and takes advantage of techniques applied in the field of pattern recognition. The algorithm provides a list of correlationvalues indicating the extent of geometric match between the surfacesofthe molecules. The procedure is equivalent to a six-dimensionalsearch, but is much faster by design, and the computations are only moderately dependent on the molecular size. The procedure was tested and validated by using five known complexes for which the relative position of the molecules in the respective adducts was successfully predicted. The molecular pairs were the a,p subunits of deoxyhemoglobin and methemoglobin, tRNA synthetase-tyrosinyl adenylate, aspartate proteinase-peptide inhibitor, and trypsin-trypsin inhibitor. The algorithm developed is being extendedto includeelectrostaticmatch and hydrophobicinteractions.In view of the above findings, the parameters determining biological specificity on the molecular level are discussed and evaluated.

1. INTRODUCTION It is a great pleasure to participate in the Mosbach Symposium on Biochemical Technology and to celebrate the 60th birthday of my young colleague and good friend Klaus Mosbach. I have followed Klaus’s scientific career for more than twenty years, and have been deeply impressed with his originality, creativity, and enthusiasm. Klaus is one of the founders of modern biotechnology, and his studies on immobilization of bioactive substances and cells, and on affinity chromatography, have stimulated a considerable amount of work, both basic and applied, in academia and industry. Together with your many students and colleagues, I congratulate you, Klaus, and wish you many years of happiness and creativity. The world-famous school of pure and applied biochemistry that you have established at the Lund University will undoubtedly continue to develop under your leadership

On the Geometric Fit Between Protein Molecules and Their Ligands

625

and guidance and open up new avenues in the fascinating new disciplinesof modem biotechnology. In my lecture I intend to discuss some aspects of proteiwligand recognition at the molecular level. I have chosen this topic because of the interest in the field displayed by my colleagues and collaborators at the Weizmann Institute of Science and my understanding that molecular recognition is one of the subjects to be discussed in this Symposium.

II. PARTICIPATION OF PROTEINS IN BIOLOGICALLY SPECIFIC INTERACTIONS Some proteins serve as building blocks for complex cellular macromolecular structures, wlule others represent intricate macromolecular machines capable of carrying out highly sophisticated biochemical reactions. All of these activities involve specific interactions on a molecular level. It is the task of the experimentalist and the theoretician to elucidate the structure of the ligand-protein complexes formed, the energy of binding, the conformational alterations involved, and the kinetics of complex formation. Proteins associate with each other in specific ways, Some of these associations are permanent, as for example in multimeric proteins, multienzyme complexes, ribosomes, and virus coat proteins. Other associations are more transient and dynamic, as revealed in the interaction between polypeptide hormones and their receptors, in calmodulin regulatory functions, in the association and dissociation of actin and tubulin polymers, in protein kinases, and in the immune system. Interaction between proteins and their corresponding ligands may be highly specific, as in many of the monoclonal antibody-protein antigen interactions or in the various interactions of protein hormones with their receptors. Protein-protein interactions can, however, show broad specificity, as in the case of interaction of chaperones or ubiquitines with a great variety of proteins to prevent denaturation or to initiate degradation, respectively. Proteins may also exhibit narrow or broad specificity towards low- molecularweight ligands. Some enzymes interact specifically with their corresponding lowmolecular-weight substrates and inhibitors, whereas others can interact with a rather broad ensemble of reactants. Of particular interest is the recent finding that MHCI and MHCII can bind a large number of peptides, each composed of eight or nine amino acids.

111. THE MEANING O F “BIOLOGICAL SPECIFICITY” Biochemists usually describe the strength of binding between a receptor or protein (P) and its corresponding ligand (L) by means of the binding constant,K, describing the equilibrium attained in the system:

626

EPHRAIM KATCHALSKI-KATZIR ET AL.

P+Lt,PL; &=K [PI [LI The free energy of the interaction can readily be calculated from the expression AG = RTlnK A high binding constant usually indicates high specificity of interaction, whereas a low binding constant usually suggests lack of specificity. However, the specificity concept, as it relates to protein-ligand binding, seems to require a somewhat more elaborate consideration. The system to be considered usually contains a well-defined protein or receptor and an ensemble of potential ligands. When the binding of each of these ligands to the receptor protein is examined and if an apriori binding constant is chosen, it is usually found that some of the ligands will bind to the receptor with a higher binding constant and others with a lower binding constant than the chosen one. If the number of ligands with binding constants above the chosen one is relatively small, this suggests high specificity for the receptor under consideration.It therefore seems that the concept of specific binding, when applied to proteins and receptors, requires a description of the protein or receptor, a description of the ensemble of ligands being tested, and a preselected binding constant determining the free energy of binding below which ligand binding is ignored. Let us assume, for example, that two receptors, A and B, are exposed to 10' hexapeptides synthesized at random and that the first (A) binds 10 peptides with binding constants above 10' M-', whereas the second (B) binds only one peptide with K > 10' M-'. There are therefore different specificities in the system being considered, receptor B showing higher binding specificity than receptor A.

IV. THE IMPORTANCE OF MOLECULAR SURFACE RECOGNITION IN SPECIFIC PROTEIN-LIGAND INTERACTIONS The association of proteins with their ligands involves intricate inter- and intramolecular interactions, solvation effects, and conformational changes. Because of this complexity, we do not yet have a comprehensive and efficient approach for predicting the formation of proteidigand complexes from the structure of their free components. However, if certain assumptions are made, such predictions become feasible and attempts at such prediction, based on energy minimization, have been partially successful (Wodak and Janin, 1978; Goodford, 1985; Billeter et al., 1987; Warwicker, 1989; Goodsell and Olson, 1990; Yue, 1990). Another simplifying approach that may alleviate some of these difficulties is based on geometric considerations. The 3D structures of most protein complexes reveal a close geometric match between those parts of the surfaces of the protein and the ligand that are in contact.

On the Geometric Fit Between Protein Molecules and Their 1igands

627

Indeed, the shape and other physical characteristics of the surfaces largely determine the nature of the specific molecular interaction in the complex. Furthermore, in many cases, the 3D structureof the components in the complex closely resembles that of the molecules in their free, native state. Geometric matching thus seems to play an important role in determining the structure of a complex, Crystallographic data showing a close fit between the contiguous molecular surfaces of proteins and their corresponding ligands have been recorded for protein-protein and protein-ligand complexes that could be crystallized. Molecular surface complementarity was observed, for example, in complexes consisting of trypsin and its corresponding pancreatic trypsin inhibitor (Marquart et al., 1983; Wlodawer et al., 1987), the two subunits of citrate synthetase (Remington et al., 1982), the a and p subunits of human hemoglobin (Fermi et al., 1984), lysozyme and its correspondingFa, fragments (Amit et al., 1986), aspartic proteinase and its corresponding peptide inhibitor (Suguna et al., 1987a, 1987b), and t-RNA synthetase and tyrosyl adenylate (Brick et al., 1989). Furthermore, it is worth noting that in those cases in which conformational alterationsoccur in the separate components of the complex before final complexation (by an induced fit or any other molecular mechanism), they take place in order to produce complex components possessing geometrically complementary contiguous surfaces. Several investigators have exploited the advantages of a geometric approach to find shape complementarity between a given protein and its ligand (Greer and Bush, 1978; Kuntz et al., 1982; Zielenkiewicz and Rabczenko, 1984; Zielenkiewicz and Rabczenko, 1985; Fanning et a]., 1986; Novotny et al., 1986; Connolly, 1986; DesJarlais et al., 1988; Chirgadze et al., 1989; Lewis and Dean, 1989; Wang, 1991; Jiang and Kim, 1991; Schoichet and Kuntz, 1991). These investigatorsconsidered geometric match between molecular surfaces as a fundamental condition for the formation of a specific complex (Connolly, 1986). In this approach, which treats proteins as rigid bodies, the complementarity between surfaces is estimated. Furthermore, the geometric analysis can serve as the foundation for a more complete approach that includes energy considerations.

V. GEOMETRIC ALGORITHM FOR IDENTIFICATION OF MOLECULAR SURFACE COMPLEMENTARITY A geometric recognition algorithm to identify molecular surface complementarity was developed by our group at the Weizmann Institute (Katchalski-Katzir et al., 1992). The algorithm is based on a purely geometric approach and employs techniques applied in the field of pattern recognition. The algorithm involves an automated procedure, which includes (i) a digital representation of the molecules (derived from atomic coordinates) by 3D discrete functions that distinguishes between the surface and the interior, (ii) the calculation, using Fourier transformation, of a correlation function that assesses the degree of molecular surface overlap

EPHRAIM KATCHALSKI-KATZIR ET AL.

628

and penetration during relative shifts of the molecules in three dimensions, and (iii) a scan of the relative orientations of the molecules in three dimensions. The algorithm provides a list of correlation values indicating the extent of geometric match between the surfaces of the molecules; each of these values is associated with six numbers describing the relative position (translation and rotation) of the molecules.The procedure is thus equivalent to a six-dimensional search but much faster by design, and the computation time is only moderately dependent on the molecular size. In the development of our algorithm, we began with a geometric description of the protein and the ligand molecules derived from their known atomic coordinates. The two molecules, denoted by a and b, are projected onto a 3D grid of N x N x N points where they are represented by the discrete fimctions 1 inside the molecule 0 outside the molecule and

{ 1 inside the molecule

blmn= 0 outside the molecule

where 1, m, and n are the indices of the 3D grid (l,m,n= { 1. . . N } ) . Any grid point is considered to be inside the molecule if there is at least one atom nucleus within a distance r from it, where r is of the order of van der Waals atomic radii. Examples of two-dimensional cross-sections of these functions are presented in Figures l a and lb. Next, to distinguish between the surface and the interior of each molecule, we retain the value of 1 for the grid points along a thin surface layer only, and assign other values to the internal grid points. The resulting hnctions thus become

1 1

1 on the surface of the molecule a,,m,n= p inside the molecule 0 outside the molecule

(2a)

1 on the surface of the molecule bl,m,n = 6 inside the molecule 0 outside the molecule

(2b)

-

and -

where the surface is defined as a boundary layer of finite width between the inside and the outside of the molecule. The parameters p and 6 describe the value of the points inside the molecules, and all points outside are set to zero. Two-dimensional cross-sections of these functions are shown in Figures l c and Id. In our method, matching of surfaces is accomplishedby calculationof correlation functions. The correlation between the discrete functions 2and 5 is defined as

On the Geometric Fit Between Protein Molecules and Their Ligands

629

a

d

C

Q figure 1. Typical cross-sections through the three dimensional grid representations of the molecules. a; Cross-section (at I= 46) through the function a/,m,n, derived by projection of the a subunit of hemoglobin (from ZHHB) onto a 3D grid (N = 90). The values 0 and 1 are represented in white and black, respectively. b; The cross-section b46,m,n was similarly derived for the j3 subunit (from 2HHB). Other details are as in a. c; The cross-section (at I= 46) through the function a/,m,n obtained by distinguishing the surface layer from the interior of the molecule in the function a/,m,n. The large negative value for p is represented in gray. d, Cross-section&6/n,n, similarly derived from b/,m,n. The small positive value for 6 is represented in a different shade of gray. The values for rand q were 1.8 and 1.2 A, respectively.

N

N

N

I=1 m=l n=l

where a,p, and y are the number of grid steps by which molecule b is shifted with respect to molecule a in each dimension. If the shift vector {a, P,?} is such that there is no contact between the two molecules (see Figure 2a), the correlation value is zero. If there is contact between the surfaces (Figure 2b) the contribution to the

EPHRAIM KATCHALSKI-KATZIR ET AL.

630

a

b

C

d

figure 2. Different relative positions of molecules a and b, illustrated by the crosssections ;&,m,n and 6&,m,n from Figure. 1 . The relative orientation of the molecules is as in the known a-P dimer. a; No contact. b; Limited contact. c; Penetration. The penetrated area is represented in black. d, Good geometric match, as indicated by the extensive overlap of complementary surface layers.

correlation value is positive. Nonzero correlation values can also be obtained when one molecule penetrates the other (Figure 2c). Since such penetration is physically forbidden a distinction between surface contact and penetration must be clearly formulated. Accordingly, we assign a large negative value to p in a and a small positive value to 6 in g. Thus, when the shift vector {a,P,y}is such that molecule b penetrates molecule a, the multiplication of the negative numbers (p) in a by the positive numbers (1 or 6) in 6 results in a negative contribution to the overall correlation value. Consequently, the correlation value for each displacement is simply the score for overlapping surfaces, corrected by the penalty for penetration. Positive correlation values are obtained when the contribution from surface contact outweighs that from penetration. Thus, a good geometric match (such as in Figure 2d) is represented by a high positive peak, whereas low values reflect a poor

O n the Geometric Fit Between Protein Molecules and Their Ligands

63 1

figure 3. Cross-section (at a = 0) through a 3D correlation function Ta,p,v. The correlation function shown was calculated for the a and p subunits of hemoglobin, oriented as in the dimer (from ZHHB, see Figures 1 c and 1 d). The correlation value at each shift vector (O,p,y) is represented by the height of the graph. The prominent peak at (a= 0, p = 14, y = 17) corresponds to the correct match between the molecules (see Figure 2d). Other intermolecular surface contacts (such as in Figure 2b) give rise to the low positive correlation values around the center of the graph. The negative correlation values caused by penetration(see Figure2c) are omitted, leaving the empty area at the center.

match between the molecules. A cross-section of a typical correlation function for a good match is presented in Figure. 3. The coordinates of the prominent peak denote the relative shift of molecule b yielding a good match with molecule a. The location of the recognition sites on the surface of each molecule can readily be determined from these coordinates. In addition, the width of the peak provides a measure of the relative displacement allowed before matching is lost. Direct calculation of the correlationbetween the two functions (see Equation 3) is a rather lengthy process because it involves @ multiplications and additions for each of the N3 possible relative shifts {a, p,y} resulting in an order of N6 computing steps. We therefore chose to take advantageof Fourier transformationsthat allowed us to calculate the correlation function much more rapidly. Finally, in order to complete a general search for the match between the surfaces of molecules a and b, the correlation function has to be calculated for all relative

c

EPHRAIM KATCHALSKI-KATZIR ET AL.

632

orientations of the molecules. In practice, molecule a is fixed, whereas the three Euler angles defining the orientation of molecule b (xyz convention in Goldstein, 1980) are varied at fixed intervals of A degrees. This results in a complete scan of 360 x 360 x 180/A3 orientations for which the correlation fkction 2 must be calculated. Each high and sharp peak found by this procedure indicates geometric match and thus represents a potential complex. The relative position and orientation of the molecules within each potential complex can readily be derived from the coordinates of the correlationpeak and from the three Euler angles at which the peak was found.

VI. IMPLEMENTATION OF THE ALGORITHM DEVELOPED A N D ITS APPLICATION T O SEVERAL K N O W N COMPLEXES In order to implement our algorithm, it is necessary to assign specific values to the various parameters involved, that is, the surface layer thickness, Y , A, p, 6, N and the grid step size denoted by q. The choice of these values is based on a number of considerations outlined in our original article (Katchalski-Katziret al., 1992). The values assigned to these parameters in our computer program were: Y = 1.8 A, A = 20°, p = -15,6 = 1, N = 90 (q z 1.0 to 1.2 A) for the scan stage and N = 128 (q z 0.7 to 0.8 for the discrimination stage. The program was run on a Convex C-220 computer using the Veclib Fast Fourier Transform subroutine. The computation time for each iteration (steps 3-8 in the summarized algorithm) in the scan stage was nine seconds. The total computation time for matching two molecules in the range of 1100 atoms each, including both the initial scan and the discrimination stage, was typically 7.5 hours. Our algorithm was applied to several known complexes whose coordinates are given in the Brookhaven Protein Data Bank (Brookhaven National Laboratory, Upton, NY) in order to test its ability to predict correct structures of protein complexes. We chose complexes that represent a wide variety of relative sizes for molecules a and b. These are two hemoglobin variants: human deoxyhemoglobin (Fermi et al., 1984) (2HHB) and horse methemoglobin (Ladner et al., 1977) (2MHB), representing naturally occurring heterodimers; and three complexes: tRNA synthetase-tyrosinyladenylate (Brick et al., 1989) (3TSl), aspartic proteinasepeptide inhibitor (Suguna et al., 1987a) (3APR),and trypsin-trypsin inhibitor (Marquart et al., 1983) (2PTC). In these tests, we treated the component molecules as separate entitiesby using their respective atomic coordinateswithin the complex. Additional tests were performed with native aspartic proteinase (Suguna et al., 1987a) (2APR) and its peptide inhibitor as well as with trypsin and native trypsin inhibitor (Wlodawer et al., 1987) (4PTI). The relative position of the molecules yielding the best geometric fit in a complex, as determined by the algorithm, was finally compared with the corresponding known complex.

A)

On the Geometric Fit Between Protein Molecules and Their Ligands

633

The results are summarized in Figure 4 in the form of histograms of ten correlation peaks for each pair of molecules. The left side of each panel presents the ten highest peaks obtained at the scan stage, whereas the right side shows the peaks reevaluated for the same ten orientations in the discrimination stage. It can be seen that the correlationpeak for the known complex (shaded) is not necessarily the highest in the scan stage. However, the highest peak that was obtained after discrimination represents the correct orientation and position of molecule b with respect to a, and it is significantly higher than the other peaks. Application of the algorithm to the a and p subunits of human hemoglobin (2HHB, Figure 4a) revealed that the highest peak at the scan stage (score 312) corresponds to the well known a-p dimer. However, in the horse methemoglobin variant (2MHB, Figure 4b) the correct position for the dimer is represented by the third peak (score 290) in the sorted histogram for the scan stage. Nevertheless,both of these peaks became predominant at the discrimination stage (scores 302 and 347 for 2HHB and 2MHB, respectively). We next applied the algorithm to the tRNA synthetase-tyrosinyl adenylate pair (3TS1, Figure 4c), which serves as an example of a complex between a highmolecular-weight protein and a small ligand. In this case, the correlation peak corresponding to the correct position of the ligand in the complex was not the highest one at the scan stage. Discrimination, however, yielded the expected result, that is, the correct orientation was associated with a peak distinctly higher than the other peaks. Further assessment of the procedure was carried out by analysis of the complex between aspartic proteinase and its peptide inhibitor (3APR). This system illustrates a case in which the structure of the protein in the complex closely resembles that of the native protein (Suguna et al., 1987a, 1987b). It is thus possible to look for the best match between the structureof the complexed peptide and the protein either in its complexed (3APR) or in its native (2APR) structure. With the complexed protein the correct relative position of the ligand yielded the highest peak already at the scan stage (Figure 4d), whereas with the native protein the peak describing the correct position was only the fourth in the sorted list (Figure 4e). Analysis of the complex trypsin-trypsin inhibitor (2PTC) was chosen because the native structure of one ofthe components,the inhibitor, differs from its structure in the complex. Specifically, conformational changes involving the side chains of three amino acids located in the binding site of the inhibitor occur upon complex formation (Marquart et al., 1983; Wlodawer et al., 1987). When the structure of the inhibitor in the complex was used (Figure 40, the highest peak after discrimination corresponded to the correct position ofthe inhibitor in the complex. However,when the native structure of the inhibitor (4PTI) was used (Figure 4g), the algorithm did not yield a distinct correlation peak either in the scan stage or in the discrimination stage. This result indicates that the extent of the conformational change occurring at the surface of the inhibitor upon binding to trypsin exceeds that tolerated by the algorithm.

EPHRAIM KATCHALSKI-KATZIR ET AL.

634

a

b

2HHB

300

2MHB

300

8

$2 200

200 100

100

1 3 5 7 9 1 3 5 7 9 Peak ID

C

Peak ID

3TS 1

Peak ID

d

e

3APR

300

2APRnAPR

300

*:

2 $200

200

100

f

1 3 5 7 9 1 3 5 7 9 Peak ID

100

1 3 5 7 9

1 3 5 7 9

2pTc

Peak ID

Peak ID

Figure 4. Correlation results for different pairs of molecules. The pairs are identified by their respective codes (see text). In each panel, the histogram on tbe left shows the ten highest correlation peaks obtained in the scan stage (q = 1 .O-1.2 A) sorted by their score. Each of these peaks was obtained at a different relative orientation of the molecules and corresponds to a potential geometric match. The shaded peak in each histogram correspondsto the known complex between the molecules. The histogram on the right side of each panel shows the scores obtained at the discrimination stage (q = 0.7A-0.8 A), for the ten orientations singled out in the scan stage. Note that in the discrimination stage the spurious peaks (plain) are suppressed, whereas the correct peak (shaded) becomes prominent.

On the Geometric Fit Between Protein Molecules and Their Ligands

635

V. CONCLUDING REMARKS Further theoretical considerations indicate that it might be possible to include in the algorithm, in addition to van der Waals interactions determining the molecular geometric fit discussed above, hydrophobic and electrostatic interactions. This would increase the reliability of the prediction of the molecular structure of the complexes formed between proteins and their ligands whose 3D structures are known and which do not undergo a marked change in their conformation on complexation. One might therefore expect that in the f h r e it will be possible to predict the structure of weak complexes formed between known proteins under well-characterized conditions. It is also reasonable to assume that extension of our algorithm will facilitate the prediction and understanding of the formation of multiprotein structures composed of a considerablenumber of subunits whose 3D structure is known. When complex formation is accompanied by conformational alterations in the participatingcomponents,energy considerationshave to be taken into account.One should bear in mind, however, that in most of the cases investigated so far a good geometric fit between the molecular surfaces in contact prevails in the final complexes. Van der Waals and electrostatic interactions, and hydrogen bonding and hydrophobic interactions within the region of contact between a protein and its corresponding ligand determine the energy of complex formation. In a system consisting of a given protein or receptor and an ensemble of ligands, as discussed above, it is the energy of binding that determines the specificity of the protein or receptor under the conditions specified. As van der Waals forces play a major role in many of the complexes investigated,it is plausible to assume that molecular surfacerecognition, that is, geometric fit between proteins and their ligands, plays an important role in determining biological specificity.

REFERENCES Amit, A.G., Mariuzza, R.A., Phillips, S.E.V., & Poljak, R.J. (1986). Three-dimensional structure of an antigen-antibody complex at 2.8A resolution. Science 233,747-753. Billeter, M., Havel, T.F.. & Kuntz, I.D. (1 987). Anew approach to the problem ofdockingtwo molecules: The ellipsoid algorithm. Biopolymers 26,777-793. Brick, P.,Bhat, T.N., & Blow, D.M. (1989). Structure of tyrosyl-tRNA synthetase refined at 2.3A resolution. Interaction of the enzyme with the tyrosyl adenylate intermediate. J. Mol. Biol. 208, 8-8. Chirgadze, Y.,Kurochkina, N., & Nikonov, S. (1989). Molecular cartography ofproteins: Surface relief analysis of the calf eye lens protein gamma-crystallin. Protein Engineering 3, 105-110. Connolly, M.L. (1986). Shape complementarity at the alp, subunit interface. Biopolymers 25, 12291247. DesJarlais, R.L., Sheridan, R.P., Seibel, G.L., Dixon, J.S., Kuntz, I.D., & Venkataraghavan, R. (1988). Using shape complementarity as an initial screen in designing ligands for a receptor-binding site of known three-dimensional structure. J. Med. Chem. 31,722-729.

636

EPHRAIM KATCHALSKI-KATZIR ET AL.

Fanning, D.W., Smith, J.A., & Rose, G.D. (1986). Molecular cartography of globular proteins with application to antigenic sites. Biopolymers 25, 8634383. Fermi, G., Perutz, M.F..,Shaanan, B., & Fourme, R. (1984). The crystal structure ofhumandeoxyhaemoglobin at 1.74 A resolution. J. Mol. Biol. 175, 15S174. Goldstein, H. (1980). In: Classical Mechanics, p. 608. Addison Wesley Publishing Co., Reading, MA. Goodford, P.J. (1 985). A computational procedure for determining energetically favorable binding sites on biologically important macromolecules. J. Med. Chem. 28, 849-857 Goodsell, D.S. & Olson, A.J. (1990). Automated docking of substrates to proteins by simulated annealing. Proteins 8, 195-202. Greer, J. & Bush, B.L. (1978). Macromolecular shape and surface maps by solvent exclusion. Proc. Natl. Acad. Sci. USA 75, 30>307. Jiang, F. & Kim, S.H. (1991). “Soft Docking”: Matching of molecular surface cubes. J. Mol. Biol. 219, 7%102. Katchalski-Katzir, E., Shariv, I., Eisenstein, M.. Friesem, A.A., Aflalo, C., & Vakser, LA. (1992). Molecular surface recognition: Determination of geometric fit between proteins and their ligands by correlation techniques. Proc. Natl. Acad. Sci. USA 89,219S2199. Kuntz, I.D., Blaney, J.M., Oatley, S.J., Langridge, R., & Ferrin, T.E. (1982). A geometric approach to macromolecule-ligand interactions. J. Mol. Biol. 161, 26%288. Ladner, R.C., Heidner. E.G., & Perutz, M.F. (1977). The structure of horse methaemoglobin at 2.OA resolution. J. Mol. Biol. 114, 385-414. Lancet, D., Horovitz, A., & Katchalski-Katzir, E. (1994). Molecular recognition in biology: Models for analysis of protein-ligand interactions. In: The Lock-and-Key Principle (Behr, J.-P., ed.). John Wiley & Sons Ltd. New York. Lewis, R.A. & Dean, P.M. (1989). Automated site-directed drug design: The formation of molecular templates in primary structure generation. Proc. R. SOC.Lond. B236, 141-162. Marquart, M., Walter, I., Deisenhofer, J.. Bode, W., & Huber, R. (1983). The geometry of the reactive site and of the peptide groups in trypsiwtrypsinogen and its complexes with inhibitors. Acta Crystallog. Sect. B 39,48&490. Novotny, J., Handschumacher, M., Haber, E., Bruccoleri, R.E., Carlson, W.B., Fanning, D.W., Smith, J.A., & Rose, G.D. (1986). Antigenic determinants in proteins coincide with surface regions accessible to large probes (antibody domains) Proc. Natl. Acad. Sci. USA. 83,226230. Remington, S.J., Wiegand, G., & Huber R. (1982). Crystallographic refinement and atomic models of two different forms of citrate synthase at 2.7 and 1.7A resolution. J. Mol. Biol. 158, I 1 1-152. Schoichet, B.K. &Kuntz, I.D. (1991). Proteindockingandcomplementarity.J. Mol. Biol. 221,327-346. Suguna, K., Bott, R.R., Padlan, E.A., Subramanian, E., Sheriff, S.,Cohen,G.H., & Davies, D.R. (1987a). Structure and refinement at 1.8A resolution ofthe aspartic proteinase from Rhizopus chinensis. J. Mol. Biol. 196, 877-900. Suguna, K., Padlan. E.A.. Smith, C.W., Carlson. W.D., & Davies, D.R. (1987b). Binding of a reduced peptide inhibitor to the aspartic proteinase from Rhizopus chinensis: Implications for a mechanism ofaction. Proc. Natl. Acad. Sci. USA 84,700%7013. Vakser, I.A., & Aflalo, C . (1994). Hydrophobic docking: A proposed enhancement to molecular recognition techniques. Prot. Struc. Funct. Genet. 20(4), 32&329. Vakser, I.A. 9 1995). Protein docking for low-resolution structures. protein engineering Vol. 8, pp. 371-377. Vakser, LA. (1996). Main-chain complementarity in protein-protein recognition. Protein Engineering. In Press. Wang, H. (1991). Grid-search molecular accessible surface algorithm for solving the protein docking problem. J. Comput. Chem. 12,746-750. Warwicker, J. (1989). Investigating proteivrotein interaction surfaces using a reduced stereochemical and electrostatic mold. J. Mol. Biol. 206, 381-395.

On the Geometric Fit Between Protein Molecules and Their Ligands

637

Wlodawer, A., Deisenhofer, J., & Huber, R. (1987). Comparison of two highly refined structures of bovine pancreatic trypsin inhibitor. J. Mol. Biol. 193, 145-156. Wodak, S.J. & Janin, J. (1978). Computer analysis of protein-protein interaction. J. Mol. Biol. 124, 323342 Yue, S.-Y (1990).Distance-constrained moleculardocking by simulatedannealing.Protein Engineering 4, 177-184. Zielenkiewicz, P. & Rabczenko, A. (1984). Protein-protein recognition: Methods for finding complementary surfaces of interacting proteins. J. Theor. Biol. 111. 17-30, Zielenkiewicz,P.& Rabczenko, A. (1985). Searchingfor interactingsurfacesofprotein-he improved method. J. Theor. Biol. 116.607-612

This Page Intentionally Left Blank

MODELS OF THE BINDING SITES OF ENZYMES: TEMPLATE INDUCED PREPARATION OF SPECIFIC BINDING SITES IN CROSSLINKED POLYMERS

G u n ter Wu Iff

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . THE IMPRINTING CONCEPT. . . . . . . . . . . . . . . . . . . . . . . . . FURTHER EXAMPLES FOR IMPRINTING . . . . . . . . . . . . . . . . . PRESENT UNDERSTANDING OF THE IMPRINTING PROCEDURE . . . EXACT PLACEMENT OF FUNCTIONAL GROUPS ON RIGID MATRICES VIA A TEMPLATE APPROACH . . . . . . . . . . . . . V. OUTLOOK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I. 11. 111. IV.

639 640 643 645 645 647

ABSTRACT Various attempts to produce specific binding sites in synthetic polymers analogous to those of biological receptors or natural enzymes are described here. To this end, an imprinting procedure was used with the aid of templates in cross-linked polymers.

Advances in Molecular and Cell Biology Volume 15B, pages 63-49. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

639

GUNTER WULFF

640

Suitable polymerizable binding groups were bound to a template molecule. This complex was copolymerized into highly cross-linked polymers. After the removal of the templates the remaining cavities possessed a shape and arrangement of their functional groups corresponding to that of the template. New developments in the application of this technique and experiments towards the understanding of the mechanism of the molecular recognition are discussed in detail.

1. THE IMPRINTING CONCEPT During recent years many approaches have been used to construct enzyme models. Successful experiments used low-molecular-weight ring systems as the binding site, viz. crown ethers (Cram, 1988),cryptands (Lehn, 1988),cyclodextrins(Bender and Komiyama, 1978; Breslow, 1982), concave molecules (Rebek, 1987), etc. and attached catalytically active functional groups. An extremely interesting approach is that based on the use of monoclonal antibodies (Lerner et al., 1991). These antibodies were generated against the transition state analogue of a reaction and they showed remarkably high catalytic activity.

Figure 1. Schematic representationof a cross-linking polymerization in the presence of a template (T)to obtain cavities of a specific shape and a defined spatial arrangement of functional groups (Wulff and Sarhan, 1972; Wulff et al., 1973; 1977).

Models of the Binding Sites of Enzymes

1

641

2

3

Figure 2. Cross-linked polymers prepared from 1 and ethylene dirnethacrylate with ( 2 )and without (3) the template (Wulff et al., 1977).

In our group we have tried to develop a directed synthesis of models of the active site of enzymes in synthetic polymers. In this respect, it is necessary to create a cavity or a cleft with a defined shape and with an arrangement of functional groups (acting as analogues of coenzymes, binding sites, or catalytic functions) with the right stereochemistry. In 1972, a novel approach for the generation of such models was first published (Wulff and Sarhan, 1972; Wulff et al., 1973; Wulff et al., 1977; for reviews see: Wulff, 1986; Wulff, 1991a; Wulff, 1995; see also Andersson et al., in this volume). Polymerizable vinyl monomers containing functional groups (binding sites) were attached to suitable template molecules. Subsequent copolymerization in the presence of solvents and relatively large concentrationsof cross-linking agents produced relatively rigid macroporous polymers. Removal of the template molecules (see Figure 1) left behind cavities in the polymer whose shape and arrangement of the functional groups were determined by the template molecules. Polymers of this type show a high selectivity in binding their template molecules. Consider for example template monomer 1 shown in Figure 2 in which two molecules of 4-vinylphenylboronic acid have been bound by ester linkages to the free hydroxy groups of phenyl-a-D-mannopyranoside as the template. Monomer 1 was copolymerized with 90% ethylene dimethacrylate in the presence of inert solvent to yield a macroporous polymer. About 90% to 95% of the template was split off the macroporous polymer, thus leaving chiral cavities, each of which bore a pair of boronic acid groups (see Figure 2). Polymers of this type exhibit a particular ability for resolution of the racemate of the template. Under batch equilibrationthe original template enantiomer is preferably incorporated and bound similarly as shown in Figure 2. The separation factor a (the ratio K,/K, of the polymer/solvent partition coefficients for the D- and the L-forms) was as high as 5 to 6 (Wulff, 1991a) corresponding to an enantiomeric mrichment of 60% of the D-compound on the polymer in the batch procedure.

642

GU NTER WU LFF

0

30

Time [rnin] Figure 3. MHPLC of racemic a-phenylrnannoside on polymer 3. Flow 0.5 ml/rnin. (5% NH3) in 6 min, curve 8 (Wulff and Minarik, 1990).

T = 60 "C; gradient mobile phase 92/8 + 50/50 acetonitrile (So/, NH3) /water

Using this material as a chromatographic adsorbent, baseline separation of a racemate of the template molecule was obtained (see Figure 3) (Wulff and Vesper, 1978; Wulff and Minarik, 1988; Wulff and Minarik, 1990). With a similar methodology the racemic resolution of free sugars is also possible (Wulff and Schauhoff, 1991). The enantiomer selectivity of these polymers is strongly dependent on the type and amount of cross-linking agent used during the polymerization (Wulff et al., 1982; Wulff et al., 1987). With ethylene dimethacrylate as the cross-linking agent, it was observed that for polymers containing
Models of the Binding Sites of Enzymes 4@(

643

P-I.IU.

3%

30( I

2.K

2.M

15c

i

“I,

Figure 4. Dependence of the specificity of the polymers for racemate resolution (expressedas the separation factor a)upon the kind and amount of cross-linking agent (Wulff et al., 1982).

In addition to the requirement of rigidity for cavity stability, the polymers should, at the same time, possess some degree of flexibility. This is necessary to enable a fast reversible binding of the substrates within the cavities. Cavities of accurate shape but without any flexibility present kinetic hindrance to reversible binding.

II. FURTHER EXAMPLES FOR IMPRINTING Many examples of imprinting with templates have been provided by our group and other groups (see reviews Wulff, 1986; 1991a; 1995), including Shea (Shea and Thompson, 1978), Neckers (Damen and Neckers, 1980), and Mosbach (see this volume). An interesting extension of the concept of molecular imprinting was introduced by Mosbach and coworkers (Arshady and Mosbach, 1981: Anderson et al., 1984; Sellergren et al., 1985, 1988; Ekberg and Mosbach, 1989). Only noncovalent interactions are used during imprinting and the resulting equilibration studies (see

644

GUNTER WULFF

L- PheNHPh

I

EDMA

OH

+D.L- PheNHPh

+O-FheNHPh

0

Figure 5. Molecular imprinting using noncovalent interactions.Template: L-Phenylalanine anilide (L-PheNHPh). A chromatographic racemic resolution of the racemate of the template is shown (Sellergren et al., 1988; Ekberg and Mosbach, 1989).

Models of the Binding Sites of Enzymes

645

review Ekberg and Mosbach, 1989). This approach offers new possibilities for binding because not too many practical linkages are available for fast and reversible covalent binding, and because it has the advantage of easy preparation of the template assemblies. Mostly racemic amino acids were separated in this case, and high selectivity was obtained (see Figure 5). It was expected that this type of interaction facilitates a quicker mass transfer in chromatographic separations. Surprisingly, these separations show kinetics very similar to those of covalent interactions.The reason for this is that the slowest step in mass transfer is the embeddingofthe molecule in the cavity and not the formation of covalent or noncovalent bonds (see review Wulff, 199la).

111. PRESENT UNDERSTANDING OF THE IMPRINTING PROCEDURE Surprisingly, these highly cross-linked, insoluble polymers are optically active and this activity can be measured with suficient precision (Wulff and Kirstein, 1990). The method of measurement involves suspending the polymer in a solvent whose refractive index is exactly identical to that of the polymer. Under this condition the suspension completely transmits the light of the polarimeter. The observed optical rotation is attributed to the boundaries of the empty chiral cavities (the imprints) because no groups with defined stereogenic centers are present. Detailed studies pertaining to the mechanism of molecular imprinting (Wulff and Schauhoff, 1991; Wulff, 1991a) revealed that the integrity of the chiral cavities is stabilized by means of the cross-linking points in the polymer chains. This type of chirality can arise from asymmetric conformations of the polymer chains that are stabilizedby cross-linking.Chiral configurationsof the linear portions ofthe chains are not expected to contributeto the asymmetry ofthe cavity because no asymmetric cyclopolymerizationis possible in this example (Wulff, 1989; 1991b; 199lc). The observed selectivity is due to a combined outcome of the correct spatial orientation of the functional groups inside the cavity and the asymmetric shape of the cavity. Studies involving the selectivity of imprinted polymers for racemates other than that of the template have shown that the orientation of the functional groups is the predominant factor for molecular recognition, whereas the shape of the cavity further enhances the selectivity (Wulff and Schauhoff, 1991). Therefore, racemic molecules other than the template can be separated if they possess a similar orientation of their functional groups.

IV. EXACT PLACEMENT OF FUNCTIONAL GROUPS ON RIGID MATRICES VIA A TEMPLATE APPROACH Since the selectivity in the racemic resolution is mainly determined by the orientation of the functional groups, it was of special interest to examine whether the arrangement of the functional groups alone can give rise to selectivity due to the

646

CUNTER WULFF

Figure 6. Introduction of two amino groups in a distinct distance apart on a plane surface of silica (WuIff et al., 1986).

distinct distance between two functional groups. In other words, the question is whether a two-dimensional instead of a three-dimensional information transfer will be adequate to bring about selectivity. Two amino groups arranged at a distinct distance were introduced into crosslinked polymers with the aid of a template (Wulff et al., 1986). Similarly, Shea and Dogherty (1986) introduced two diol groupings at a distinct distance with the aid of diketones as templates. Furthermore, a similar imprinting procedure on the surface of silica was performed through the formation of siloxanebonds (see Figure 3) (Wulff et al., 1986; Norrlow et al., 1987). Two amino groups were attached to the surface of silica at a distance of 0.72 and 1.05 nm from one another using the template monomers 4 and 5 (Table 1). The attachment to the surface is achieved through the formation of siloxane bonds by condensation between the methoxy silane groups of 4 and 5 and the silanol groups on the surface of the silica. Most of the remaining silanol groups were capped afterwards by reaction with hexamethyldisilazaneto avoid nonspecific adsorption. Over 95% of the templates were split off (see Figure 6). Unlike the behavior of polymers, the position of the two amino groups is not expected to be changed as a result of chain mobility, swelling, or shrinking. The distance can only be altered by conformational changeswithin the functional group part. For comparison purposes, a silica with randomly distributed amino groups was prepared from template

Models of the Binding Sites of Enzymes

647

Table 7. Selectivity of Modified Silicas with Each of the Two Amino Groups in a

Defined Distance

Monomer 6

H2N-(CH2),-SSi(OCH3)3

Apparent Binding Constants of Percentage Splitting

Silica modified with 4 Silica modified with 5 Silica modified with 6 (at random)

Distance Groups (nm) r of o w a c y o H c ~ c H , + c H ; ~ ~ ‘ ~ ~ ? ’

>95%

0.72

4.91

2.58

1 .?4

>95%

1.05

9.07

13.77

1.67

-

-

2.26

2.05

-

monomer 6 (Table 1). In order to elucidate the role of distance accuracy the selectivity was determined by equilibration with an equimolar mixture of the two template dialdehydes 7 and 8 (see Table 1). Both silicas showed a significant difference in binding, preferring their own templates with cx values of 1.74and 1.67. This result clearly suggests that by using distance selectivity alone and with differencesof only 0.33 nm (between 7 and 8) substrate selectivity can be observed. Silicas of this type were used to separate by chromatography dicarboxylic acids that differ in the distance between their carboxyl groups. Good separations were observed compared with those under identical conditions on silica in which randomly distributed amino groups were used (Wulff and Gorlich, 1992). Imprinting on the surface of silica is thus a further extension of the original imprinting method.

V. OUTLOOK In addition to the ability of the polymers to effect racemic resolution, they have also shown their potential as chiral supports for asymmetric synthesis (Wulff and

648

G U NTER WU LFF

Vietmeier, 1989a; 1989b). The main application for these polymers in the future will be their use as novel asymmetric catalysts. With molecular imprinting there is the possibility that analogues of the active site of enzymes and of synthetic antibodies can be prepared. Similarly, polymers imprinted with transition state analogues of a reaction should function as catalysts.Work along these lines is under way in a number of different laboratories (see chapter by Anderson et al. in this volume).

ACKNOWLEDGMENT These investigations were supported by financial grants from Minister fiir Wissenschaft und Forschung des Landes Nordrhein-Westfalen, Fonds der Chemischen Industrie, and Deutsche Forschungsgemeinschaft.

REFERENCES Andersson, L., Sellergren, B., & Mosbach, K. (1984). Imprinting of amino acid derivatives in macroporous polymers. Tetrahedr. Lett. 25,5211-5214. Arshady, R. & Mosbach, K. (198I). Synthesis of substrate-selective polymers by host-guest polymerization. Makromol. Chem. 182,687-692. Bender, M.L. & Komiyama, M. (1978). Cyclodextrin Chemistry. In: Reactivity and Structure Concepts in Organic Chemistry, Vol. 6, pp. 1436. Springer Verlag, Berlin. Breslow, R. (1982). Artificial enzymes. Science 218,532-537. Cram, D.J. (1988). The design of molecular hosts, guests, and their complexes. Science 240,760-767. Damen, J. & Neckers, D.C. (1980). Stereoselectivesynthesis via a photochemical template effect. J. Am. Chem. SOC.102,3265-3267. Ekberg, B. & Mosbach, K. (1989). Molecular imprinting: A technique. Trends Biotech. 7.92-96. Lehn, J.-M. (1988). Supramolecular c h e m i s w o l e c u l e s , supermolecules, and molecular functional units. Angew. Chem. Int. Ed. Engl. 27,89-I 14. Lemer, R.A., Benkovic, S.J., & Schultz, P.G. (1991). At the crossroads of chemistry and immunology: Catalytic antibodies. Science 252,659-667. Norrlow, O., Mansson, M.-O., & Mosbach, K. (1987). Improved chromatography: Prearranged distances between boronate groups by the molecular imprinting approach. J. Chromatogr. 396, 374-377. Rebek, J. Jr. (1987). Model studies in molecular recognition. Science 235, 1478-1484. Sellergren, B., Ekberg, B., & Mosbach, K. (1985). Molecular imprinting of amino acid derivatives in macroporous polymers. Demonstration of substrate- and enantio-selectivityby chromatographic resolution of racemic mixtures of amino acid derivatives.J. Chromatogr. 347, 1-10, Sellergren, B., Lepisto, M., & Mosbach, K. (1988). Highly enantioselective and substrate-selective polymers obtained by molecular imprintingutilizing noncovalent interactions.NMR and chromatographic studies on the nature of recognition. J. Am. Chem. SOC.110,5853-5860. Shea, K.J. & Dougherty, T.K. (1986). Molecular recognition on synthetic amorphous surfaces. The influence of functional group positioning on the effectiveness of molecular recognition. J. Am. Chem. SOC.108, 1091-1093. Shea, K.J. & Thompson, E.A. (1978). Template synthesis of macromolecules. Selective functionalization of an organic polymer. J. Org. Chem. 43,4253-4255. Wulff, G. (1986). Molecular recognition in polymers prepared by imprinting with templates. In: Polymeric Reagents and Catalysts (Ford, W.T., Ed.), ACS Symp. Ser. 308, 186230.

Models of the Binding Sites of Enzymes

649

Wulff, G. (1 989). Hauptkettenchiralitat und optische Aktivitiit von Polymeren aus C-C-Ketten. Angew. Chem. 101,22-38; Main-chain chirality and optical activity in polymers consisting of C-C chains. Angew. Chem. Int. Ed. Eng. 28,21-37. Wulff, G. (199 la). Polymer assisted molecular recognition: The current understanding ofthe molecular imprinting procedure. In: Bioorganic Chemistry in Healthcare and Technology (Pandit, U.K., Aldenveirelt, F.C., Eds.), pp. 5548,Plenum Press, New York. Wulff, G. (1991b). Optically active vinyl polymers. ChemTech 364370. Wulff, G. (1991~).The chirality and optical activity of vinyl polymers. Polym. News 16, 167-173. Wulff, G. & Gorlich, T.Unpublished results. Wulff, G. (1995). Molekulares Pragen (Imprinting) in veruetzten Materialien mit Hilfe von Matrizenmolekiike-uf dem Weg zu Kunstlichen Antikorpem. Angew. Chem. 107,195w979. Molecular imprinting in cross-linked materials with the aid of molecular templateway towards artificial antibodies. Angew. Chem. Int. Ed. Engl. 34, 1812-1832. Wulff, G. & Kirstein. G. (1990). Die Messung der optischen Aktiviat von chiralen Abdriicken in unloslichen hochvernetzten Polymeren. Angew. Chem. 102, 706708; Measuring the optical activity of chiral imprints in insoluble highly cross-linked polymers. Angew. Chem. Int. Ed. Eng. 29,684686. Wulff, G. & Minarik, M. (1986). Pronounced effect of temperature on racemic resolution using template-imprinted polymeric sorbents. J. High Res. Chrom.. Chrom. Comm. 9. 607408. Wulff, G. & Minarik, M. (1988). Tailor-made sorbents. A modular approach to chiral separation. In: Chromatographic Chiral Separations (Zief, N. &Crane, L.J., Eds.), pp. 15-52, Dekker, New York. Wulff, G. & Minarik, M. (1990). Template imprinted polymers for HPLC separation of racemates. J. Liqu. Chrom. 13,2987-3000. Wulff. G. & Sarhan. A. (1972). Uber die Anwendung von enzymanalog gebauten Polymeren zur Racemattrennung. Angew. Chem. 84,364: Use of polymers with enzyme-analoguous structures for the resolution ofracemates. Angew. Chem. Int. Ed. Engl. 11,341 I. Wulff, G., Sarhan, A,, & Zabrocki, K. (1973). Enzyme-analogue built polymers and their use for the resolution of racemates. Tetrahedron Lett. 43294332. Wulff, G. & Schauhoff, S. (1991). Racemic resolution of free sugars with macroporous polymers prepared by molecular imprinting. Selectivity dependence on the arrangement of functional groups versus spatial requirements. J. Org. Chem. 56, 395400. Wulff, G. & Vesper, W. (1978). Enzyme-analogue built polymers, part VIII. On the preparation of chromatographic sorbents with chiral cavities for racemic resolution. J. Chromatogr. 167, 17 1186. Wulff, G. & Vietmeier, J. (1989a). Enzyme-analogue built polymers, 25. Synthesis of macroporous copolymers from a-amino acid based vinyl compounds. Makromol. Chem. 190, 1717-1 726. Wulff, G. & Vietmeier, J. (1989b). Enzyme-analogue built polymers, 26. Enantioselective synthesis of amino acids using polymers possessing chiral cavities obtained by an imprinting procedure with template molecules. Makromol. Chem. 190. 1727-1735. Wulff, G., Heide, B., & Helfmeier, G. (1986). Molecular recognition through the exact placement of functional groups on rigid matrices via a template approach. J. Am. Chem. SOC.108, 108%1091. Wulff, G.. Kemmerer, R.. & Vietmeier, J., Poll, H.-G. (1982). Chirality of vinyl polymers. The preparation of chiral cavities in synthetic polymers. Nouv. J. Chim. 6,681487, Wulff, G., Vesper, W., Grobe-Einsler, R., & Sarhan, A. (1977). Enzyme-analogue built polymers, IV. On the synthesis of polymers containing chiral cavities and their use for the resolution of racemates. Makromol. Chem. 178,2799-2816. Wulff, G., Vietmeier, J., & Poll, H.-G. (1987). Enzyme-analogue built polymers, 22. Influence of the nature of the crosslinking agent on the performance of imprinted polymers in racemic resolution. Makromol. Chem. 188.731-740.

This Page Intentionally Left Blank

MOLECULAR IMPRINTINC: THE CURRENT STATUS AND FUTURE DEVELOPMENT OF POLYMER-BASED RECOGNITION SYSTEMS

Lars I. Anderson, Ian A. Nicholls, and Klaus Mosbach

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . I. INTRODUCTION: A BRIEF HISTORIC ACCOUNT . 11. IMPRTNT PREPARATION AND STUDIES ON MOLECULAR RECOGNITION IN IMPRINTS . . . A. Preparation . . . . . . . . . . . . . . . . . . . . . B. Recognition in Molecular Imprints . . . . . . . . . 111. APPLICATIONS . . . . . . . . . . . . . . . . . . . . A. Enantiomeric Separations . . . . . . . . . . . . . B. Peptide and Protein Separation . . . . . . . . . . . C. Artificial Antibodies and Receptor Mimics . . . . D. Selective Synthesis and Artificial Enzyme Systems E. Substrate Selective Sensors . . . . . . . . . . . . . IV. SUMMARY AND OUTLOOK . . . . . . . . . . . . .

Advances in Molecular and Cell Biology Volume 15B,pages 651670. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

65 1

. . , . . . . . . . . . 652 . . . . . . . . . . . . 652 . .... .... .. ........... ........... ....... .... . .. ........ ........... . .. .... .... ... ........ . .... .... .. .... .......

,

. .

. . . .

. .

.

653 653 655 656 656 660 661 662 664 665

652

LARS I . ANDERSON, IAN A. NICHOLLS, and KLAUS MOSBACH

ABSTRACT Molecular imprinting is becoming increasingly recognized as a technique for the ready preparation of polymeric materials containing recognition sites of predetermined specificity. In many instances molecularly imprinted polymers show binding affinities approaching those demonstrated by antigen-antibody systems. The preparation of molecular imprints in synthetic polymers is presented. The experimental parameters affecting the success of imprint formation and their later use in ligandselective recognition are discussed. The application of molecular imprinting to chromatography, sensor technology, immunoassay, directed synthesis, and enzyme and receptor mimicry is described.

1. INTRODUCTION: A BRIEF HISTORIC ACCOUNT To mimic Nature’s capacity for constructing receptors that are specific for given ligands has been a significant challenge to research efforts in the second half of this century. Linus Pauling (1940) proposed, albeit incorrectly, that it was the impression of an antigen onto the surface of an antibody which accounted for antigenic activity and their astonishing diversity in the human body. Over the past few decades attempts have been made by many groups to produce specific recognition sites in polymeric materials. Dickey (1949), a former student of Pauling’s, reported a strategy similar to that proposed by his mentor for the production of silica-based specific adsorbents. This concept was further developed by other groups (Bemhard, 1952; Curti and Colombo, 1952; Haldeman and Emmett, 1955; Beckett and Anderson, 1957; Morrison et al., 1959; Erlenmeyer and Bartels, 1964). More recently, the crown ether, cryptand, spherand, and other recognition systems have been designed and synthesized for the binding of specific compounds, or compound classes, based upon sets of predefined interactions (Lehn, 1988; Cram, 1988; Pedersen, 1988). These ‘captivand’ systems are characterized by a carefully engineered cavity specific for a particular atom or molecule, but their development has required monumental feats in synthetic organic chemistry in order to obtain the correct host for the desired guest. Nature abounds with examples of remarkable complementarity utilizing the macromolecular structures inherent to biological systems. The cyclodextrins,in particular, bear a strong structural correlation to the synthetic ‘captivand’systems (Bender and Komiyama, 1978; Breslow, 1982; Tabushi, 1986). In parallel with the research endeavors described above, the use of organic polymer materials for molecular imprinting has been investigated by several groups (Mosbach, 1994; Ramstrom and Mosbach, 1996). The development of recognition sites of predetermined selectivity in molecularly imprinted polymers (MIPS)has since led to a vast range of polymeric recognition systems compatible with a virtually infinite array of molecules. Molecular imprinting entails the polymerization of monomeric units in the presence of a print molecule. The interactions

653

Molecular Imprinting

between complementarity fimctionalities present in the print molecule and monomer@)prior to the initiationof polymerization are conserved in the product polymer after removal of the print species. The recognition site thus formed constitutes a physical memory selective for the print molecule. Two distinct approaches have been developed for the generation of MIPS: one utilizing covalent interactions between print species and monomeric species, pioneered by G. Wulff and K. Shea, and the other employing reversible noncovalent interactions between the two species developed by Mosbach and his co-workers. Although both offer many interesting possibilities, it is the latter which has come to prominence through the diverse range of interactions which may be employed and its ready adaption to different imprint species.

II. IMPRINT PREPARATION AND STUDIES ON MOLECULAR RECOGNITION IN IMPRINTS A. Preparation

Imprint preparation is basically a three-step procedure (Figure 1). The first step is the formation of specific and definable interactions between the monomer(s) and the print molecule. These interactions are responsible for the subsequent recognition of the print molecule by the MIP. The intermolecularinteractions can be either noncovalent bonds, e.g., ionic and hydrogen bonds, or reversible covalent bonds, e.g., esters and ketals. The second step involves co-polymerization of the print molecule-monomer complexes in the presence of a high percentage of cross-linking monomer. Finally, the print molecule is removed from the rigid and insoluble polymer. This process renders a polymer matrix with a series of “imprints” complementary in both shape and chemical functionality to the print molecule. These imprints enable the polymer to later selectively rebind the print molecule from a mixture of closely related compounds. The degree of specificity is reflected in the fact that imprinting against a single enantiomer of a chiral compound endows the polymer with the ability to selectively bind the imprinted enantiomer from a mixture of the optical antipodes (Figure 1). The optimal choice of functional monomers is a consequence of at least three criteria: (1) a stable complex or adduct must be formed between the print molecule and the functional monomer(s) in the pre-polymerization mixture and should endure during the subsequent polymerization phase; (2) it must be possible to quantitatively remove the print molecule from the resultant polymer, and (3) the rebinding reaction should be rapid. These three criteria are to some extent contradictory, as the formation of a stable complex between the print molecule and the functional monomers is required for the creation of the recognition site, whereas the removal of the print molecule and subsequent application,e.g., chromatography, often require weak binding interactions.

654

LARS I. ANDERSON, IAN A. NICHOLLS, and KLAUS MOSBACH

3d

-I-’

+

POLYMERlSAllON

figure 1. Preparation of molecular imprints against the @-blockingagent S-propranolol (1) using rnethacrylic acid (MAA, 2) as the functional monomer and ethylene glycol dimethacrylate (EGDMA, 2 ) as the cross-linking agent.

Of the two imprinting variations, the noncovalent approach is the most readily employed. The interactions (complexes) are formed simply by mixing the print molecule with a suitable mixture of monomers prior to the polymerization. This approach was first successfully employed with dyes (Arshady and Mosbach, 1981). In the majority of the published studies, ionic interactions were extensively utilized, e.g., between amino groups of the print molecule and carboxylic acid groups of the functional monomers (Sellergren et al., 1985, 1988; O’Shannessy et al., 1989a, b, c; Andersson et al., 1990b;Kempe and Mosbach, 1991;Fischer et al., 1991a).More recently, molecular imprints have been prepared using neither covalent nor ionic interactions between functional monomers and the print molecule (Andersson and Mosbach, 1990; Ramstrom et al., 1994). These works have relied upon hydrogen bonding, hydrophobic and dipoldipole interactions,to engender the polymer with cognitive capacity. An additional approach has utilized metal ion chelation effects

Molecular Imprinting

655

to arrange the functional groups within the recognition sites (Fujii et al., 1985; Leonhardt and Mosbach, 1987; Rosatzin et al., 1991; Dhal and Arnold, 1991). Although restricted by the necessity for print molecule functionality compatible with reversible covalent bond formation, covalent imprinting has been utilized to good effect especiallywhen applied to sugar recognition (Wulff, this volume). The use of reversible covalent bonds requires chemically selective excision of the print molecule, which often necessitates drastic conditions, such as treatment with mineral acids at elevated temperatures. An additional problem associated with covalently imprinted polymers lies in the frequentIy slow rates of rebinding of the substrate at the recognition sites. Reported rebinding reactions times have ranged from several hours, to days, though a catalyst can be added to enhance the rate of rebinding for polymers using boronic ester mediated recognition (Wulff and Minarik, 1990).This is contrasted by the ready removal of the print molecule when noncovalent interactions are used for imprinting. Print species may be quantitatively removed simply by solvent extraction under comparatively mild conditions. Additionally, substrate binding is significantlyfaster for imprinted polymers which use noncovalent interactions, making them more compatible for application in many areas such as column chromatography and in sensors. As noncovalent molecular imprints may be made against a great number of molecules, we believe that the noncovalent approach provides more versatility for the preparation of synthetic polymers with preselected affinity. A combination of the two themes will, however, be extremely powerful in certain imprinting applications,e.g., preparation of polymers with enzyme-like capacities (Leonhardtand Mosbach, 1987;Robinson and Mosbach, 1989; Anderson and Mosbach, 1989). B. Recognition in Molecular Imprints

Investigationsinto competitive binding have provided valuable, though indirect, information about the structure of the sites and the mechanisms of binding. In the case of chiral print molecules, enantiomeric separation abilities are recorded. The dissociation constant, Kd, and the number of accessible sites, N, for the binding of a substrate to the chiral recognition sites of a MIP can be estimated by frontal zone analysis in the HPLC-mode. Enantiomers have been found to bind to the same sites, but with different affinities (Kempe and Mosbach, 1991). The strength of binding is highly dependent upon the conditions used; optimization of binding conditions gives Kd-values as low as lo4 M (Vlatakis et al., 1993). The limitations of molecular “size and shape” and chemical complementarityfor recognition have been determined for MIPS made against a number of related L-amino acid aromatic amide derivatives(O’Shannessy et al., 1989b,c; Anderson et al., 1990b). The results are summarized as a proposed model for recognition, schematically depicted in Figure 2. The shape and positioning of the recognition site functionalities(functional complementarity) are defined by the print molecule, e.g., L-phenylalanine anilide (Figure 2A). A schematic representation of enan-

656

LARS I.ANDERSSON, IAN A. NICHOLLS, and KLAUS MOSBACH

A. After imprinting

B. Recognition

figure 2. Molecular recognition anilide (A).

(B) in

an imprint made against L-phenylalanine

tiomeric recognition of a substrate by this imprinted site is shown in Figure 2B. The studies reveal that the amino function must possess the same degree of substitution as the print molecule (arrow a), the amide function must be present (arrow b) and the amino and amide functions must possess the correct spatial geometry around the chiral carbon (arrow c). These studies also show that the substitution and positioning ofthe substrate polar functionalities, which are capable of participating in e.g., hydrogen bonding and ionic forces with polymeric complementary functionalities of the recognition sites, are the most critical for the (enantiomeric) recognition process. In contrast, there is a certain tolerance to the bulkiness of the aliphatic or aromatic (side) groups R, and R2 in this system, as studied by varying the degree of cross-linking. Studies of other print molecules have confirmed these findings (Anderson and Mosbach, 1990; Vlatakis et al., 1993; Ramstrom et al., 1994; Anderson et al., 1994).

111. APPLICATIONS A. Enantiorneric Separations

Studies on the use of MIPs as chiral stationary phases in the high-performance liquid chromatographic (HPLC) mode serve two purposes. First, the analytical and preparative resolution of enantiomers is highly interesting due to the growing need for optically pure drugs and fine chemicals (Taylor and Maher, 1992). Secondly, the analysis of the polymers in the chromatographic mode provides a convenient way of determining the recognition capabilities of MIPs. The mechanical stability of EGDMA-based co-polymer systems (Figure l), the most widely used for imprinting, is sufficient for routine use under HPLC-conditions. The MIPs can be dried under vacuum and stored for at least several years in the dry state without any detectable loss of enantiomeric separation capacity (Anderson, 1991). MIP col-

Molecular Imprinting

657

Table 7. Representative Examples of MIPS Used as Chiral

Stationary Phases in HPLC Separation Factor (a)“

Print Molecule

References

~

N- Acetyl-L-Phe-L-Trp-OMe L-Phenylalanine anilide

17.8 4. I

N-Pyridylmethyl-L-phenylalanineanilide N,N-Dimethyl-L-phenylalanine anilide Cbz-L-aspartic acid Cbz-L-glutamic acid Boc-L-tryptophan Cbz-L-tryptophan methyl ester Cbz-L-tyrosineb Dansyl-L-phenylalanineb L-Tryptophan ethyl ester L-Leucine-P-naphthylamide Timolol Timolol‘ Propranolol Atenolol Metoprolol 1(-)-Mandelic acid R-Phenylsuccinic acidd

8.4 3.1 2.2 2.5 I .9 1.5 4.3 3.2 1.8 3.8 2.9 2.5

1.4 3.6

Ramstrom et al., 1994. Sellergrenet al., 1988; O’Shannessy et al., 1989a,b.c; Andersson et al., 1990b. Anderson et al., 1990b. Anderson et al., 1990b. Anderson & Mosbach, 1990. Anderson & Mosbach, 1990. Anderson & Mosbach, 1990. Anderson & Mosbach, 1990. Ramstrom et al., 1993. Ramstrom et al., 1993. Sellergrenet al.. 1985. Anderson et al., 1990b. Fischer et al., 1991a. Fischer et al., 1991a. Fischer et al., 1991b. Fischer et al., 1991b. Fischer et al., 1991b. Anderson & Mosbach. 1990. Ramstrom et al., 1993.

Notes: aSeparation factor (a) is defined as the ratio ofthe capacity factor for the imprinted enantiomer to the capacity factor for the other enantiomer. bPrepared using methacrylic acid and 2-vinylpyridine as the functional monomers. ‘Prepared using itaconic acid as the functional monomer. dPrepared using 2-vinylpyridine as the functional monomer.

umns have been used up to at least 100 times over 9 month periods (O’Shannessy et al., 1989c) and loadings of up to 400 pg/g polymer have been applied with acceptable resolution (Fischer et al., 199la). Molecular imprints have been prepared against a number of chiral compounds and applied to enantiomeric separations in the chromatographic mode (Table 1). Furthermore, MIPSmade against acetylated aminophenyl-pyranosides have demonstrated a capacity for anomeric differentiation (Mayes et al., 1994). In most instances methacrylic acid was the functional monomer of choice, as the monomer carboxyl functionalitycould interact both ionically and through hydrogen-bonding interactions,with a wide array of functionalities.Molecular imprints in methacrylic acid type polymers have also been successfully prepared utilizing purely non-ionic interactions (Anderson and Mosbach, 1990). Recently, vinylpyridines and

658

LARS I. ANDERSON, IAN A. NICHOLLS, and KLAUS MOSBACH

vinylimidazole, weakly basic monomers, have been introduced and allowed improved complexation with print molecules possessing weakly acidic functionality (Kempe et al., 1993; Ramstrom et al., 1993). More interesting, however, was the demonstration that these monomers were compatiblewith the methacrylic acid type system such that the formation of molecular imprints containing both acidic and basic bctionalities was possible (Ramstrom et al., 1993). In all systems studied, MIPs made against a single enantiomer have been shown to provide efficient resolution of a racemate of the print molecule (Figure 3). In all instances, the imprinted enantiomer of a specific chiral compound was eluted last on a column packed with a MIP made against that enantiomer. In this context, a polymer imprinted against a racemate of the print molecule, did not display any enantiomeric separation.The chromatographic traces in Figure 3B and C illustrate the degree of specificity and selectivity possible in MIPs (Andersson and Mosbach, 1990). A polymer prepared in the presence of Cbz-L-aspartic acid was shown to preferentially recognize and bind the print species relative to its D-configuration antipode (Figure 3B). Further, the polymer was not capable of resolving a racemate of Cbz-glutamic acid. Similarly, imprinting of Cbz-L-glutamic acid led to a polymer with distinct selectivity for the print species over its enantiopode and their homologues (Figure 3C). This result indicated that the recognition site was sufficiently well defined to allow differentiation of structures both adjacent in a homologous series and with the presence of a chiral center. The noncovalent imprinting technology has been successfully applied to the resolution of optically active pharmaceuticals, currently an area of major interest (Taylor and Maher, 1992). A polymer system was developed capable of resolving racemic mixtures of the j3-adrenergic receptor blocking agent timolol and the related P-blockers atenolol, metoprolol and propranolol (Fischer et al., 1991a). Separation of the timolol enantiomers was achieved with a separation factor of 2.9 on a methacrylic acid based MIP made against ( S X - ) timolol. Slightly lower enantio-separation, though greater substrate-selectivity, was obtained using itaconic acid as monomer (a= 2.5). For sufficiently similar substances, enantio-separation is possible on single polymers made against one of the compounds. One such example is an L-phenylalanine anilide MIP, on which the separation of the enantiomers of some amide derivatives, structurally very closely related to the print molecule, has been demonstrated (O’Shannessy et al., 1989~).This concept may prove valuable in situations where the compound(s) needed to be separated are difficult to prepare, or are available in small quantities or only available as a racemate. In such situations, the design of a polymer with a print molecule of similar structure that is readily available would mitigate this limitation. In conclusion, the polymer preparations presented here were all shown to provide efficient separation of the enantiomers of the print molecule (enantio-selectivity) in the HPLC-mode. The separation factors (a) recorded on MIP stationary phases are, on the average, rather high a-values in the range of 4-8, when ionic interac-

Molecular imprinting

659

I

I

I

0

12

24

TIME (rnin)

24t4

C

B

3. Cbr-L-lsplrdc acid 4. Cbr-D-aspdc acid

-

1

1

I

I

1

0

4

8

12

VOLUME h l l

Figure 3. Enantio- and substrate-separations on MIP-columns (packed with molecularly imprinted polymers). (A) Separation of the enantiomers of N,N-dimethyl-phenylalanine anilide on a polymer made against N,N-dimethyl-L-phenylalanine anilide. (6 and C) Separation of a mixture of equal amounts of (1) Cbz-L-aspartic acid, (2) Cbz-D-aspartic acid, (3) Cbz-L-glutamic acid and (4) Cbz-D-glutamic acid on polymers made against (B) Cbz-L-aspartic acid and (C) Cbz-L-glutamic acid.

660

LARS I . ANDERSSON, IAN A. NICHOLLS, and KLAUS MOSBACH

tions are utilized, and in the range of 1.%18, when only non-ionic interactions are employed. These figures compare well with those reported for other commercially available chiral stationary phases (Taylor and Maher, 1992). With the predictable enantiomeric elution order and some improvement in column performance (peak shape) molecularly imprinted chiral stationary phases might, in the very near future, be considered as an adjunct to the Pirkle phases (Pirkle and Pochapsky, 1989). ligand exchange phases (Davankov, 1980), immobilized cyclodextrins (Ward and Armstrong, 1986), cellulose-derived phases (Mannschreck et al., 1985; Blaschke, 1986; Okamoto et al., 1989) and immobilized protein phases (Allenmark, 1986; Hermansson, 1989) presently in use. €3.

Peptide and Protein Separation

The recognition capabilities of MIPs qualify them for consideration as affinity phases for the isolation of high-value peptides and proteins. Recent efforts in this area are very encouraging. The first reported attempt to imprint peptides involved the dipeptide derivative L-phenylalanyl-glycine anilide in conjunction with methacrylic acid as a functional monomer (Andersson et al., 1990b). The resultant polymers were capable of resolving (separation factor 5.1) a racemate of phenyl-. alanyl-glycine anilide. In the case of the dipeptide system N-Ac-Phe-Trp-OMe (Ramstrom et al., 1994), very high separation factors, up to 17.8, were achieved for not only the L-L configuration dipeptide relative to its D-D antipode, but also for the corresponding diastereomeric species. The separation was shown to improve with an increasing number of differing configurations at the chiral centers. The methacrylic acid type polymeric system was successfully applied to the preparation of imprints against the endogenous neuropeptide Leus-enkephalin and some derivatives (Andersson et al., 1995). Evaluation of the MIPs was performed both in the HPLC mode and by ligand binding (MIA) mode (see below). The MIPs expressed selectivity not only for the print molecule (Leu5-enkephalinanilide) but also for free Leu5-enkephalin. Two D-amino acid containing analogues D-Ala2showed only low cross-reactivity. Leu5-enkephalinand D-Ala2-D-Leu5-enkephalin Furthermore, neither of four different tetra- and penta-peptides, with unrelated amino acid sequences, were recognized by the polymer. In preliminary studies, imprints against proteins have been described for the first time (Glad et al., 1985; Kempe et al., 1992). This work employed a mixture of variously functionalized silanes (e.g., amino-, hydroxy-, dodecyl-, phenyl- and boronic) on the surface of silica particle pores (“surface imprinting”). The imprinted polymers showed selectivity for the glycoprotein transferrin (print molecule), relative to blank polymers prepared both with and without BSA as the print molecule to account for non-specific interactions. A specially synthesized boronate silane, added prior to polymerization in order to interact with the glyco-moiety of the protein, most likely accounted for a major part of the recognition observed.

Molecular Imprinting

661

Present investigations include a more thorough characterization of the various parameters involved in the recognition process. Prompted by the protein surface-imprintingdescribed above, imprints have been made against two bis-NAD-derivatives which differed in spacer length between the NAD-moieties (Norrlow et al., 1987). Similarly, the imprinting was accomplished by the polymerization of a mixture of silanes, identical to the system described in connection with the imprinting of proteins. The boronate residues, in particular, facilitated the formation of the recognition sites due to formation of boronate ester linkages with the ribose units of the coenzymes as shown by Wulff (this volume) with other systems. A correlation between the chromatographic elution times for NAD and the two bis-NAD-derivatives was demonstrated. C. Artificial Antibodies and Receptor Mimics

The affinity and selectivity of MIP binding sites can become sufficiently high, to allow their use as antibody mimics (artificial antibodies). Several drug compounds, e.g., theophylline and diazepam, have been investigated as model systems (Vlatakis et al., 1993). Under favorable binding conditions the artificial antibodies show Kd-values as low as lo4 M. Similar cross-reactivity profiles have been demonstrated by artificial antibodies to those of biological antibodies. Theophylline polymers, for example, showed excellent selectivity for theophylline (Figure 4) in the presence of the structurally related compound caffeine (Figure 4). Despite their close resemblance, caffeine showed less than 1% cross-reactivity. The applicability of the artificial antibodies was exemplified by their use in a new ligand binding assay for the accurate determination of drug levels in human serum (Vlatakis et al., 1993). The assay relies on the competition of radiolabeled ligand and analyte for binding to a limited amount of polymer. The fraction of radioligand bound to the polymer is inversely related to the concentration of drug present in the sample. Prior to the actual assay, performed under optimized solvent conditions,the analyte was extracted from the serum using standard protocols. This assay, for which the name Molecularly Imprinted sorbent Assay (MIA) has been proposed, accurately measures drug levels in human serum, with results comparable to those obtained using a well established immunoassay technique, the EMIT (Enzyme Multiplied Immunoassay Technique)(Oellerich, 1980).Data from studies

Thcophyllinc

Caffeine

figure 4. Structures of theophylline and caffeine.

662

LARS I. ANDERSSON, IAN A. NICHOLLS, and KLAUS MOSBACH

with an emphasis on recognition in aqueous systems, using compounds such as opiates, e.g., morphine (Anderson et al., 1995), and biologically active peptides, e.g., enkephalins (Anderson et al., 1995), have been promising, allowing further simplification of the MIA analysis protocol. The results obtained demonstrate the ability to use chemically prepared macromolecules with preselected specificity, instead of the traditional biomolecules, as receptors in competitive binding assays. A great advantage of the MIPs is their cheap, simple and rapid preparation, and their stability. As a great number or low molecular weight compounds can be used as print molecules, there is potential for using molecularly imprinted artificial antibodies as an alternative to biological antibodies. Furthermore, the MIA approach does not involve the use of laboratory animals, nor any material of biological origin. D. Selective Synthesis and Artificial Enzyme Systems

The step from receptor to enzyme in vivo is paralleled by the move from the MIP derived artificial antibodies described previously to polymers showing both substrate specificity and catalytic capacity. This area is fascinating, both in terms of the molecular recognition processes involved and the almost infinite number of potential applications of such systems. Although only a handful of reports have to date appeared, endeavors to produce genuine catalytic systems are being explored by several groups around the world. MIPs can be utilized as both stoichiometric and catalytic reaction mixture components. Although the two are intrinsically linked, it is best to first consider their stoichiometric use in directing reaction outcomes. Preliminary studies have been reported on the use of MIPS for example, regio- and enantioselectivesynthesis (Mosbach et al., 1992). A covalently imprinted system, based upon reversible boronate ester and Schiffs base chemistry was used as a model system for the preparation of the amino acid threonine (Wulff and Vietmeier, 1989).Althoughthe yields were low, chiral induction resulting in enantiomeric excesses of up to 36% were obtained.These studieshave indicated that the selectiverecognitionproperties of MIPs can be effectively used to alter the natural course of a chemical reaction (Mosbach et al., 1992; Bystrom et al., 1993). Early work on the co-immobilization of histamine and octylamine bound to a Sephadex matrix yielded an enzyme-like system capable of the selectivehydrolysis of hydrophobic esters (Nilsson and Mosbach, 1979).This inspired the construction of polymers by molecular imprinting using 4-(5)-vinylimidazole as functional monomer, which during the polymerization was linked to an amino acid derivative template via metal ion chelation (Leonhardt and Mosbach, 1987). Such polymers were shown capable of selectively hydrolyzing amino acid ester substrates related to the print species with a modest turnover number. Perhaps one of the most promising avenues toward artificial enzymes lies in the application of transition state analogues to MIP production in a manner similar to

Molecular Imprinting

663

Figure 5. Hydrolysis of pnitrophenylacetate and the corresponding transition state analogue.

their use for the production of catalytic antibodies (Lerner et al., 1991). In a preliminary study (Robinson and Mosbach, 1989), a polymer was imprinted with p-nitrophenylmethylphosphonate,a transition state analogue of the hydrolysis of p-nitrophenyl acetate (Figure 5). The polymer demonstrated preferential binding of the transition state analogue and induced a fair increase in the rate of hydrolysis of p-nitrophenyl acetate to p-nitrophenol and acetate. This rate enhancement was specifically inhibited by the transition state analogue, providing evidence that the catalysis achieved was a function of supplying specific binding sites by molecular imprinting. Non-peptidic catalytic auxiliaries, enzyme cofactors, have been perceived as a complement to the construction of artificial enzyme systems. Imprinting of the enzyme cofactor pyridoxal has been carried out using a stable analogue of the Schiffs base between pyridoxal and phenylalanine anilide (Anderson and Mosbach, 1989). The MIP was found to modestly enhance the rate of the pyridoxalcatalyzed a-proton exchange on tritium-labeled phenylalanine anilide. More recently, efforts in two laboratories, have led to the development ofpolymer systems showing modest catalytic activity for the @-eliminationof hydrogen fluoride from 4-fluoro-4(p-nitrophenyl)-2-butanone (Figure 6 )(Miiller et al., 1993; Beach and Shea, 1994). The first truly successful attempt to create biocatalyststhat are like artzjkiul enzymes, as characterized by substrate selectivity, rate enhancement and significant site turnover, has recently taken place in Professor Mosbach’s laboratory (Matsui et al., 1996). The potential of enzyme-like MIPSlies not only in their use as mimics ofenzymes present in nature, but also for carrying out reactions either not observed in natural

Figure 6. Elimination of HF from 4-fluoro-4 (pnitrophenyl)-2-butanone.

664

LARS I. ANDERSON, I A N A. NICHOLLS, and KLAUS MOSBACH

systems, or not possible due to the presence of water. With current knowledge we can design and prepare MIPs capable of directing reactivity; both regio- and stereoselective synthetic reactions on unprotected substrates might become possible. The development of MIPS capable of selectivelydirecting carbowarbon bond formation and cleavage, perhaps the pinnacle of potential achievement with such systems, should be made an objective of all those working in the area.

E. Substrate Selective Sensors Substrate-selectivesensory devices, ideally interact specificallywith a predetermined compound or compounds from amongst a complex mixture and provide a signal which may be monitored externally. This concept has been widely established in the area ofbiosensors where a biomolecule, such as an enzyme or antibody, has been used in conjunction with an electronic transducer (See other chapters in this volume) (Figure 7). It was conceived that MIPs may be employed in place of biomolecules. The greater inherent physical and chemical stability of MIPs make them ideally suited for this r81e. In principle, the capacity exists to produce recognition sites for any substrate, thus reinforcing the potential for MIPs in this area. We describe below attempts to apply MIPs to detection systems. Flow through column electrodes have been used for the detection of molecules utilizing streaming potential measurements. These measurements can be regarded as a general method for the detection of binding interactions, though necessitating a change in the charge distribution at the surface at which they occur. The electrode system adopted consisted of a glass column in which the MIP, made against L-phenylalanine anilide, was packed and where the end frits constituted the electrodes (Andersson et al., 1990a). The column resolved the enantiomers of

Output signal

pzs

j

Figure 7. Principle underlying a biosensor configuration.

Molecular Imprinting

665

phenylalanine anilide and the recorded potentiometric signals could be correlated with the respective concentrationsof D- and L-phenylalanine anilide in the sample. In an earlier study, though not strictly within the area of imprinted polymers, the binding of vitamin K1 to the surface of a silicon wafer coated with an octadecylsilane was monitored by ellipsometric detection (Anderson et al., 1988). In this instance, hydrophobic interactions constitute the dominant forces leading to the specific binding of the guest molecule and the transduction of an optical signal and represents the first direct approach to using molecular imprints instead of biomolecules in (bio)sensor devices. More recently, MIPs have been utilized in conjunction with field effect electronic devices; again, the response on binding of a particular molecular species induced a detectable signal (Hedborg et al., 1993). A sandwich polymerization procedure was developed for the fabrication of thin (1-3 pm in thickness)polymer membranes covalently attached to the silicon oxide surface of the capacitor structure. In the most successful configuration, the gate consisted of a porous film of platinum evaporated directly on top of the polymer membrane. MIP membranes, made against L-phenylalanine anilide, were prepared and used as a sensing layer in the field effect capacitors. This sensor measured the compounds phenylalanine anilide and tyrosine anilide in ethanol solution and distinguished them from phenylalaninol. The potential for such devices as “artificial noses” in single and multiple compound determinations appears promising, and would constitute a valuable extension to existing chemical- and bio-sensors.

IV. SUMMARY AND OUTLOOK We have described the historical development and concepts underlying the molecular imprinting technique and the areas in which MIPs have been utilized. Although the concept and initial investigations leading to the development of molecular imprinting have been spread over nearly half a century, it has not been until relatively recently that the full potential of molecularly imprinted polymers has become realized. The utility of the technique is reflected in the diverse array of chemical classes that have been used for preparing MIPs (Table 2 ) and the range of applications in which they have been employed. The use of MIPs as CSPs should prove of great benefit to the pharmaceutical industry in their response to pressure from regulatory authorities for the sole use of enantiomericallypure pharmaceuticals. At present only 25% of all optically active pharmaceuticalsare administered as a single enantiomer, indicating huge scope for development. As specific recognition site structures, MIPs have already demonstrated their ability to accurately mimic the affinity and selectivity of natural antibodies, a major breakthrough for the area. Their cheap and ready production, as compared to the raising of antibodies in living systems, should see them used for the commercial development of new drug and biomolecule assays not requiring the use of laboratory animals, nor restricted to immunologically significant mate-

LARS I. ANDERSSON, IAN A. NICHOLLS, and KLAUS MOSBACH

666

Table 2. Print Molecules Successfully Used for Molecular Imprinting Print Molecule Dyes Amino acid derivatives

Peptides Proteins Carbohydrate derivatives Carbohydrates Aromatic diketones Aromatic bisaldehydes Aromatic bisimidazoles P-Blocking agents (aryloxipropanolamines) Bronchodilators (theophylline) Tranquilizers (benzodiazepines) Opiates (naloxone, morphine) NAD-derivatives Metal ions

References Arshady & Mosbach, I98 I . Sellergrenet al., 1985; O’Shannessy et al., 1989a,b,c; Andersson et al., 1990b; Andersson & Mosbach, 1990. Ramstrom et al., 1994; Andersson et at., 1995. Glad et al., 1985; Kempe et al., 1992. Wulff, this volume. Wulff. this volume. Shea, 1986. Wulff, this volume. Dahl & Arnold, 1991. Fischer et al., 1991a.b Vlatakis et al., 1993. Vlatakis et al., 1993. Andersson et al., 1995. Norrlow et al., 1987. Rosatzin et al., 1991.

rials. These selective recognition characteristics should also prove valuable for combinatorial peptide library screening, immuno affinity style chromatography, environmental clean up operations and as active components in extracorperal shunt systems in clinical detoxification processes, and so forth. The use of molecularly imprinted polymers as substitutes for biomolecules in biosensors is already close at hand. Preliminary studies on the imprinting of large bio-molecules and their aggregates, e.g., large peptides, proteins and enzymes, have already furnished interesting results. MIP derived substrate selective catalytic polymers, or enzyme mimics, have already been reported possessing high substrate specificity, although catalytic rates worthy of direct comparison with naturally occurring enzymes have yet to be achieved. With the full potential of molecular imprinting only just being realized, many tantalizing new applications and developments are sure to be found. It is already apparent that for this potential to be fulfilled, concerted multidisciplinary efforts will be required in areas such as: 1. The development of new polymer systems suitable for imprinting in aqueous media; 2. The improvement of MIP capacities,possibly through the use of alternate or new polymerization protocols;

Molecular Imprinting

667

3. Furthering our understanding of the molecular recognition phenomena underlying the formative and cognitive nature of these polymers. There seems little doubt, with the rapidly growing interest in molecular imprinting, that the discipline will continue to develop and become a major area at the forefront of chemistry and chemical technology.

ACKNOWLEDGMENTS The authors thank Dr. Bjorn Ekberg for valuable discussions during the preparation of this manuscript.

REFERENCES Allenmark, S. (1986). Optical resolution by liquid chromatography on immobilized bovine serum albumin. J. Liq. Chromatogr. 9,425442. Andersson, L.I. (1991). Molecular recognition in synthetic polymers. A study of the preparation and use of molecularly imprinted polymers. Doctoral Thesis, Lund. Anderson, L.I. & Mosbach, K. (1989). Molecular imprinting of the coenzyme substrate analogue N-pyridoxyl-L-phenylalanineanilide. Makromol. Chem., Rapid Commun. 10.49 1-495. Anderson, L.I. & Mosbach, K. (1990). Enantiomeric resolution on molecularly imprinted polymers prepared with only non-covalent and non-ionic interactions. J. Chromatogr. 516.3 13-322. Anderson, L.I., Mandenius, C.F., & Mosbach, K. (1988). Studies on guest selective recognition on an octadecyl silylated silicon surface using ellipsometry. Tetrahedron Lett. 5437-5440. Andersson, L.I., Miyabayashi. A., O’Shannessy, D.J., & Mosbach. K. (1990a). Enantiomeric resolution of amino acid derivatives on molecularly imprinted polymers as monitored by potentiometric measurements. J. Chromatogr. 5 16,323-33 1. Anderson, L.I., Muller, R., Vlatakis, G., & Mosbach, K. (1995). Mimics of the binding sites of opioid receptors obtained by molecular imprinting of enkephalin and morphine. Proc. Natl. Acad. Sci. USA, 47884792. Anderson, L.I., O’Shannessy. D.J., & Mosbach, K. (1990b). Molecular recognition in synthetic polymers. Preparation of chiral stationary phases by molecular imprinting of amino acid amides. J. Chromatogr. 513, 167-179. Arshady, R. & Mosbach, K. (1981). Synthesis of substrate-selective polymers by host-guest polymerization. Makromol. Chem. 182.687492. Beach, J.V. & Shea, K.J. (1994). Designed catalysts. A synthetic network polymer that catalyzes the dehydrofluorination of 4-fluoro-4-p-nitrophenylbutanone.J. Am. Chem. SOC.116,37%380. Bender, M.L. & Komiyama, M. (1978). Cyclodextrin Chemistry. Springer, Berlin. Beckett, A.H. & Andersson, P. (1957). A method for the determination of the configuration of organic molecules using “stereo-selective adsorbents.” Nature 179, 1074-1 075. Bemhard, S.A. (1952). The preparation of specific adsorbants. J. Am. Chem. SOC.74,494fA947. Blaschke. G . (1 986). Chromatographic resolution of chiral drugs on polyamides and cellulose triacetate. J. Liq. Chromatogr. 9,341-368. Breslow, R. (1982). Artificial enzymes. Science 218,532-537. Bystrom, S.E., Boje, A., & Akermark, B. (1993). Selective reduction of steroid 3- and 17-ketones using LiAIH, activated template polymers. J. Am. Chem. SOC.115,2081-2083. Cram. D.J. (1988). The design ofmolecular hosts, guests, and their complexes. Angew. Chem. Int. Ed. Engl. 27, 100W020.

668

LARS 1. ANDERSON, IAN A. NICHOLLS, and KLAUS MOSBACH

Curti, R. & Colombo, U. (1952). Chromatography of steroisomers with “tailor made” compounds. J. Am. Chem. SOC.74,3961. Davankov, V.A. (1980). Resolution of racemates by ligand-exchange chromatography. In: Advances In chromatography. (Giddings, J.C., Grushka, E., Cazes, J., & Brown, P.R., eds.), 18, pp. 13%195. Marcel Dekker, New York. Dhal, P.K. & Amold, F.H. (1991). Template-mediated synthesis of metal-complexing polymers for molecular recognition. J. Am. Chem. SOC.113,7417-7418. Dickey, F.H. (1949). The preparation of specific adsorbents. Proc. Nat. Acad. Sci. USA 35,227-229. Erlenmeyer, H. & Bartels, H. (1964). iiber das problem der ahnlichkeit in der chemie dunnschichtchromatographie mit spezifisch adsorbierenden silikagelen. Helv. Chim. Acta 4 7 , 4 6 5 1. Fischer, L., Muller, R., Ekberg, B., & Mosbach. K. (1991a). Direct enantioseparation of P-adrenergic blockers using a chiral stationary phase prepared by molecular imprinting. J. Am. Chem. SOC.113, 935S9360. Fischer, L., Miiller, R., Ekberg, B., & Mosbach, K. (1991b). Patent application PCT-SE 92/00751. Fujii, Y., Matsutani, K., & Kikuchi, K. (1985). Formation of a specific co-ordination cavity for a chiral amino acid by template synthesis of a polymer Schiff base cobalt (111) complex. J. Chem. SOC., Chem. Commun. 415-417. Glad, M., Norrlow, N., Sellergren, B., Siegbahn, N., & Mosbach, K. (1985). Use of silane monomers for molecular imprinting and enzyme entrapment in polysiloxane-coated porous silica. J. Chromatogr. 347, 11-23. Haldeman, R.G. & Emmett, P.H. (1955). Specific adsorption of alkyl orange dyes on silica gel. J. Phys. Chem. 59, 103%1043. Hedborg, E., Winquist, F., Lundstrom, I., Andersson, L.I., & Mosbach. K. (1993). Some studies of molecularly imprinted polymer membranes in combination with field effect devices. Sensors and Actuators A 37-38, 796799. Hermansson, J. (1989). Enatiomeric separation of drugs and related compounds based on their interaction with al-acid glycoprotein. Trends Anal. Chem. 8,25 1-259. Kempe, M. & Mosbach, K. (1991). Binding studies on substrate- and enantio-selective molecularly imprinted polymers. Anal. Lett. 24, 1137-1145. Kempe, M., Fischer, L.. & Mosbach. K. (1993). Chiral separation using molecularly imprinted heteroaromatic polymers. J. Mol. Recog. 6, 2 5 2 9 . Kempe, M., Glad, M.. & Mosbach, K. (1992). Swed. Pat. Appl. P9102622-9. Lehn, J.-M. (1988). Supramolecular chemistry-Scope and perspectives, molecules, supermolecules, and molecular devices. Angew. Chem. Int. Ed. Engl. 27,8%112. Leonhardt, A. & Mosbach, K. ( 1987). Enzyme-mimicking polymers exhibiting specific substrate binding and catalytic functions. Reactive Polym. 6,285-290. Lemer, R.A., Benkovic, S.J., & Schultz, P.G. (1991). At the crossroads of chemistry and immunology: Catalytic antibodies. Science 252, 65-67, Mannschreck, A., Koller, H., & Wemicke, R. (1985). Microcrystalline cellulose triacetate, a versatile stationary phase for the separation of enantiomers. Kontakte 1 , 4 0 4 8 . Matsui, J., Nicholls, I.A., & Karube, I., & Mosbach, K. (1996). Substrate selective catalytic polymers prepared by molecular imprinting:: An artificial aldolase. J. Org. Chem. 61, in press. Mayes, A., Andersson, L.I., & Mosbach, K. (1994). Sugar binding polymers showing high anomeric and epimeric discrimination obtained by non-covalent molecular imprinting. Anal. Biochem. 222, 48-88, Morrison, J.L., Worsley, M., Shaw, D.R., & Hodgson, G.W. (1959). The nature of the specificity of adsorption of alkyl orange dyes on silica gel. Can. J. Chem. 37, 19861995. Mosbach, K., Nicholls, I.A., & Ramstrom, 0. (1992). Framstiillning av polymerer genom molekylavhyck fdr anvandning vid stereo- och enantioselektiva syntheser styrda primart av icke-kovalenta interaktioner. Swedish Patent Application 9203913-0.

Molecular Imprinting

669

Mosbach, K., Nicholls, LA., & Ramstrom. 0. (1993). The use of molecularly imprinted polymers for stereo- and regio-selective synthesis primarily by non-covalent interactions. PCT/SE93/01107. Mosbach, K. (1994). Trends Biochem. Sci. 19,9-14. Miiller, R., Anderson, L.I., & Mosbach, K. (1993). Molecularly imprinted polymers facilitating a p-elimination reaction. Makromol. Chem., Rapid Commun. 14,637-641. Nilsson, K. & Mosbach, K. (1979) J. Solid-Phase Biochem. 4,271-277. Norrlow, O., MBnsson, M.-O., & Mosbach, K. (1987). Improved chromatography: Prearranged distances between boronate groups by the molecular imprinting approach. J. Chromatogr. 396, 374-377. Norrlow, 0. (1986). Ph.D. Thesis. Oellerich, M. (1980). Enzyme immunoassays in clinical chemistry: Present status and trends. J. Clin. Chem., Clin. Biochem. 18, 197-208. Okamoto, Y., Kaida, Y., Aburatani, R., & Hatada, K. (1989). Optical resolution ofarninoacidderivatives by high-performance liquid chromatography on tris(pheny1carbamate)s of cellulose and amylose. J. Chromatogr. 477,367-376. O’Shannessy, D.J., Ekberg, B., & Mosbach, K. (1989a). Molecular imprinting of amino acid derivatives at low temperature (0 “C) using photolytic homolysis of azobisnitriles. Anal. Biochem. 177, 144149. O’Shannessy, D.J., Ekberg, B., Anderson, L.I., & Mosbach, K. (1989b). Recent advances in the preparation and use of molecularly imprinted polymers for enantiomeric resolution of amino acid derivatives. J. Chromatogr. 470,391-399. O’Shannessy, D.J., Anderson, L.I., & Mosbach, K. (1989c). Molecular recognition in synthetic polymers. Enantiomeric resolution of arnide derivatives of amino acids on molecularly imprinted polymers. J. Mol. Recog. 2, 1-5. Pauling, L. (1940). A theory of the structure and process of formation of antibodies. J. Am. Chem. SOC. 62,2643-2657. Pedersen, C.J. (1988). The discovery of crown ethers. Angew. Chem. Int. Ed. Engl. 27, 1021-1027. Pirkle, W.H. & Pochapsky, T.C. (1989). Considerations of chiral recognition relevant to the liquid chromatographic separation of enantiomers. Chem. Rev. 89,347-362. Ramstram, O., Anderson, L.I., & Mosbach, K. (1993). Recognition sites combining different chemical functionalities prepared by molecular imprinting. J. Org. Chem. 58, 7562-7564. Ramstrom, O., & Mosbach, K. (1996). The emerging technique of molecular imprinting and its future impact on biotechnology. Bio/Technology. 14, 163-1 69. Ramstrom, O., Nicholls. I.A., & Mosbach, K. (1994). Synthetic peptide receptor mimics: Highly stereoselective recognition in non-covalent molecularly imprinted polymers. Tetrahedron: Asymmetry 5,649456. Robinson, D.K. & Mosbach K. (1989). Molecular imprinting of a transition state analogue leads to a polymer exhibiting esterolytic activity. J. Chem. SOC.,Chem. Commun. 969-970. Rosatzin, T., Anderson, L.I., Simon, W., & Mosbach, K. (1991). Preparation of Ca” selective sorbents by molecular imprinting using polymerisable ionophores. J. Chem. Soc.,Perkin Trans. 2 12611265. Sellergren, B., Ekberg, B., & Mosbach, K. (1985). Molecular imprinting of amino acid derivatives in macroporous polymers. Demonstration of substrate- and enantio-selectivity by chromatographic resolution of racemic mixtures of amino acid derivatives. J. Chromatogr. 347, 1-40. Sellergren, B., Lepisto, M., & Mosbach, K. (1988). Highly enantioselective and substrate-selective polymers obtained by molecular imprinting utilizing non-covalent interactions. NMR and chromatographic studies on the nature of recognition. J. Am. Chem. Soc. 110,5853-5860. Shea, K.J. & Dougherty, T.K. (1986). Molecular recognition on synthetic amorphous surfaces. The influence of functional group positioning on the effectiveness of molecular recognition. J. Am. Chem. Soc. 108, 1091-1093.

670

LARS I . ANDERSSON, IAN A. NICHOLLS, and KLAUS MOSBACH

Tabushi, 1. (1986). Chiral selection and chiral induction by the use of regiospecifically di-(or poly)substituted cyclodextrins. Pure & Appl. Chem. 58, 152W534. Taylor, D.R. & Maher, K. (1992). Chiral separations by high-performance liquid chromatography. J . Chromatogr. Sci. 30, 67-85. Vlatakis, G., Anderson, L.I., Miiller, R., & Mosbach, K. (1993). Drug assay using antibody mimics made by molecular imprinting. Nature 361, 645447. Vlatakis, G., Anderson, L.I., & Mosbach, K. (1994). Manuscript in preparation. Ward, T.J. & Armstrong, D.W. (1986). Improved cyclodextrin chiral phases: A comparison and review. J. Liq. Chromatogr. 9,407423. Wulff, G., & Vietmeier, J. (1989). Enantioselective synthesis of amino acids using polymers possessing chiral cavities obtained by an imprinting procedure with template molecules. Makromol. Chem. 190, 1727-1735. Wulff, G., & Minarik, M. (1990). Template imprinted polymers for HPLC separation of racemates. J. Liq. Chromatogr. 13,2987-3000.

AN APPROACH TOWARD THE SEMIQUANTITATION OF MOLECULAR RECOGNITION PHENOMENA IN NONCOVALENT MOLECULARLY IMPRINTED POLYMER SYSTEMS: CONSEQUENCES FOR MOLECULARLY IMPRINTED POLYMER DESIGN

Ian A. Nicholls

Abstract

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

I. INTRODUCTION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. DISCUSSION.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Advances in Molecular and Cell Biology Volume 15B, pages 671-679. Copyright 0 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 0-7623-0114-7

671

672 672 675 677

IAN A. NICHOLLS

672

ABSTRACT Noncovalently imprinted polymer systems contain recognition sites defined by complementary interactions between the imprint, or template, species, and the polymer matrix. The specificity of these sites for their imprint molecule are comparable to those observed in biological systems. These systems should prove valuable for the study of the factors governing molecular recognition phenomena. A previously described approach to the semiquantitative analysis of reversible noncovalent bimolecular association ofpeptides appears suitable for the study of these systems. The adaption of this technique to the study of molecular recognition in imprinted polymer systems will be discussed, along with implications for the design of molecularly imprinted polymer systems.

1. INTRODUCTION Understanding the energetic factors contributing to the stabilization of reversible noncovalent interactions should provide valuable insight into the nature of molecular recognition phenomena. Extending and building on the work of Jencks (Page and Jencks, 1971; Jencks, 1978), both Andrews et al. (1984) and Williams and colleagues (Williams et al., 1990; 1991) independentlyproposed factorizations of the energetic contributions to ligand-receptor interactions. In the first case, the approach was developed without the use of detailed receptor site knowledge, whereas the second utilized a well defined ligand-receptor system (the complex formed between members of the vancomycin class of antibiotics and bacterial cell wall mucopeptide precursor analogues) for the evaluation of component energetic contributions.Williams et al. (1991) presented a general expression for the estimation of binding constants for bimolecular associations in aqueous solution (Equation l). They concluded that for systems in which the complex formed between the two species displays good molecular complementarity and if the ligand-receptor interaction process takes place with each component near its global minimum energy conformation, the expression may be simplified to the form shown in Equation 2.

AGbind= AG,,, + AGr + AG, + ZAG,

(2)

Equations 1 and 2: Where AGbindis the Gibbs free energy change involved in formation of a complex; AG,,, the sum of changes in translational and rotational free energies; AGr, the energy change resulting from restriction of rotors upon complexation; AGh, the free energy change due to the hydrophobic effect; ZAG,, the intrinsic free energies of binding for each set of interactinggroups summed over all polar interactions (residual vibrational modes are encompassed by this term); AG,,, the free energy contribution resulting from adverse conformational

Semiquantitation of Molecular Recognition Phenomena

673

changes; and AG,,, the free energy penalty resulting from unfavorable van der Waals interactions. The principles developed from the factorization approach were successfully applied to the prediction of binding constants in aqueous solution (Williams et a]., 199l), to the determinationof a number of biologically significantintrinsic binding energies (Cox et al., 1991; Williams, 1991; Williams et al., 1991), and to improve understanding of other terms in Equation 1 (Searle and Williams, 1992). By comparison with binding energies for closely related chemical structures (with knowledge of the free energy contributions resulting from changes in translational and rotational free energies, internal rotational,and hydrophobic effect free energies and granted the two assumptions with respect to complementarity and binding conformations), it is possible to determine the free energy change due to the contribution of a particular polar functional group interaction,the intrinsic binding energy AGp. It should be noted that soft vibrational modes present in the complex are attributed to the polar term, which yields a higher value for a given interaction than is the case if these are factored out or attributed to the AG,,, term. The molecular recognition capabilities of molecularly imprinted (or template polymerized) polymers (Mosbach, 1994) have found use in a diverse array of applications: antibody combining site mimics (Vlatakis et al., 1993), chiral chromatographic stationary phases (Fischer et al., 1991; Ramstrom et al., 1994), mediators of organic synthesis (Mosbach et al., 1992),enzyme mimics (Leonhardt and Mosbach, 1987; Matsui et al., 1996), and in electronic-sensortype configurations (Hedborg et al., 1993). To date, two types of polymer systems have been developed, the first utilizing reversible covalent interactions between the print species and polymer, covalent imprinting (Wulff, 1986), and the second relying upon noncovalent interactions to define the recognition site in the polymer, noncovalent imprinting (Anderson et al., 1993; 1994). The latter noncovalent approach has proven the more versatile as a direct consequence of the types of interactions involved in the recognition process and is potentially amenable to evaluationusing the approach described above. The principles underlying the preparation of noncovalent molecularly imprinted polymer systems have been described elsewhere (Anderson et al., 1993; 1996)and are summarized in Figure 1. A monomer with chemical functionality complementary to that of the imprint molecule is mixed with the imprint species in the presence of a suitable cross-linking agent. The complementarily interacting functionalities form predictable solution structures (Sellergren et al., 1988),which after polymerization and extraction of the template species, yield a recognition site of complementary steric and functional topography to the imprint molecule. The resultant polymer recognition sites possess a cognitive capacity sufficient for the selective rebinding of the imprint species from a range of closely related structures. The possibility ofproducing recognition sites ofpredetermined selectivitymakes molecularly imprinted polymer systems interesting candidates for hndamental mdies in molecular recognition and for investigation of the polymers themselves

P

.

Polymerisation

\

Print molecule

Extraction

n

Incubation with print molecule

. Figure 1. Schematic representation of the molecular imprinting process, where A and X, and B and Y are sets of complementarily interacting functional groups.

Semiquantitation of Molecular Recognition Phenomena

675

using the semiquantitativeapproach described above. Molecularly imprinted polymer recognition site populations are in many ways analogousto those ofpolyclonal antibody samples with a distributionof binding modes and affinities. Accordingly, the benefits of generating sites of higher affinity and enhanced regularity will yield more selective recognition (c.f. a high affinity monoclonal and polyclonal antibody). These issues must be addressed in order to see major improvements in the many molecular imprinting applications currently being developed.

II. DISCUSSION The two assumptions necessary to apply the approach of Williams et al. (1991) to the study of liganck-eceptor systems using Equation 2 are, firstly, that conformational strain is not introduced into either component upon complex formation and, secondly, that good molecular complementarity is achieved in the complex. With respect to the first assumption, previously reported solution phase NMR studies identified solution structuresinvolving the coordination of the functional monomer by polar sites in the template species (Sellergren et al., 1988). By virtue of the fact that the final recognition site topography reflects that of the prearrangement solution adduct, and granted that polymerization and rebinding take place in the same solvent, no significant conformational compromise is necessary for rebinding at the polymer recognition site. In reference to the second assumption, a wealth of recent data covering highly refined levels of ligand differentiation has highlighted the degree of molecular complementarity inherent in these systems (Ramstrom et al., 1994). The consequences of fulfilling the complementarity assumption are reflected in the sensitivity of noncovalently imprinted polymers to subtle differences in ligand electronicand steric effects, as demonstrated by polymers imprinted against a variety ofpeptides (Ramstrom et al., 1994) and pharmaceuticals (Vlatakis et al., 1993; Fischer et al., 1991). Granted the assumptions discussed above, Equation 2 may be applied to noncovalently imprinted polymer systems. Careful consideration must, however, be given to the other terms and their relevance to molecularly imprinted polymer recognition systems. For the binding of a ligand to a polymer-bound receptor, three degrees of translational free energy and three degrees of rotational free energy will be lost. This energetic penalty is calculated using the log relationship between AG,,, and molecular weight as described elsewhere (Williams et al., 1991). For many imprinted polymer systems, the imprinting and rebinding experiments are conducted in nonpolar solvents; in such cases the contribution of the hydrophobic type dielectric effects, AG,,, are minimal, though attempts to account for different dielectrics within the polymer and in bulk solution may prove useful. The energetic penalty associated with the freezing of a rotor, AGr, has been the subject of some conjecture, with system-dependent values in the range 1.6 to 9 kJ mol-' (TAS at 300 OK)having been cited (Williams, 1991; Searle and Williams, 1992).A value of 5 to 6 kJ mol-' is currently considered a reasonable estimate. The remaining term, the sum of the polar group

676

IAN A. NICHOLLS

intrinsic binding energies, ZAG,, is, therefore, the only significant term favorable to binding in these systems. It should be noted that in this treatment, the intrinsic binding energy, AG,, for a particular interaction in a given system is a direct measurement of that individual interaction. To apply this thermodynamic factorization to a functionally simple, though nonhomogeneous, receptor site population like that found in molecularly imprinted polymers, the intrinsic binding energy, AG,, calculated for a given interaction is a measure of the average strength of that interaction throughout the polymer matrix under investigation. This figure will, therefore, encompass the contribution of nonspecific binding modes, for example, hydrophilic interactions between ligands and functional monomer residues randomly oriented in the polymer matrix. For binding in polymer systems prepared and operating in a nonpolar environment, assuming negligible hydrophobic contribution, Equation 2 may be simplified as shown in Equation 3.

With the recent development of methods such as the molecularly imprinted sorbent assay (MIA) (Vlatakis et al., 1993) and solvent frontal zone analysis (Kempe and Mosbach, 1991) for the determination of binding constants for ligand interactions with molecularly imprinted polymer recognition sites, suitable data sets should become available for application of this approach to the study of individual terms and their contribution to binding. For example, the study of differences in the free energy of binding between closely related systems, AGbind, should yield individual functional group contributionsto binding, AG, terms. These systems should also prove useful tools for the investigation of rotor and translational and rotational free energy terms, AG, and AG,,, respectively. This will, in turn,provide a better understanding of the recognition mechanisms in molecularly imprinted polymers and improved imprinted polymer system design. It should be stressed that the numerical values obtained by such an approach will possess inherently large errors due to the uncertainties associated with the parameters employed,hence the use ofthe term semiquantitative.This fact does not detract from the approach for, as shown by Williams et al. (1991), it is the order of magnitude of the values that is significantwhen comparing the energetic contributions. The ramifications of Equation 3 for the design of new imprinted polymer systems are manifold. Firstly, for the development of high affinity systems, the impact of the entropically unfavorable freezing of rotors, the AG, term, can be minimalized through the selection of less flexible imprint species. Amore rigid imprint structure will yield a better defined solution adduct through not having to contend with as many conformation distributions. This will, in turn,result in higher imprint molecule selectivity due to improved average geometric fit. An additional benefit

Semiqoantitation of Molecular Recognition Phenomena

677

resulting from increasing the number of functional monomers involved in defining iinprint sites is a reduction of nonspecific interactions arising from random functional monomer residue orientation throughout the bulk polymer. Consequently, the use of a more rigid imprint species should lead to polymers with improved affinity and selectivity. Such a working hypothesis, namely the use of rotor restricted ligands, was successfully applied to the rational design and study of ligands with enhanced affinity and efficacy for biological receptors (for a recent comprehensive review see Giannis and Kolter, 1993; for recent examples see Nicholls, 1990; Nicholls et al., 1990; 1993a; 1993b; 1994; Nicholls and Alewood, 1994). Recent reports (Vlatakis et al., 1993; 1994) ofpolymers imprinted with relatively rigid structures, such as opioid alkaloids and xanthine derivatives, support the argument presented above. These systems showed both very low dissociation constants and high selectivity for the rebinding of the print species. Importantly, polymers selective for opioid peptides, for example [Leu’]-enkephalin, demonstrated consistently lower binding aflfinities (Vlatakis et al., 1994) than polymers specific for opioid alkaloids, for example morphine. The opioid alkaloid’s rigidity must contribute significantly to the superior binding because relatively few polar binding features are present in morphine capable of contributing to the favorable ZAG, term. The choice of imprint molecules with polar functionalities most compatible with those present in the functional monomer will further enhance the contribution from the ZAG, term. Thus, an astute selection of both imprint species, functional and cross-linking monomers and solvent, is necessary to produce systems displayingimproved recognition characteristics. Furthermore, it is critical that monomers be selected to ensure that functionality present to complement that of the print species is compatible with subsequent rebinding conditions. Finally, for systems that will be employed as synthesis mediators, the polymer functionality must not react adversely with any reagents subsequently employed.

111. CONCLUSIONS Molecularly imprinted polymers provide a potentially valuable tool for the study of molecular recognition phenomena. The combination of molecular imprinting with a suitably adapted semiquantitative analysis method, such as that described here, should allow the study of thermodynamic factors contributing to recognition. On the basis of recent results comparing molecularly imprinted polymer antibody mimics to biologically derived antibodies, the application of such an approach to imprinted polymers should prove useful for the study of biologically significant interactions. Finally, the principles enunciated here should prove useful for a more rational design of highly selective imprinted polymer systems.

678

IAN A. NICHOLLS

ACKNOWLEDGMENT This manuscript is dedicated t o my friend and mentor Professor Klaus Mosbach o n the occasion of his 60th birthday.

REFERENCES Andersson, L.I., Nicholls, LA., & Mosbach. K. (1993). Molecular i m p r i n t i n e a versatile technique for the preparation of separation materials of predetermined selectivity. In: Separations in Biotechnology (Street, G., Ed.), in press, Chapman-Hall, London. Andersson, L.I., Nicholls, LA., & Mosbach, K. (1996). Molecular imprinting. In: this volume. Andrews, P.R., Craik, D.J., & Martin, J.L. (1984). Functional group contributions to drug receptor interactions. J. Med. Chem. 27, 164W657. Cox, J.P.L., Nicholls, LA., & Williams, D.H. (1991). Molecular recognition in aqueous solution: An estimate of the intrinsic binding energy of an amide-hydroxyl hydrogen bond. J. Chem. Soc., Chem. Commun. 12951297. Fischer, L., Muller, R., Ekberg, B., & Mosbach, K. (1991). Direct enantioseparation of P-adrenergic blockers using a chiral stationary phase prepared by molecular imprinting. J. Am. Chem. SOC.I 13. 935g9360. Giannis, A. & Kolter, T. (1993). Peptidomimetics for receptor IigandsDiscovery, development, and medical properties. Angew. Chemie Int. Ed. Engl. 32, 1244-1267. Hedborg, E., Winquist, F, Lundstom, I., Andersson, L.I.. & Mosbach, K. (1993). Some studies on molecularly imprinted polymers in combination with field effect devices. Sensors and Actuators A 37-38,76%799. Jencks, W.P. (1978). On the attribution and additivity of binding energies. Proc. Natl. Acad. Sci. USA 78,4046-4050. Kempe, M. & Mosbach, K. (1991). Binding studies on substrate- and enantio-selective molecularly imprinted polymers. Anal. Lett. 24. 1137-1 145. Leonhardt, A. & Mosbach, K. (1987). Enzyme mimicking polymers exhibiting specific substratebinding and catalytic functions. Reactive Polymers 6.285-290. Matsui, J.. Nicholls, LA., Karube, I., & Mosbach, K. (1996). Carbon-carbon bond formation using substrate selective catalytic polymers prepared by molecular imprinting.: An artificial aldolase. J.. Org. Chem. 61, in press. Mosbach, K. (1994). Molecular imprinting. Trends Biochem. Sci. 14, S l 4 . Mosbach, K., Nicholls, LA., & Ramstrom, 0. (1992). Framsttillning av polymerer genom molekylavtryck fdr anvanding vid stereo- och enantioselektiva synteser styrda primart av icke-kovalenta interaktioner. Swedish Patent Application 92039 13-0, Worldwide patent pending. Nicholls, I.A. (1990). The design synthesis and biological evaluation of novel central nervous system active agents. Ph.D. thesis, University of Melbourne, Australia. Nicholls, LA. & Alewood, P.F. (1994). Design, synthesis and opioid receptor binding of some novel benzazepine constrained leucine enkephalin mimetics. I. Chem. Research, in press. Nicholls, LA., Alewood, P.F., & Andrews, P.R. (1990). Rational CNS drug design - potential antihypertensive agents. Aust. J. Hosp. Pharm. 20,334-338. Nicholls, LA., Alewood, P.F., Brinkworth, R.I., Morrison, S.F., &Andrews, P.R. (1993a). 2-substituted 1,3-benzodiazocines: Design, synthesis and evaluation as potential central nervous system active agents. J. Chem. Research (M) 281 1-2826, (S) 4 0 W 0 9 . Nicholls, LA., Craik, D.J., & Alewood, P.F. (1994). 'H-NMR and molecular modeling based conformational analysis of some N-alkyl- l- and 2-benzazepinones: Useful central nervous system agent design motifs. Biochem. Biophys. Res. Commun. 205,98-104.

Semiquantitation of Molecular Recognition Phenomena

679

Nicholls, LA., Morrison, S.F., Brinkworth, R.I., Alewood, P.F., & Andrews, P.R. (1993b). Central nervous system receptor binding profiles of some 2-amino-4-phenyl quinolines: A novel class of a*-adrenoceptor selective ligands. Life Sci. 53, PL34L347. Page, M.I. & Jencks, W.P. (1971). Entropic contributions to rate accelerations in enzymic and intramolecular reactions and the chelate effect. Proc. Natl. Acad. Sci. USA 68. 1678-1683. Ramstrom, O., Nicholls, LA., & Mosbach, K. (1994). Synthetic peptide receptor mimics: Highly stereoselective recognition in non-covalent molecularly imprinted polymers. Tetrahedron: Asymmetry 5, in press. Searle, M.S. & Williams, D.H. (1992). The cost ofconformational order: Entropy changes inmolecular associations. J. Am. Chem. SOC.114, 10690-10697. Sellergren, B., Lepisto, M., & Mosbach, K. (1988). Highly enantioselective and substrate selective polymers obtained by molecular imprinting utilizing non-covalent interactions. NMR and chromatographic studies on the nature of recognition. J. Am. Chem. SOC.110,585>5860. Vlatakis, G., Anderson, L.I., Miiller, R., & Mosbach, K. (1993). Drug assay using antibody mimics made by molecular imprinting. Nature (London) 361,645-647. Vlatakis, G., Miiller, R., Anderson, L.I., & Mosbach, K. (1994). Unpublished results. Williams, D.H.(1991). The molecular basis of biological order. Aldrichimica Acta 24, 71-80. Williams, D.H., Cox, J.P.L., Doig, A.J., Gerhard, U., Kaye, P.T., Lal, A.R.. Nicholls, LA., Salter, C.J., & Mitchell, R.C. (1991). Towards the semiquantitative estimation of binding constants. Guides for peptide-peptide binding in aqueous solution. J. Am. Chem. SOC.113,7020-7030. Williams, D.H., Doig, A.J., Cox, J.P.L., Nicholls. LA., & Gardner, M. (1990). Molecular basis of the activity of antibiotics of the vancomycin group: Guides for peptide-peptide binding. In: Chirality in Drug Design and Synthesis (Brown, C., Ed.), pp. 101-113. Academic Press, London. Wulff, G.I. (1 986). Molecular recognition in polymers prepared with templates. In: Polymeric Reagents and Catalysts (Ford, W.T., Ed.), pp. 18&230. ACS symposium series 308, American Chemical Society, Washington DC.

This Page Intentionally Left Blank

INDEX

Abzymes see; Catalytic antibodies Acetylcholinesterase (AChE), 25 Acetyldimethyl(phenyl)silane, reduction of, 70 Acylsilane 4, 70-73 Affinity adsorbents, 606-607 Affinity chromatography, 156, 51 15 14,604-605 adsorbent design and construction, 515-517 analytical affinity chromatography, 553-554 immobilized metal ion affinity chromatography, 521-523 modern requirements for, 5 14-5 15 process development, 5 17-519 upstream engineering, 515 Affinity partitioning, 541-542 separation and biorecognition, 507-509 temperature-induced phase separation and, 542-543 Agarose, superporous for chromatography, 583-586 Akrilex C, 2 1 1 Alcohol oxidate (AOD), 436-438 Alcohol sensor, 423,432-436 Aldehyde derivative, 167-168 Aldolase, 2 12 Aldose sensor, 422-423,430-432 Alkaline phosphatase, 465

Alphachymotrypsin, 1 11-1 13 Alumina silicate, hydroxylapatite, 248 Aminoacylase, 212 immobilized for optical resolution, 150-151 Amphiphatic gels, 599-601 Animal cell production technology, 195,200 high cell density and, 200 long term cell maintenance, 20 1-206 physiological features, 196-200 Antibodies, 184 anti-idiotypic antibodies, internal imagery of, 22-28 artificial antibodies, 657-658 ligand-based approaches, 22-24 thiophilic adsorption and, 545-549 transition state analogs, 23 Apoenzyme, biosensing of heavy metal ions and, 459460,465469 Aquaphilicity parameter (Aq), 9 ArsanylazoCPA derivatives, 36 Arthrobacter simplex, 144 Artificial seed system, 95-96 Ascorbate oxidase, 466 Aspergillus niger, 53 Aspoxicillin, 155 Back-propagation neural network algorithm, 4849 Bacterial luciferace, 302-303 68 1

682

Batch ligninase production, 242-244 Bed reactors, fluidized, 200, 205 Benzene oxidation, 144 Benzoate derivatives, 330 Beta-galactosidase, 4, 5 1-52, 249, 300 water-miscible organic solvents, 108 Biocatalysts, 239, 253 immobilization of, 135-139, 4 10, 606-607; see also enzyme immobilization Biochemical engineering, 45, 197, 564 history of, 4, 116-121 Bioimprinting, 3, 15-18 Biological specificity, definition of, 623-624 Biomolecules, molecular recognition and, 352-353 Biosensors, 347-350, 363-364, 559564; see also Recycling sensors alcohol sensor, 423, 432-436 aldose sensor, 422-423, 430-432 analytical chromatography and, 6 10-6 I2 application of, 359-360 chemical signals transfer to transducer, 354-356 D-amino acid sensor, 423-425, 438-443 L-blutamate sensor, 444-445 L-lactate sensor, 422,425-430 micromachining and, 373-374 miniaturized thermal biosensors, 497-503 new areas of research, 364-367 optimization of immobilization/ transducers, 353-354 organic phase biosensors, 356-359 status of biosensor sales and research, 368-370

INDEX

technology overview, 35 1-352 thermal biosensor, 489-49 1 Biotransformation, 68, 139 Bis-diazotized silica, 167 Bovine serum albumin, 19 Brevibacterium ammoniagenes, Lmalic acid and, 153 Broad host range vector, 96 Calvin cycle enzymes, 274-276 Capillary gas chromatography, 72 Carbon dioxide fixation, 96 Carbon paste electrodes, 42 1-422 Carbon-carbon bond synthesis, 145 Carbonic anhydrase, 467 Carboxylic groups, activation of, 21 1 Carboxypeptidase A arsanylation of, 33-36 optical properties, 37-42 modulation of catalytic pathway, 32-33 polyvinyl alcohol conjugates and, 33-35 Carboxypeptidase B, 212 Carrageenan, 152 Catalytic antibodies (abzymes), 3, 21-24 therapeutic applications, 27 Cephalosporin C, 288 Chelate mediated immobilization, 247-249 bioorganic application, 254 materials and methods, 249-250 Chemiluminescence detector, enzymatic reactor and, 375-376 Chinese hamster ovary (CHO) cells, 194 recombinant CHO, 197 Chloramphenicol acetyltransferase, 97 Chromatium, 98

Index

Chromatography chromatographic experiments, 588-590 chromatographic gels, 102 superporous agarose and, 583-586 Chymotrypsin, 3 bioimprinting of, 15-18 mobility of polypeptide chains, 18 Citation analysis, 127 Citrate synthase (CS), 264, 298 Conjugation (as method for gene transfer), 96 Conserved orientation transfer, 265 13 C NMR techniques, 268-270 Counter current phase partitioning, 276 Coupling efficiency, 178-179 Cross-linking agents, 20 Cryptic plasmid, 96 Cyanobacteria, 94-95 Cyanogen bromide, 166-167 Cyanophage, 97 Cyclodextrin glycosyltransferase (CGTase), 212, 522 D-amino acid sensor, 423-425 D-aspartic acid, 154-155 Dalziel theory, 219 Dehalogenation, 146 Denaturants, 109,223 Dextran sucrase, 120 mathematical modeling of kinetics, 121-127 DH FR see; Dihyrofolate reductase Diazotization, 172-174 Diffusion distance, 605 Dihydrofolate reductase (DHFR) characterization of, 88 effects of sulfate and chlorine on enzymic activity/ stability, 84-88 refolding process, 78, 82-83 two-step refolding, 83-84

683

DnaK complex formation, recombinant proteins and, 3 12-315 Electrochemical flow cell, 374 enzyme immobilized column and, 374-315 Electron connections, and redox enzymes, 389-391 Electrophoresis, 529 Embryogenesis, 95 Enantiomeric separations, 652-656 Energy of maintenance (Em), 197I99 Enterostatin, 340 Enzymatic reactor, chemiluminescence detector and, 375-376 Enzyme activities, determination of, 212-213 Enzyme based diffusion badge formaldehyde and, 449-450 principles of operation, 450-452 production of, 455-456 Enzyme electrodes, 382-383 construction of, 420-421 Enzyme engineering, 101-106 Enzyme immobilized column, electrochemical flow cell and, 374-375 Enzyme optode, 383-384 Enzyme organization, 280-282 Enzyme support material, 161-164 Enzyme thermistor, 386,410-413 Enzyme-coenzyme-substrate reactions, 218 Enzymes artificial, 658 binding sites and, 635-639 chemical model systems, 298-299 covalent binding, 22 entropy and geometric effects, 22 enzyme columns, 314-375,491-493 enzyme kinetics (Km),181- 182, 216-219, 232

684

fusion enzymes, 299-304 in homogeneous solutions, 11-13 immobilization of, 4-7, 32, 119, 230-23 1,247-250,298 catalytic properties, 213-220 characteristics, 103-104, 180-181 inorganic supports and, 160164, 177-182 technology development, 140142, 182-184 microenvironment and, 5-7, 22 neural networks and, 45-46 organization in the cell, 296-298 properties of enzymatic reaction, 22 role of polarizing groups, 22 solid support and, 7-1 1 solubilization in microemulsions, 12, 83 solvatation substitution, 22 specificity and, 22 stabilization of enzymes, 453 structural complementarity of active site, 22 structure-activity relationship, 27 three-enzyme system, 301-304 water-poor media and, 5-7 Enzymic flow (micro)calorimetry (EFMC), 410-416 Escherichia coli, 58-60, 78, 83, 96 chaperone DnaK, 309-3 1 1 L-aspartic acid produced by immobilized E. coli, 151-153 production of native DnaK, 3 I I 312 Ethanol, 156 Expert systems, 46 Fatty acids, 96 Fermentations, split-flow modified thermal biosensor and, 489493

INDEX

Flavans, 230 Flavoenzyme redox centers, electron connections in, 394-399 Fluorescence polarization, 275 Fluorescense spectroscopy, 82-84 Flurohydrolase, 20 Flux-Control connectivity/ summation theorem, 283 Foreign gene expression, 97 Formaldehyde, 449-450 Fusion enzymes, 299-304 analytical applications, 304-306 Fuzzy logic, 46, 118 GI (GO) phase, 197 Galactose oxidase, 467 Gel electrophoresis, 82 Gel filtration chromatography, 82 Gene fusion, 296 Genetic engineering, 96, 280 Glucoamylase, 4, 51-53, 211, 250253 Glucose isomerase, 119, 142 Glucose-phosphate isomerase, 212 Glutamic acid, 96 Glyceraldehyde-3-phosphate dehydrogenase, 273-274 Glycogen synthesis, 288-293 Green algae, 94-95 Guanidinium hydrochloride, 223 Haptens, 23-24 Heat shock, 309 Heavy metal ions biosensing and, 459-460 procedure for biosensing, 464-465 Hemeprotein peroxidases, electrical connection, 401-402 Hemoglobin, 287-289 Heterobifunctional reagents, 247 Hexokinase, 212, 298 High cell density, 200

685

Index

High-annealing-temperature (HAT) PCR primers, 471-474 amplification protocol design, 476, 480-482 detection of HCMV, 482-484 materials and methods, 474-478 primer design, 478 Histidine, immobilized, 156-157 Horse liver alcohol dehydrogenase (HLADH), 9-1 1 Hsp70,310 Human immunodeficiency virus

(HI V), 555-559 Human interleukin 5 , 564 Hydrolysis, subzero temperature and, 12-13 Hydrophilic gels, 596-599 Hydroxybisphosphonic acids, 248 Idiotypic network, 24 IgG affinity chromatography, 309 Immobilization of cells macroporous matrix and, 201-206 optimization of, 353 Immobilized biocatalysts (IMB), 239,253,410 characterization of, 606-607 mathematical model, 41 1-413 Immobilized enzymes; see Enzymes, immobilized Immobilized ligands, recognition analysis and, 559-564 Immobilized metal ion affinity chromatography (IMAC), 522525 Cu(I1)-IDA and, 525-529 materials and methods, 523-525 recombinant cyclodextrin glycosyl transferase and, 521-525 Immune system, antibodies and, 2223 Immunological sensors, 364-367 Imprinted polymer systems, 668-671

Imprinting; see Molecular imprinting Inorganic carriers, 248 Inorganic support material capacity (enzyme immobilization), 162-163 chemical durability, 162 coupling efficiency, 178-179 pH profile, 179-180 pore size, 161 preparation of biocatalyst supports, 164-169

technology, 182-184 Interesterification of fats, 143 Ion exchange chromatography, 310 Km, 181-182, 216-219, 232 Knowledge-based systems, 46,54 Krebs citric acid cycle, 263-265 Krebs tricarboxylic acid cycle (TCA) enzymes, 264 isolated enzymes and mitochondrial membranes, 266-267 TCA cycle metabolon, 265-267 yeast cell mutants, 267-268 L-alanine, 153-154 L-amino acids, 150 L-aspartic acid, 15 I L-malic acid, 19, 153 Laccase immobilization, 230-234 Lactate dehydrogenase, 2 12 Leucrose, 123-125 Ligate molecule density, 609-610 Lightdiffusing optical fibers, 96 Ligninase (lignin peroxidase), 238, 244 production scale-up, 240 Lipases, 143, 250 Living cell systems, immobilization, 155-156 Log A% 9 Log P, 7, 10

686

Lower critical solution temperature (LCST), 538 Macroporous carriers, 205 Malate dehydrogenase (MDH), 264, 298 Marine biotechnology, 93-98 Mass transfer limited systems, 607-609 Mechanistic view of biochemical reactions, 116-1 17 Metabolic control analysis, 280-284 Metabolic engineering, 287, 301 Metabolon, 265 Metal bridge activation, 168-169 Metalloenzymes, characteristics, 46 1462 Microenvironment, role in performance of enzymes, 5-7, 32, 103 Micromachining, biosensors and, 373-374 Miniaturized thermal biosensors, construction of, 498-503 Molecular imprinting concept of, 636-639 examples for, 639-641 historic overview, 648-649 present understanding, 64 1 studies on molecular recognition, 649-652 Molecular recognition, 6 19-620, 624625, 667-673 geometric algorithm for identification, 625-630 Molecular tailoring strategy, 70 Molecular weights determinations, 82 Multienzyme systems, 296 Mung bean (Vigna rodiate), 339 N-acetyl-D-tryptophan, 16 NAD(P)H assay, 425-430 quantitative translation of Hz02, 403-404

INDEX

Neural networks, 4,45 application in Beta-galactosidase production, 5 1-52 application in glucoamylase production, 51-53 architecture, 47 programming environment, 49-5 1 topology, 48 training, 48-49 Nonsilane methods, carrier activation, 166-169 Nylon web, 239 Oligosaccharides, 125-127 On-line monitoring system for cyanobacterial cells, 96 Operational stability, 224-225 Organic media, biotransformations Of, 143-144 Organosilicon compounds, 68, 164 reduction of, 68-74 Packed beds, 140 Penicillin acylase, 119, 142, 250 Peptides drug design and, 572-574 multiple uses of, 574-576 peptidyl-L-aminocidhydrolase, 32 secondary structure of, 569-572 synthetic peptide libraries, 568569 Perfused systems, 200 Peroxidase; see Plant peroxidases PH dependence, 2 14-216 P H levels, 179-180 PH stability, 222-223 Phanerochaete chrysoporium, 238 Phenolic compounds, 230 Phosphoglycerate kinase, 273-275 Photobioreactor, 96 Phycocyanin, 96 Phytochemicals, 3 19

Index

Plant cell cultures, 3 18-319, 323 biochemistry studies and, 325-328 biosynthetic capacity, 319-321 biotechnological applications and, 317-319, 321-325 inhibition by tetrapeptide des-arg enterostatin (VPDP), 339-343 Plant peroxidases, 4, 57, 238, 250 gene cloning, 60 gene structure, 62-63 molecular structure of, 4, 58-61 properties of, 61-62 refolded peroxidase, 60 Pneumocystis carinii, 4 Pneumocystis carinii dihydrofolate reductase (DHFR), 78 cloning of gene, 79 gene expression in E. coli, 79 Polyacrylamide bead polymer, 210 Polyethyleneglycol, 9 Poly(NAT) gels, 593-596 Polyurethane, 240 Polyvinyl alcohols, 33-34 Pore flow, 588 Porous carrier matrices, 20 1 Procolipase protein, 340 Proteases, 3 10 Protein coupling activation of supports for, 169-177 alkylalmine and sulfhydryl group, 171- 172 alkylamine and amino group, 170-171 arylamine and diazotization, 172-174 sulfhydryl and sulfhydryl protein, 174-177 Protein engineering, 106, 145 Protein layer, 7 , 9 Protein recognition, 552-553 Protein-ligand interactions, molecular surface recognition and, 624-625

68 7

Proteins biologically specific interactions, 62 1-623 catalytic modified, 19-20 storage stability of, 10-1 1 Pseudanabaena, 96 Pseudomonas dacunhae, D-aspartic acid and, 154-155 Pseudomonas putida, 144 Publications, criteria for evaluation, 127-129 Purple bacteria, 97 Pyrimidine bases, 78 Pyrogen, 156 Pyrroloquinoline quinone redox centers, electrical connection, 39940 1 Pyruvate kinase, 212 Quinoproteins, 98-99 (R)-selective reductase, 73 RDNA technology, 145 Recycling sensors, 379-38 1 analytical reactor, 384-386 the enzyme electrodes, 382-383 the enzyme optode, 383-384 enzyme thermistor, 386 the reaction, 381-382 Redox enzymes, 389-391 amperometric biosensors and, 419-42 Redox hydrogels, electron diffusion in, 391-394 Reductionism, 117-1 18 Regulatory agencies, 195 Replicon, 97 Response surface, 240 Reverse engineering techniques, 107 Rhodobacter capsulatus, 97 Rhodobacter sphaeroides, 97 Rhodospirillaceae,98 Ribonuclease, 19

688

Saturated calomel electrode (SCE), 420 Secondary metabolites, 319 Sensors; see Biosensors Separation media, 591-593 Shake cultures, 239 Shuttle vector, 96-98 Sila-drugs, 68 Silane coupling, 160, 164-166 new applications, 185-186 6-aminopenicillanic acid, 142 Somatic embryogenesis stimulating factors, 95 Specific metabolic flux rates (QX), 196 Spectroflorometry, 84 Staphylococcal protein A, 309 Statistical experimental designs, 239 Storage conditions, 10, 103 Storage stability, 224, 25 1 Streptococcus salivarius, 5 1 Stress response, 3 I0 Structure-function relationship, 58,60 Substrate specificity, 6 1,2 19 Substrate-selective sensory, 660-661 Superporous agarose, 583-586 preparation of, 587-588 Synechococcus, 95-96 Tannin, immobilized, 156 Temperature, 12-13, 18, 85, 88, 109, 180-181, 216 Temperature-induced phase separation, 537-539 protein purification and, 539-541 Thermal assay probe, 410-413 Thefmal stability, 220-223 tests of, 213 Thiophilic adsorption antibodies sites and, 545-549 chromatography (TAC) of, 546

INDEX

3-phosphoglycerate kinase, 2 12 Thymidylate, synthesis of, 78 Transconjugation, 96 Trigonopsis variabilis, 4, 68-69, 73 microogranism cultivation, 7 1 reductase of, 70 Trypsin, chemical modification, 108109 Two-dimensional electrophoresis, 205 Tyramine, 327 Tyrosine decarboxylase, 326 Tyrosine metabolism, 326-330 Urea, 223 Urease, 212 Vanillin acid, 330-335 Vasoactive intestinal peptide (VIP), 27 Veratryl alcohol, 238 Verfahrenstechnik, I 18 Water, 16-18 as factor in biocatalysis, 8, 17, 140 as a reactant, 12 waste water treatment, 97 Water-poor media, 3, 5-7 rationale for study, 7 Water-soluble carbodiimide, 2 10 Wine-making technology, 230-235 Yeast glutamate metabolism, 269-270 propionate metabolism, 268-269 Yeast extract, purification of enzyme from, 543 ZZ-proteins, 3 12-315

Advances in Organ Biology Edited by E. Edward Bittar, Department of Physiology, University of Wisconsin Medical School Volume 1, Pregnancy and Parturition 1996,248pp. ISBN 1-55938-639-8

$109.50

Edited by Tamas Zakar, Perinatal Research Centre, University of Alberta CONTENTS: Preface, Tamas Zakar. Late Pregnancy and Parturition in the Sheep, Wendy J. McLaren, Ian R. Young, and Gregory E. Rice. Regulation of Oviposition, Frank Hertelendy, Kiyoshi Shimada, Mikl6s Toth, Mikos Molnar, and Kousaku Tanaka. Initiation of Parturition in Non-human Primates, Jonathan J. Hirstand GeoffreyD. Thorburn.The Physiology of Human Parturition, Jane E. Mijovic and David M. Olson. The Trophoblast as an Active Regulator of the Pregnancy Environmentin Health and Disease: An EmergingConcept, Donald W. Morrish, Jamaf Dakour, and Hongshi Li. The Endocrinology of Late Pregnancy and Parturition, Tamas Zakar and Brian F. Mitchell. Steroid Modulationof Myometrial Structure and Function During Pregnancy, Charles A. Ducsay and Joon W. Rhee. Determinants of Reproductive Mortality and Preterm Childbirth, John A. McCoshen, f . Audrey Fernandes, Michael L. Boroditsky, and James G. Allardice.

FACULTYlPROFESSlONALdiscounts are available in the U.S. and Canada at a rate of 40% off the list price when prepaid b personal check or credit card and

JAI PRESS INC.

55 Old Post Road No. 2 - P.O. Box 1678 Greenwich, Connecticut 06836-1678 Tel: (203)661- 7602 Fax: (203)661-0792

J A I P R E S S

J A I P R E S S @

Advances in Cell Aging and Gerontology Edited by Paola S. Timiras, Department of Molecular and Cell Biology, University of California at Berkeley and E. Edward Bittar, Department of Physiology, University of Wisconsin, Madison Volume 1, 1996,264 pp. ISBN 1-55938-631-2

$109.50

CONTENTS: The Cell Aging Process, faola S. Timiras. Genes, Hormones and Aging, Peter J. Hormsby. Erythrocyte Aging, G. Bartosz. The Influence of Human Growth Hormone in Elderly Men, K. R. Shew and David Rudman. The Cardiovascular System in Aging, James R. Doche*. Neuronal Calcium Regulation in Aging Brain, Mary L. Michaelis and J. Huschenbeft. Neuronal Aging and Alzheimer’s Disease, G.J.C.G.M. Bosman and W.J. Grip. The Neurobiologyof LateLife Psychosis, Frank W. Brown. Medicine in Elderly People, Peter H. Millard. Geriatric Psychiatry, D.G. Wilkinson.

FACULTYlPROFESSlONAL discounts are available in the U.S. and Canada at a rate of 40% off the list price when prepaid b personal check or credit card and

JAl PRESS INC.

55 Old Post Road No. 2 - P.O. Box 1678 Greenwich, Connecticut 06836-1678 Tel: (203)661- 7602

Fax: (203)661-0792

Advances in Developmental Biochemistry Edited by Paul Wassarman, Depattment of Cell Biology and Anatomy, Mount Sinai Medical School Volume 4, 1996,208pp. ISBN 1-55938-968-0

$109.50

CONTENTS: Preface, Paul M. Wassarman. The wingless/ Wnt-1 Signaling Pathway-New Insights Into The Cellular Mechanisms of Signal Tranduction, Amy Sejsovec and Mark Peifer. Cell Interactionsin the Sea Urchin Embryo, Charles A. Ettensohn, Kristen A. Guss, Katherine M. Malinda, Roberta N. Miller, and Seth W. Ruffins. The Drosophila Sex-Peptide: A Peptide Pheromone Involved In Reproduction, Eric Kubi. Vertebrate Homologs of the Neurogenic Genes of Drosophila, Thomas Gridley. Pleiotropic Roles of CSF-1 in Development Defined by the Mouse Mutation Osteopetrotic, Jeffery W. Pollard and €. Richard Stanley. Index. Also Available: Volumes 1-3 (1992-1994)

$109.50each

FACULTWPROFESSIONALdiscounts are available in the US. and Canada at a rate of 40% off the list price when prepaid by personal check or credit card and ordered directly from the publisher.

JAl PRESS INC. 55 Old Post Road No. 2 - P.O. Box 1678 Greenwich, Connecticut 06836-1678 Tel: (203)661- 7602 Fax: (203)661-0792

J A I

P R E S S @

J A I P R E S S @

Advances in Developmental Biology Edited by Paul Wassarman, Department of Cell and Developmental Biology, Roche Institute of Molecular Biology Volume 3, 1994,194 pp. ISBN 1-55938-853-6

$109.50

CONTENTS: Preface, Paui M. Wassarman. Mechanisms of Neurogenesis in Drosophila Melanogaster,Jose A. CamposOrtega. The Role of Growth Factors in Mammalian Pregastrulation Development, Daniel A. Rappolee andZena Werb. Retinoid Signaling in Mouse Embryos, Elwood Linney and Anthony-Samuel LaMantia. RNA Localization During Oogenesis in Drosophila, Elizabeth R. Gavis and Ruth Lehmann. Actin as a Tissue-Specific Marker in Studies of Ascidian Developmentand Evolution, William R. Jeffery. Index. Also Available: Volumes 1-2 (1992-1993)

$109.50each

FACULTYPROFESSIONAL discounts are available in the U.S. and Canada at a rate of 40% off the list price when prepaid by personal check or credit card and ordered directly from the publisher.

JAl PRESS INC.

55 Old Post Road No. 2 - P.O. Box 1678 Greenwich, Connecticut 06836-1678 Tel: (203)661-7602 Fax: (203)661-0792

Perspectives on Bioinorganic Chemistry Edited by Robert W. Hay, Department of Chemistry, University of St. Andrews, Jon R. Dillworth, Department of Chemistry, University of fssex, and Kevin B. Nolan, Division of Chemistry, Royal College of Surgeons, Dublin, Ireland This series presents state of the art review articles in the rapidly developing area of bioinorganic chemistry. Bioinorganic chemistry is, by its very nature, an interdisciplinaryarea, and as a result there is a considerable need for review articles covering the many different aspects of the subject. In a diverse and rapidly developing field, the series will be of assistance to all those wishing a rapid update in a wide variety of specific areas.

Volume 1,1991,284 pp ISBN 1-55938-184-1

$109.50

CONTENTS: Introductionto the Series: An Editor's Foreword, Albert Padwa. Introduction, Robert W. Hay. Complex Formation Between Metal Ions and Peptides, Leslie D. Petit, Jan E. Gregor and H. Kozlowski. Metal-Ion Catalyzed Ester and Amide Hydrolysis, Thomas H. Fife. Blue Copper Proteins, S.K. Chapman. Voltammetry of Metal Centres in Proteins, Fraser A. Armstrong. Gold Drugs Used in the Treatment of Rheumatoid Arthritis, W.E. Smith and J. Reglinski. Iron Chelating Agents in Medicine: Application of Bidentate Hyroxypyridine-4-Ones, R.C. Hider and A.D. Hall. New Nitrogenases, Robert R. Eady. Volume 2, 1993, 292 pp.

$109.50

ISBN 1-55938-272-4 CONTENTS: Introduction, Robert W. Hay. Dynamics of Iron (11) and Cobalt (11) Dioxygen Carriers, P. Richard Warburton and Daryle H. Busch. Homodinuclear Metallobiosites, David R. Fenton. Transferrin Complexes with Non-Physiologicaland Toxic Metals, David M. Taylor.Transferrins, Edward M. BakeK Galactose Oxidase, Peter Knowles and Nobutoshi /to. Chemistry of Aqua Ions of Biological Importance, David T. Richens. From a Structural Perspective: Structure and Function of Manganese-Containing Biomolecules, David C. Weatherburn, Index.

J A I

P R E S S @

J A I

Advances in Metals in Medicine Edited by Michael J. Abrams, Materials Technology Division, Biomedical Research, Johnson Matthey, West Chester, Pennsylvania and Barry A. Murrer, Johnson Matthey Technology Centre, Reading England. Volume 1, 1993,196 pp. ISBN 1-55938-352-8

P R E S S @

$109.50

CONTENTS: Preface, Michael J. Abrams and Barry A. Murrer. Technetium Heart and Brain Perfusion Imaging Agents, Timothy R. Carroll. Diagnosis and Therapy with Antibody Conjugates of Metal Radioisotopes, Karl J. Jankowski and David Parker. Metal Radionuclides in Diagnostic Imaging by Position Emission Tomography (Pet), Mark A. Green. Bone-Seeking Radiopharmaceuticals in Cancer Therapy, Wynn A. Vokert and Edward A. Deutsch. Radiation Synovectomy, Sonya Shortkroff, Alun G. Jones, and Clement 5. Stedge. Index.

FACULTWPROFESSIONALdiscounts are available in the US. and Canada at a rate of 40% off the list price when prepaid b personal check or credit card and ordered directly lorn the publisher.

JAI PRESS INC.

55 Old Post Road No. 2 - P.O. Box 1678 Greenwich, Connecticut 06836-1678 Tel: (203)661- 7602 Fax: (203)661-0792