Biotechnology Annual Review Volume 2
This Page Intentionally Left Blank
Biotechnology Annual Review Volume 2
Editor:
M. Raafat El-Gewely Department of Biotechnology, University of Tromso, Tromsg, Norway
1996
ELSEVIER Amsterdam - Lausanne - New York - Oxford - Shannon - Tokyo
0 1 9 9 6 Elsevier Science B.V. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher, Elsevier Science B.V., Permissions Department, P.O. Box 521, lo00 AM Amsterdam, The Netherlands. No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, the Publisher recommends that independent verification of diagnoses and drug dosages should be made. Special regulations for readers in the USA - This publication has been registered with the Copyright Clearance Center Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01293, USA. Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made in the USA. All other copyright questions, including photocopying outside the USA should be referred to the copyright owner, Elsevier Science B.V., unless otherwise specified. ISBN 0 444 82444-8 This book is printed on acid-free paper. Published by: Elsevier Science B.V. P.O. Box 21 1 lo00 AE Amsterdam The Netherlands Library of Congress Cataloging-in-PublicationData.
In order to ensure rapid publication this volume was prepared using a method of electronic text processing known as Optical Character Recognition (OCR). Scientificaccuracy and consistency of style were handled by the author. Time did not allow for the usual extensive editing process of the Publisher.
Printed in the Netherlands
V
Scope This new series, as the title implies, aims at covering the development in the field of biotechnology in the form of comprehensive, illustrated and well-referenced reviews. With the expansion in the field of biotechnology both in industry as well as in education, coupled with the increase in the number of new journals reporting new results in the field, the need for a publication that is continuously providing reviews is urgent. The goal of Biotechnology Annual Review is to fill this gap. Naturally, all aspects of biotechnology cannot be reviewed extensively in each issue every year, but each volume will have a number of reviews covering different aspects of biotechnology. Reviewed topics will include biotechnology applications in medicine, agriculture, marine biology, industry, bioremediation and the environment. Fundamental problems dealing with enhancing the technical knowledge encountering biotechnology utilization, regardless of the field of application, will be emphasized. Examples of such vital topics are promoters, vectors, media, induction, genetic stabilization during heterologous gene expression and any relevant new technique. Essential information dealing with the utilization of data banks, such as protein and nucleic acid data banks, will be reviewed. Homology studies as related to biotechnology, as well as issues dealing with the characterization of motifs and motif data bases will be also dealt with. New developments in protein engineering, optimization of protein function and protein design will be addressed. Problems dealing with protein functionality are important not only for the production of active recombinant proteins and enzymes, but also for the purpose of drug development and design based on screening using such proteins, whether by employing in vitro or in vivo assays. Newly discovered open reading frames or protein identified by two-dimensional gel electrophoresis will be updated whenever possible. Other issues, dealing with policy and regulation of biotechnology as well as the problems of development in developing countries, as related to biotechnology, will be included in the various issues. The “Editorial Board” of Biotechnology Annual Review encourages suggestions and contributions of articles from industry or from academic institutions that would constitute a comprehensive covering of a relevant topic in biotechnology. Please contact me for any suggestions about chapter contributions. M. Raafat El-Gewely, PhD Professor of Biotechnology Institute of Medical Biology University of Tromsa 9037 Tromsa, Norway Tel.: +47-776-44654. Fax: +47-776-45350. E-mail:
[email protected]
This Page Intentionally Left Blank
vii
Biotechnology Annual Review Volume 2 Editor Dr. M.R. El-Gewely Department of Biotechnology Institute of Medical Biology University of Tromsa MH-Bygget 9037 Tromsa Norway Tel.: 47-77-6446-54 Fax: 47-77-64-53-50
Associate editors Dr. Thomas M.S. Chang Artificial Cells & Organs Research Centre McGill, 3655 Drummond St. Room 1005 Montreal, Quebec Canada H3G 1Y6 Tel.: +1-5 14-398-3512 Fax: +1-514-398-4983
Dr. F. Felici IRBM P Angeletti Via Pontina km 30.600 00040 Pomezia, Roma, Italy Tel.: +39-6910931 Fax: +39-691093225 Dr. Shigehiro Hirano Department of Agricultural Biochemistry and Biotechnology Tottori University Tottori 680, Japan Tel.: +81-857-280321 (ext. 5200) Fax: +81-857-315347 Dr. Kuniyo Inouye Department of Food Science and Technology Faculty of Agriculture Kyoto University Sakyo-ku, Kyoto 606-01 Japan Tel.: +81-75-753-6267 Fax: +81-75-753-6265
Dr. Thomas T. Chen Biotechnology Center University of Connecticut 184 Auditorium Road U-149 Stoms CT 06269-3149, USA Tel.: +1-203-486-5011/5012 Fax: +1-203-486-5005
Dr. Guido Krupp Institut fur Allgemeine Mikrobiologie Christian-Albrechts-Universitat Am Botanischen Garten 9 D-24118 Kiel, Germany Tel.: +49-43 1-880-4330 Fax: +49-43 1-880-2194
Dr. Roy H. Doi Section of Biochemistry and Biophysics University of California, Davis Davis, CA 95616-8535, USA Tel.: +1-916-752-3191 Fax: +1-9 16-752-3085
Dr. Eric Olson Department of Biotechnology Warner-Lambed 2800 Plymouth Road Ann Arbor, MI 48105, USA Tel.: +1-3 13-998-5961 Fax: +1-313-998-5970
...
Vlll
Dr. Steffen B. Petersen SINTEF, UNIMED 7034 Trondheim Norway Tel.: +47-73-99-77-00 Fax: 47-73-99-77-08
G. Kristin Rosendal The Fridtjof Nansen Institute P.O. Box 326 1324 Lysaker Norway Tel.: +47-67-53-89-12 Fax: 47-67- 12-50-47
Dr. Jack Preiss Department of Biochemistry Michigan State University Biochemistry Building East Lansing, MI 48824-1319 USA Tel.: +1-517-353-3137 F a : +1-517-353-9334
Mark Tepfer Laboratoire de Biologie Cellulaire INRA - Centre de Versailles F-78026 Versailles Cedex France Tel.: +33-1-30-83-30-29 Fax: +33-1-30-83-30-99
ix
Contributors Dr.Rohini Acharya The Royal Institute of International Affairs Chatham House 10 St. James's Square London SWlY 4LE, UK Tel.: +44-0171-957-5700 Fax: +44-017 1-957-5710 Dr. P. Ainsworth Department of Zoology and Division of Medical Genetics The University of Western Ontario 307 Western Science Centre London, Ontario, Canada N6A 5B7 Tel.: +1-5 19-661-3135 Fax: +1-519-661-2014 Dr. Javier Barrios-Gonzalez Departmento de Biotechnologia Universidad Autonoma Metropolitana Av. Michoach y La Purisima Iztapalapa, 09340 Mexico, D.F. Tel.: +52-724-47-11 (724-47-12 and 724-47- 13) Fax: +52-724-47-12. Anthio Baptista SINTEF, UNIMED 7034 Trondheim, Norway Tel.: +47-73-99-77-00 Fax: +47-73-99-77-08
'
Dr. Ton Bisseling Department of Molecular Biology Agricultural University 6703 HA, Wageningen, The NetherlancdS Tel.: +31-8370-82036 Fax: +31-8370-83584
Douglas S. Burdette Department of Biochemistry Michigan State University East Lansing, MI 48824, USA Tel.: +1-517-353-4614 Fax: +1-517-353-5556
Dr. Thomas T. Chen Biotechnology Center University of Connecticut 184 Auditorium Road U-149 Stems, CT 06269-3149, USA Tel.: +1-203-486-5011/5012 Fax: +1-203-486-5005 Rex A. Durham Department of Fisheries and Allied Aquacultures Auburn University, Auburn Alabama, USA
Dr. Matthias Ehrmann Lehrstuhl fur Technische Mikrobiologie Technische Universitat Munchen 85350 Freising-Weihenstephan Germany Tel.: +49-8161-71-3663 Fax: +49-8 161-71-3327 Dr. Shigehiro Hirano Department of Agricultural Biochemistry and Biotechnology Tottori University Tottori 680, Japan Tel.: +81-857-280321 (ext. 5200) Fax: +81-857-315347 Dr. Rawle Hollingsworth Department of Biochemistry Biochemistry Building Michigan State University East Lansing, MI 48824-1319, USA Tel.: +1-5 17-353-0613 Fax: +1-517-353-9334 Dr. J.H. Jung Department of Zoology and Division of Medical Genetics The University of Western Ontario 307 Western Science Centre London, Ontario Canada N6A 5B7 Tel.: +1-5 19-661-3135 Fax: +1-5 19-661-2014
X
Chun-Mean Lin Biotechnology Center University of Connecticut Storrs, Connecticut, USA
Carmen L.A. Paiva Rua Custodio Serrfio, 14 apto. 701 Lagao, Rio de Janeiro, RJ Brazil, 22470-230
Jenn-Kan Lu Biotechnology Center University of Connecticut Storrs, Connecticut, USA and Department of Biological Sciences University of Maryland at Baltimore County Baltimore, Maryland, USA
Anita D. Panek Rua Custodio Serrfio, 14 apto. 701 Lagao, Rio de Janeiro, RJ Brazil. 22470-230
Paul0 J. Martel Instituto de Tecnologia Quimica e Biol6gica Apartado 127, P-2781 Oeiras Portugal Armando Mejia Departmento de Biotechnologia Universidad Autonoma Metropolitana Av. Michoackn y La Purisima Iztapalapa 09340 Mexico, D.F. Dr. John Mugabe ERA Witmakerstraat 10 621 1 JB Maastricht The Netherlands Fax: +31-43-25-69-17 R.N. Ott Department of Zoology and Paediatrics Molecular Genetics Unit Division of Medical Genetics University of Western Ontario and Molecular Medical Genetics Program Child Health Research Institute, Children’s Hospital of Western Ontario and Victoria Hospital London, Ontario Canada
Katharina Pawlowski Department of Molecular Biology Agricultural University 6703 HA Wageningen The Netherlands Tel.: +31-8370-82036 Fax: +31-837043584 Dr. Steffen B. Petersen SINTEF, UNIMED 7034 Trondheim Norway Tel.: +47-73-99-77-00 Fax: +47-73-99-77-08 Mariagrazia Pizza (Instituto di Recerche Immunobiologiche Siena) via Fiorentina 1 - 53100, Siena Italy
Dr. Jack Preiss Department of Biochemistry Michigan State University Biochemistry Building East Lansing, MI 48824-1319 USA Tel.: +1-5 17-353-3137 Fax: +1-517-353-9334 Dr. Rino Rappuoli IRIS (Instituto di Recerche Immunobiologiche Siena) via Fiorentina 1 - 53100, Siena Italy Tel.: +39-577-293-414 Fax: +39-577-293-564
xi Renate Reimschuessel Aquatic Pathobiology Group Department of Pathology University of Maryland at Baltimore Baltimore, Maryland, USA h a Ribeiro Department of Molecular Biology Agricultural University 6703 HA Wageningen The Netherlands
Dr. Christophe Robaglia Laboratoire de Biologie Cellulaire INRA - Centre de Versailles F-78026 Versailles Cedex France Tel.: +33-1-30-83-30-29 Fax: +33-1-30-83-30-99 Dr. D.I. Rodenhiser Department of Zoology and Division of Medical Genetics 307 Western Science Centre The University of Westem Ontario London, Ontario Canada N6A 5B7 Tel.: +1-5 19-661-3135 Fax: +I-519-661-2014 Dr. Nobufusa Serizawa Biomedical Research Laboratories Sankyo Co. Ltd. No. 2-58, Hiromachi-1-chome Shinagawa-ku, Tokyo 140 Japan Tel.: +81-3-3492-3131 (ext. 3330) Fax: +81-3-5436-8565 Dr. Shiva M. Singh Department of Zoology and Division of Medical Genetics 307 Western Science Centre The University of Western Ontario London, Ontario Canada N6A 5B7 Tel.: +1-519-661-3135 Fax: + 1-519-661-2014
Dr. G. Stranzinger Department of Animal Sciences Eidgenossische Technische Hochschule Tannenstrasse l/ETH-Zentrum CH-8092 Zurich Switzerland Tel.: +41-1/632-32-56 Fax: +41-1/632-11-67 Mark Tepfer Laboratoire de Biologie Cellulaire INRA - Centre de Versailles F-78026 Versailles Cedex, France Tel.: +33- 1-30-83-30-29 Fax: +33- 1-30-83-30-99 Claire Vieille Department of Biochemistry Michigan State University East Lansing, MI 48824, USA Dr. Rudi F. Vogel Lehrstuhl fur Technische Mikrobiologie Technische Universitat Munchen 85350 Freising-Weihenstephan Germany Tel.: 4 9 - 8 161-71-3663 Fax: +49-8161-7 1-3327 Nick Vrolijk Biotechnology Center University of Connecticut Storrs, Connecticut USA Dr. Dirk F. Went Institut fur Nutztierwissenschaften Gruppe Zuchtungsbiologie Tannenstrasse l/ETH-Zentrurn CH-8092 Zurich, Switzerland Dr. J. Gregory Zeikus Department of Biochemistry Michigan State University East Lansing, MI 48824 USA Tel.: +I-517-353-4614 Fax: +1-517-353-5556
This Page Intentionally Left Blank
xiii
Contents Preface
Thermozymes C. Vieille, Douglas S . Burdette and J . Gregory Zeikus Production of secondary metabolites by solid-state fermentation J . Barrios-Gonzales and A. Mejia Genetics of lactobacilli in food fermentations R.F. Vogel and M . Ehrmann Nitrogen fixing root nodule symbioses: legume nodules and actinorhizal nodules K. Pawlowski, A. Ribeiro and T. Bisseling Using nonviral genes to engineer virus resistance in plants C. Robaglia and M . Tepfer Transgenic fish and its application in basic and applied research T.T. Chen, N. Vrolijk, J.-K. Lu, C.-M. Lin, R. Reimschuessel and R A . Dunham Chitin biotechnology applications S. Hirano ADPglucose pyrophosphorylase: basic science and applications in biotechnology J. Preiss The chemical degradation of starch: old reactions and new frontiers R. Hollingsworth Biotechnological applications of the disaccharide trehalose C.L.A. Paiva and A.D. Panek Protein electrostatics P J . Martel, Antbnio Baptista and S.B. Petersen Biochemical and molecular approaches for production of pravastatin, a potent cholesterol-lowering drug N. Serizawa Novel molecular biology approaches to acellular vaccines R. Rappouli and M . Pizza Strategies and applications of DNA level diagnosis in genetic diseases: past experiences and future directions S.M. Singh, D.I. Rodenhiser, R.N. Ott, J.H. Jung and P J . Ainsworth Molecular genetics as a diagnostic tool in farm animals G. Stranzinger and D.F. Went
V
1 85 123
15 1 185
205 237
259 28 1 293 315
373 39 1
409
447
xiv Biotechnology in developing countries: critical issues of technological capability building R . Acharya and J . Mugabe
465
Index of authors
505
Keyword index
507
81996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R. El-Gewely, editor.
1
Thermozymes Claire Vieillel, Doug S. Burdette' and J. Gregory Z e i k u ~ ' * ~ 'Department of Biochemistry, Michigan State University, East Lansing, Michigan; and 'Michigan Biotechnology Institute, Lansing, Michigan, USA
Abstract. Enzymes synthesized by thermophiles (organisms with optimal growth temperatures >60°C) and hyperthermophiles (optimal growth temperatures >8OoC) are typically thermostable (resistant to irreversible inactivation at high temperatures) and thermophilic (optimally active at high temperatures, i.e., >6OoC). These enzymes, called thermozymes, share catalytic mechanisms with their mesophilic counterparts. When cloned and expressed in mesophilic hosts, thermozymes usually retain their thermal properties, suggesting that these properties are genetically encoded. Sequence alignments, amino acid content comparisons,and crystal structurecomparisons indicate that thermozymes are, indeed, very similar to mesophilic enzymes. No obvious sequence or structural features account for enzyme thermostability and thermophilicity. Thermostability and thermophilicity molecular mechanisms are varied, differing from enzyme to enzyme. Thermostability and thermophilicity are usually caused by the accumulation of numerous subtle sequence differences. This review concentrates on the mechanisms involved in enzyme thermostability and thermophilicity. Their relationships with protein rigidity and flexibility and with protein folding and unfolding are discussed. Intrinsic stabilizing forces (e.g., salt bridges, hydrogen bonds, hydrophobic interactions) and extrinsic stabilizing factors are examined. Finally, thermozymes' potential as catalysts for industrial processes and specialty uses are discussed, and lines of development (through new applications, and protein engineering) are also proposed.
Key words: Archaea, biocatalysis, core model, electrostatic interactions, hydrophobic interactions, hyperthermophiles, industrial enzymes, proline zipper, protein flexibility, protein rigidity, specialty enzymes, thermophiles, thermophilicity, thermostability.
Introduction The world market for industrial and research enzyme sales has grown to about US $1 billion annually. Industrial enzyme uses are growing because they: 1) offer less polluting processes than chemical catalysts; 2) perform reactions with higher specificity than chemical catalysts; and 3) perform reactions for which chemical catalysts are not known. Nonetheless, most described enzymes, unlike chemical catalysts, suffer from instability at high temperatures and pH extremes. Thermozymes are proteins evolved by thermophiles and hyperthermophiles to perform catalysis from 60°C to above 110°C. This feature makes the study of thermozymes a "hot" topic for protein science and biotechnology.
Address for correspondence: J. Gregory Zeikus, Michigan Biotechnology Institute, 3900 Collins Road, Lansing, MI 48909, USA. Tel.: +1-517-337-3181. Fax: +1-517-337-2122.
2 Intrinsically stable and active at high temperatures, thermozymes offer major biotechnological advantages over mesophilic enzymes: 1) once expressed in mesophiles, thermozymes are easier to purify by heat treatment; 2) they are easy to crystallize; 3) their thermostability is associated with a higher resistance to chemical denaturants (such as a solvent or guanidine-hydrochloride); 4)performing enzymatic reactions at high temperatures can allow higher reaction rates, higher substrate concentrations, and lower viscosity; and 5 ) there is a higher product yield during certain reactions due to chemical equilibrium shifts with high temperature. While discovering thermophilic organisms that thrive at higher and higher temperatures, different terms (e.g., thermophiles, caldoactive organisms, moderate thermophiles, hyperthermophiles) have been used over the years to characterize thermophiles. Our review uses only three terms: 1) mesophiles are organisms that optimally grow at temperatures between 25 and 60°C (the optimal growth temperature is defined as the temperature corresponding to the highest growth rate), 2) thermophiles are organisms that optimally grow at temperatures between 60 and 80°C, and 3) hyperthermophiles comprise the organisms whose optimal growth temperature is above 80°C. In certain cases an organism's optimal growth temperature can depend on the growth conditions. The effect temperature has on an organism's growth rate is typically determined under the best known growth conditions. Our classifications differ slightly from previously published classifications. Before the 1970s, it was generally thought that thermophilicity was a property associated with spore forming bacteria, and the general hypothesis of the time was that thermophilic enzymes were not inherently stable, but rather they were rapidly turned over. This high turn over explained why thermophiles did not grow faster than mesophiles [ 11. However, since the discovery of Thermus aquaticus, a nonsporulating thermophile that retained an inherently thermostable protein synthesizing apparatus and enzymes [2], most thermophiles and hyperthermophiles have been shown to possess inherently stable enzymes that function at temperatures above the organisms optimal growth temperature. Enzyme thermostability is the enzyme's capacity to resist irreversible thermal inactivation, and is commonly indicated by the enzyme's half-life at a given temperature. We define enzyme thermophilicity to be the temperature at which the enzyme is optimally active. (An enzyme is thermophilic if it is optimally active at high temperatures from 60°C to above 100°C). Thermozymes are both thermostable and thermophilic enzymes. Although thermozymes originate from thermophiles and hyperthermophiles, some mesophiles can produce enzymes active and stable above 60°C. For example, Bacillus licheniformis produces a stable aamylase, optimally active at 75-80°C (see [3]). We use the term mesophilic enzymes, or mesozymes, to describe enzymes optimally active at moderate temperatures (20-60OC). These enzymes typically originate from mesophiles, and include most eukaryotic enzymes and most enzymes from mesophilic bacteria and archaea. Thermophiles and hyperthermophiles belong to the extended group of prokaryotes called extremophiles. Extremophiles include all organisms that thrive in extreme conditions; temperatures above 60"C, halophilic conditions (saturated NaCl), extremes of pHs (below 4.0 or above 10.0), and conditions of substrate stress (e.g., limited
chemical free energy and toxic compounds). These organisms typically produce enzymes (extremozymes) that are optimally active at extreme temperatures, salt concentrations, or pHs. Extremophiles and thermophiles have been the subject of several reviews [4,5]; they are not discussed in this review. Our review updates enzyme thermophilicity and thermostability knowledge. This review includes a general discussion of thermozyme properties, supported by the information presented in Tables 1 and 2. Due to the breadth of this field, our efforts concentrate on enzymes with potential biotechnological applications. Some enzymes, regarded as not useful in todays applications may, however, prove important in future processes. The molecular mechanisms that account for enzyme thermophilicity and thermostability are discussed in detail, and our review is illustrated with examples taken from the literature and from thermozymes studied in our laboratory. The different methods for improving enzyme thermostability and designing enzymes to work at elevated temperatures are presented. Lastly, current and potential applications of thermozymes are reviewed. We include where high-temperature enzymes are now used, and what potential applications they may have. Thermozyme potentials are assessed with regard to new technologies and to different markets. Already the object of extensive reviews, thermophiles and hyperthermophiles are only briefly described (for detailed reviews, see [4,6-9]). No exhaustive descriptions of all enzymes isolated and characterized from hyperthermophiles are presented here, since this information is available elsewhere (see [9,10]).
General properties of thermozymes Thermozyme sources Thermophilic Almost all thermophiles are bacteria and archae. While some blue-green algae can grow at temperatures up to 60°C, eukaryotes typically do not grow at temperatures above 50°C. (No eukaryotic rDNA was detected in the screening of a Yellowstone National Park hot spring algal mat ecosystem [ 113). Numerous thermophiles have been isolated from hot environments that include: 1) natural volcanic environments (continental solfataras, hot springs, soils, shallow marine and deep-sea hot sediments, submarine hydrothermal vents), 2) microbially self-heated environments (e.g., manure, coal refuse piles, compost piles), and 3) industrial environments (e.g., food industry effluents, hot water lines, sewage sludge systems, oil drilling injection water systems), and have been the object of excellent reviews [ 1,4,7,12,13]. All nutritional categories are represented among thermophiles: aerobes and anaerobes, heterotrophs, chemoorganotrophs, chemolithotrophs, autotrophs, phototrophs, etc. Thermophiles show as much diversity as their mesophilic counterparts. Usually, their nutritional category reflects the environment from which they originate. Soil, manure, and sewage sludge are often rich in complex organic material (especially polysaccharides and proteins); amylolytic, cellulolytic, and proteolytic heterotrophic thermophiles have been isolated
4
from these sources (see [4]). These anaerobic environments are also rich in thermophilic methanogens (see [4]). Hot springs and solfataras are rich in H, and CO,. The primary producers isolated from these biotopes are usually obligate or facultative aerobes, either photosynthetic (cyanobacteria are abundant in hot springs [4]) or chemoautotrophs (mainly from the archaeal sulfolobales [8,14]). The growth of moderately thermophilic algae (optimum growth at 5 0 4 5 ° C ) at a hot spring perimeter provides thermophilic heterotrophs with a variety of organic materials. Due to low oxygen solubility at high temperatures, lower layers of hot spring and solfataras are anaerobic. Most heterotrophs isolated from these biotopes are thermophilic anaerobes. These organisms thrive on peptide-containing substrates (derived from the primary producers decomposition), on organic acids, or on saccharides and polysaccharides (derived from blue-green algal decomposition). See Lowe et al.'s 1993 review [4] for an extensive description of heterotrophic thermoanaerobes. Since most studies of organisms that thrive in hot, submarine environments have focused on hyperthermophiles, only three thermophiles have been characterized from these environments; two chemoautotrophs, Thiothrix and Beggiatoa (abundant in bacterial mats at the base of hot vents) [12,15], and one heterotroph, Thermosipho africanus [ 161. Hyperthermophilic After pioneering characterizations in 1972 of the thermoaerobic archaeum Sulfolobus by Brock et al. [17] and of the thermoanaerobic archaeum Methanobacterium thermoautotrophicum by Zeikus and Wolfe [ 181, the first hyperthermophile anaerobes growing at temperatures higher than 100°C were isolated in 1982 from a submarine volcanic area by Karl Stetter [ 191 and from hot deep-sea waters by Baross et al. [20] and Zillig et al. [21]. Since that ground-breaking work, approximately 50 species, 20 genera, and 1 1 orders of hyperthermophiles have been described [ 131. Most hyperthermophiles have been isolated from hot natural environments, including continental solfataras [22,23], deep geothemally heated oil containing stratifications [24], shallow marine and deep-sea hot sediments [25], and hydrothermal vents located up to 4,000 m below sea level [8,13,26]. Hyperthermophiles have also been isolated from hot industrial environments (e.g., outflow of geothermal power plants and sewage sludge systems). In deep-sea environments, organisms also have to resist high hydrostatic pressures ranging from 200 to 360 atm; some hyperthermophiles are bqotolerant [27] or even barophilic [28,29]. With the exception of Thermotogales and Aquifex, all organisms which thrive at temperatures above 80°C are Archaea [4,6,8,9]. All hyperthermophilic primary producers are chemoautotrophs (i.e., sulfur oxidizers, sulfur reducers, and methanogens) [4,23]. Based on the high sulfur content of most hot natural biotopes, most hyperthermophiles are facultative or obligate chemolithotrophs, and reduce So with H, to produce H,S (the anaerobes) or oxidize So with 0, to produce sulfuric acid (the aerobes). Extremely acidophilic hyperthermophiles belong to the order of Sulfolobales. They are all strict (e.g., Sulfolobus) or facultative aerobes (e.g., Acidianus), and are almost exclusively isolated from continental solfataras 181. While most
5 heterotrophs are obligate sulfur reducers, Thermotoga, Pyrococcus, and Thermococcus can grow independently from So,getting their energy from fermentations [6,8,9]. Related to the extremely low organic matter content of their submarine environments, hyperthermophilic heterotrophs typically get their energy and carbon from complex mixtures of peptides derived from the decomposition of primary producers. A few species are able to use polysaccharides (e.g., starch, pectin, glycogen), and only Archeoglobus profundus uses organic acids. Most studies have focused on species of Thermotogales such as Thermotoga maritima and the archaeum Pyrococcus furiosus. Thermotogales are the deepest branch in the bacterial genealogy, representing an obvious interest in evolutionary studies [30]. Because they use a variety of carbohydrates, Thermotogales are a potential source of saccharolytic enzymes with biotechnological applications. Easily cultivated in the laboratory, P . furiosus, a starch user, is able to grow in the absence of So,making it a good model hyperthermophilic archaeum candidate. Based on recent research trends in this young field, the Small world of hyperthermophiles can be expected to grow rapidly. The large extent of hyperthermophile diversity has been suggested by lipid analysis [31] and by the study of rDNA [ l l ] in samples originating from hydrothermal vents or continental hot springs. Knowledge of the hyperthermophile world will grow by studying new geographical locations, and by developing new isolation techniques for microorganisms with different physiologies. Barns et al.’s 1994 archaeal diversity study [ l l ] yielded a new isolation procedure. Huber et al. (1995) [32] used a fluorescently labeled 16s rDNA sequence belonging to a yet noncharacterized archaea, as a probe in whole-cell hybridization. Cells that gave a positive signal were cloned by “optical tweezers”, and their 16s rDNA contained the exact sequence of the oligonucleotide used as a probe.
Purified thermozymes and cloned genes
’
In constant expansion, the study of thermophilic enzymes remains a recent research area. As we went through the literature, it appeared that many thermozymes were characterized without their temperature for optimal activity being determined, or without data on their thermostability. In addition, when thermostability properties are reported, they are often reported in different ways, and this difference limits comparative studies. Attempts have been made to set standards for thermostability measurements, and reporting an enzyme’s half-life at a specific temperature has been consensually adopted as a standard. Because of the broad range of thermostabilities existing among enzymes, it is impossible to define a unique temperature at which thermostability should be assayed. Thus, it remains difficult to compare similar enzymes originating from mesophiles and thermophiles. To facilitate comparisons, we propose to express thermostability as the temperature at which one enzyme’s half-life is 1 h. Tables 1 and 2 list enzymes which have been characterized and/or cloned from thermophilic (Table 1) and hyperthermophilic (Table 2) organisms; thermophilicity and thermostability properties are included, when available. Their heterogeneity
a
Table I . Thermozymes from hemophilic organisms. Enzyme
Organism (optimal growth temperature)
Enzyme thermophilicity
Enzyme themostability
Oxidoreductases Fomylmethanofuran dehydrogenase GAPDH MTHF dehydrogenase Secondary alcohol dehydrogenase Sulfite reductase Glutamate dehydrogenase
Methanobacterium wolfei (60°C) Bacillus stearothermophilus (7OOC) Clostridium thermoaceticum (60°C) Thermoanaerobacter ethanolicus (69°C) Thermodesulfobacteriumcommune (70°C) S. shibatae
65OC/pH 7.4
stable at 65°C (+I M KCI) 20 mifl5"C stable at 50°C
Transferases Adenylate b a s e DNA polymerase I 3-Phosphoglycerate kinase 3-Phosphoglycerate kinase Hydrolases PN-Acety lhexosaminidase Alcaline phosphatase a-Amylase a-Amylase
S. acidocaldarius (75OC. pH 2-3)
B. stearothermophilus Thermus sp. saain Rt41A Bacillus caldovelox (7OoC) B. stearothemphilus
Amylopullulanase Amylopullulanase Amylopullulanase Amylopullulanase o-Asparaginase ATPW Cyclodextrinase Cyclodextringlycosyltransferase Endo .0-1,4-glucanase Endo PI ,4-glucanase
Clostridiurn thermocellum C. thermocellwn
Amylopullulanase
33.34 see 10 35
C. S, Expr. 65-70°C c, 48 W5-8O"C 40 m h ~ 9 5 ~ C
s
36 37 38
C, S, Expr. C, S, Expr. C. S, Expr.
39,40 41.42 43 44
10 minn3"C 5 min/85"C
E E E, C, S, Expr.
90°C/pH 5.5 80°C/pH 5.5 70°C/pH 5.5 75"C/pH 5.5
5.5-28 minDO°C (+5 mM CaCl,) 2 minDOWpH 5.0 or 12.5 min/90'C/pH 8.0 70% active after 1 W0"C nd nd S O % active after 1 W5"C
45 46 47 48.49
E, C, S, Expr. E, C, S, Expr. E, C, S, Expr. E, C, S, Expr.
52.53
85-90°C/pH 5.6
several h/85"C
E
54
90°C/pH 5.5 80°C 75°C 70-75OC >80°C/pH 9.5 75-85WpH 6.2 65WpH 6.0 hydrolysis at 90-95"C cyclization at 80-85OC 62"ChH 5.2 65'C/PH 6.0
40 min/90"C several h/8O"C 45 mifl5"C nd 25 min/85"C
E, C, S, Expr. E E, C, S, Expr. E, C, S, Expr.
55.56 57
90°C/pH 5.3-6.0
Thermus aquaticus (7OOC) B. stearothermophilus T. thermophilus (65-72OC)
Dictyoglomus thermophilum (78°C) D. themphilum D. thermophilwn Thermoanaerobocteriwn thermosulfurigenes 4B (60°C) Clostridium thermohydrosulfuricum El01 (65°C) T. ethanolicus (69°C) ThermoanaerobacteriumTok6-B 1 (65°C) Thermoanaerobacteriumsaccharolyticum (58°C) T. thermosulfurigenes EM1 (60°C) T. aquaticus S. acidocaldarius T. ethanolicus T. thermosulfurigenesEM1
a-Amylase (AmyA) a-Amylase (AmyB) a-Amylase (AmyC) PAmylase
%4"C
General comments References
c. s
75'C/pH 6.5
70°C
75 miflOOC MB, C, S, Expr. 75% active after 5 h/90"C (+ starch) C, S,Expr. nd inactive after 30 min/800C
50 51 51
58 59
60 see 61 62.63
64 see 65 see 65 (continued)
Table I. Continued. ~~
Enzyme
Organism (optimal growth temperature)
Enzyme thermophilicity Enzyme thermostability
General comments References
Endo p- 1,4-glucanase Endo- 1.4-P-xylanase Endo- 1,CP-xylanase FGalactosidase a-Glucosidase a-Glucosidase p-Glucosidase P-Glucosidase a-Glucuronidase PHydantoinase PMannanase NeopuUulanase Proteinase Proteinase Proteinase
C . thermocellum Caldicellulosiruptor saccharolyticus (70°C)" T. saccharolyticum Thermus 41A Bacillus sp. T. ethanolicus C. saccharolyticus C . thermocellum (60°C) Thermoanaerohocterium sp. (60OC) B. stearothermophilus C. saccharolyticus B. stearothermophilus Thermus sp. strain Rt4A2 Thermus sp. strain Rt4 1A Bacillus sp. Ak.1
60"C/pH 6.4 70"C/pH 5.5-7.7 7OoC/pH 6.0
E, C, S, Expr. E. C, S, Expr.
60-65OC pH 9.0 90'C/pH 8.0 pH 7.5
Pullulanase Pullulanaseb Pullulanase Pullulanaseb Pyrophosphatase p-1 ,CXylanase f3-Xylosidase !3-Xylosidase
B. stearothermophilus Bacillus sp. Thermus sp. strain AMD-33 Thermus aquaticus YT- 1 Thermoplasma acidophilum (60'C) Thermoanaerobactenum sp. T. saccharolyticum C . saccharolyticus
65"CIpH 6.0 75oc 70°C/pH 5.5-5.7 70-85OC? 85'C/pH 6.7 80"CIpH 6.2 70"CIpH 5.5 70"C/pH 6 . e 6 . 5
C. thermosacchamlyticum T. acidophilum B. steamthermophilus
70°C 55°C
Methanobacterium thermoautotrophicum (70°C) C. thesmosacchamlyticum B. steamthermophilus T. saccharolyticum T. thermosulfurigenes T. aquaticus HB8 T. thermophilus
65"CIpH 7.4 62'CIpH 8.6 85°C 80"ClpH 7.5 80"ClpH 7.5 85"CIpH 7.0 nd
Lyases, isomerases, and ligases Fructose-1,&diphosphate aldolase Citrate synthase Glutamine synthetase Phosphoenolpyruvatecarboxylase Triose phosphate isomerase Xylose isomerase Xylose isomerase Xylose isomerase Xylose isomerase Xylose isomerase
7SoC/pH 5.5 75°C ndpH 6.W6.5 60"C/pH 5.4 65OCIpH 8.0
1 N85"C
20 mifl5"C. , pH 6.0 (+BSA) 35 m i n / S O T 40 W5"C or 8 min/W"' 10 mifl5"C 35 minI75"C 70 mi@O"C 60% active after 7 N60"C 1 W62"C 30 min/80°C
90% active after 1 hI60"C 90 mi@O°C (+ 5 mM CaCI,) 20 min/9ooC (+ 5 mM CaC1,) 13 N8O"C or 19 min/9O"C (+5 mM CaCI,) stable for I N65"C 96 hn0"C
C, S . Expr.
c, s C, S . Expr. E E E E E E, C, S
S O % active after 10 h at 85°C c, 1 W0"C 55 minI75T 45 min/8O0C
stable 4 N57"C stable after 10 mini78'C stable 5 WO°C (+ Mn2+/M$, glutamine, and NH4C1)
s
nd
79 80 81 82 83.84 85
C, S, Expr. C, Expr.
86 87
C, S. Expr., CS
88 see 89 see 10
10 min/64"C
42 W0"C (+ MgCI, and CoC1,) 4dnO"C nd
see 65 66.67 68.69 see 10 70 71 see 10 see 65 72 13 74 75 76 77 78
C, S , Expr. C, S,Expr. C, S . Expr.
90 91 92 93.94 93,95 96 97,98
C: cloned; S: sequenced; CS: crystal structure available; E: extracellular; MB: membrane associated; Expr.: expressed in a mesophile; MTHF. 5,lO-methylenetetrahydrofolate;GAPDH Glyceralhehyde-3-phosphatedehydrogenase. "Formerly Caldocellum saccharolyticum [991; bthe activity of this pullulanase on starch has not been characterized.
00
Table 2. Thermozymes from hyperthermophilic organisms. Enzyme Oxidoreductases Alcohol dehydrogenase (NAD-specific) Alcohol dehydrogenase (NADP-specific) Aldehyde ferredoxin oxido-reductase Ferredoxin Formaldehyde oxidoreductase Glutamate dehydrogenase (NADP-specific) Glutamate dehydrogenase (NADDIADP) Glutamate dehydrogenase (NADDIADP) Glutamate dehydrogenase (NADP-specific) GAPDH GAPDH GAPDH (NAD) GAPDH (NADP) GAPDH Hydrogenase (F,,-reactive) Hydmgenase (F,,-nonreactive) Hydrogenase (H2producing) Hydrogenase Hydrogenase (H, producing) L-Lactate dehydrogenase Malate dehydrogenase CH, = H,MF'I dehydrogenase (F,,-dependent)
Organism (optimal growth temperature) Enzyme thermophilicity Enzyme thermostability Sulfolobus solfataricus (70-85T)
Themcoccus litoralis (88°C) Pyrococcus furiosus ( 100OC) P. furiosus T. litoralis ES4 (100°C) P. furiosus S. solfataricus T. litoralis Thennotoga maritima (80OC) Pyrococcus woesi ( 100°C) Thermoproteus tenax (88OC) T. tenax Methanothermus fervidus (82OC) Methanococcusjannaschii (85°C) M . jannaschii P . furiosus Pyrodictium brockii (105OC) T. maritima T. maritima M . fervidus Methanopyrus kandleri (98°C)
95"C/pH7.5-8.5 8O0C/pH8.8 >90°C/pH9.0-10.0 >95T 95°C
General comments References
5 h/7O0C
2 h/85"C 6 h/80°C stable for 12 h at 95°C 2 h/80"C 3.5 h/105"C 2-12 h/loO"C 15 hI8O"C
7535'C/pH9.0 70°C/pH 10.0 65OC/pH9.0 >95'C/pH8.0 >75'C/pH6.0-8.0 nd nd nd nd 80-90°C/pH7.CF10.0 80°C/pH9.0 >95T >90T >95'C/pH8.69.5 >95'C/pH7.0
2 N98"C >2 h/10O0C/pH6.0 44 min/100"C 220 min/100"C 35 mid100"C 60 min/83"C 1 6 2 5 mifl5"C 107 mW75"C 2 N100"C 15 W 6 ' C 50 m i W C 30 min/85"C
75"C/pH4.5-6.5/2 M
100% active after 1 h/90°C
100
cs
101 102,103 104 see 9 105 106-108
109 C. S,Expr C, S,Expr
C, S, Expr. c, s
101 110-112 113,114 115 1 I5 114 1 I6 116 117 118 1 I9 120.12 1 122 123
salt
CH, = H,MFT dehydrogenase (H,-forming)
M. kandleri
>90"C/1.3M salt
CH, = H,MF'T reductase (F,,,-dependent)
M . kandleri
90"C/pH6.5-7.0 2.2-2.5 M SO:-, P O : -
F420 dependent NADP reductase
Archaeoglohus fulgidus (83'C)
80°C/pH8.0
Pyruvate: ferredoxin oxidoreductase Pyruvate: ferredoxin oxidoreductase Pyruvate: ferredoxin oxidoreductase Rubredoxin
A. fulgidus P. furiosus T. maritima P. furiosus
>9O0C/pH7.5 >90'C/pH8.0 >90OC/pH6.3 >95T
crude extract 100% active after 1 h/9o"C pure enzyme rapidly inactivated at 90°C rapidly inactivated at 90°C (no salt) 100% active after 1 h/9O0C (100 mM KZHPO,) 1 6 1 7 min/9O0C (no salt) stable at 90°C (I M K,HPO,) 60 min/9O0C (+2M KC1) 23 min/EO"C or 18 min/90"C 15 N80"C or 11 hD0"C stable for 24 h at 95°C
124 125 126 127 128,129 129 130 (Continued)
Table 2. Continued. Enzyme
Organism (optimal growth temperature) Enzyme thermophilicity
Transferases A. fulgidus ATF' sulfurylase 4-a-Glucanouansferase T. maritima DNA polymerase II (vent po1ymerase)T. litoralis DNA polymerase II P. furiosus CHO-H,M€T formyltransferase M.kandleri (98°C) 2-Phosphoglyceratekinase Hydrolases a-Amylase (extracellular) a-Amylase (intracellular) a-Amylase a-Amylase P-Amy1ase a-Amylase/glucoamylase Amylopullulanase Amylopullulanase Amylopullulanase ATPase complex Carboxypeptidase Endo- 1.4-gglucanase Endo-1.4-pxylanase Exo- 1,4-P-cellobiohydrolase PGalactosidase Exo- 1,4-p-glucanase p-Glucosidase p-Glucosidase a-Glucosidase CH, = H 4 m cyclohydrolase
90°C, pH8.0 70°C 75°C >75"C 90'C/pH6.5/ 2 M SO,", PO4%
Enzyme thermostability
3 h/80"C 7 W5"C 20 h/95"C unstable at 90°C (no salt) almost 100% stable with 1.5 M K,HF04
M. fervidus
P. furiosus P. furiosus P. woesi Themcoccus profundus (80°C) T. maritima T. maritima P . furiosus ES4 T. litoralis Pyrodictium occultum (105°C) S. solfntaricus T. maritima Thermotoga sp. (80°C) T h e m t o g a sp. S. solfataricus T. maritima Thermococcus celer (87°C) Thermotoga sp. P. furiosus A. fulgidus
10O0C/pH5.5
100"C/pH6.5-7.5 100'C/pH5 .5 8OoC/pH5.5-6.0 95"C/pH5.0 90"C/pH6.0 125'C/pH5.5 (+ 5mM CaCI,) 110-125°C 117'C/pH5.5 (+ 5mM CaCI,) 100°C 85'C/pH5.5-9 95'C/pH6.0-7.5 105"C/PH5.0-5.5 105°C/PH7.0 9YC/pH6.0-7.5 NDPH7.0 lIWC/pH5.0-6.0 85OC/pH8.5
3.2 h/llO°C 85% active after 3h at 100°C 4 h/lIO"C 4 h/9O"C (+ 5mM CaCI,) 30 min/90°C 30 min190"C 12 min/120°C (+ 5mM CaCl, and PG7) 20 N98"C (+ 5mM CaCI,) 5 min/120°C (+ 5mM CaCI, and PG7) 30 midl 10°C 13-14 min/90°C (holoenzyme) 14 min/80°C (apoenzyme) 2 h/95"C 90 minj95"C 70 min/lO8"C 24 W5'C or 3 W85"C 30 min/95"C 20 min/lO5"C 2.5 h/98"C (+4Opg/ml BSA) 4 6 4 8 W8"C unstable at 90°C. 100% stable after 50 min/9o"C with 1 M K,HPO,
General comments References
C, S . Expr. C, S, Expr.
131 see 9 see 9, 132 see 9. 133 134
C, S, Expr.
135
E C, S, Expr. E E
136 137,138 139
E E E
E E
MB
140
3 3 141 142 141 143 144,145 146 147 148 see 9 146 see 9 149 150 151 (continued)
W
L
0
Table 2. Continued. Enzyme Hydrolases (cntd.) CH, = H,M€T cyclohydrolase Protease Serine protease (pyrolysin) Sucrose a-hydrolase Thiol protease
Organism (optimal growth temperature) Enzyme thermophilicity
Enzyme thermostability
M.kandleri
95°ClpH8.011-2 M salt 10 0 T 115°C 105OC 1IO°C/pH7.0 ND/pH7.0
100% active after 60 min/90"C 1.5 W5"C 4 N100"C or 20 min/lO5T 48 W5"C 60 min/100"C 4 W8"C (+4Opg/ml BSA)
active up to 100'C 85-90OC nd 105-110~C 9 7 ~ 1 p wI.
2 h/100"C nd 10 min/120°C 24 rmn/95"C
Desulfurococcus mucosus (85°C) P. furiosus P. furiosus Pyrococcus sp. (95°C) ~Xylosidaselarabmofuranosidase Thermofoga sp. Lyases, Isomerases, and Ligases DNA topoisomerase V M. kandleri Glutamine synthetase P. furiosus Glutamine synthetase P. woesei Xylose isomerase T. maritima Xylose isomerase T. neapolitana
General comments
References 152 see 9
MB
153 see 9
E MB
154 149,155 156
c,s C, S, Expr.
157 158
159 160
C: cloned, S: sequenced E extracellular; MB: membrane associated, Expr.: expressed in a mesophile; CHZ=H4MPT N5,N'a-methylenetetrahydromethanopterin;CHO-H,MPT: N5-formylmethanofuran-methanopteM, GAPDH Gl yceraldehyde-3-phosphate dehydrogenase.
11 reflects the lack of consensus on the way to measure these properties. These lists are not exhaustive (i.e., in most cases, only examples of each enzyme type are mentioned), and they focus on enzymes with potential biotechnological applications. Thus, some enzymes involved in membrane transport (amino acids, ions, etc.), cellular energy production (methanogenesis,respiration), heat-shock proteins, and wall proteins are omitted. An extensive list of enzymes purified from thermophiles (optimum growth at temperatures above 65°C) was published by Coolbear et al. in 1992 [ 101. For detailed descriptions of the individual thermozymes, see references [9,10,41]. Clostridium thermocellum's cellulolytic system has been extensively studied; fifteen endoglucanase genes, two xylanases genes, and two P-glucosidase genes have been characterized. Only three endo-p- 1,Cglucanases and one pglucosidase are listed in Table 1. However, all C. thermocellum cellulolytic enzymes are optimally active at 60-70°C and are thermostable [161]. Thermal properties of thermozymes Thermostability and thermophilicity are inherent properties of thermozymes. As seen in Tables 1 and 2, most thermozymes characterized from thermophiles and hyperthermophiles are optimally active at temperatures close to the host organism's optimal growth temperature. This well-established trend is particularly noticeable when comparing enzymes from thermophiles and from hyperthermophiles; while most enzymes characterized from thermophiles are optimally active at temperatures of 60-90°C (Table l), enzymes purified from hyperthermophiles are optimally active at 70-125°C (Table 2). This trend holds more stringently for enzymes within a single structural family (i.e., a-amylases, proteases, alcohol dehydrogenases (ADHs), glyceraldehyde-3-phosphatedehydrogenases (GAPDHs), etc.) than across a range of proteins, suggesting that stability is partly a function of the protein structural fold. The temperature at which some thermozymes, oxidoreductases in particular, are optimally active has not been determined because of substrate or coenzyme (e.g., NAD, NADP) instability. Table 3 shows the thermal and kinetic properties of xylose Table 3. Comparison of xylose isomerase thermal parameters with the optimal growth temperature of the respective microorganism".
Microorganism
Mesophile E. coli Thermophile T. thermosuljiurigenes B. stearothermophilus Hyperthermophile T. neapolitana
Organism
Xylose isomerase
Top: ("C)
T,,,'
31
55
<1.0
52
53
60 65
80 85
45
nd
80 93
80 85
80
95
330
120
90
("C)
half-life (min at 85°C) T$ ("C)
Tp; ("C)
'Adapted from [92]; borganism optimal growth temperatures (Tq) are from the DSM catalog; 'temperature of maximal activity; dtemperature of 50% enzyme unfolding; "temperature of 50% enzyme precipitation. nd: not determined.
12 isomerases isolated from a mesophile, two thermophiles, and a hyperthermophile, as well as the host organism's optimal growth temperature. All the thermal characteristics presented in this Table (i.e., temperature for maximal activity, half-life at 85"C, precipitation initiation temperature, temperature of 50% unfolding) increase with the host organism's optimal growth temperature. Since the four enzymes studied are recombinant enzymes expressed in Escherichia coli, their thermal properties are clearly genetically encoded, and have adapted to the thermal environment of their bacterial source. All extracellular and cell-bound thermozymes (i.e., saccharidases and proteases) listed in Tables 1 and 2 are optimally active at temperatures above or far above the host organism's optimum growth temperature, and are, as a rule, highly stable. For example, Thermococcus litoralis amylopullulanase is optimally active at 117OC, 29°C above T. litoralis' optimum growth temperature (88°C). While they are usually less thermophilic than extracellular enzymes purified from the same host, intracellular enzymes (such as xylose isomerases, in Table 3) are often still optimally active at the organisms optimal growth temperature. Few exceptions to this trend exist for intracellular enzymes, such as Sulfolobus solfataricus glutamate dehydrogenase [ 1091 and Methanopyrus kandleri methylenetetrahydromethanopterin (CH, = H,MPT) dehydrogenase [ 1231which are optimally active 20°C below the organism's optimum growth temperature. In their comparison of mesophilic, thermophilic, and hyperthermophilic xylose isomerases, Tchemajenko et al. (Table 3 and [92]) showed that while xylose isomerases thermal properties were closely related to the organisms thermal environment, their kinetic parameters (i.e., specific activity, activation energy, and preexponential factor) showed no correlation whatsoever with the thermal properties of both the enzyme and the organism. This observation suggests that enzyme thermal properties evolve independently from the kinetic properties. The temperature range for significant thermozyme activity is often narrow. The two T. maritima amylases characterized by Schumann et al. (1991) [3] show 50% maximal activity or more between the ranges 73-97°C and 83--100"C, respectively. Their narrow activity range makes thermozymes typically inactive or poorly active under mesophilic conditions. At 40"C, the two T . maritima amylases show ( t 5 % activity [3]; T. maritima lactate dehydrogenase shows less than 5% activity [162]; and P. furiosus and T. litoralis amylopullulanases show no detectable pullulanase or amylase activity [141]. While most pure enzymes are intrinsically very stable, some intracellular thermozymes get their high thermostability from intracellular environmental factors such as salts, high protein concentrations, coenzymes, substrates, activators, or general stabilizers such as thermamine (see Tables 1 and 2). The stability of Thermus caldophilus lactate dehydrogenase at 100°C was increased 4-5 times in the presence of its activator fructose 1,6-bisphosphate [ 1631. M. kundleri cells contain high concentrations of potassium and 2,3-diphosphoglycerate [134] (>1 M and 1.2 M, respectively). Thauer and his colleagues have done detailed studies of salts effects on the stability and activity of M. kandleri methanogenic enzymes [123--125,134,1521.
13 The five enzymes characterized show a strong activation in the presence of salts (see Table 2). In that respect, the F,,,,-dependent CH, = H,MPT dehydrogenase and N5formylmethanofuran: tetra hydromethanopterin (CHO-H,MPT) formyltransferase activities show an extreme behavior, since they are completely inactive in the absence of salts [123,134].(See section - Multiple substitutions and protein stabilization - for a more detailed analysis of the effect of salts on the M . kandleri methanogenic enzymes). The discovery that thermophilic organisms did not grow faster than mesophilic organisms suggested that these organisms may have an added metabolic burden from the high turnover of nonthermostable macromolecules (e.g., proteins, lipids, and nucleic acids). The isolation of highly stable thermophilic enzymes from these organisms indicates that the factors limiting growth rates at high temperatures have to be found elsewhere, such as in membrane stability and function at high temperatures [ 1641.
Catalytic mechanisms It is commonly accepted that high temperatures decrease enzyme catalytic efficiencies. It seems clear that an increased temperature causes the decrease of one enzyme's affinity for its substrates. Hecht et al. (1989) [162] have shown a dramatic decrease of T. maritima lactate dehydrogenase's affinity for both NADH and pyruvate with a rising temperature. The same effect is seen with Thermotoga neapolitana xylose isomerase [ 1601 and P.furiosus glutamate dehydrogenase (see Table 4). But the effect increasing temperature has on enzyme affiity for its substrate does not mean that all thermophilic enzymes will have low affinities at high temperatures. Table 4 compares the Michaelis constants of mesophilic and thermophilic glutamate dehydrogenases. Table 5 does the same with mesophilic, thermophilic, and hyperthennophilic type I1 xylose isomerases. In Table 4, the four glutamate dehydrogenases characterized from mesophiles (E. coli, Salmonella typhimurium, Nitrosomonas europaea, and Table 4 . Kinetic constants for glutamate dehydrogenases purified from mesophiles and thermophiles.
Organism
E . coli S . typhimurium N . europaea T. novellus "S.solfataricus 'Thermococcus NA 1 "P.furiosus
Temperature tested ("C)
25 25 23-25 30 60 80 60 85
Km (mM) L-glutamate 2-0xoglutarate NH:
NADPH NADP
1.3 50 6.7 35.5 2.5 9.12
0.64 4.0 4.3 7.4 1.4 1.7
0.04 0.019 0.049 0.077 0.01 0.066
0.6 1.6
0.33 4.2
1.1 0.29 16 7.5 4.2 15.5
6-27 0.012 9.5 nd
Specific Ref. activity (pmol/ midmn)
0.042 0.013 b0.0079 0.08 0.05 0.038
217 340 1030 nd 35 nd
165 166 167 168
0.02 0.18
nd 130
106 107
109
169
These glutamate dehydrogenases utilize both NADPH and NADH as cofactors, but utilize preferentially NADPH. %e reverse reaction (NADP' reduction) is inhibited by NADPH and NH,'.
14 Table 5 . Kinetic constants for type I1 xylose isomerases purified from mesophiles and thermophiles".
Organism
Temperature Km for xylose Vmax Kcat (mM) (pmol/min/mg) (l/min)
("C)
E. coli B. stearothermophilus T . thermosuljiirigenes T. saccharolyticum T. maritima T. neapolitana
22-25 60 65 65 90 90
10 100 20 16 74 16
6 44.5 15.7 17.6 68.4 52.2
300 4900 3140 3520 8550 2654
Kcat/Km Ref. (l/mMmin) 30 49.0 157.0 220.0 72.5 166.8
170 171 93 93 159 160
"Activitiesfor conversion of xylose to xylulose are reported.
Thiobacillus novellus) show high heterogeneous affinities for their different substrates. The constants calculated for the thermophilic enzymes (S. solfataricus, Thermococcus, and P . furiosus) - even determined at much higher temperatures - are still in the range of the constants of the mesophilic enzymes. As seen in Table 5, a similar observation can be made for xylose isomerases; no significant decrease in the hyperthermophilic xylose isomerase affinities for xylose is observed when comparing them to their thermophilic counterparts. The second intervening factor in the determination of an enzymes's catalytic efficiency is k,, (indirectly, the reaction maximum velocity, V,,,). Following the Arrhenius equation, V,, increases with temperature. The kinetic constants of T . neapolitana xylose isomerase have been determined for the substrates xylose, glucose, and fructose at different temperatures [160]. As is the case with T. muritima lactate dehydrogenase, T. neapolitana xylose isomerase's K,,, for substrates increases with temperature. However, significant V,, and k,, increases compensate for the K,,, increase to a great extent, keeping the catalytic efficiency in a similar range at different temperatures. While an enzyme's K,,, and V,, increase with temperature, this observation can not be extrapolated to the comparison of related enzymes k,,, behaviors at different temperatures. The comparison of the V, determined for mesophilic and thermophilic adenylate kinases illustrates this remark well. E. coli's adenylate kinase has a V,, of 1020 pmol/min/mg protein at 30°C [172], Bacillus stearothermophilus and S . solfataricus enzymes have a V,, of 288 at 37°C [ 1721 and 433 at 70°C [39], respectively. E. coli and B. stearothermophilus adenylate kinases have similar specific activities (2,400 and 3,000 pmol/min/mg, respectively) at their respective optimal activity temperatures (45 and 65°C) [172]. In Table 5, B . stearothermophilus xylose isomerase has also almost the same V,, at 60°C (44.5 pmol/min/mg) as the T. neapolituna enzyme at 90°C (52 pmol/min/mg). As more kinetic data accumulate on thermophilic enzymes, it becomes evident that, despite their activity at high temperatures, themophilic enzymes catalyze reactions with V,, and K,,, values similar to those of their mesophilic counterparts at their respective optimal temperatures. Thermophilic enzyme-catalyzed reactions do not show the high catalytic efficiencies that were first expected for high temperature reactions. T. maritima and T. neapolitana xylose isomerase catalytic efficiencies (72.5
15 and 166.8 l/mM min, respectively) are in the range of the catalytic efficiencies of their less thermophilic counterparts (see Table 5). While disappointing from a biotechnological standpoint, the fact that thermozymes are not more active than mesozymes at high temperatures still does not explain the lower growth rates of thermophilic organisms. This lack of catalytic rate optimization in thermophiles is still not well understood. Is it due to the increasingly difficult molecular recognition at high temperatures? Are reaction rates limited by a minimum time requirement for adequate fidelity in protein molecular discrimination? Do membrane transfer rates or other cellular compounds’ instabilities eventually limit the rate at which cytoplasmic metabolic processes can proceed, practically negating the advantage of further maximizing enzyme efficiency? It is now currently accepted that thermophiles represent the most ancient form of life on Earth, and that mesophiles evolved later [30]. According to evolution rates calculated from 16s rDNA comparisons, hyperthermophiles, including the eubacteria Thermotogales, are among the slowest evolving organisms [ 11,301. Thermophilic and hyperthermophilic proteins might represent less-evolved proteins; or thermostabilization of these enzymes might also limit their catalytic efficiency at optimal temperatures.
Cloning, and expression in rnesophiles An increasing number of genes from thermophilic and hyperthermophilic organisms have been cloned and expressed in mesophiles. Cloning genes from thermophilic bacterial species is facilitated by their direct expression in mesophilic hosts: 1) B. stearotherrnophilus xylanase, Therrnoanaerobacter ethanolicus amylopullulanase,and numerous C . therrnocellurn cellulases were cloned by activity screening of genomic libraries expressed in E. coli [56,173,174]; 2) B. stearotherrnophilus neopullulanase was cloned by activity screening of a genomic library expressed in Bacillus subtilis [75]; and 3) other genes, like Therrnoanaerobacterium saccharolyticurn xylose isomerase, were cloned by direct complementation of E. coli mutants [94]. The absence of sequences recognizable as E. coli promoters has made cloning of Therrnus genes difficult. A typical example was the cloning of T . aquaticus Taq DNA polymerase [42]. Most Therrnus genes were expressed in E. coli as fusion genes (T. aquaticus L-lactate dehydrogenase) or fusion proteins (Taq1 restriction endonuclease) (reviewed in [41]). Koyama and Furukawa (1990) [175] cloned the T. therrnophilus tryptophan synthase genes by direct plasmid transfer of a genomic library from E. coli to a T. therrnophilus trpB strain and complementation. T. therrnophilus trpBA genes could only complement the E. coli trpB or trpA mutants when expressed under the control of the E. coli lac promoter [168]. With the exception of the T . rnaritirna L-lactate dehydrogenase gene cloned by the complementation of a D-lactate dehydrogenase E. coli mutant [ 1211, most genes from hyperthermophiles have been cloned by hybridization. The typical procedure is to determine one or more amino acid sequences in the native protein. The oligonucleotides reverse-translated from these peptides are used either as direct probes or as primers to synthesize a PCR fragment (which is used to probe a genomic library). While Therrnotoga genes are usually directly expressed in E. coli (T. maritima L-
16 lactate dehydrogenase [ 1211, T. neapolitanu xylose isomerase [ 160]), archaeal genes are not, and are only expressed in E. coli under the control of strong promoters (plac, ptac, or T7 RNA polymerase promoter) [113,135,138,176]. The P. furiosus DNA polymerase gene was directly expressed from a 7.0 kb clone, but not from the archaeal promoter [ 1331. Cloning the T. litoralis DNA polymerase (Vent polymerase) gene in E. coli allowed Perler et al. (1992) [ 1321 to discover the protein intervening sequences (IVSs) in an archaeal gene. These IVSs are removed from the protein precursor by protein splicing [177]. Vent polymerase could be expressed in E. coli only under the control of a tightly regulated promoter and after the deletion of IVS1 [ 1321. Despite being synthesized in a mesophilic host, recombinant thermophilic enzymes usually have kinetic and thermal stability characteristics identical to those of the native protein. T. maritima and Pyrococcus woesei recombinant GAPDHs were extensively studied. The T. maritimu recombinant GAPDH had the same fluorescence emission and circular dichroism spectra, the same guanidine-hydrochloride-dependent denaturation and heat-inactivation behaviors, the same Michaelis constants, the same allosteric inhibitor effect, and same specific activity as the native enzyme [ 1 1 11. P . woesei recombinant GAPDH had the same heat-inactivation behavior, the same Michaelis constants and V,, and the salt effect on stabilization as the P. woesei native enzyme [113]. Thus, proper thermophilic protein folding is not strongly temperature dependent, and all the information necessary for protein thermophilicity and thermostability is encoded in the primary structure of the enzyme. Chaperones have been described in hyperthermophilic organisms and they appear to play a role in the heat-shock responses of thermophiles [ 143,1781. In a few cases recombinant thermophilic proteins are expressed in E. coli, but are completely inactive. The two Clostridium thermoaceticum dimeric recombinant proteins, carbon monoxide dehydrogenase and corrinoid/Fe-S protein, were inactive, and their subunits were not properly assembled [ 1791. P. furiosus a-amylase expressed in E. coli was active, but had an aberrant molecular weight, suggesting an improper subunit assembly at low temperatures [ 1381. Since other thermophilic polymeric enzymes have been successfully expressed in E. coli [160], these few failures might be related to the effect the different E. coli intracellular environment (salt concentration, pH, 0,) has on specific protein assembly. An attempt was made to express hyperthermophilic ES4 glutamate dehydrogenase in vitro in a eukaryotic cell-free expression system. Most of the protein synthesized remained monomeric, and the fully assembled hexameric form was completely inactive [105]. Cloning genes from thermophiles has sometimes been extremely helpful when studying individual proteins. The enzymes from the C . thermocellum cellulosome are a well known example. The cellulosome of C. thermocellum is a complex extracellular structure composed of many different proteins. The drastic conditions required to dissociate the different cellulosome constituents irreversibly denature most of them. Cloning the individual genes allowed the characterization of 15 endoglucanases, two P-glucosidases, two xylanases, one lichenase, as well as the cellulosome scaffolding protein, CipA (see [174]). Another advantage of cloning and expressing a ther-
17 mophilic protein in a mesophilic host is the possibility of introducing a powerful heat-treatment step in the purification procedure to denature most mesophilic host proteins. Heat treatments of recombinant E. coli strains for 10 min at 70°C and 30 min at 80°C allowed a 7-fold and a 13-fold purification of T. saccharolyticum endoxylanase [69] and M. fervidus 2-phosphoglycerate kinase [ 1351, respectively. Heat treatment at 15°C for 15 min of T. thermosulfurigenes xylose isomerase expressed in E. coli resulted in a 5-fold purification of the T. thermosulfurigenes enzyme [93]. Thus, heat treatment of certain thermozymes expressed in mesophiles can yield “industrially pure” enzymes (i.e., S O % pure) in a single step. Sequence comparison and amino acid composition As more and more sequence information is gathered on thermophilic and thermostable proteins, most evidence indicates that thermophilic proteins are homologous to their mesophilic counterparts. Thermozymes contain the same catalytic consensus regions [ 105,122,130,160], and, in many cases, thermophilic enzyme structure computer modeling (or direct structural analysis) reveals that they share the same structural backbone as their mesophilic counterparts [ 121,1301. No discrete structural features unique to thermozymes can account for their activity at elevated temperatures and their resistance to heat denaturation. Thermozymes are composed of the 20 natural amino acids, and their thermal properties reside in subtle sequence differences. Several attempts have been made to define amino acid “traffic rules” for high temperature molecular adaptation. Several hypotheses try to explain the increased stability of thermostable proteins: 1) a-helix stabilization by substituting helixstabilizing residues (such as alanines) for helix-destabilizing residues (most often glycines) [ 1801; 2) short-loop and p-turn rigidification by introducing prolines [1813; 3) average protein hydrophobicity increase [47,113]; 4) salt-bridge number increase [182]; and 5 ) decrease in cysteine, lysine, and deamidable residues. However, as more genes from thermophiles and hyperthermophiles are cloned and sequenced, examples accumulate of thermozyme sequences contradicting these rules. Protein alignments and amino acid composition comparisons, more statistically significant with todays data, clearly show that no single, simple protein thermostabilization traffic rule can be defined. No trend in thermophilic GAPDHs’ amino acid compositions could be identified from the thorough comparison of GAPDH sequences from 26 different organisms spanning an 82°C growth temperature range [183]. In Table 6 we compare the amino acid contents of mesophilic, thermophilic, and hyperthermophilic enzymes from nine different enzyme families. No prominent general trend is observed and it is highly probable that features that account for protein thermostabilization differ from one enzyme class to another. Enzymes from hyperthermophiles, in particular extracellular enzymes stable at temperatures above 100°C, show a high resistance to amino acid covalent modification. In Table 6, with a few exceptions, enzymes from hyperthermophiles show a slight decrease in their cysteine or deamidable residues contents compared to their mesophilic and thermophilic counterparts. This trend is particularly noticeable among type I1 xylose isomerases. In Fig. 1, the Asn + Gln contents of four type I1 xylose isomerases was plotted against their temperature for
Table 6. Amino acid composition comparison of enzymes from mesophilic, thermophilic, and hyperthermophilic prokaryotes. Protein Class
Gly Pro Ala
Hydrolases a-amylases 8.5 4.5 mesa (13) them (4) 8.9 4.8 6.8 4.3 hyper (1) neutral proteases rneso (11) 7.7 3.7 12 3.3 them (2) Oxidoreductases 10 ADHS meso ( 5 ) 8.2 7.4 them (1) 10 4.7 11 3.5 hyper ( I ) 2" ADHs meso (2) 11 3.5 them(1) 12 6.2 Glu DHs 11 3.6 meso (4) hyper (3) 8.1 4.6 LDHs rneso (10) 6.8 4.6 them (5) 9.4 3.6 9.6 3.5 hyper ( I ) MDHs meso (1) 11 4.5 8.2 4.9 them (1) 9.7 4.4 hvwr (1)
Val
Leu
Ile
F'he
Trp
Tyr
His
Cys
Met
Ser
Thr
Lys
Arg
Asp
Glu
Asn
Gln
Asn
3.7 5.2 3.6 4.0 10 3.7
4.5 3.3 2.3
9.7 7.3 6.0
8.8 8.1 4.0
5.7 6.4 9.2
8.0 6.9 9.1
4.4 3.8 3.8 4.4 6.3 5.7
2.9 4.0 2.6
4.1 5.9 6.3
2.5 2.3 2.2
1.4 2.0 0.75 1.9 0.31 1.8
7.6 5.6 5.1
6.8 8.3 1.8
4.1 5.1 7.4
5.5 4.1 5.4
5.8 7.4 5.4
8.4 9.4
6.6 7.7
7.4 7.0
4.2 3.2 4.3 3.0
0.90 5.2 1.3 6.9
2.3 2.0
0.57 1.2 0.37 1.9
8.6 6.2
6.7 6.3
7.3 3.7
4.4 4.8
5.9 6.0
4.7 4.2
6.8 5.4
4.3 4.3
11 6.4 9.9 12 8.4 7.5
7.4
1.4 0.88 0.63
2.1 2.3 2.1
1.6 2.0 2.3
7.2 3.4 4.8
5.3 3.5
5.0 7.5
7.9
4.9 2.7 6.4 1.9 8.9 3.2
5.1
5.5
8.4 3.8 4.9
5.0 4.7 5.7
4.0 7.0 4.9
3.8 2.9 5.3
2.5 3.2 2.7
8.4 7.5 9.9 10
7.9 6.5
8.9 3.2 7.4 4.0
0.63 3.0 0.14 0.17
2.1 2.8
2.3 3.0 0.11 4.3
4.8 5.1 0.26 3.7
5.5 6.8
4.9 3.1
5.7 5.7
4.9 6.0
5.3 2.8
9.5 9.4
7.0 5.5
4.9 4.0 7.5 2.3
1.6 2.4
3.1 4.8
1.5 1.2 3.6 0.87 0.32 3.4
4.9 3.4
5.0 5.7
6.1 8.1
4.5 4.7
4.6 6.0
10 6.8 8.8 12 10 8.9 8.6 9.5 10
7.3 3.8 6.6 3.7 7.5 4.0
1.1
0.95 nd
3.1 3.3 2.8
2.4 2.2 2.0
6.1 4.1 4.1
5.4 4.1 4.9
6.3 3.5 6.5
4.9 6.8 4.3
5.1 5.6 nd
0.96 0.96 0.13 5.8 5.4 0.12 0.30 3.4 0.28 4.3 0.18 0.59 0.26 4.7 5.0
7.0
0.26 5.8 0.24
3.8 5.2 5.6
8.8 8.4
8.5
r
00
2.6 3.5 3.0
12 10 10 6.4 0.29 0.0 0.13 15 8.6 9.2 5.2 3.0 0.12 0.21 7.7 6.5 8.0 11 0.26 0.29 0.24
2.8 1.7 3.0
0.94 2.4 0.38 1.3 2.0 nd
5.5
9.7
+ GW-phob
Polar Charged
11
29 28 34
47 55
19 17
33 28
6.3 6.1 8.0
52 54
19 13 17
29 29 29
2.7 0.11
8.0 2.9
54 60
17 14
29 26
7.8 4.3 8.3 3.6
3.5 2.5
7.8 6.1
53 53
16 14
31 33
6.0 6.8
4.5 3.2 nd
3.6 2.6 nd
8. I 5.8 10
52 59 56
17 12 13
30 28 32
3.5 3.4 5.3
4.2 3.7
7.7 7.1
58
57
14 12
28 31
52
15
33
nd
6.4 7.3 8.8
0.88
11
9.7
51
53 54
58
20 19
~~
(continued)
Table 6. Continued. Protein Class
Isomerases D-xylose isomerases meso (8) therm (3) hyper (1) Ligases Gln synthetases meso (12) therm (1) hyper (2)
Phe Tm Tvr
His Cvs Met Ser Thr Lvs
Are
Asu
3.4 4.1 2.0 2.5 5.1 8.0 1.1 4.8 5.6 7.9 0.11 4.3
3.2 2.6 2.5
1.5 2.1 4.4 5.1 0.53 2.7 3.8 5.3 0.68 0.18 2.7 5.6
4.4 8.6
9.6 3.9 4.7
5.6 7.7 7.2
5.6 5.0 5.5 4.7 7.2 5.5
2.7 2.1 2.4
1.2 3.0 5.7 5.3 0.21 2.3 5.5 5.9 0.34 2.6 4.4 4.0
6.9 6.6 7.3
4.3 5.1 4.7
6.0 6.8
Glv
Pro
Ala
Val Leu Ile
8.7 6.5 7.6
5.8
11
5.5
3.0 3.6
9.0 4.8 7.9 5.0
8.8 7.4 9.0
7.0 7.2 7.3
5.3 6.8 5.9
8.1 6.2 6.8 6.6 7.1 6.3
8.3 8.7 8.2
1.1 3.8 0.13 5.9 1.4 5.1
8.5
5.8
Asn Gln
Asn
5.3 3.1 3.9 7.5 5.1 2.4 9.0 3.4 0.18
+ Gln
H-uhob Polar
Chard
7.0 6.5 3.6
52 50 52
13
32 35 35
7.2 4.4 2.7 7.1 7.0 2.7 2.1 4.8 10 3.5 0.91 4.4
50 54
18 16 14
32 30 32
Glu
54
16 15
meso: enzymes from organisms with optimal growth temperatures below 60°C; therm: enzymes from organisms with optimal growth temperatures from 6040°C; hyper, enzymes from organisms with optimal growth temperatures above 8 0 T . Numbers in parentheses indicate the number of proteins represented in the category. GenBank database accession numbers or specific references for the proteins used in the comparison: Primary alcohol dehydrogenase (1' ADH): meso - D9ooo4, L02104. M91440, X17065, X59263; therm - D90421; hyper - S51211. Secondary alcohol dehydrogenase (2' ADH): meso - J03362, M84723; therm - [36]. Lactate dehydrogenase (LDH): mesa - D13405, L29327, M22305, M72545, M82881, M95919, X01067, X.55118, X55119.222737; therm - w O 5 8 5 , M19394, M19396, M28336, X04519; hyper - ref #320. Glutamate dehydrogenase (Glu DH): meso - K02499, M24021, M65250, M76403; hyper - L12408, L19995, M97860. Malate dehydrogenase (MDH): meso - M95049 them - J02598; hyper - X51714, Glutamine spthetase (Gln synth): meso - wO513, D10020, M5609, L08256, M14536, M16626, M18966, M22811, M26107, M57275, XO4880, X.53509: therm - X53263; hyper - X60160, X60161. Neuwal protease: meso - D10773, K01985, K02497, M36694, M36723, M62845, M64809, M83910, X61286, X73315, X75070 therm - Ml1446, M63575. a-amylase; mesa - J01542, K00563, L19299, M18244, M24516, M25263, U04956, X07796, X12725, X12726, X52755, X.52756, X.55799; therm - M11450, M34957, M57457, X02769 hyper - L22346. Dxylose isomerase ( X I ) : meso L12967, M15050, M36269, M73789, M84564, X00772, X02795, X61059; therm - J05650,LO9699, M91248; hyper - L38994. n d not determined.
20
0 20, 50
.
,
I
I
I
60
70
80
90
.
100
Temperature for maximal activity (“C) Fig. I. Number of asparagine + glutamine residues per xylose isomerase monomer as a function of the temperature for maximal enzyme activity.
maximal activity [160]. Despite the small number of enzymes plotted, Fig. 1 clearly shows an inverse relationship between Asn + Gln content and temperature for maximal activity. More information is needed on proteins from thermophiles and hyperthermophiles to confirm this trend in a statistically significant way. Contrary to the early belief that thermophilic enzymes gained their thermostability from a different amino acid composition, enzyme thermostability is enhanced by numerous subtle sequence differences. These differences protect the highly labile residues and bonds (by modifying their immediate environment), they limit destabilizing interactions, and they restrain protein flexibility. Other environmental elements can influence how protein thermostability develops. Whereas the bacterial cell cannot control its intracellular temperature, it actively controls its intracellular pH and osmotic pressure. For example, thriving in acidic conditions (pH 2-3), Sulfolobus acidocaldarius maintains its intracellular pH between 5.3 and 6.5 [184]. Thus, extracellular and intracellular enzymes, while facing the same temperatures, are not in the same pH and osmotic environments. At the same temperature, proteins are relatively less susceptible to cysteine degradation in an acidic environment than in a neutral environment [ 1851. Nonenzymatic hydrolysis of Asp-Xaa bonds is more likely to happen at acidic pHs [185]. These factors, as well as salt concentrations, might influence the way extracellular and intracellular proteins have evolved in the same organism.
21
Molecular mechanisms of enzyme thermostability and thermophilicity General thermal properties of thermozymes Protein jlexibiliry and rigidity It is currently believed that enzymes should be flexible enough to allow the adequate resonance or conformational changes associated with efficient catalysis. Thus, the total system energy favors or hinders catalytic efficiency (based on whether the enzyme is below, at, or above the temperature necessary to maintain the optimal enzyme catalytic structure). Beyond an enzyme-specific temperature limit, excessive intramolecular motions result in inactive structures. Thermophilicity - the optimal activity temperature - is defined by the upper limit of the energy range in which an enzyme’s average conformation and active dynamic states are preserved. With few exceptions, Arrhenius plots for enzyme activity are linear for all psychrophilic, mesophilic, and thermophilic enzymes. For example, the four xylose isomerases studied in Table 3 (i.e., E. coli, T. thermosulfurigenes, B . stearothermophilus, and T . neapolitana enzymes) show linear Arrhenius behaviors (data not shown) [92]. As for chemical reactions, linear Arrhenius plots predict that, in an enzyme’s active range, enzymatic activity is driven by the temperature-dependent substrate kinetic energy variation alone. If enzyme structures changed in a catalytically significant manner with increasing temperature, one would expect to find: 1) nonlinear Arrhenius plots for most enzymes, and 2) different types of plots for different enzyme classes. The nature and extent of the catalytically significant structural changes are expected to be different for different enzyme classes. In their respective temperature ranges, psychrophilic enzymes, mesozymes, and thermozymes are more rigid at low temperatures than at higher temperatures (G. Petsko, personal communication). Despite thermozymes being significantly more rigid than their mesophilic counterparts at room temperature [ 1861, increased rigidity at low temperatures and linear Arrhenius behavior are still general mechanisms affecting all enzyme categories. Rigidity is an essential factor in protein thermostability. Optimally active under more denaturing conditions (i.e., at higher temperatures) than mesozymes, thermozymes need to be more rigid than mesozymes. Their higher rigidity is essential for preserving their catalytic active structure, and protects them from unfolding. With a similar amino acid composition, thermozyme and mesozyme total molecular energies are the same. We propose that at elevated temperatures in the presence of higher kinetic energy, thermozymes channel more of their molecular energy toward increasing their rigidity (rather than toward catalytic rate optimization). This thermodynamic trade-off would explain why thermophilic enzymes show lower catalytic rates than is expected from extrapolating the catalytic rates of their mesophilic counterparts. While thermozymes optimize their rigidity for hightemperature stability, they might not be catalytically optimized enzymes. One current hypothesis is that at low temperatures the active site is “frozen” in a conformation not optimized for activity. Temperature-dependent increased active site
22
flexibility is required to reach a catalytically optimal structure. Since they are highly complicated catalysts, enzyme activity depends on local vibrations, or movements, that involve a whole protein region surrounding the active site residues. These movements are supposed to be directly related to temperature-dependent increased flexibility. According to this hypothesis, it is enzyme flexibility, in addition to substrate energy variations, that controls the activity in a temperature-dependent manner. Whereas rigidity (which limits excessive flexibility and unfolding) is directly involved in enzyme thermostabilization, no experimental evidence exists correlating poor enzyme activity at low temperatures with high structural rigidity. In an alternative hypothesis, substrate kinetic energy variations, by themselves, can explain linear enzyme Arrhenius plots. Since molecular mechanisms involved in enzyme stability and flexibility are not well characterized, and since their effects on enzyme activity are unknown, it is premature to consider any role of enzyme rigidity/flexibility in the control of temperature-dependent enzyme activity. Protein thermostability Protein thermostability is the protein's resistance to irreversible thermal inactivation. Inactivation is due either to protein chemical modification which leaves the overall protein structure intact, or to protein denaturation (i.e., irreversible unfolding). Denaturation is believed to be a rapid cooperative process, once a threshold molecular energy is reached. Enzyme-specific,this threshold is related to the protein characteristic stabilizing energy in a particular environment. Thermostability is typically measured as the fraction of activity remaining (determined by standard enzyme assay in the enzyme's activity temperature range) in a sample after heating it at a constant temperature for a specific period of time. While these measurements do not predict enzyme inactivation mechanisms, studies have shown that partial inactivation of a thermophilic secondary (2") ADH corresponds to the complete inactivation of the precipitated enzyme fraction. The soluble fraction remains fully active [36].These studies are consistent with the strong cooperativity of protein folding and unfolding and with protein folding intermediate instability [187,188]. They also agree with the general model proposed by Tomazic and Klibanov for enzyme thermoinactivation [189]: native enzyme e nonnative (active) (inactive)
+ scrambled structures
(scheme 1)
(often precipitate)
Tomazic and Klibanov's model is also consistent with an intramolecular ratedetermining step in thermal inactivation. (The representation of the residual activity natural log plotted vs. time is usually linear, fitting a pseudo-first-order equation (Eqn. 1, (with k = rate constant and t = incubation time)). ln(residua1 activity) = -kt
(1)
No protein concentration term is present in Eqn. 1, suggesting that the rate-
23 determining step is intramolecular. The only way to ensure that a pseudo first-order rate law reflects a truly intramolecular process, however, is to measure thermal inactivation at various initial protein concentrations, verifying that k remains constant. Protein stability is due to numerous ionic and nonionic interactions within the protein molecule and between the protein and the environment [186]. While some thermozymes have been shown to be stabilized by glycosylation [190,191], most recombinant thermozymes - expressed in mesophilic hosts, in the absence of glycosylation, and in different cellular environments - remain as thermophilic and thermostable as the native enzymes [ 111,1131. This observation demonstrates that thermophilicity and thermostability are encoded in the peptide sequence; they are not a consequence of posttranslational modifications, or of noncovalent interactions with cellular components. Mesophilic protein inactivation results from irreversible denaturation, due to the breakage of numerous noncovalent interactions rather than to protein chemical decomposition [ 1921. The original protein conformational information remains intact in many denatured proteins. The protein stabilizing energy (usually of the order of 3 M 5 kJ/mol [193]) is 10-fold smaller than the magnitude of the opposing forces [194]. Since each additional hydrogen bond or salt bridge can contribute approximately 2-20 kJ/mol to the stabilizing energy, and since the possibilities to add hydrogen bonds or salt bridges in a protein are vast, it is believed that protein covalent modifications, rather than unfolding, determine the theoretical upper limit of protein thermostability. Ahern and Klibanov [ 1951 reported that peptide depolymerization becomes significant at temperatures above 100°C, but oxidative decomposition of cysteine, asparagine, and glutamine residues has been shown to limit the thermostability of some enzymes to temperatures below 1OOOC. The identification of enzymes active at temperatures above 12OOC suggests that the protein structural noncovalent interactions are strong enough to allow stability to approach the theoretical maximum temperatures defined by peptide bond destruction.
Protein thermophilicity Because of subangstrom variations in atom positions known to be tolerated in a functional active site, it is believed that, within its active temperature range, an enzyme maintains its average structure within strict limits. Therefore, while protein denaturation is not directly involved in enzyme thermophilicity, structural interactions which oppose denaturation are very important for thermophilicity. Arrhenius plots for thermophilic and mesophilic enzymes are typically linear, suggesting that mesophilic and thermophilic enzyme functional architectures are tightly controlled throughout their respective temperature ranges (significant structural changes that alter the functional architecture are expected to cause non-Arrhenius behaviors). Differential scanning calorimetry measurements indicate that the protein’s heat capacity remains reasonably constant within the catalytically active temperature range, suggesting the absence of significant structural changes. Biphasic Arrhenius plots reported for some mesozymes and thermozymes [ 110,115,1961represent an important exception to the typical Arrhenius-like behavior. In the cases of yeast and Thermoproteus t e r n GAPDHs, it has been proposed that the shift extent indicates the degree of enzyme
24 thermostability and thermophilicity [1151. The thermophilic T. ethanolicus 2" ADH also shows biphasic Arrhenius plots. This enzyme's behavior, however, is substrate specific, since it occurs with ethanol as a substrate, but not with 2-propanol [36]. Biphasic Arrhenius plots have been generally attributed to functionally significant enzyme structure alterations, in addition to other mechanisms which do not involve altered catalyst structures [197]. Interpretations of Arrhenius data must be verified by careful biophysical and kinetic experimentation. Similar Arrhenius plots obtained for thermozymes and mesozymes suggest that all proteins respond similarly to temperature. Discontinuities are not a specific trait of thermozymes. While the system energy increase is the same between 25 and 30°C and between 65 and 70"C, the total molecular kinetic energy is greater at high temperatures. Thermophilic and mesophilic enzymes similarly resist the structural changes associated to variations in their molecular energy, but thermophilic enzymes must do so under conditions of a higher total kinetic energy. In other words, thermozymes must be more resistant to denaturation, while having active dynamic states close to the active dynamic states of their mesophilic counterparts. Classical reaction-rate theory states that the rate of a chemical reaction increases with temperature, since more molecules have energies greater than the reaction activation energy. Catalysts enhance reaction rates by stabilizing reaction intermediates, thus reducing reaction activation energies. The rates of enzyme-catalyzed reactions should increase with temperature as long as the enzyme structure maintains the original activation energy and productivity of molecular collisions. If thermozymes and mesozymes similarly control these kinetic parameters, the maximal rate of a thermozyme should be higher than the rate of its mesophilic counterpart. Furthermore, solving the Arrhenius equation for activity relative to that at the maximal temperature yields Eqn. 2, where K,, = maximal reaction rate, KT = reaction rate at temperature (TT), E, = activation energy, R = gas constant, and T = reaction temperature.
Linear Arrhenius plots indicate that E, does not vary with temperature, so the relative activity at a temperature (TT)below the temperature of maximal activity (T,,,,,) is less than the maximal activity by a factor of ( l D T - l/Tma,).For a same temperature difference Tmax-TT, the (l/T, - 1/Tmm)will be smaller for high temperatures, predicting a broader active range for thermozymes than for mesozymes (Fig. 2). The prediction that, at higher temperatures, thermozymes maximal catalytic rates are higher than their mesophilic counterparts is not supported by experimentation, since maximal reaction rates reported for thermophilic enzymes are typically similar to those reported for their mesophilic counterparts, not higher (see section on: General properties of thermozymes (Tables 4 and 5)). Based on Eqn. 2, low activation energy reactions are predicted to retain a higher percentage of their activity at temperatures far below their maximal activity temperature. This prediction has been experimentally corroborated. B. sreurother-
25
1.o
a Ea=8okJhno/Tmw=m 0 Eia=80kJhn0&Tmw=37~C
0.8
0
I
~=mkJhnoLTmax=m h
3
3 Y
>
Ea=ZOkJhnol.T1~~=370C
0.8
I
11
a
I
P
0.4
0.2
0.0
-1 50
-1 00
-50
0
Change in temperature from maximal ("C) Fig. 2. Arrhenius dependence of reaction rate on the temperature and activation energy for mesophilic and hyperthermophilic enzymes. Tmax is defined as the temperature for maximal enzyme activity and Ea is the reaction activation energy.
mophilus D-xylose isomerase, for example (maximal temperature 85"C, E, = 50 kJ/mol), retains 10% of its activity at 40°C (45°C below its maximal temperature), whereas the T. neapolitana enzyme (maximal temperature 95OC, E, = 80 kJ/mol) retains less than 10% of its maximal activity at 65°C (30°C below its maximal temperature) [92]. T. ethanolicus 39E 2' ADH (maximal temperature 85-95OC, E, = 20-25 kJ/mol) retains more than 10% of its maximal activity at 25°C (6(t7OoC below its optimal temperature) [36], demonstrating that such behavior is not specific to D-XylOSe isomerases. Designing thermozymes that perform a desired reaction with a low activation energy and that will be active in a broader temperature range, represent an important potential for engineering biocatalytic processes. '
Protein folding Protein folding begins as the peptide is synthesized at the ribosome and involves the rapid condensation of particular regions, or nuclei, into native-like states. Folding is believed to be driven primarily by hydrophobic interactions, and the native protein structure is stabilized by a variety of hydrophobic, covalent, and coulombic interactions between parts of the protein and between the protein and the solvent [ 1981. Despite the staggering number of theoretical structures they may occupy, the fact that proteins fold rapidly into their active conformation indicates that they reach their final structure through a series of well-defined intermediate states. It also
suggests that folding is cooperative, and that partially folded intermediates are not stable. Thus, it is expected that the native state includes very few structural conformers of similar energy. This conclusion is supported by the conserved structures seen in X-ray crystallography and nuclear magnetic resonance (NMR). Extending this conclusion to protein folding energetics suggests that the free-energy well containing the protein native state is steep-walled and deep, comprising most of the free energy range that describes the actual folding pathway. Freire et al. [ 1991 proposed that at least two factors are specifically related to the unfavorable energetics of exposing complementary surfaces to the solvent in the partially folded/unfolded states: 1) the driving energy in protein folding, and 2) the protein stabilizing forces. They define complementary surfaces as surfaces which are not solvent-exposed in the native state, but which become solvent-exposed in the partly unfolded states. These regions compose parts of the folded core surfaces that remain condensed in partly foldedhnfolded proteins, and that are uniquely present in these intermediate states. Their model predicts that protein folding intermediates are highly unstable in aqueous solutions due to the exposure of hydrophobic complementary surfaces to the solvent, and that these states are poorly populated. This model relies on universal proteins characteristics and not on structural motifs (ahelices and P-sheets) specific to individual protein folds. This core model is supported by the fact that most proteins, regardless of their secondary structural composition, fold in a two-state process. Freire et al. [199] also noted that protein condensed phases that retain “...significant percent of the secondary structure content of the native state, exhibits considerable flexibility and has a highly disrupted tertiary structure...” are stable in certain conditions. The existence of these stable condensed cores indicates that secondary structural element interactions cannot, by themselves, control the protein folding pathway and folded protein stability. Haynie and Freire [ 188,200] proposed a thermodynamic model to determine the conditions maximizing the stability of protein intermediate folded states. Their model predicts that, in the presence of a denaturant, intermediate-state stability is independent from the freeenergy variation between the intermediate and the native state, and that, in the absence of denaturant, the enthalpy change of the intermediate state is a function of the thermodynamic parameters of the unfolded state alone. This model implies that, thermodynamically, protein folding is de facto a one-way process. The successive steps depend only on their thermodynamic properties, relative to the previous state and not to the subsequent one. Based on this model, they further proposed that, contrary to the current belief, the two-state protein folding process relies on a small entropy contribution to the intermediate state stability. This protein folding scenario is further complicated when the complete protein folding process is considered as a function of the total energies of the folded and unfolded states [201]. At 25°C and high dilution it has been established that nonpolar solute transfer into water was opposed mainly by entropy and not enthalpy. Under these conditions, protein folding can be considered to be entropy driven. However, because of the large, positive heat-capacity change (ACp) of non-polar-solute transfer to water [201,202], the enthalpy variation (AH) and the entropy variation (AS) for this
27 process cannot be considered temperature-independent. The free energy change (AG) of non-polar-solute transfer into water is predicted to be highest between 130°C and 160"C, where TAS=O [201,202]. At these temperatures the transfer is proposed to be completely enthalpy driven. Considering that the cell cytoplasm approximates a highdilution water solution, and that a protein approximates a small, nonpolar molecule, protein folding can be considered entropy driven only at mesophilic temperatures. Above these temperatures, both entropy and enthalpy contributions become significant. At temperatures where proteins are usually folded (37-9OoC), entropy and enthalpy contributions to the folding free-energy are expected to be approximately equal [202,203]. These contributions are consistent with experimental results [203], and may help explain why, despite the temperature dependence of entropy and enthalpy partitioning, thermophilic proteins are correctly folded when expressed in mesophilic hosts. These results suggest that protein folding is a robust process resistant to entropy and enthalpy variations. Also supported is the conclusion that the free-energy well describing the protein native state is deep (relative to the surrounding free-energy range describing accessible folding intermediates). Since folding intermediates do not necessarily contain local, condensed conformations identical to their counterparts in the native enzyme, the intermediates' free energies cannot predict the partial free energy of the corresponding region in the native protein [204]. Protein unfolding Protein unfolding is a process of fundamental importance in understanding protein stability. AG of unfolding directly measures the stabilizing energy of a folded protein. Protein destabilization by thermal energy and chemical denaturing agents has been well documented [205]. Extensive irreversible denaturation (i.e., loss of active architecture that is not recovered by the removal of the denaturing force) is far more common than extensive reversible denaturation (where active structure is regained upon removal of the denaturing force). Since the system has to be chemically or thermally altered to initiate protein unfolding, reversible unfolding is not necessarily a mirror image of protein folding. The existence of irreversible protein denaturation also indicates that irreversible unfolding is not a mirror image of protein folding. Thus, unfolding intermediates do not necessarily represent folding intermediates. Not a process at equilibrium, the microreversibility principle describing equilibrium processes does not apply. Many structural properties of the states and processes involved in protein folding and unfolding can be accurately modeled without this consideration. Focusing on the processes of thermal denaturation, the conditions under which a protein folds and unfolds are often significantly different. The molecular aggregation and precipitation common in protein thermal denaturation did not occur for these protein molecules during proper folding. The AG between the completely unfolded and native states in the same environment, however, must be identical for both folding and unfolding. Studies of protein folding energetics using data from unfolding experiments can yield direct insights into the fundamental nature of enzyme stability (defined as the energy which sustains the active enzyme architecture in a
28 system). Beware of extrapolations to the mechanism of protein folding: they may be misleading due to the potential folding and unfolding asymmetry. Protein denaturation seems to be directly related to thermophilicity and thermo stability. Enzyme activity usually increases with temperature until it falls precipitously above the temperature of maximal activity. This rapid loss of activity is consistent with the loss of an enzyme’s active structure by denaturation. Many enzymes, however, have long half-lives at temperatures above their highest active temperature. For these enzymes the activity recovery once they cool into their active temperature range indicates that some reversible, incomplete unfolding is occurring [206,207]. Completely unfolded enzymes usually do not refold correctly. Complete denaturation is often accompanied by aggregation and precipitation. These two facts are indirect evidence of this partial unfolding. Molecular mechanisms involved in protein thermostability
Many of the molecular mechanisms responsible for protein thermostabilization are presented in detail. Their description is subdivided into two parts; intrinsic factors (specific amino acid replacement, altered entropy of unfolding, hydrophobic core packing, loop region engineering, etc.), and extrinsic factors (glycosylation, immobilization, stabilization by salts, pressure effect, etc.). Intrinsic mechanisms The characterization of thermophilic and thermostable enzymes, structurally and functionally similar to well characterized mesophilic enzymes, allows us to do comparative studies and to determine the factors involved in stabilizing enzyme architecture against denaturation. No universal mechanism explains the differences between thermozymes and mesozymes. Thermostability and thermophilicityproperties are believed to be due to subtle changes in the whole amino acid sequence of the thermophilic enzymes. An extensive comparative amino acid analysis by Argos et al. [208] led to the conclusion that thermal stability was related to: 1) increased internal hydrophobic amino acids and decreased external hydrophobic amino acids; 2) the replacement of Gly, Ser, Ser, Lys, and Asp by Ala, Ala, Thr, Arg, and Glu, respectively; and 3) to helix stabilization by more exclusive use of amino acids commonly found in the helices. Not all amino acid substitutions alter the function or stability of a protein. Specific interactions and residues, rather than all amino acids, contribute significantly to protein structural stability [209]. While single substitutions can increase the stability of an enzyme over 10°C [210], the thermostability intrinsic to thermozymes most often results from multiple amino acid substitutions [211-2131. The effects of single and multiple amino acid substitutions on enzyme thermostability are shown in Table 7. The examples were chosen from site-directed mutagenesis experiments in which the mutation effects were thoroughly studied by structural analyses ( e g , crystallography, hydrogen exchange measurements, Fourier transform infrared spectroscopy, calorimetry). These examples are also representative of the known
Table 7. Representative site directed mutagenesis studies of enzyme themostability mechanisms. Protein (Reference)
Mutation
a,-antitrypsin [2141
Wt
F51C L I
V A
Methanothermusfervidus (Mf) and Methanobacterium hryantii (Mh) glyceraldehyde-3-phosphate dehydrogenases PI51
Mesophilic (M-wt) and thermophilic (T-wt) kanamycin nucleotidyl transferases [2161
Mesophilic (M-wt) and thermophilic (T-wt) kanamycin nucleotidyl transferases ~171
Site directed mutants of M. fervidus Y323S Y323W
Effect
Conclusions
Stability at 57OC (min) 3 42 40 34 33 6.9
Low volume highly flexible hydrophobic sidechains allow better regional packing and prevent aggregation
'AT50 ("C) = -4.5 1.3
Chimeric mutants between Mf and Mh parental Mf Mfl1-294)/Mh(295-336) MA1-242)/Mb(243-336) MA 1-176)/Mb(177-336) MA 1-176)/Mh(177-294)/Mf(295-336) parental Mb
Increased hydrophobicity of residues involved in interdomain contact increased thermostability
0.0 -9.0 - 10.0 -10.5 -4.6 -12.3
9, ("C) solution M-wt T-wt M-D80Y T-Q102K
L252 K130L252 K130 Y80L252 Y 80 YSO/Kl30L252 Y80K 130
51.0
55.3 59.4 57.2 'ATopt ("C) n.d. 5.0
8.0 10.0
I)
immobilized 58.2 62.9 66.5 60.9
'AT, ("C) -7.2 -1.7 3.4 3.8 7.4 7.3
Immobilization stabilized the enzymes
2) Stabilizing mutations in solution do not necessarily stabilize the immobilized enzyme
I)
Multiple mutations can have a greater effect on thermostability that single mutations
2) Increasing enzyme thermophilicity by mutation reduced specific activity
10.8
(continued)
w
Table 7. (Continued)
0
Protein (Reference)
Mutation
Effect
Conclusions
Human lysozyme R181
Cl7NC95A
‘ATm (“C) = -14.4
Stability is reduced by the loss of a disulfide bond
AT^
Human lysozyme [211.219]
(“C)
W I G P I 03G WIG P 103G A47P D9lP VllOP T4 lysozyme [2101
E128A Vl31A L133A Dl 27AlEl28A E128A/V131A V 13IA/NI32A E128A/V131A/N132A Dl27AiE128A/V13 l W 1 3 2 A E128A/V131A!N132AjL133A D127AlE128A/V131~132~133A
T4 lysozyme (2201 T59N S
D G
V A
-4.5
-4.7 4.I 0.3 -1.1
I .2
%AGu (kcal mol-’) -1.1 -1.2 0.0 1.1 -0.8 2.4
ATm (“C) = 0.6 f 0.25 1.0 f 0.25 -17.0 f 2.0 0.9 f 0.25 1.5 f 0.25 2.3 f 0.25 3.4 f 0.22 4.0 f 0.25 -10.3 f 0.5 -9.4 f 0.5 pH 2.0 -2.1 -2.6 -3.1
-1.7 -10.0 -10.1
ATm (“C) pH 6.5 -2.8
Stability is enhanced by addition of a less flexible. residue (Pro) and reduced by addition of a more flexible residue (Gly)
I)
a-helix stabilization by Ala
2) Site mutation effects are independent and additive
I)
Better H-bond and charge interactions stabilize helix capping
-0.4
-3.1
2) Better local core packing increases stability
4.1
-4.0 -4.0 (continued)
Table 7.(Continued) Rotein (Reference)
Mutation
Effect
T4 lysozyme ~141
L46A L99A L118A L121A L133A F153A L99AiFI 53A
ATm ("C) =
T4 lysozyme [2211
T26S
ATm ("C) = 1.35 0.13 0.93
A93T T151S
Conclusions
-8.6 -15.7 -12.2 -9.3 -8.9 -12.3
Cavity creating mutants destabilize proteins
-41.8 1)
Local packing and release of strain increase stability
2) Addition of a peptide ligand to bound water increases stability ATm ("C)
T4 lysozyme [2221 C54T/C97A (disulf. = 0) I3C/C54T (disulf. = 1) 19C/C54T/C97A/L164C (disulf. = I ) T2lC/C54T/C97A/T142C (disulf. = 1) 13C/I9C/C54T/L164C (disulf. = 2)
IScn;!lC/C54T/C97A/T142C/L164C(disulf. = 2) 13C/I9C/TZ1c/C54T~142C/L161C (disulf. = 3) T4 lysozyme W31
L99A L99V L99I L99M L99F F153A F153V F153I F153L F153M L99A/F153A
+DlT 0.0 -1.9 4.5 -2.7 -10.3 -5.9
-DTT n.d. 4.8 6.4 11.0 15.7 17.0 23.4
ATm ("C) =
Addition of disulfide bonds enhanced moderate temperature protein stability and more bonds made the protein more stable
-8.5 -1 1.4
-5.2 -3.7 -1.5 -0.7 -9.3 -4.5 -0.5 0.8 -1.6 -22.8
Stability is enhanced by: 1)
Increased hydrophobicity of the amino acid
2) Increased core packing efficiency
(continued)
w
L
Table 7. (Continued) Protein (Reference)
W
N
Mutation
T4 lysozyme "2241
Effect pH
G77A A82P T4 lysozyme ~251
pH Ql05A Ql05E Q 105G
T4 lysozyme 12121
T4 lysozyme P261 T4 lysozyme 12271
Conclusions
ATm ("C) 2.0 pH 6.5 -1.4 0.9 0.8 2.1
Stability is enhanced by addition of a less flexible residue (Pro) and reduced by addition of a more flexible residue (Gly)
ATm ("C) 2.1 pH 5.8 -3.5 -1.6 -2.1 -3.0 -7.2 -3.9
1) Altered core packing alters thermostabliity
M102L VlllF VI 111 L99FM102L L99F/V1 I11 L99FF153L MIO2L/VlllI VlllW153L L99F/M102L/V11 1I L99F/M102L/F153L L99F/V 1 1 1UF153L L99FM102L/V11 IIF153L
ATm ("C) = -2.54 477 -2.32 -2.34 -2.70 0.09 -5.49 -3.52 4.02 4.68 -1.73 -1.82
Q86DlA92D
ATopt = 17'C
A41S A42S A49S A73S A82S A93S A98S A130S A134S V75T V87T V 149T
ATm ("C) =
-
2) The (-) of Glu at high pH destabilized the mutant
1)
Effect of multiple mutations is less than the added effects of individual mutations (cooperative)
2) Compensatory local core packing accounts for the cooperativity
Added Ca" binding site increases maximal active temperature -I .77 -7.49 -1.53 -1.27 4.99 -0.52 -7.47 -2.89 4.44 -3.70
Burying a polar group in the protein core is more destabilizing than a solvent exposed substitution so hydrophobicity is important to maintaining folded protein stability
-4.55
-10.08 (continued)
Table 7. (Continued) Protein (Reference)
Mutation
Effect
Conclusions Proteins can accomodate structural perturbations by local core repacking making them robust to random mutation 1) Internal cavity creating, destabilizing mutants can be stabilized by binding of a small hydrophobic molecule so core packing is important to protein stability
T4 lysozyme 12281 SI 17F
ATm (“C) pH 3.0 pH 5.4 4.8 2.8
T4 lysozyme 12291
L99A L99A + benzene (7.5mM) F153A F153A + benzene L99A/F153A L99A/F153A + benzene
ATm (“C) = -15.3 -9.3 -12.0 -12.1 -9.8 4.8
B. stearothermophilus neutral protease (W-ste)
A69P T63P S65P Y66P A69P
hAT50 (“C) =
A166S
AT50 (“C)= 1.2
T63R T63K T63F T63I T63Y T63M T63V T63G T63S T63A T63D T63P
AT50 (“C) =
WOI B . stearothermophilus neutral protease (NP-ste) 12311 B. stearothermophilus neutral protease (NP-ste) [2321
5.6 -10.0 4.2 -16.0 5.6
7.1 6.7 6.2 4.1 3.6 1.5 4 9 -1.4
Addition of Pro to restrict local flexibility can dramatically alter enzyme stability but depending on added conformational strain this can be negative or positive Release of a buried water molecule and enhanced H-bondingincreased folded protein stability Increased hydrophobic contacts at the protein surface stabilize the protein
4.0
-7.5 -7.5 -10.0 (continued)
W
w
w
Table 7.(Continued Protein (Reference) E . cereus oligo-1,6-glucosidase PI31
E. coli RNase HI
P331
P
Conclusions
Mutation
Effect AT50 (“C) 1.4 1.8 2.6 3.4 3.1 3.1 3.6 3.7 5.1
‘At1/2 (min)
K121P K121PE175P KI 21P/E175P/E290P K121P/El75P/E290P/E208P K12 lP/E175P/E290P/E208P/E27OP K 12 1P/E 175P/E290P/E208P/E270P/E378P IP K 121P/E 175P/E290P/E208P/E27OP/E378P/T26 K12 lP/E175P/E290P/E208P/E27OP/E378PK261 P/E2 16P P/E2 16P/N109P KI 2 I P/E175P/E290P/E208P/E27OP/E378PK261
ATm (“C) =
1.2 -1.5 2.7 -2.9
Q8O-W81 -+ Q8O-G81-W82 QSO-WSl -+ Q80-A81-W82 QSO-W8l -+ Q80-G81-W82/G77A G77A
’19
95 595:2.3 ‘7.3 k4.3 ‘5.2 ‘12.5 ’1 1.5 ‘38.5
1) Added prolines on the surface of a protein increase thermostability 2) 2* sites in !%turns and N-caps of ahelices provide the largest effect
3) The effects of Pro residues are additive and not cooperative 1) G81 stabilizes C-terminus of an ahelix (paperclip) 2) A77 in an a-helix adds bond strain and alters local packing to destabilize
3) G81 alters local packing around A77 and their dual stabilization is cooperative E. coli RNase HI
pH
P341 RI = ML -+ MNPSPR R2 = YRGR + FHAH R3 = YTR -+ EAC R4 = H62P R5 = VRQGITO --t LKKAFEG R6 = KTADK + REG R7 = LGQHQIKWEW -+ MAPRVRFHF R8 = A125T R9 = TGYQVEV -+ CPPRAPTLFHEEA GW = WQ -+ QGW P”’ = Q113P R7E’I9 = LGQHQIKWEW -+ MAPRVRFEF
ATm (“C) 3.0 pH 5.5 -2.0 -1.7 -2.3 -3.2 3.4 2.7 4.0 -5.7 0.4 -18.9 1.2 4.6 -1.1
4.1 4.5 5.6 2.4 0.0 -13.8
1) Mutagenic effects may be independent, or show (+) or (-) cooperativity. 2) Pro substitution even based on sequence comparison is not always stabilizing 3) Pro insertion in loop regions was stabilizing in both cases
0.8
-2. I I .9 (continued)
Table 7. (Continued) Protein (Reference)
Mutation
E. coli RNase HI
P341 (continued)
pH RI/R2 RIB4 Rl/RZ/R4 R4/R7 R5iR-I R5K6 R4K6 R4/R5/R7 R5/R6/R7 R4/R5/R6/RI
E. coli RNase H1 W31
pH V74L V74I V74A
E. coli RNase HI W51
B . amyloliquefaciens subtilisin BPN’ W71
ATm (“C) 3.0 pH 4.5 0.4 -2.1 -2.0 -1.9 6.0 6.7 2.0 1.9 6.2
K95G K95N
Y45W W59Y Y45WW59Y
gband -1.6 4.2 4.5
V26C/A232C A29ciMl19c D36C/P210C V148C/N243C D41C/G80C
5.5
4.9 1.9 -0.1 6.9 8.2 9.2 8.8 12.5 12.4 16.7
ATm (“C) 3.0 pH 5.5 3.7 3.3 2.4 2.1 -7.6 -12.7
pH 3.0 5.7 2.9
Aspergillus oyzae RNase TI P361
Conclusions
Effect
ATm (“C) pH 5.5 6.8 3.2 ATm (“C) Tyr-hand -1.9 4.5
4) Thermostabilizationof mesophilic proteins by region replacement based on sequence comparisons between functionally similar thermophilic and mesophilic proteins may not yield the expected effect
1)
Cavity filling mutants slightly stabilize the folded protein
2)
A cavity creating mutant significantly destabilized the protein
Replacement of f h mresidues in left handed conformations with Gly or Asn stabilizes proteins 1)
Altered local core packing destabilizes mutants
2)
Loss of H-bond to bound water in the W59Y mutant further destabilizes it
1)
Disulfide bond stabilization of folded proteins does not correlate to their resistance to reduction
2)
Disulfide bonds may not stabilize irreversibly unfolded proteins because of kinetic, not thermodynamic factors
-4.7
Atln (min) at 61OC -DTT +DTT 4 8 -34 4 2 -81 3 -101 ad. -113 -1 1s
(continued)
w wl
w
Table 7. (Continued) Protein (Reference) B. amyloliquefaciens subtilisin BPN' W81
subtilisin E [2391
o\
Mutation
T22c S24C S81C T22ClS8lC S24ClS8lC G6 1ClS98C
Effect
Conclusions
A t l n (min) at 58-59°C -DTT -98 0
-16 -5 I -3
Thennoanaerobacterium rhermosulfurigenes xylose isomerase ~421 L
.
For Table footnote, see next page.
-23
I ) disulfide engineering did not significantly stabilize the protein relative to the wild-type 2) the disulfide disrupted an intrapeptide Hbond causing as much destabilization as stabilization
Addition of a disuffide bond enhanced moderate temperature protein stability
A t W (h) K253R K294R K253Q K294Q A13S GlOSlG74T GlOS/A73S/G74T
sol (85°C) 0.8 -1.4 n.s. n.s. 4.1 0.6 1.1 At112 (h) +gluc.(60"C)
A. missouriensis
xylose isomerase ~4 11
4%
ATm ("C) = 4.5 AtIR at 55°C = 50 min
Actinoplanes missouriensis
xylose isomerase [2401
+DTT n.d. n.d. n.d.
K253R K309R K319R K323R K309RlK319RlK323R W49R W 139F W 139M W139A F60H
160
20 10 0 50
A1112 (min) =
immob (70°C) I76 -14 -39 n.d. -15 -19
1) Effect of mutations on soluble vs. immobilized enzymes may differ
2) Glycation of Lys by glucose (substrate) is major pathway of protein destabilization
17
-glut.(84°C) 1 .o I .8 1.8 0.0 0.4
20 32 23 15 -10
1) Arg for Lys substitutions prevent destabilizing glycation
2) Arg residues add stabilizing H-bonds
Reduced water accessible surface area increases protein thermostability
37 Table 7 notes: n.d. = not determined 'AT50 ("C) = the change in temperature where 50% of the activity remained after 10 min incubation bTk is defined as the temperature at which the inactivation rate constant = 0.5 'ATopt = the change in the temperature for maximal activity dTkis defined as the temperature at which the inactivation rate constant = 0.1 "ATm= the change in melting temperature fATd= the change in denaturation temperature gMG, = change in free energy of unfolding hAT50= the change in temperature where 50% of the activity remained after 30 min incubation at pH 5.2 'At,n = the time at which half of the activity remained at the temperature indicated 'determined at 45°C 'determined at 48°C
mutation effects. Examples corroborating every rational substitution prediction in thermophilic proteins are observable in nature. Nature uses all types of thermostabilization, and usually uses several types to stabilize a single protein. Multiple substitutions and protein stabilization. The effect multiple mutations have on enzyme thermostability is varied and sometimes unpredictable. Multiple substitutions can have additive effects (their effect is equal to the sum of the effects of the single mutations) or cooperative effects (their effect is greater or smaller than the sum of the effects of the single mutations). These observations are consistent with the view that: 1) the effect of a substitution is a local phenomenon, and 2) cooperativity of mutagenic effects is due to constructive or destructive interference between the individual spheres of mutational effect [212]. During their studies of T4 lysozyme, Matthews et al. observed both effects. When they introduced multiple alanines in an a-helix, the mutation effects were independent and additive (Table 7). When they constructed multiple mutations affecting a single internal cavity (where the side-chains of the target residues interacted with each other), the effects of the multiple mutations were cooperative and less predictable (Table 7). A continuum of thermostabilities should be attainable by protein engineering, bounded only by the thermodynamic limits of the protein's total molecular energy. The effects of substitutions, however, would be difficult to predict. '
Substitutions and modification of the thermodynamics of unfolding. Since entropy has traditionally been considered the main factor driving protein stability, it has been proposed that reducing the entropy gained by protein unfolding stabilizes the native structure [224]. Eliminating a naturally occurring disulfide bond in a protein generally destabilizes the protein. Kuroki et al. [218] showed that, despite the fact that the mutants preserved the wild-type tertiary structure, flexibility was increased in the mutant folded structures. Based on protein alignments, crystal structures, and molecular modeling, additional disulfide bonds were engineered with varying success in different proteins (T4 lysozyme, subtilisin BPN, and subtilisin E) [222,237-2391
38 (see Table 7). Disulfide bonds generally stabilize folded proteins. For some mutants the net stabilizing effect of an engineered disulfide bond was reduced to zero by the additional strain put on the local protein structure by each newly introduced cysteine [238]. To optimize the stabilizing effect of an engineered disulfide bond, the protein structure must be known. Engineered cysteines should be introduced in somewhat flexible regions that are limiting to protein stability, where they will not add excessive strain on the structure. Since, in all cases, the stabilizing effect disappeared when the disulfide bond was reduced [222,239], this stabilization strategy may be of little value for proteins folded and used in reducing conditions. The chemical instability of cysteine residues at temperatures nearing 100°C, also makes this strategy potentially ineffective for high temperature applications [243], and does not appear as an evolutionary strategy for thermozymes. The thermodynamics of the stabilization provided by disulfide bonds remains unclear. Enthalpy appears to be an important factor, even more important than entropy [218]. The entropic effect of amino acid substitutions was discussed by Matthews et al. [224]. Glycine lacks a P-carbon. In solution it has more conformational flexibility and entropy than any other amino acid. Substitutions of glycines with alanines or other residues containing a P-carbon should stabilize proteins by reducing their entropy of unfolding (as long as the engineered residue does not introduce unfavorable strain on the protein structure). Matthews et al. [224] predicted that replacement of glycine by any other amino acid would stabilize the folded structure by -4 kJ/mole, relative to the unfolded state. Results of their subsequent exhaustive mutagenic study of T4 lysozyme were not completely consistent with the theory. While mutations of poorly mobile amino acids (as determined by low average crystallographic b values) with reduced solvent accessibility were far more likely to have a significant effect on protein thermostability, stability mutants primarily affecting the free energy of the unfolded state (i.e., mutations in flexible or solvent accessible regions) were rare [244]. It is, therefore, local packing efficiency in folded proteins, rather than an entropy difference between the folded and unfolded states, that is the major determinant of thermostability. Matthews et al. also considered the introduction of prolines to add extra constraint on unfolded proteins and to decrease the entropy of unfolding [224]. The extra steric constraint of proline residues present in both the folded and unfolded protein would similarly stabilize the folded protein by -4 kJ/mole due to the decreased favorable ASunfolding. In each case, the extent of the effect depends at least on the extent of the conformational constraints introduced by the mutated residue in the folded protein. Introduction of stabilizing prolines in proteins has been extensively documented. The specific minimum-energy configurations prolines can occupy have been carefully calculated [245]. It has further been demonstrated that prolines play an important role in directing the local conformations surrounding them [245], and that proline insertions are only well tolerated in specific parts of a protein structure [245]. Proline substitutions have been shown to stabilize mesophilic proteins [211,219,224,234] and to further stabilize thermophilic proteins [230]. These results suggest that careful proline substitution is a general technique which can enhance enzyme thermostability, and that the practical limits of enzyme
39 stability are higher than the limits engineered by nature. Careful analysis of the thermodynamic effect of proline substitutions on stability indicated the importance of an enthalpic component (in addition to an entropic one) in thermostabilization [211,246]. Due to the steric constraints of proline relative to other amino acids, only a small energy gain is expected from proline introduction into a protein. The gain observed is often higher because of this additional enthalpic component.
Prolines in loop regions. Analogous to the stop on a zipper, prolines are used in constrained loop regions to prevent the sequential dissociation of numerous coulombic stabilizing interactions between the two adjacent core elements [247]. This prolinezipper model predicts a crucial role for loop regions in protein thermostability (Fig. 3). This observation contradicts traditional knowledge that considers loop regions only as links maintaining a continuous peptide between two properly positioned core elements. This traditional view was supported by the observations that loop regions can withstand the accumulation of more neutral substitutions than core elements can [247], and that there is usually a greater sequence variability in loop regions than in core elements in an enzyme family. The traditional view of the
Fig. 3. Role of proline residues in protein structural stabilization.
molecular basis for protein stability arose from observations related to enzyme catalysis which depends more on the relative positions of core elements. Emerging data indicates that the role of loops in protein stabilization should be more carefully examined. Studies focusing on p-turns suggest that constrained loop regions play a major role in protein resistance to denaturation. Manipulating turn structures may emerge as a common method for protein thermostabilization [235,248]. Kimura et al. [235] constructed E. coli RNase H site-directed mutants, using the results of a previous mesophilic/thermophilic enzyme sequence comparison [234]. Mutations K95G and K95N that targeted a p-turn clearly stabilized the protein, as shown by activity assays and circular dichroism measurements. Based on the enzyme X-ray structure and molecular modeling, Kimura et al. proposed that the K95G mutation eliminated some structural strain created by the lysine residue, and allowed a better interaction between the two neighboring core elements. The K95N mutation allowed the formation of an intra-residue hydrogen bond in the asparagine residue. This bond created a structural constraint analogous to the constraint introduced by prolines [235]. Two highly (85%) similar 2" ADHs from the mesophile Clostridium beijerinckii and the thermophile T. ethanolicus have been sequenced and compared [36]. Ten percent of the nonconservative substitutions corresponded to additional prolines in the thermophilic enzyme. These prolines were either in short (2-4 residues) loops or in a longer loop containing multiple prolines. In the thermophilic enzyme the presence of multiple prolines in a long loop suggests that specific constraints on this loop are required to stabilize it and stabilize the protein [36]. Additional prolines were also observed in short and long, constrained loops in two thermostable and one hyperthermostable xylose isomerases. These prolines were absent in the mesophilic enzymes (C. Vieille, unpublished data). These data predict that loop stabilizing mutations are more effective when introduced in constrained loops. Protein thermostabilization by proline introduction in turn regions, therefore, is an indication of the general importance of turn regions in protein stability, representing a naturally occumng thermostability control mechanism that can be used in protein engineering. Salt bridges. Salt bridges comprise one type of specific interactions proposed to stabilize proteins. Activities of ions immobilized on the same molecule are extremely high, even compared to those in bulk solution at molar concentrations [249]. This fact suggests that intramolecular salt bridges may be very stable, even at the surface of a protein in a highly polar solvent environment. Comparing the B. stearothermophilus and yeast phosphoglycerate kinases' crystal structures led to the conclusion that extra salt bridges present in loops in the thermophilic protein contribute to its thermo stability [250]. Crystal structure analysis of ribonuclease H [251] also allowed the identification of salt bridges as important factors in this enzyme's thermostability. An intra-peptide salt bridge also stabilizes the Actinoplanes missouriensis xylose isomerase [241]. Tomazic and Klibanov [252] proposed that additional salt bridges in B. lichenformis a-amylase reduced reversible unfolding, thus reducing the probability of the partially unfolded enzyme forming scrambled structures and
41 preventing irreversible denaturation. A careful study of chaotropic salts’ effect on S. solfataricus carboxypeptidase similarly concluded that salt bridges contributed significantly to an enzyme resistance to denaturation [ 1441. Electrostatic interactions also can explain protein thermostabilization when surface residues are substituted with arginines [241].
Hydrogen bonds. Hydrogen bonds (H-bonds) are another type of coulombic interactions contributing to protein stabilization. Their role was first recognized by Mirski and Pauling in 1936 [192], and has been thoroughly studied and reviewed (See [253]). A recent publication by Cleland and Kreevay addressed the importance of low-barrier H-bonds in catalysis [254]. Low-barrier H-bonds are formed between residues carrying functional groups with similar pKa’s, and are significantly stronger than ordinary double-well H-bonds. They can be detected crystallographically by either their characteristic short bond distances (<2.5 A) or by signal shifts in NMR (See [255]). It has long been known that the pKa’s of ionizable groups are strongly dependent on environmental factors; in proteins, the environment surrounding Hbonding partners is often quite different from their environment in solution. These observations, combined with recent X-ray crystallography showing that the core of thermophilic proteins contains less and smaller cavities than their mesophilic counterparts [89], suggests that both enzyme thermostability and thermophilicity can be enhanced by the formation of stronger H-bonds. Thus, thermophilic proteins may substantially increase their stabilization energy with only small differences in structure or amino acid content. The excess enthalpic component of protein stabilization created by proline substitution in constrained loops may be explained by the geometric constraint strengthening the H-bonds between adjacent core elements. Substantiating or refuting this hypothesis will require careful analysis of exact protein structures. Hydrophobic interactions and core packing. Hydrophobic interactions are believed to provide the energy needed to fold proteins in aqueous solutions. Thermostable Bacillus caldovelox a-amylase [47] and ES4 glutamate dehydrogenase [ 1051 contain more hydrophobic-interaction-formingresidues than their mesophilic counterparts. It has been proposed, therefore, that hydrophobic interactions play a significant role in enzyme thermophilicity and thermostability. Matsumura et al. [256] created multiple substitutions of Ile3, a residue which contributed to the major hydrophobic core in T4 lysozyme. The hydrophobicity of the engineered residue (measured as the free energy of transfer from water to ethanol) appeared clearly related to protein stability (measured as the difference between the free energy of unfolding of the mutant and the wild-type protein). Residue water/ethanol free energy of transfer contributed to approximately 80% of the difference in overall protein stability, indicating that hydrophobic interactions were the main stabilizing factor at position 3. Other amino acid substitutions increasing the hydrophobic content of the protein core can increase protein thermostability [223,227,229,257]. Furthermore, the stability cost of burying a polar hydroxyl group in the core of T4 lysozyme can be significant [227]. It is
42 generally believed that hydrophobic residues at the surface of proteins are unfavorable for protein stability. Some surface residues, however, have protruding hydrophobic surfaces creating a hydrophobic pocket in the protein surface, where the presence of a hydrophobic residue will favor protein stability. The B. stearothermophilus neutral protease was significantly stabilized either by introducing bulky hydrophobic residues or by Arg or Lys residues in such a surface pocket ([232], and Table 7). Extensive X-ray crystallographic studies have shown: 1) that folded proteins contain significant cavities (some filled by water molecules, some not) [258]; 2) that protein local core packing is surprisingly able to compensate for deformations due to amino acid substitutions [259,260];and 3) that aromatic-aromatic interactions in the hydrophobic core environment may stabilize the folded structure [212,228]. Other crystallography-based experiments have demonstrated the potential to increase protein thermostability by filling cavities in the folded structure [214,223,233,261],and by displacing buried H,O molecules with either amino acid side chains ([226], and Table 7) or benzene ([229], and Table 7). Sequence analysis of T. maritima GAPDH was used to argue that this enzyme’s high thermostability was attributable to local packing interactions specifically involving its higher percentage of large hydrophobic groups [ 1121. Crystallographic studies have also shown the link between increased core hydrophobicity and better packing efficiency in thermostable proteins, supporting the importance of protein hydrophobicity in thermostability [262,263]. Some thermophilic enzymes are significantly stabilized by an overall increase in hydrophobicity: this trait, however, is not universal among thermostable enzymes since, in many enzyme families, no significant differences in hydrophobic residue content exist between thermostable and thermolabile proteins. This observation indicates that increased hydrophobicity is not the only stabilizing tool used by naturally occurring proteins. While better core packing is often linked to increased hydrophobicity, in other cases it can affect stability through different factors. Reduction of bond strain due to altered core packing has also been shown to stabilize proteins [259,264,265]. Not necessarily stable in solution, secondary structural elements are still important in maintaining active enzyme architecture [202]. Based on the ability of high short peptide densities to form a-helices in crystals, it has been proposed that increasing peptide concentration increases the stability of helices [202], and that helix stabilization in core domains is implicated in protein thermostability [210,233,2661. Internal packing, therefore, may be involved in protein thermostability either as a general stabilizing force or as a factor altering the stability of secondary structures. Covalent destruction and irreversible denaturation. While the intrinsic thermostability factors discussed so far are potentially related to thermophilicity mechanisms, enzyme irreversible denaturation by temperature-induced covalent modifications (i.e., Asn or Gln deamidation, or the oxidation of Cys to cysteic acid) directly affects protein thermostability. These covalent alterations are likely to affect thermophilicity through catalytic residue decomposition (e.g., Cys residue in ADHs or GAPDHs), through their role in irreversible denaturation due to protein aggregation, or through other mechanisms associated with these covalent modifications. Several studies [ 195,243,
43 260,267-27 13 indicated that deamidation was probably involved in the formation of scrambled structures after significant protein unfolding occurred. Klibanov et al. showed that deamidation was the major mechanism controlling thermoinactivation in Bacillus a-amylases [ 1891, hen egg-white lysozyme [268], and glucose isomerases [272]. Partial unfolding (followed by irreversible denaturation and then by molecular scrambling and aggregation) can affect different enzymes at different temperatures, but most thermophilic and hyperthermophilic enzymes are structurally stable and resist unfolding at moderate temperatures. A cause of enzyme inactivation, amino acid covalent degradation becomes significant only at elevated temperatures (usually above 90°C) [ 1851. Tomazic and Klibanov showed that Bacillus amyloliquefaciens and B . stearothermophilus a-amylases are inactivated by monomolecular conformational scrambling, whereas the more thermostable B. licheniformis enzyme is inactivated by Asn/Gln deamidation [252]. An inverse relationship was seen between the percent of Asn plus Gln residues in D-XylOSe isomerases and their respective temperature for maximal activity (ranging from 55 to 95OC) [160] (see Fig. 1). Deamidation destabilizing effects have been proposed to be limited to the deamidation of structurally important Gln and Asn residues. In any case, the general trend of fewer Cys, Asn, and Gln residues in the more thermophilic proteins provides indirect support for the hypothesis that the thermally induced decomposition of these residues is undesirable for protein function at very high temperatures. In a comparative study of mesophilic and thermophilic type I1 D-XylOSe isomerases, we extrapolated the enzyme melting temperatures (Le., temperature for 50% unfolding) from melting experiments performed in the presence of guanidine-hydrochloride (see Table 3). The mesophilic E. coli, thermophilic T. thermosulfurigenes, and thermophilic B . stearothermophilus xylose isomerases were 50% unfolded at temperatures (53, 80, and 95"C, respectively) close to their respective maximal activity temperatures (55, 80 and 85"C, respectively) and their precipitation initiation temperatures (52, 80, and 85°C). T. neapolitana enzyme's melting temperature ( 120°C), however, was 25°C higher than its temperature for maximal activity (9SOC) and for initiation of protein precipitation (95°C). Thus, T. neapolitana xylose isomerase's inactivation mechanism appears to be different from the other three xylose isomerases', and does not involve a high amount of initial unfolding. We suspect that T. neapolitana xylose isomerase inactivation occurs through deamidation (or another covalent destruction mechanism) at temperatures where the enzyme is barely unfolded [92]. These data argue that the engineering of hyperthermostable proteins will require either the removal or the modification of such thermolabile residues. Extrinsic mechanisms It is well known that in vitro enzyme activity is strongly affected by the nature and concentration of effector molecules present in solution as well as by environmental factors. Allosteric effectors, substrates, and ions (such as Ca2+ and PO,%) are specifically bound by proteins. It is not surprising, therefore, that the thermostability of a number of enzymes is altered by such effectors [273]. Substrate molecules have
44 long been known to stabilize enzymes, presumably by interaction in specific binding sites [226,242], and these effects are not reviewed here. Chemical cross-linking or immobilization, glycosylation, inorganic salts, and high pressure also stabilize enzymes. We limit our discussion of these topics to the current mechanistic theories and provide representative examples supporting or opposing these theories.
Chemical cross-linking and immobilization. Chemical cross-linking has been shown to stabilize proteins by a thermodynamic mechanism similar to the one involved in stabilization by disulfide bonds. The conversion of thermolabile amino acids (such as amines) to more stable residues (see [274]) participates in this mechanism. Aside from the obvious fact that peptide regions maintained in their folded state by covalent bonds will only lose their native conformation when these bonds are broken, the reduced entropy of the unfolded state relative to the folded state also applies here. Enzyme covalent immobilization may be considered in this context to be similar to cross-linking, and often leads to thermostabilization of active enzymes (as reported for a Thermus sp. proteinase) [275]. Enzyme long-term stability has also been dramatically improved by immobilization (e.g., industrial glucose isomerases are immobilized to maintain a half-life of several months at 60°C [276]). However, immobilization does not thermostabilize all enzymes. A thermophilic kanamycin nucleotidyl transferase is stabilized by immobilization. A punctual mutation that stabilized this soluble enzyme destabilized the immobilized enzyme ([216], and Table 7). Forecasting immobilization or cross-linking effects on a protein will require detailed structure-function information about the native and cross-linked enzymes. Precise control over the cross-linking chemistry is also required. Noncovalent changes in the protein, induced by cross-linking, complicate the predictability of enzyme stabilization by cross-linking. These noncovalent changes are also expected to alter the modified enzyme’s stability. Cross-linking, for example, can destabilize some interactions between two core elements, depending on where the cross-linkage occurs. Glycosylation. Protein glycosylation is widespread among eukaryotic inzymes. An increasing number of bacterial extracellular enzymes have been shown to be glycosylated (for example, C. thermoceffum cellulosome components and T. saccharolyticum endoxylanase [277-2791). Most of these glycosylated enzymes (bacterial, as well as eukaryotic) retain their catalysis and stability properties when expressed in mesophilic hosts (not known to extensively glycosylate proteins). A major source of destabilization for the A. missouriensis glucose isomerase has been shown to be the glycosylation of a Lys residue by glucose - a glycoside as well as the glucose isomerase substrate [240]. While some glycosylated proteins have been shown to be more stable than their nonglycosylated forms [46,206], glycosylation is not a thermostabilization method commonly found in nature [ 190,2801. Exogenous salts. Two ways by which inorganic salts stabilize proteins have been described: 1) a specific effect, where a metal ion interacts with the protein in a conformational manner; and 2) a general salt effect, which mainly affects the water
45 activity. Examples of salt specific effects are abundant in the literature. a-Amylases specifically bind Ca2'. The a-amylase catalytic site is located in a cleft between two domains (an (alp), barrel and a large loop). Coordinated by ligands belonging to these two domains, Ca2+ is essential for the a-amylases catalytic activity and thermostability [281]. Xylose isomerases bind two metal ions (chosen from these three Co2+,Mg2+,and Mn2+).One cation is directly involved in catalysis (the catalytic metal); the second stabilizes the protein (the structural metal) [282,283]. The two metal-binding sites have different specificities, and replacing one cation with another often significantly alters enzyme activity, substrate specificity, and thermostability [282,284]. Ca2' stabilization of lysozyme was proposed to be due to a decrease in the entropy of the Ca2'-bound unfolded protein relative to the entropy of the unfolded protein in the absence of Ca" [273]. In this specific salt effect the salt specificity was attributed to the characteristics of the specific binding sites (i.e., size, coordination number, and nature of the liganding residues) [285]. A study of GAPDH thermostabilization by salt indicated that the relative effects of &PO,, Na,PO,, K,SO,, NqSO,, KC1, and NaCl were consistent with their respective abilities to reduce the enzyme solubility in aqueous solvent. Their action was attributed to their decreasing effect on water activity [ 1961. Ammonium sulfate protein precipitation is an application of the general salt effect for enzyme precipitation. Once it is ammonium-sulfate-precipitated,the enzyme is stabilized. This procedure is a standard, convenient way to store enzymes. Thauer et al. studied the effect of salts on the thermostability and activity of five M . kandferi methanogenic enzymes [ 124,125,134,151,1521. While the five enzymes are activated and stabilized by salts, the extent of the salt effect varies from enzyme to enzyme: CHO-H,MF'T formyltransferase is optimally stabilized in the presence of 1.5 M K,HPO,; F420dependent CH, = H,MPT reductase is stabilized by 0.1 M K,HPO,. K' and NH; cations typically improve enzyme stabilities more efficiently than other cations. Of all the anions, SO," and HP0;- have the strongest activating effect [134]. Enzyme salt requirements are not always satisfied by the intracellular salt concentration. M . kandferi's intracellular salt concentration (>1 M potassium plus 1.2 M 2,3-diphosphoglycerate) [ 1341 seems to favor CH, = H,MPT reductase activity (optimal at 2.0-2.5 M salt), but this intracellular salt concentration is far from being the optimal enzyme stabilizing concentration (optimal between 0.1 and 1.5 M salt) [125]. The effect of salts on CHO-H,MPT formyltransferases from M . kandferi,M . thermoautotrophicum, Archaeoglobus fufgidus, and Methanosarcina barkeri were compared. Breitung et al. (1992) [ 1341 found that the difference in formyltransferase activation by salts was directly correlated to the intracellular 2,3-diphosphoglycerate concentration in the different organisms. These studies illustrate the variety of situations encountered: 1) similar enzymes from different organisms do not have the same stabilization and activation requirements; 2) enzymes from the same organism are not equally affected by an environmental (i.e., cellular) factor; and 3) thermophilicity and thermostability are not necessarily controlled or favored by the same factors. The hypothesis that the general salt effect was due to direct electrostatic interactions between charged amino acids and salt counterions was not corroborated
46 by experimental data [286]. Instead, a smeared charge repulsion mechanism was proposed: Salt-induced protein destabilization was due to nonspecific charge repulsion and to ionization changes at the interface between complementary peptide surfaces in the partially unfolded protein. This hypothesis agrees with the well documented protein stabilizing effect of intermediate salt concentrations, in particular with the relatively high salt concentrations (1.0-2.0 M) required to stabilize some thermophilic enzymes [286]. It also agrees with the reduced polar uncharged residue content in some thermophilic enzymes, when compared to their mesophilic counterparts [47,208]. At high temperatures, the system’s higher kinetic energy increases the potential for exposing buried surfaces to the solvent. Salt interactions with the less polar residues in the partially unfolded thermozyme are less favorable than the interactions with the more polar mesozyme residues. Greater solvent ionic character is expected to stabilize the folded thermozyme relative to its partially unfolded forms. Pressure efsects. The ability of high pressure to stabilize proteins has been reported [287,288]. The observation that folded proteins have densities similar to crystalline solids is the theoretical basis for pressure as a general stabilizing force. Unfolding a protein would increase its volume, making a partially unfolded protein less stable than the folded protein at increased pressures. Thus, the higher the pressure is, the more compact the protein is. Hei and Clark [287] have studied the effect pressure has on enzymes from barophilic organisms. They showed that some barophilic thermophiles have enzymes which are also barophilic (i.e., optimal activity at pressures from 200 to 400 atm). Hei and Clark even characterized barophilic enzymes that had been isolated from organisms that are not, themselves, barophilic [287]. Since numerous chemical reactions are performed at high temperatures and pressures, enzyme stability at high pressures is potentially important for biocatalysis at high pressures [289]. Molecular mechanisms involved in protein thermophilicity The molecular basis for enzyme thermophilicity is not well understood, and there is limited literature on this subject. Linear Arrhenius plots for numerous enzymes indicate that if there are structural changes in the catalyst throughout the active temperature range, they are either not catalytically significant or they are exactly compensatory. The discontinuous Arrhenius curves observed for catalysis by some enzymes [ 1 15,1961 indicate that, in these specific cases, temperature dependent, functionally significant stnktural changes may occur. These discontinuous Arrhenius curves were believed to be characteristic of thermozymes, but, as mentioned previously (see section on: Protein thermostability), some mesophilic enzymes show this behavior [ 1 lo]. Explanations for this occurrence have been proposed that do not involve altering the catalyst’s structure (e.g., change in the slow step of the reaction) [197,290]. Enzyme activity in its active temperature range is well described by the Arrhenius equation, and can be derived from the temperature of maximal activity and the activation energy for the reaction. The Arrhenius equation, however, cannot predict the optimal temperature for an enzymatic reaction. An enzyme’s active
47 temperature range appears to be determined at the lower end by the theoretical limitation on activity defined by the Arrhenius equation. Partial protein unfolding or covalent chemical modifications appear to determine the upper limit, independently from the Arrhenius equation. Most studies on enzyme thermostabilization (i.e., by mutagenesis or by extrinsic factors) do not comment on the mutant enzyme thermophilicity. Only a few reports [226,2913 describe mutations that altered both enzyme thermostability and thermophilicity. Kuroki et al. demonstrated that adding a Ca” binding site in human lysozyme by site directed mutagenesis both stabilized the protein and increased its thermophilicity from 70 (native enzyme) to 80°C (mutant holoenzyme) [226]. In the absence of Ca”, the mutant enzyme thermophilicity was reduced to 65°C. The native and mutant enzyme catalytic rates were superimposable up to the temperature for maximal activity of the wild-type enzyme (-70°C). Above that temperature the mutant enzyme activity continued to increase, with the same apparent Arrhenius behavior, until it reached its own temperature for maximal activity. The mutant behavior suggests that the wild-type enzyme thermophilicity was limited by its thermostability, and that the mutant higher thermophilicity was due to an increase in its thermostability. In a study using TAB-linker mutagenesis, insertion of 2-4 amino acids into Caldicellulosiruptor saccharolyticus xylanase generated stability-only mutants and stability-plus-optimal temperature-altering mutants [29 11. Insertion of a single Pro-Arg sequence reduced the enzymes specific activity more than 2-fold, the thermostability 4-fold, and the optimal temperature by 20°C. These mutations support the hypothesis, in this case, that thermophilicity is limited by protein stability. Other C . saccharolyticum xylanase mutants, displaying reduced activity and reduced stability (similar to the Pro-Arg mutant) while retaining native-like thermophilicity (70”C), clearly demonstrated that the molecular bases for protein thermostability and thermophilicity can be structurally distinct. The difference between the two classes of C . saccharolyticum xylanase mutants may be explained using the Tomazic and Klibanov thermostability model (scheme 1, section on Protein thermostability and [ 1891). The mutations which only reduce enzyme thermostability would increase the rate of scrambled structure formation from the enzyme nonnative state, while the mutations that reduce both thermophilicity and thermostability would result from the shift of native to nonnative enzyme at lower temperatures. This analysis predicts that a class of thermophilicity-reducing mutants that do not alter thermostability could be generated if the nonnative state (resulting from reversible unfolding) were stabilized with respect to the scrambled structure (the enzymes would only reversibly inactivate). An N-terminal deletion in the T. ethanolicus 39E amylopullulanase caused a change in the temperature activity optimum from a wild-type enzyme with a sharp 80°C optimum to a mutant with a broad 55-80°C optimum (Park et al., manuscript in preparation) without affecting thermostability. C-terminal deletion mutants were also constructed which retained the wild-type thermophilicity but were significantly destabilized (Park et al., manuscript in preparation). None of the enzymes could be purified to homogeneity, so activity curves for the mutant and the wild-type enzyme
48
were compared using relative activities (usually referring to 100% activity at the temperature of maximal activity). Thus, it was not possible to conclude if the mutations altered the thermophilicity, or if they just changed the protein thermo stability. These data suggest that different amylopullulanase regions distant from the active site account for thermostability and thermophilicity. The broadened thermophilicity curve of the amylopullulanase N-terminal deletion mutant can be explained by the superimposition of two separate mechanisms. Whereas the enzyme remains stable up to 80°C, a structural element in the catalytic site becomes more flexible after 55"C, progressively inhibiting activity. At the same time the substrate energy keeps increasing (Arrhenius equation), and compensates for the activity loss. While the relationship between enzyme kinetics and protein thermophilicity/thermostability remains unclear, it has been proposed that the poor activity of thermozymes at low temperatures is the result of excessive rigidity. This rigidity is necessary to maintain active enzyme architecture at high temperatures [ 186,194,2011. Data from crystallography, deuterium exchange, proteolytic susceptibility, and other experimental approaches have demonstrated that folded thermozymes are indeed more rigid at low temperatures than their mesophilic counterparts [ 186,194,2011. Tchemajenko et al. [92], however, reported linear Arrhenius plots for the four xylose isomerases they studied. Since these plots were perfectly fit by the Arrhenius equation, increased low-temperature protein rigidity was not required to explain the poor low-temperature activity of thermophilic enzymes [92]. Thermozyme activity might be, instead, only determined by temperaturedependent substrate energy variations. The activation energies for protein stabilization determined from Arrhenius plots of both thermophilic and mesophilic protein thermoinactivation are similar to each other (generally in the order of 1.0-10 kJ/mol). They are typically similar to, or lower than, the activation energies for catalysis by the same enzymes. If protein unfolding were only a thermodynamically driven process, it would be expected that proteins unfold below or at their temperature for maximal activity. Since numerous proteins unfold at much higher temperatures, protein unfolding must also be a kinetically driven process. Initiation of unfolding might be limited by the time needed to reach the unfolding activation energy.
Biotechnological applications of thermozymes Many commercialized industrial and specialty enzymes do not require high temperature activity (e.g., in medicine, foods, detergents). Most current commercial enzymes need stability, but have to display activity in a certain temperature range. Sales of industrial enzymes in 1995 exceeded US $1 billion in the world, and that figure is growing at 5-1596 per year. The market for specialty enzymes is growing faster than the market for industrial enzymes. Industry enzymes are used primarily for making detergents; processing starch, food, and feed, and producing chemicals and ethanol. This section reviews applications where both thermostability and activity
49 are important. Due to the lack of knowledge on thermophiles and the lack of thermozyme sources, most of todays industrial enzymes originate from mesophilic organisms. This section reviews major industrial and specialty enzyme applications and assesses the interest thermozymes hold for these applications.
Industrial thermozymes Starch-degrading enzymes Corn starch is the most widely used carbohydrate feedstock in the United States. Of 800 million bushels of corn industrially used in 1992/93, one-half was used to produce fuel alcohol [292]. The other 50%were used to produce High Fructose Corn Syrup (HFCS) and maltose syrups [293] - ingredients for the food industry - and to produce dextrin syrups to be used as feedstocks for various fermentations (enzyme, beer, citric acid, lysine, etc.) [293,294]. Most industrial uses of starch require the hydrolysis of starch into glucose, maltose, or maltodextrin syrups. These syrups are used to produce fuel ethanol, citric acid, lysine, enzymes, and other products by microbial fermentations. HFCS is produced by the enzymatic isomerization of high glucose syrup. Starch bioprocessing usually involves two steps, liquefaction and saccharification, both run at high temperatures. Typically not water-soluble, starch starts gelatinizing at temperatures close to 100"C, thus becoming more accessible to hydrolysis by liquefying aamylases. During liquefaction, starch granules are gelatinized at 105°C for 5 min in an aqueous solution (pH 6.5), and then they are partially hydrolyzed at the a-1,4 linkages with a thermostable a-amylase at 95°C for a period of 2 to 3 h. The pH is then adjusted to 4.Cb5.5 and the temperature lowered to 55--60°C for the saccharification step. During saccharification (24 to 72 h), the liquefied starch is converted into low molecular weight saccharides, and, ultimately, into glucose or maltose. Glucose syrups (up to 95-96% glucose) are produced using pullulanase and glucoamylase in combination, while maltose syrups (up to 8 0 4 5 % maltose) are produced using pullulanase and P-amylase [295,296]. Except for a-amylases (e.g., B. licheniformis a-amylase is highly thermostable at 90"C), the enzymes used today in industrial starch processing are not highly thermostable. Pullulanase, isoamylase, P-amylase, and glucoamylase (originating from Klebsiella pneumoniae, Pseudomonas, plants (barley, soybean, .wheat), and Aspergillus niger, respectively) are only marginally stable at 60°C. Engineering or isolating new pullulanases, P-amylases, and glucoamylases that are stable at high temperatures and environmentally compatible (i.e., temperature, pH, and cation requirement) should represent a significant improvement for the starch processing industry. Increasing the saccharification process temperature would result in many benefits: 1) higher substrate concentrations, 2) decreased viscosity and lower pumping cost; 3) limited risks of bacterial contaminations; 4) increased reaction rates, and decrease of operation time; 5 ) lower costs for enzyme purification; and 6) longer catalyst half-life, due to increased enzyme thermostability.
50
a-Amylases. Different sources of a-amylases are used commercially, including a recombinant product (MegaDex) comprised of B. stearothermophifus a-amylase, cloned and produced in B. subtifis. This enzyme and the B. licheniformis a-amylase display high activity at 95°C (the liquefaction process temperature). The B. licheniformis a-amylase is activated and stabilized by Caz+,and has a half-life of 30 min at 55°C (in the absence of substrate and Ca2+)[3]. The extremely thermostable P.furiosus and P. woesei extracellular a-amylases (optimally active at 100°C and pH 5.0 and 5.5, respectively, see Table 2) represent additional a-amylase sources. With these two enzymes, starch liquefaction could be performed at an even higher temperature (100°C) and at a lower pH (pH 5.0). Since these two archaeal enzymes do not require Ca” for their activity [ 136,1391,starch liquefaction could be performed in the absence of Ca”. (Ca” strongly inhibits xylose isomerase, and has to be removed before the isomerization step during HFCS production.) Since P. furiosus and P. woesei extracellular a-amylases have not been cloned, more work (i.e., cloning and over expression) is needed to make their industrial production economically feasible. P-Amylases. Maltose syrup production uses mesophilic enzymes, and is typically performed at 5 5 6 0 ° C . Saccharification is not performed at higher temperatures due to: 1) an increasing risk of unwanted side-reactions (browning) at alkaline pHs; 2) the reducing size of the dextrins present in the mixture; and 3) the length of the saccharification processes (48-72 h). Thermophilic P-amylases that have high activity at acidic pHs may improve saccharification. Unless thermozymes can significantly reduce the saccharification time by increasing the reaction rates, no significant temperature increase can be expected. Two sources of thermophilic P-amylases exist. If intermediate temperatures are required to limit the browning side-reactions, using T. thermosulfurigenes P-amylase [52], optimally active at 75°C and pH 5.5, is an option. Optimally active at 95”C, pH 5.0, and in the absence of Ca” [3], T. muritima P-amylase could be a good candidate if its specific activity is high enough for testing the impact high temperatures have on the length of the saccharification process. Production of maltose syrups using these thermozymes would still require a compatible debranching enzyme. Glucoamylases and a-glucosidases. No glucoamylases have been purified from starch-degrading thermophiles or hyperthermophiles. In these organisms starch is usually hydrolyzed extracellularly by an amylase ( a or P), amylopullulanase, or both; oligosaccharides are further degraded intracellularly by an a-glucosidase. It is possible that glucoamylase can be replaced with a thermostable a-glucosidase in glucose syrup production. Among the thermostable a-glucosidases characterized, only the P. furiosus enzyme seems to have some potential for glucose production. Optimally active between 105 and 115”C, and showing only 15 and 35% maximal activity at 80 and 9OoC, respectively [150], it still has a much higher specific activity than T. ethanoficus a-glucosidase [7 11, which is optimally active at 75°C. Moreover, P.furiosus a-glucosidase does not seem to have any transglucosylation activity [ 1501.
51
Bacillus sp. SAM1606 a-glucosidase’s high transglucosylation activity (optimally active at 75°C and pH 5.5) represents a major drawback for its use in glucose production [70]. Other starch-utilizing thermophiles and hyperthermophiles probably contain an a-glucosidase. More understanding is needed before the use of thermophilic a-glucosidases in glucose production from starch can be assessed. Pullulanases and amylopullulanases. Pullulanases are used as debranching enzymes in starch saccharification. Because of the thermal instability of current industrial enzyme sources, todays saccharification processes are performed at 55-60°C. It is interesting to note that, while thermophilic and hyperthermophilic anaerobes typically degrade pullulan with an amylopullulanase,pullulanases have only been characterized from thermophilic aerobes. True pullulanases (those with no activity on a-1,4glucosidic linkages) were characterized from B. stearothermophilus and a Thermus strain. Pullulanases were also purified from a Bacillus strain and from T . aquaticus YT-1, but their activity on starch has not been characterized (see Table 1 for references). Bacillus and Thermus true pullulanases are thermostable. While B. stearothermophilus pullulanase’s optimum pH (6.0) and instability at pH 5.0 [79] is barely compatible with the optimum pH of the thermophilic P-amylases and aglucosidases discussed above, the Thermus enzyme shows 75% activity at pH 5.0. Amylopullulanases’ dual specificity for both a-1,4- and a-l,6-glucosidic linkages in starch [297] does not make them good sources for industrial starch debranching enzymes. However, amylopullulanases have been suggested as alternative enzymes to replace a-amylases during starch liquefaction. Since certain amylopullulanases specifically produce maltose, maltotriose, and maltotetraose (DP2 to DP4) as the major end-products (T. thermosulfurigenes EM1 and T. succharolyticum amylopullulanases produce maltose as the major end-product [58,298], whereas the T. ethanolicus enzyme produces mainly maltotriose and maltotetraose [299]), they have been suggested as catalysts in a one-step liquefactiotdsaccharification process for the production of high DP2-DP4 syrups [300]. The amylopullulanases purified from the hyperthermophiles T . litoralis, P . furiosus, and ES4 are strong candidates for starch liquefaction because of their high temperature activity (110 to 125°C) and their exceptional thermostability (see Table 2). However, their specific activities on starch (35,49, and 24 pmol/min/mg, respectively) [141,142] are 13 to 47 times lower than the P . woesei and T. profundus a-amylases-specific activities (667 and 1,143 pmol/min/mg, respectively) [ 139,1401, and represent drawbacks for their industrial application. Cyclodextrin glycosyltransferases. Cyclodextrin glycosyltransferases (CGTases) are able to convert oligodextrins into cyclodextrins. Cyclic compounds, a-,p-, and 7cyclodextrins are composed of 6,7, or 8 a-1,Clinked glucose molecules, respectively. Cyclodextrins’ internal cavities are hydrophobic, and can accommodate or “encapsulate’’ hydrophobic molecules. This property makes cyclodextrins suitable for numerous applications in the food, cosmetic, and pharmaceutical industries, where they are used to capture undesirable tastes or odors, stabilize volatile compounds,
52 increase a hydrophobic substance’s water-solubility, and protect a substance against unwanted modifications. Improving the bioavailability of poorly soluble medical substances and increasing their therapeutic effects while allowing a decreased drug usage, is another major use for cyclodextrins (see [301]). Used for the annual production of more than 100 tons of cyclodextrins, CGTases can be considered industrial, not specialty, enzymes. Cyclodextrin production typically involves aamylase-catalyzed high-temperature starch liquefaction, followed by cyclodextrin formation using a mesophilic CGTase. CGTases have been characterized from the thermoanaerobes Thermoanaerobacter and Thermoanaerobacterium (Table 1; see [64] for references). Besides producing cyclodextrins in a single enzymatic step, these enzymes also have a-amylase activity, and are thermophilic enough to replace aamylases in the starch liquefaction process. Cellulases and hemicellulases Cellulose represents the most abundant, renewable, nonfossil carbon source on our planet. For this reason, it has been considered a feedstock substrate for alcohol and chemical productions, as well as a feed source for animal nutrition. Embedded in a network of hemicellulose and lignin, cellulose is a fundamental structural component of plant cell walls, and is therefore not easily accessible for degradation. At present, industrial uses of cellulases are limited to detergents; hemicellulases are used in animal feeds and bread making. Current procedures for ethanol production from cellulose involve three consecutive steps: 1) fermentation for cellulase production; 2) enzymatic cellulose saccharification (the cellulose has been preheated in alkaline or other conditions to remove the lignin, limiting the enzyme access to cellulose); and 3) ethanologenic fermentation (yeast fermentation). This three-step process is not economically feasible, largely because cellulase cost is high due to low enzyme activity. The best-characterized cellulolytic thermophile is C . thermocellurn (reviewed in [302]). This organism has been suggested for one-step ethanol production, but the production of other carbon endproducts (i.e., lactate, acetate) and the low ethanol concentration achieved (ethanol causes growth inhibition) may limit its industrial fermentation applications [303,304]. The simultaneous cellulose saccharification and ethanol fermentation by a yeast culture in the presence of added cellulases were also investigated. The incompatible optimal temperatures for yeast growth (25-30°C) and for cellulases action (45-50°C) are a major limitation to this process. For a more comprehensive description of the different processes investigated, see Margaritis and Merchant’s (1986) review [65]. The new thermophilic cellulase sources may provide improved industrial enzymes for cellulose saccharification and other applications. Several cellulases (i.e., endoglucanase, cellobiohydrolase, P-glucosidase) [ 146,148,149,1551(see Table 2) were characterized from different Thermotoga strains - the only hyperthermophiles known to use cellulose [8]. Multiple hemicellulases were also characterized from the same Thermotoga strains [ 146,147,149,155,3051 as well as from the thermophile C. saccharolyticus [66,67,74,87] (see Table 1 and 2). The concomitant use of highly thermostable cellulases and hemicellulases for cellulose and hemicellulose sac-
53 charification should provide a net yield-improvement. While several C. saccharolyticus xylanase genes have already been cloned and expressed in E. coli [66,74] most Thermofoga cellulases and xylanases have not. However, since numerous Thermotoga genes have been successfully cloned and expressed in E. coli, cellulase and xylanase cloning should be achieved soon. Among the other cellulase applications reviewed by BCguin and Aubert (1994) [174], several would probably profit from the use of highly thermostable enzymes. Due to bacterial fermentations, temperatures in silage can rise considerably, and using thermostable cellulases - to partially hydrolyze the plant cell walls - can lengthen cellulase action during silage fermentations. Thermophilic cellulases could also be used as detergent additives instead of mesophilic cellulases for high-temperature cleaning. With the pollution associated with paper pulp bleaching in paper mill effluents, the use of xylanases for paper pulp prebleaching to reduce the amount of chlorine required for bleaching is of interest. In the current process, lignin is prehydrolyzed in high-temperature alkaline conditions. After cooling and pH adjustment, hemicellulose is hydrolyzed using mesophilic hemicellulases produced by a noncellulolytic fungus. Since lignin is tightly associated with hemicellulose, a significant amount of it is washed from the pulp with the hydrolyzed hemicellulose. Substituting thermophilic hemicellulases for the mesophilic ones would improve the process in several ways: 1) less cooling would be required prior to hemicellulose hydrolysis; 2) better lignin-washing is expected since ligneous derivatives become more soluble at high temperatures; and 3) once the thermophilic xylanases are cloned in mesophilic hosts (required to achieve high expression levels), the balance between the different enzymes can be adjusted to reach optimal hydrolysis efficiency. Thermotogales are the only hyperthermophiles known to produce xylanases. The multiple Thermotoga hemicellulases characterized so far (endo-1,CP-xylanase, P-xylosidase, a - ~ arabinofuranosidase,glucuronidase, laminarinase/P-xylosidase) [ 146,147,149,155,3051, and Table 2) are candidates for paper pulp bleaching because they are optimally active in the temperature range 8CF 105"C, highly thermostable, and lack detectable cellulase activity. Thermotoga hemicellulases still need to be cloned in a convenient system to achieve high expression levels in the absence of contaminating cellulase activity. Thermostable xylanases from C . saccharolyticus and T . saccharolyticum have been characterized. Although less stable than the Thermotoga enzymes, different specificity enzymes (see Table 1 for references), such as C. saccharolyticus Pmannanase, endo- 1,4-P-xylanase,and P-xylosidase,or T. saccharolyticum endo- 1,4-Pxylanase and P-xylosidase (most of them have been cloned and expressed in E. coli) could add to the efficiency of the Thermofoga xylanase system. Use of xylanases in animal feeds is a recent but rapidly growing industrial enzymes market [301]. Adding xylanase to animal feed improves animal nutrition; it permits using a wider range of cereals and results in better digestibility. Xylanases can enhance the digestion of complex animal feed, or they can remove antinutritional animal feed factors for monogastric animals (e.g., poultry and swine). P-glucanase and pentosanase are used to hydrolyze P-glucan and pentosan, respectively. In
54 excessive concentrations in cereals, these two polysaccharides have a negative action on poultry digestion. More thermostable enzymes are needed to process the feed at high temperatures. Proteases The use of microbial alkaline serine proteases in detergent formulations represents the largest market for an industrial enzyme [306]. The currently used proteases are reasonably active at 60-70°C, but are rapidly inactivated at these temperatures, impeding their use at higher temperatures. With a constantly changing detergent market demand, the tendency today is to use phosphate-free liquid detergents. New constraints are associated with these formulations: 1) proteases are less stable in an aqueous environment than as a dry powder, and 2) the surfactants and bleaching agents used in phosphate-free detergents are typically serine-protease inhibitors. Thermophilic serine proteases may prove useful in detergent formulations. Their increased thermostability and their optimal activity at elevated temperatures should expand the use of enzymes in high-temperature detergents. Compared to mesophilic proteases, their increased resistance to denaturation should favor their stability in liquid detergents. More research is needed to characterize and clone more thermophilic proteases and to characterize their resistance to inhibitors and denaturants. Thermus strain Rt4A2 alkaline serine protease, has already proven to be highly thermostable (half-life of 90 min at 90"C), and is chelator-resistant [76]. Proteases optimally active at temperatures above 100°C have already been characterized from hyperthermophiles; A serine and a thiol protease were characterized from P. furiosus and Pyrococcus sp. strain KOD1, respectively [ 153,1541.Pyrococcus sp. strain KODl was shown to produce at least three extracellular proteases [154], and, since hyperthermophilic heterotrophs.typically get their energy and carbon from complex mixtures of peptides (see section on! General properties of thermozymes), more proteases active and stable in the range 100-110°C should be characterized in the near future. Such proteases should prove useful in specialized, industrial cleaning procedures. Preliminary investigations into Thermus serine protease use for cleaning ultrafiltration membranes fouled during whey processing have been performed [307]. The second group of major industrial proteases, aspartic proteases, are used in the cheese-making industry (see [306]). The enzymatic renneting process requires mesophilic, heat-sensitive enzymes. Other protease applications (i.e., leather processing, brewing, baking) represent marginal development areas for thermophilic proteases, but these processes, by themselves, would not justify expensive research and development efforts. Xylose isomerases Xylose isomerases (D-xylose isomerase EC 5.3.1.5) represent the first industrial use of immobilized enzymes [308]. Used in HFCS production, glucose isomerases catalyze the equilibrium isomerization of glucose into fructose. The conversion rate at equilibrium is shifted toward fructose at high temperatures; at 60 and 90"C, the fructose contents at equilibrium are 50.7 and 55.6%, respectively. The isomerization
55
process is typically run in packed-bed reactors at 58-60°C for 1 to 4 h, and the converted syrup reaches 42% fructose. An additional strong acid cation-exchange chromatographic step further increases the fructose concentration to 55%, the concentration required by most of todays HFCS applications. The limited thermal stability of the currently used glucose isomerases determines the moderate temperatures used is this process. Due to enzyme inactivation, the reactors need to be repacked every 2 months. Highly thermophilic and thermostable glucose isomerases have been characterized from T. thermophilus, T. aquaticus, T . maritima, and T. neapolitana (see Tables 1 and 2). T. thermophilus and T. neapolitana enzymes have been cloned and expressed in E. coli [97,98,160]. Despite their optimal activity at elevated temperatures (95--100°C) and their attractive, high catalytic efficiency at 90°C (see [ 160]), T. maritima and T. neapolitana glucose isomerases require Co2+for maximal activity, a major limitation to their HFCS production applications. Assayed in the presence of Mg2+,T. neapolitana glucose isomerase showed only 10% maximal activity (Vieille, unpublished data). (Not as catalytically efficient, the T. aquaticus enzyme shows a similar requirement for Co2+or Mn2+see [ 1601). The mechanisms of glucose isomerase thermal inactivation have been thoroughly investigated by several laboratories. Volkin and Klibanov [272] have studied the inactivation of immobilized glucose isomerase, and other authors have stabilized the enzyme by site directed mutagenesis [240,241,2691. Because of glucose isomerase’s industrial importance, abundant structural data have been accumulated [30+3 113. Professor Blow et al. (Imperial College, London) have also determined the structure of B. stearothermophilus, T. thermosulfurigenes,and T. neapolitana xylose isomerases (unpublished data). Based on the structural data, Meng et al. [312j have mutagenized the catalytic site of the T. thermosulfurigenes glucose isomerase and switched its substrate preference from xylose to glucose. They also thermal stabilized the enzyme by reducing the water-accessible hydrophobic surface in the active site [242]. Numerous protein engineering studies have shown that, once structural data are available, increasing enzyme thermostability becomes possible. With the increasing structural information available on glucose isomerases’ metal-binding sites and on other metal-binding proteins [309,313-3 161, altering one enzyme’s metal affinity might become a realistic task. Because of their industrial importance and because of the abundance of data, glucose isomerases represent a particularly attractive engineering model. ’
Specialty uses of thermozymes Molecular biology reagents DNA polymerases. The cloning and expression of T. aquaticus Taq DNA polymerase in E. coli was of major importance in developing the PCR technology. PCRs uses in different scientific areas (i.e., molecular biology, forensic and diagnostic medicine, taxonomy) have been the object of numerous reviews and will be not be covered here (see [41,317]). The biological significance of PCR technology is reflected in the
56 number of thermophilic DNA polymerases (see Table 8) now available through major molecular-biology product companies (e.g., New England Biolabs, Stratagene, Pharmacia, Boehringer Mannheim Biochemicals, U.S. Biochemicals). Thermophilic DNA polymerases are now available both with (Vent polymerase, Deep vent polymerase, etc.) and without (Taq, vent (Exo-), Deep vent (Exo-)) 3 ' 3 ' proofreading exonuclease activity. Proofreading enzymes are preferred when highfidelity DNA amplification is required (for example, in direct gene cloning). Enzymes without proofreading activity are required in sequencing procedures. Since Deep vent polymerase, unlike the Taq polymerase, does not have a terminal transferase activity, it is efficiently used in direct PCR cloning procedures. While thermophilic DNA polymerases have partially replaced mesophilic enzymes in a few applications, most applications are tightly linked to their thermophilicity and thermostability properties, and were developed after their characterization. In this respect, thermophilic DNA polymerases are an excellent example of a scientific discovery creating new applications.
DNA restriction endonucleases. Numerous thermophilic restriction endonucleases have now been characterized. At least 48 enzymes are already commercialized. All optimally active in the range 50-65"C, the majority of these enzymes have been isolated from thermophilic eubacteria (26 come from Bacillus strains and seven come from Thermus strains). The genes encoding the two restriction endonucleases PI-ThiI and PI-PspI were identified in protein intervening sequences that were discovered in a Pyrococcus and in the T. litoralis DNA polymerase genes [ 132,1771. Other archaeal thermophilic restriction endonucleases have been characterized [3 18,3191. Other enzymes. Three DNA ligases, isolated from T. aquaticus, T. thermophilus, and P. furiosus, are commercially available (New England Biolabs and Stratagene). Optimally active in the range 37-80°C, with half-lives of 30 min or more at 95"C, they represent an excellent addition to the PCR technology. The recent ligasedependent DNA-amplification reaction (LAR) used in identifying gene defects makes full use of their properties (see [320]). The T. aquaticus Taql methylase, optimally active at 65"C, is available from New England Biolabs. Thermophilic proteases have been proposed for protease treatments during DNA and RNA preparations. Some mesophilic proteins, resisting proteolytic digestion at moderate temperatures (20 to 37"C), unfold at higher temperatures and become more sensitive to proteolytic attack. Since DNA and RNA preparations are usually further treated with other enzymes, the best protease candidates should be easily inactivated. Rapidly inactivated by EGTA, Thermus Rt41A serine protease was used successfully in DNA and RNA preparation procedures (see [77]). This enzyme was also proposed as an adjunct to PCR for diagnostic applications, to break down cellular structures prior to the amplification reaction (see [41]). Thermus Rt41A serine protease (PRETAQ)is commercialized by Life Technologies.
Table 8. Commercialized thermophilic DNA polymerases. Enzyme
Source
Activity
Optimal temperature ("C)
Stability
Taq polymerase
Thermus aquaticus YTI
75
Vent DNA polymerase
Thermococcus litoralis
unknown
can withstand temperatures up to 95°C t,, = 6.7 h at 95°C
Vent (Exo-) DNA polymerase
Engineered vent DNA polymerase
unknown
t,, = 6.1 h at 95°C
Deep vent DNA polymerase
Pyrococcus furiosus
15
t,, = 23 h at 95°C
75
t,, = 23 h at 95'C
unknown
t,, = 8 h at 95°C
9"N,DNA polymerase
Engineered DNA polymerase from Thermococcus sp. strain 9'N-7
5'-3' DNAdependent DNA polymerase, no proof-reading activity 5'-3' DNA-dependent DNA polymerase, proof-reading activity 5'-3' DNA-dependent DNA polymerase, no proof-reading activity 5'-3' DNA-dependent DNA polymerase, proof-reading activity 5'-3' DNA-dependent DNA polymerase, no proof-reading activity 5'-3' DNAdependent DNA polymerase, I-5% proof-reading activity
Tth DNA polymerase
Thermus thermophilus HB8
5'-3' DNAdependent DNA polymerase
unknown
unknown
Heat-TUFF DNA polymerase
not available
5'-3' DNAdependent DNA polymerase
unknown
unknown
Hot Tub DNA polymerase
Thermus ubiquitous
5'-3' DNA-dependent DNA polymerase
75
TET-z DNA polymerase
Thermus thermophilus
5'-3' DNAdependent DNA polymerase plus 5'-3' RNA-dependent DNA polymerase activity (reverse transcriptase)
unknown
can withstand temperatures up to 95°C unknown
Deep vent (Exo-) DNA polymerase Engineered deep vent DNA polymerase
58 Organic synthesis reagents While industrial organic syntheses have typically been dominated by chemical processes, the rising environmental public awareness (reflected by new legislation) plus the recent developments in enzyme biotechnology, suggest that enzymatic and mixed chemo-enzymatic processes will progressively be substituted for chemical processes. Three enzyme groups - lipases, proteases, and oxidoreductases - have a high potential for synthesizing peptides (e.g., neuropeptides, peptides for research purposes) [32 11; flavors and fragrances (e.g., benzaldehyde, naringin) [322]; polymers (e.g., polyesters, polyphenols, polyacrylates) [323]; chiral compounds [324]; and surfactants (e.g., monoglycerides, sugar fatty acid esters, alkyl glucosides) [325]. Typically known for their hydrolytic activity, lipases and proteases can, in specific environments, be used as synthetic enzymes. Enzymatic syntheses present numerous advantages over chemical syntheses: they are usually highly specific (i.e., enantio-, regio-, and stereo-specific), environmentally friendly, and their products are usually easily biodegradable. Thermophilic proteases. Protease use for peptide synthesis is determined by enzymatic peptide synthesis’ advantages (i.e.. structural, regio-, and stereo-specificities, effect on reaction rates, mild conditions of reaction) over chemical synthesis (See [321,326] for details). In nature, the zwitterionic character of peptides and free amino acids hinders protease-catalyzed peptide synthesis. Adding organic solvents to the reaction mixture lowers the dielectric constant and diminishes the hydration of amino acids ionic groups, ultimately affecting their pKa values. According to Le Chatelier’s principle, the endothermic reaction of peptide bond synthesis should be favored by rising temperatures. This effect has been experimentally observed using B. thermoproteolyticus neutral protease (thermolysin) to synthesize a dipeptide from carboxybenzoxy-glycine and L-phenylalanine amide [327]. Optimally active at high temperatures and more resistant to organic solvents than their mesophilic counterparts, thermophilic proteases should prove excellent candidates for enzymatic peptide synthesis. Thermolysin, frequently used in experimental enzymatic peptide syntheses, has proven to be a valuable catalyst among the metalloproteases investigated (see [321]). Thermus Rt41A alkaline serine protease (optimally active at 90OC) has been assayed for peptide synthesis in multiple conditions; rising temperature, varying pH, varying ionic strength, substrate concentrations, different solvents, varying solvent concentrations, and free and immobilized enzyme [275,328]. Protease stability at extreme temperatures, pHs, and solvent concentrations allowed peptide synthesis under conditions not possible with mesophilic enzymes [329]. The only thermophilic protease currently used on an industrial scale is thermoly sin, used to produce the dipeptide aspartame (L-aspartyl-L-phenylalaninemethyl ester) [330,33 11. Active at elevated temperatures and highly resistant to solvent denaturation, proteases isolated from hyperthermophiles are strong candidates for synthesis applications where the highly temperature-dependent solvent viscosity and substrate solubility affect the reaction rate.
59 Lipase reagents. Although lipases' uses are limited to a small number of specialized applications (particularly in the food industry for flavor development), they hold huge potential for enzymatic organic syntheses. By enantioselectively hydrolyzing a wide range of nonnatural esters, lipases are able to generate a variety of pure enantiomers that can be used as starting materials in pharmaceutical production [332]. Some of their potential applications for synthesizing polymers (e.g., polyesters, polyacrylates) and surfactants (e.g., monoglycerides, amino acid esters) have been recently reviewed [323,325]. Worldwide, annual surfactant production is expected to reach four million tons by the year 2000. Because major surfactant applications (i.e., detergent, personalcare products, and food industries) directly affect the environment and/or consumers, the surfactant manufacturing industry feels pressure to substitute biologically produced surfactants for the traditional chemically produced ones. To avoid using solvents, some lipases are directly used in a mixture of their substrates. Due to fats high melting-point, reactions are often run at high temperatures (50-80") [333]. The potential of thermophilic lipases has not received much research attention in either academia or industry. Oxidoreductase reagents. The potential biotechnological applications of oxidoreductases in the production of chiral compounds has become the focus of tremendous interest. However, their expensive cofactor requirement (NAD, NADP, FAD) may limit their use [324,334]. Several systems, including hydrogenase use [335], have been proposed for cofactor recycling, but current applications are typically performed in whole cells, where the cells regenerate the cofactor themselves [324,334]. With the top 10 optically active drugs representing sales of US $10 billion annually and with the recent FDA policy focused on pharmaceutical pure enantiomers [324], more attention centers on the use of oxidoreductases to perform synthetic transformations. 2" ADHs, active on a wide range of substrates ( 1 and 2" alcohols, aldehydes, and ketones), have great industrial potential. So far, 2" ADHs have been characterized from several Thermoanaerobacterium and Thermoanaerobacter species [36,336]. Thermoanaerobacter brockii 2" ADH has already been used in the analytical scale production of a chiral constituent of civet fragrance [337] and of an insect pheromone [338]. Its potential value for other syntheses has also been examined [339]. These enzymes have high temperature optima (80 to 95"C), high thermostability, and reduced sensitivity to oxygen (they are only reversibly inactivated, unlike 1" ADHs) [336]. The extent of T. brockii 2" ADH stability in the presence of organic solvent is still controversial. In Lamed et al.'s 1981 study [339] the enzyme remained 80% active after 15 min at 52°C in the presence of 40% 2-propanol, whereas it was reviewed by Cowan [320] to rapidly inactivate at temperatures above 45°C and in the presence of 10% organic solvent. A 1" ADH with a broad substrate specificity has been purified from the hyperthermophile S. solfataricus [loo]. This enzyme has an increased stability in the presence of solvents [320], but is low catalytic efficiency on 2" alcohols and ketones might limit its applications.
60 Other applications Diagnostics. Because of their high specificity and catalytic efficiency, enzymes are abundantly used in diagnostics (reviewed in [301]). The most common diagnostic enzymes (i.e., diaphorases, oxidases (in association with peroxidase), P-galactosidase, and alkaline phosphatase) catalyze easily monitored reactions. With the popular immunoassay use of enzymes (taking the place of many radioimmunoassays), enzyme’s use in diagnostics is increasing. Reagent stability and consequent extended shelf-life are essential in developing a reproducible diagnostic assay. Thermozymes’ impressive stability under suboptimal conditions makes them attractive diagnostic reagents. Since thermozymes can be stably stored at room temperature, they are not subject to freeze-thaw denaturation as happens with mesozymes. Since enzyme cost is not a major issue in diagnostic medicine, thermozymes’ lower activity at mesophilic temperatures can be compensated by using more enzyme. P-galactosidase and alkaline phosphatase have been characterized from the hyperthermophiles S. solfataricus (see [lo]) and T. neapolitunu (Dong and Zeikus, unpublished results), respectively. The discovery of slightly different metabolic pathways and enzymes in the hyperthermophilic archae can also lead to new diagnostic assay developments by allowing new reactions. Waste treatment-and minimization. With our planet’s population explosion, waste accumulation has become a serious environmental problem necessitating biological solutions (both enzymatic and microbial) for waste minimization or treatment. Using thermophilic microbial systems is a standard practice for treating animal waste and composting natural material. The use of enzymes for waste-treatment is much more limited. Proteases currently convert meat by-product wastes (feathers, fish byproducts, animal hair, and blood) into nutritional sources [301]. Because animal waste sources potentially contain pathogens, these wastes are sterilized before being mixed with other feedstuffs. Adding highly thermostable proteases to the sterilization process may allow simultaneous proteolytic digestion and sterilization, and, since proteins are usually more susceptible to proteolytic degradation at elevated temperatures, this process would increase proteolytic efficiency. Hydrogenase-mediated hydrogen isotope exchange has been proposed to treat thermal tritiated water (see [335]) in nuclear power plants. In radiocontaminated applications, enzyme stability represents a particularly important isSue, since it determines the frequency of reactor repacking. Many industry chemical processing wastes are radioactive, and need to be treated before discharge. Therefore, these industries might be interested in thermostable nitrilases or dehalogenases. The high tech solution to European animal waste pollution involves altering feed so that treatment demands are eliminated. Over half the phosphorus present in grains is in phytin (myo-inositol hexaphosphate) - not degradable by monogastric animals. Adding the enzyme phytase to the feedstuff has a double benefit: 1) since more phosphate is available, there is no need for a phosphate complement in the diet, and 2) pig and poultry feces contain less phosphates and are less polluting. Since much
61 feedstuff is pelleted at temperatures up to 90°C (to limit pathogenic contamination), the use of thermozymes as food additives to minimize waste treatment would be particularly appropriate. The growth of industrial composting systems for treating yard and paper wastes represents an opportunity to evaluate the utility of cellulolytic thermozymes in waste treatment. Genetic engineering of thermozymes Modification of enzyme catalytic properties New protein structural data are constantly accumulating, and help in understanding the stereospecific mechanisms of enzymatic reactions. This new knowledge incites the use of enzyme engineering as a rational research approach, to make enzyme catalytic properties fit industrial processes. Knowing the L-lactate dehydrogenase structure and its catalytic mechanism, Holbrook et al. attempted to modify L-lactate dehydrogenase's substrate specificity. In an early work [340], B. stearothermophilus L-lactate dehydrogenase was changed into malate dehydrogenase by a point mutation that increased the substrate pocket size to better accommodate oxaloacetate. The mutant enzyme showed better malate dehydrogenase efficiency than the native B. stearothermophilus malate dehydrogenase. In a later work, Holbrook et al. used the thermophilic B. stearothermophilus L-lactate dehydrogenase again, and redesigned its substrate binding site to generate a nonspecific a-hydroxy acid dehydrogenase [341]. The mutant enzyme remained thermostable, and could be used for various chemical syntheses. A similar approach was used by Meng et al. [312] to switch the substrate preference of T. thermosulfurigenes xylose isomerase from xylose to glucose. Combined with the characterization of highly thermostable enzymes from hyperthermophiles, this site-directed mutagenesis approach should extend thermozyme applications. Arnold et al. [342,343] used PCR-mediated random mutagenesis to enhance subtilisin E activity in polar organic solvents. They developed a sensitive plate-assay screening method, and, by using several screening steps, they selected a multiple mutant that was 256 times more efficient than the wild-type enzyme in 60% dimethylformamide. Thermostability and thermophilicity engineering Research has shown that while thermophilicity and thermostability can be altered independently, they are often structurally related. Numerous molecular mechanisms might be responsible for these properties; only a few, however, are of genetic engineering value: 1) protein sequences contain redundant information for proper folding, and protein structures can often accommodate amino acid substitutions without significantly altering the catalytic efficiency; 2) a small number of amino acid substitutions throughout the protein can significantly alter its thermophilicity and thermostability; and 3) certain regions in a protein are more labile than others stabilizing these regions (rather than the more stable ones) is critical to improve protein thermal properties. Four general strategies for thermal stabilization have been
62 identified: more efficient protein core packing, a-helix stabilization, surface loop driven core stabilization, and prevention of chemical degradation. Recent crystallographic analysis has indicated that more rigid core packing is a hallmark of thermophilic proteins [89,186]. However, systematic engineering of core rigidity without altering protein function is beyond the capabilities of current technology. Stabilizing a-helices by capping or by introducing alanines has had spotty success, and suffers from the same engineering problems as do the core packing strategies. Loop stabilization has successfully stabilized numerous proteins. Sequence analysis methods now reliably predict surface loop regions, even in the absence of threedimensional structural information. Specific stabilizing amino acid substitutions have been identified (e.g., substitution of surface lysines with arginines and introduction of prolines in short loops), making this approach a good engineering tool. Prevention of peptide chemical degradation is also important for very high temperature applications (>80°C). Trends toward reduced Cys, Asn, and Gln contents in hyperthermophilic enzymes have been identified in nature [ 1601, and deamidation is a main factor responsible for irreversible a-amylase inactivation [ 1891. The appropriate strategy for engineering protein thermostability or thermophilicity depends on the project goal, on the available structural information, and on the thermal mechanisms limiting protein stability or activity (see Fig. 4). Selecting an initial protein with thermal properties as close to the target as possible clearly increases the chances of successful engineering. Chances for engineering success are further enhanced by having the initial proteins high-resolution three-dimensional structure as well as a structurally similar protein that has thermal characteristics close to the desired protein. Comparing thermophilicity and thermostability properties indicates whether the limiting step in protein inactivation is the initial partial unfolding or irreversible structural changes (such as precipitation or chemical degradation). If, in the temperature range of interest, the protein thermostability halflife is longer than the activity half-life, the reversibly partially unfolded protein is not immediately irreversibly inactivated. If, however, these two half-lives are similar, irreversible inactivation follows the initial unfolding almost instantly. The protein concentration in the target application should be estimated based on the protein function (receptor, enzyme, antibody, etc.), to provide a concentration range for these characterizations. Determination of the thermostability half-life at different initial protein concentrations within this range will indicate whether inactivation is intramolecular (concentration independent) or not. Having increased thermostability allows applications of repeated thermal cycling (e.g., PCR, LCR), as well as permitting preapplication high-temperature processing steps (e.g., processing food or feed additives). To increase thermostability engineering efforts should focus on preventing irreversible inactivation. Strategies to prevent aggregation (e.g., chemical modification of lysine residues with succinic acid, addition of polyalcohols to the solution, or immobilization on an inert matrix) might significantly thermostabilize the protein. This approach would need to be undertaken if 1) enzyme precipitation coincides with inactivation and the precipitated protein does not retain complete activity, or 2) if aggregation is undesirable for the target
63 Design goal 'Ihumoslability
Thermophilicity
I
RevaSible w i n inactivation
Irreversible protein inactivation
Precipitation
Recipitarc
Reciiw
Revent suucturalchange:
c3soa8L 1) Random mutagemis with selection 2) wpence. comparisondirected mutagcncsis 3) substitute Pm midues inlo loops 4) r e p h e Lys with Arg in loops 5) chemicallyCrOSSiinL 1) immobilize
6) immobilize 7) add s t a b i l i i chemicals to the system a) W i c I) salts (NaCI, KCI, etc.) ii) S u b s I r a ~ f f ~ a U s b) reduce solvent e n m y i) glywml ii) Hofmeister swies ions
2) Chemically modify a) succinylateLys residues b) intramolecular uoss-link 3) add polyalcohols to system
1) add divalent cation binding site 2) reduce bond stran by mutagenesis 3) fffl intend cavities by mulagenesis 4) stabilim helices i) M l y to cap helix C-terminus ii) Glu/Asp to cap hclix N-tcrminus iii) Ala to srabilize helix
1) add Gly to loops
OW 80°C~
-WC* 1) reduce the number of A.m. Gln, and Cys residues
t -3-D hS
1) add disulfide bonds
Fig. 4. Flow chart of potential steps toward engineering enzyme thermophilicity and thermostability.
application. If these measures are insufficient or inappropriate for the design goal, then prevention of inactivating conformational changes through peptide modifications (such as intramolecular cross-linking or through system additives, salt or glycerol, for example, that stabilize the folded protein structure) should be examined. Salts have
64 been shown to stabilize proteins through specific binding, charge shielding, and modification of the system entropy. The addition of organic chemicals such as glycerol and polyethyleneglycol has also been shown to stabilize some proteins, but they often reduce activity and can destabilize some proteins. The specificity of intramolecular cross-linking is difficult to control and often reduces protein activity but it covalently tethers protein regions analogously to disulfide linkages. Either random or site directed mutagenesis of the initial protein may be used to thermostabilize proteins. Random mutagenesis has been successfully used to increase enzyme thermostability and resistance to solvent-induced denaturation. Two research groups simultaneously developed [344,345] similar systems in which a gene encoding a mesophilic enzyme is introduced into a thermophilic host, and variant enzymes with increased thermostability are selected during growth at increasing temperatures. In both studies, heat-stable kanamycin nucleotidyltransferase mutants were obtained using a B. stearothermophilus strain that expressed a mesophilic kanamycin nucleotidyltransferase to select kanamycin resistant variants at increasing temperatures. Although attractive at first glance, this approach is limited by the absence of genetic tools (i.e., shuttle vectors, transformation methods) available for most thermophiles and hyperthermophiles, as well as by the limited enzymatic activities directly selectable in B. stearothermophilus cultures. The current lack of thermophilic cloning hosts makes also classic genetic complementations by thermophilicity mutants a significant challenge. Practical use of random mutagenesis requires a powerful mutant selection procedure or specific structural information (to limit the size of the target, and limit the total number of possible mutants to be tested). Site-directed mutagenesis usually requires protein structural information. Strategies that include substitutions with residues present in structurally similar proteins with thermal properties similar to the desired mutant require sequence information on both proteins. If the enzyme targeted for hyperthermophilic applications (>80"C) is irreversibly inactivated, reduction of the number of noncatalytic Gln, Asn, and Cys residues can be used to stabilize the protein structure using the sequence information alone. If the enzyme is only reversibly inactivated, substituting Gly residues with more constrained residues in loop regions is another strategy which does not require extensive structural information. Loop stabilization by introducing Pro or Arg residues requires only knowing the protein sequence. Exact structural information from crystallography, NMR, or homology modeling allows the engineering of multiresidue motifs such as metal binding sites or disulfide bridges (for moderate temperature stabilization) to stabilize the enzyme structure, and allows specifically tailored design schemes such as cavity filling mutations, helix stabilization by alanine insertions or capping, and alleviation of specific bond strains. Deuterium exchange, measured by NMR, can precisely identify the more labile regions to stabilize [346]. Muheim et al. [347] created a dihistidine metal-chelating site on a surface P-sheet of cytochrome c, and cross-linked it with the metal complex Ru" (2,2'-bipyridine). This cross-linking increased the melting temperature of the mutant enzyme by more than 23°C. Increasing enzyme thermophilicity requires preventing thermally induced unfolding. Since enzyme thermophilicity is often limited by thermostability,
65 engineering thermophilicity usually requires enzyme thermostabilization. The two studies [226,2911 described in the section: Molecular mechanisms involved in protein thermophilicity - are nice examples of thermostabilization studies leading to a significant thermophilicity increase. However, for an enzyme that precipitates upon heating, if precipitation is not the first inactivation step, preventing aggregation is unlikely to enhance thermophilicity.
Conclusion Data are rapidly accumulating on thermozymes characterized from thermophilic and hyperthermophilic microbes. Studying archaeal and bacterial hyperthermophiles is expected to provide elements essential to understanding the origins of life on Earth. With their remarkable thermostability and activity, thermozymes are a major model for studying protein thermostability and biocatalysis at high temperatures. Easy to purify and crystallize, thermozymes represent excellent models for studying enzyme structure/function relationships. Excellent candidates for industrial enzymatic applications (which may require high-temperature processes and highly stable enzymes), thermozymes also attract considerable attention from biotechnology. Long believed to be different from mesophilic enzymes (thermozymes are intrinsically thermostable and thermophilic), thermozymes amino acid compositions, their sequences, and their structures are strikingly similar to those of mesophilic enzymes. Typically active in a temperature range that includes the host organism’s optimal growth temperature, thermozymes catalyze the same reactions, and show similar Arrhenius behaviors as their mesophilic counterparts. We believe that, while their higher intrinsic rigidity allows thermozymes to resist denaturation at elevated temperatures, thermozymes do not significantly differ from mesozymes. One current hypothesis is that, in addition to substrate energy variations, catalytically significant enzyme conformational changes govern the activity variations with temperature. This hypothesis would justify the thermozymes poor activities at low temperatures. At these temperatures, excessive enzyme rigidity “freezes” the catalytic site in a poorly active conformation. Since the nature and sequence of events involved in temperature-dependent flexibility increase is enzyme specific, all types of Arrhenius plots (mostly nonlinear) would be expected if this hypothesis were correct. The linear Arrhenius plots observed for most enzymes do not support this hypothesis. In addition, as of today no experimental data have identified any correlation between enzyme activity level and enzyme rigidity. Instead, we hypothesize here that substrate energy variations are the main factor controlling enzyme activity levels, and that enzyme rigidity does not affect catalytic rates to a significant extent. Thus, poor catalytic-rate optimization would explain thermozymes lower catalytic efficiency when compared to mesozymes, and their poor activity at low temperatures. From all the factors (e.g., hydrophobicity increase, introduction of more alanines in a-helices, decrease of the Lys/Arg ratio, prolines in loops) suggested to explain
increased thermozyme thermostability, none appears as a general trend. There is no reason to believe that a single, unifying mechanism accounts for protein thermostability or thermophilicity. Only a few thermozyme crystal structures have been determined, and, with the exception of one enzyme [89], no comparative data are available on the relative compactness of thermozymes and mesophilic enzymes. We believe that different factors stabilize different enzymes, and that all these factors have a convergent, compacting effect on proteins. The absence of a unique stabilization mechanism does not preclude the application of a single strategy for engineering thermostability in a broad range of enzymes. As an engineering problem, there is rich potential for a wide range of solutions. Thermozymes are currently only used in a limited number of applications. In several industrial sectors (such as the starch processing and the detergent industries, which use the so-called “bulk” industrial enzymes), cost is the major issue. Introducing a new enzyme is worthwhile only if the cost improvement provided by this enzyme justifies research and development costs as well as the necessary changes to production equipment. Thus, in the near future we do not expect to see many new thermozyme applications in the major industrial processes. More possibilities exist in the specialty areas, where cost is not an important factor when compared to process specificity. Thermozyme uses are more likely to develop in sectors such as specialty chemicals, diagnostics, research reagents, personal-care, and food additives. Many aspects of hyperthermophile diversity and metabolism are poorly studied. Characterizing new organisms, their metabolic pathways, and enzymes, should broaden pefspectives on thermozyme applications. While thermozymes will soon compete with mesozymes for a few applications, and while some enzyme applications specifically require mesophilic enzymes (e.g., the cheese-making industries), we expect thermozymes to find uses in completely new processes (for example, as systems for specific chiral chemical synthesis). The development of the PCR technique after the cloning of T. aquaticus Taq DNA polymerase is an excellent example. In most cases it is still too early to predict the biotechnological future of thermozymes. Some of them have been cloned and overexpressed in mesophilic systems, but most enzymes have only been purified and characterized from the host organism. Further development work is required before comprehensive knowledge of potential thermozyme biotechnological applications is gained. The need is there. We propose that protein chemists use thermozymes as model systems for designing industrial enzyme catalysts. Two approaches can be used: 1. Because thermozymes expressed in mesophiles are easy to purify and crystallize, they represent excellent models for studying protein stability. Since they are structurally similar to their mesophilic counterparts, understanding the molecular mechanisms involved in thermozyme stabilization can be used to stabilize commercial mesozymes while maintaining the Mesophilic activity range. 2. A desired activity could be introduced into a stable enzyme. The highly stable thermozymes can be used as skeletons on which new activities can be engineered. With this approach, new industrial enzyme catalysts can be created. New
67 processes, for which no mesophilic enzyme and no chemical catalysts exist, can be developed.
Acknowledgements This research was supported by Grant 89-34189-4299 from the U.S. Department of Agriculture, and by Grant NSF-BES-9529047-63143-Zeikus from the National Science Foundation. We gratefully acknowledge Bridgette Leftridge for compiling the data used in the protein amino acid composition table, Carol McCutcheon for her clerical assistance in preparing this manuscript, and Dr. Vladimir Tchemajenko for valuable discussions. We express our gratitude to Christopher B. Jambor for editing the manuscript. Any mistakes that might remain are ours.
References 1.
2. 3.
4.
5.
6. 7. 8.
9.
10. 11.
12. 13. 14.
15.
16.
Zeikus JG. Thermophilic bacteria: ecology, physiology and technology. Enzyme Microb Techno1 1979;1 :243-252. Zeikus JG, Brock TD. Protein synthesis at high temperatures: aminoacylation of tRNA. Biochim Biophys Acta 1971;228:736745. Schumann J, Wrba A, Jaenicke R, Stetter KO. Topographical and enzymatic characterization of amylases from the extremely thermophilic eubacterium Thermotoga muritimu. FEBS Lett 1991;282: 122-1 26. Lowe SE, Jain MK, Zeikus JG. Biology, ecology, and biotechnological applications of anaerobic bacteria adapted to environmental stresses in temperature, pH, salinity, or substrates. Microbiol Rev 1993;57:451-509. Horikoshi K. Enzymes of akalophiles. In: Fogarty WM, Kelly CT (eds) Microbial Enzymes and Biotechnology, 2nd edn. New York: Elsevier Sciences Publishers, 1990275-294. Daniel RM. Modem life at high temperatures. Orig Life Evol Biosphere 1992;22:33-42. Wiegel J, Ljungdahl LG. The importance of thermophilic bacteria in biotechnology. CRC Crit Rev Biotechnol 198*3:39- 108. Stetter KO, Fiala G, Huber G, Huber R, Segerer A. Hyperthermophilic microorganisms. FEMS Microbiol Rev 199075: 1 17- 124. Adam MWW. Enzymes and proteins from organisms that grow near and above 100°C. AM Rev Microbiol 1993;47:627-658. Coolbear T, Daniel RM, Morgan HW. The enzymes from extreme thermophiles: bacterial sources, thermostabilities and industrial relevance. Adv Biochem Eng Biotechnol 1992;45:57-98. Barns SM, Fundyga RE, Jeffries MW, Pace NR. Remarkable archaeal diversity detected in a Yellowstone National Park hot spring environment. Proc Natl Acad Sci USA 1994;91:1609-1613. Jannasch HW, Mottl JM. Geomicrobiology of deep-sea hydrothermal vents. Science 1985;229:7 17725. Stetter KO. Microbial life in hyperthermal environments. ASM News 1995;61:285-290. Zillig W, Stetter KO, Wunderl S, Schulz W, Priess H, Scholz I. The Sulfolobus-"Caldariella" group: Taxonomy on the basis of the structure of DNA-dependent RNA polymerases. Arch Microbiol 1980 125:25+269. Jannasch HW, Nelson DC, Wirsen CO. Massive natural occurrence of unusually large bacteria (Beggiutoa sp.) at a hydrothermal deep-sea vent site. Nature 1989;342:834-836. Huber R, Woese CR, Langworthy TA, Fricke H, Stetter KO. Thermosipho africanus gen. nov.
represents a new genus of thermophilic eubacteria within the “Thermotogales”. Syst Appl Microbiol
1989;12:32-37. 17. Brock TD, Brock Kh4, Belly RT, Weiss RL. Sulfolobus: A new genus of sulfur-oxidizing bacteria living at low pH and high temperature. Arch Mikrobiol 1972;84:54-68. 18. Zeikus JG, Wolfe RS. Methanobacterium thermoautotrophicum sp. n, an anaerobic, autotrophic, extreme thermophile. J Bacteriol 1972;109:707-713. 19. Stetter KO. Ultrathin mycelia-forming organisms from submarine volcanic areas having an optimum growth temperature of 105°C. Nature 1982;300:258-260. 20. Baross JA, Lilley MD, Gordon LI. Is the CH,, H, and CO venting from submarine hydrothermal systems produced by thermophilic bacteria? Nature 1982;298:366-368. 21. Zillig W, Stetter KO, Prangishvilli D, Schaefer H, Wunderl S , Janekovich D, Holz I, Palm P.
22. 23. 24. 25.
26.
27. 28.
29.
30. 31.
32. 33.
Desulfurococcaeceae, the second family of the extremely thermophilic anaerobic sulfur respiring Thermoproteales. Zb Bakt Hyg I. Abt Orig 1982;C3:304-317. Segerer A, Langworthy TA, Stetter KO. Thermoplasma acidophilum and Thermoplasma volcanium sp. nov. from solfatara fields. Syst Appl Microbiol 1988;10:161-171. Fischer F, Zillig W, Stetter KO, Schreiber G. Chemolithoautotrophic metabolism of anaerobic extremely thermophilic archaebacteria. Nature 1983;301:511-513. Stetter KO, Huber R, Blochl E, Kurr M, Eden RD, Fielder M, Cash H, Vance I. Hyperthermophilic archaea are thriving in deep North Sea and Alaskan oil reservoirs. Nature 1993;365:743-745. Huber R, Langworthy TA, Konig H, Thomm M, Woese CR, Sleytr UB, Stetter KO. Thermotoga maritima sp. nov. represents a new genus of unique extremely thermophilic eubacteria growing up to 90°C.Arch Microbiol 1986;144:324-333. Pley U, Schipka J, Gambacorta A, Jannasch H W ,Fricke H, Rachel R, Stetter KO. Pyrodictium abyssi sp. nov. represents a novel heterotrophic marine archaeal hyperthermophile growing at 1 10°C. Syst Appl Microbiol 1991;14245-253. Reysenbach A-L, Deming JW.Effects of hydrostatic pressure on growth of hyperthermophilic archaebacteria from the Juan de Fuca Ridge. Appl Environ Microbiol 199137:1271-1274. Erauso G, Reysenbach A-L, Godfroy A, Meunier J-R, Crump B, Partensky F, Baross JA, Marteinsson V, Barbier G, Pace NR, Prieur D. Pyrococcus abyssi sp. nov., a new hyperthermophilic archaeon isolated from a deep-sea hydrothermal vent. Arch Microbiol 1993;160:338-349. Nelson CM, Schuppenhauer MR, Clark DS. High-pressure, high-temperature bioreactor for comparing effects of hyperbaric and hydrostatic pressure on bacterial growth. Appl Environ Microbiol 199258:1789-1793. Achenbach-Richter L, Gupta R, Stetter KO, Woese CR. Were the original eubacteria thermophiles? Syst Appl Microbiol 1987;9:34-39. Hedrick DB, Pledger RD, White DC, Baross JA. In situ microbial ecology of hydrothermal vent sediments. FEMS Microbiol Ecol 1992;lOl:l-10. Huber R, Burggraf S , Mayer T, Barns SM, Rossnagel P, Stetter KO. Isolation of a hyperthermophilic archaeum predicted by in situ RNA analysis. Nature 1995;376:57-58. Winter J, Lerp C, Zabel H-P, Wildenauer FX, Konig WH, Schindler F. Methanobacterium wolfei sp. nov., a new tungsten-requiring thermophilic autotrophic methanogen. Syst Appl Microbiol
19845457-466. 34. Schmitz RA, Richter M, Linder D, Thauer RK. A tungsten-containing active formylmethanofuran dehydrogenase in thermophilic archaeon Methanobacterium wolfei. Eur J Biochem 1992;207:559565. 35. O’Brien WE,Brewer JM, Ljungdahl LG. Purification and characterization of thermostable 5.10methylenetetrahydrofolate dehydrogenase from Closrridium thermoaceticum. J Biol Chem
1973;248:403-408. 36. Burdette DS, Vieille C, Zeikus JG. Cloning, expression, and biochemical characterization of the thermophilic secondary-alcoholdehydrogenase from Thermoanaerobacter ethanolicus 39E.Biochem J 1996;313:(Inpress). 37. Hatchikian EC, Zeikus JG. Characterization of a new type of dissimilatory sulfite reductase present
69 in Thermodesulfobacterium commune. J Bacteriol 1983;153:1211-1220. 38. Benachenhou-Lahfa N, Labedan B, Forterre P. PCR-mediated cloning and sequencing of the gene encoding glutamate dehydrogenasefrom the archaeon Sulfolobus shibatae: identification of putative amino acid signatures for extremophilic adaptation. Gene 1994;140:17-24. 39. Lacher K, Schafer G. Archaebacterial adenylate kinase from the thermoacidophile Sulfolobus acidocaldarius: Purification, characterization, and partial sequence. Arch Biochem Biophys 1993;302:391-397. 40. Kath T, Schmid R, Schafer G. Identification, cloning, and expression of the gene for adenylate kinase from the thermoacidophilic archaebacterium Sulfolobus acidocaldarius. Arch Biochem Biophys 1993;307:405-5 10. 41. Berquist PL, Morgan HW. The molecular genetics and biotechnological application of enzymes from extremely thermophilic eubacteria. In: Herbert RA, Sharp RJ (eds) Molecular Biology and Biotechnology of extremophiles. New York: Chapman and Hall, 1992;U-75. 42. Lawyer FC, Stoffel S , Saiki RK, Myambo K, Drummond R, Gelfand DH. Isolation, characterization, and expression in Escherichia coli of the DNA polymerase gene from Thermus aquaticus. J Biol Chem 1989;264:6427-6437. 43. Davies GJ, Littlechild JA, Watson HC, Hall L. Sequence and expression of the gene encoding 3phosphoglycerate kinase from Bacillus stearothermophilus. Gene 1991;109:3+45. 44. Bowen D, Littlechild JA, Fothergill JE, Watson HC, Hall L. Nucleotide sequence of the phosphoglycerate kinase gene from the extreme thermophile Thermus thermophilus. Biochem J 1988;254:50!+-5 17. 45. Sakai K, Narihara M, Kasama Y, Wakayama M, Moriguchi M. Purification and characterization of thermostable P-N-Acetylhexosaminidaseof Bacillus stearothermophilus CH-4 isolated from chitincontaining compost. Appl Environ Microbiol 1994;60:2911-2915. 46. Hartog AT, Daniel RM. An alkaline phosphatase from Thermus sp strain Rt41A. Int J Biochem 1992;24 1657-1660. 47. Bealin-Kelly F, Kelly CT, Fogarty WM. Studies on the thermostability of the a-amylase of Bacillus caldovelox. Appl Microbiol Biotechnol 1991;36:332-336. 48. Gray GL, Mainzer SE, Rey MW, Lamsa MH, Kindle KL, Carmona C, Requadt C. Structural genes encoding the thermophilic a-amylases of Bacillus stearothermophilus and Bacillus licheniformis. J Bacteriol 1986;166:6355643. 49. Brosnan MP, Kelly CT, Fogarty WM. Investigation of the mechanisms of irreversible thermoinactivation of Bacillus stearothermophilus a-amylase. Eur J Biochem 1992;203:225-23 1. 50. Fukusumi S , Kamizono A, Horinouchi S, Beppu T. Cloning and nucleotide sequence of a heat-stable amylase gene from an anaerobic thermophile, Dictyoglomus thermophilum. Eur J Biochem 1988;174:15-21. 51. Horinouchi S, Fukusumi S, Ohshima T, Beppu T. Cloning and expression in Escherichia coli of two additional amylase genes of a strictly anaerobic thermophile,Dictyoglomus thermophilum, and their nucleotide sequences with extremely low guanine-plus-cytosine contents. Eur J Biochem 1988;176:243-253. 52. Hyun HH, Zeikus JG. General biochemical characterizationof thermostable extracellular P-amylase from Clostridiurn thermosulfirogenes. Appl Environ Microbiol 1985;49:1162-1 167. 53. Kitamoto N, Yamagata H, Kato T, Tsukagoshi N, Udaka S. Cloning and sequencing of the gene encoding thermophilic P-amylase of Clostridium thermosulfurogenes. J Bacteriol 1988;170:58485854. 54. Melasniemi H. Characterization of a-amylase and pullulanase activities of Clostridium rhermohydrosulfuricum. Biochem J 1987;246:193-197. 55. Mathupala SP, Zeikus JG. Improved purification and biochemical characterization of extracellular amylopullulanase from Thermoanaerobacter ethanolicus 39E. Appl Microbiol Biotechnol 1993;39:487-493. 56. Mathupala SP, Lowe SE, Podkovyrov SM, Zeikus JG. Sequencing of the amylopullulanase (apu) gene of Thermoanaerobacter ethanolicus 39E, and identification of the active site by site-directed
70 57. 58. 59. 60. 61. 62. 63.
64.
65. 66. 67,
68.
69.
70.
71. 72.
73.
74. 75. 76. 77.
mutagenesis. J Biol Chem 1993;268:16332-16344. Plant AR, Clemens RM, Morgan HW, Daniel RM. Active-site - and substrate-specificity of Therrnoanaerobiurn Tok6-B1 pullulanase. Biochem J 1987;246:537-541. Saha BC, Lamed R, Lee C-Y, Mathupala SP, Zeikus JG. Characterization of an endo-acting amylopullulanasefrom Thermoanaerobacter strain B6A. Appl Environ Microbiol199056:88 1-886. Madi E, Antranikian G, Ohmiya K, Gottschak G. Thermostable amylolytic enzymes from a new Clostridium isolate. Appl Environ Microbiol 1987;53:1661-1667. Guy GR, Daniel RM. The purification and some properties of a stereospecific D-asparaginase from an extremely thermophilic bacterium, Thermus aquaticus. Biochem J 1982;203:787-790. Schafer G, Meyering-Vos M. The plasma membrane ATPase of archaebacteria. A chimeric energy converter. Ann NY Acad Sci 1992;671:293-309. Saha BC, Zeikus JG. Characterization of thermostable cyclodextrinase from Clostridiurn thermohydrosulfuricum 39E. Appl Environ Microbiol 1990;56:2941-2943. Podkovyrov SM, Zeikus JG. Structure of the gene encoding cyclomaltodextrinasefrom Clostridium therrnohydrosulfuricurn 39E and characterization of the enzyme purified from Escherichia coli. J Bacteriol 1992;174:540(t5405. Wind RD, Liebl W, Buitelaar RM, Penninga D, Spreinat A, Dijkhuizen L, Bahl H. Cyclodextrin formation by the thermostable a-amylase of Thermoanaerobacteriurn therrnosulfurigenes EM1 and reclassification of the enzyme as a cyclodextrin glycosyltransferase. Appl Environ Microbiol 1995;61:1257-1 265. Margaritis A, Merchant RFJ. Thermostablecellulases from thermophilic microorganisms. CRC Crit Rev Biotechnol 1986;4:327-366. Schofield LR, Daniel RM. Purification and properties of a P-1,4-xylanase from a cellulolytic extreme thermophile expressed in Escherichia coli. Int J Biochem 1993;25:60+617. Liithi E, Jasmat NB, Bergquist PL. Xylanase from the extremely thermophilic bacterium “Caldocellum saccharolyticurn”: overexpression of the gene in Escherichia coli and characterization of the gene product. Appl Environ Microbiol 199056:2677-2683. Lee Y-E, Lowe SE, Henrissat B, Zeikus JG. Characterization of the active site and thermostability regions of endoxylanase from Thermoanaerobacterium saccharolyticum. J Bacteriol 1993;17558905898. Lee Y-E, Lowe SE, Zeikus JG. Gene cloning, sequencing, and biochemical characterization of endoxylanase from Thermoanaerobacterium saccharolyticurn B6A-RI. Appl Environ Microbiol l993;59:3 134-3 137. Nakao M, Nakayama T, Harada M, Kakudo A, Ikemoto H, Kobayashi S, Shibano Y. Purification and characterization of a Bacillus sp. SAM 1606 thermostable a-glucosidase with transglucosylation activity. Appl Microbiol Biotechnol 1994;41:337-343. Saha BC, Zeikus JG. Characterization of thermostable a-glucosidase from Clostridium thermohydrosulfuricum 39E. Appl Microbiol Biotechnol 1991;35:568-571. Shao W, Obi SKC, Puls J, Wiegel J. Purification and characterization of the a-glucuronidase from Thermoanaerobacteriurn sp. strain JW/SL-YS485, an important enzyme for the utilization of substituted xylans. Appl Environ Microbiol 1995;61:1077-1081. Lee S-G, Lee D-C, Hong S-P, Sung M-H, Kim H-S. Thermostable D-hydantoinase from thermophilic Bacillus stearotherrnophilus SD-1: characteristicsof purified enzyme. Appl Microbiol Biotechnol 1995;43:27(t276. Gibbs MD, Saul DJ, Liithi E, Bergquist PL. The P-mannanase from “Caldocellum saccharolyticum” is part of a multidomain enzyme. Appl Environ Microbiol 1992;58:3864-3867. Kuriki T, Okada S, Imanaka T. New Type of pullulanase from Bacillus stearotherrnophilus and molecular cloning and expression of the gene in Bacillus subtilis. J Bacteriol 1988;170:1554-1559. Freeman S-A, Peek K, Prescott M, Daniel R. Characterization of a chelator-resistantproteinase from Therrnus strain Rt4A2. Biochem J 1993;295:463-469. Peek K, Daniel RM, Monk C, Parker L, Coolbear T. Purification and characterization of a thermostable proteinase isolated from Thermus sp. strain Rt41A. Eur J Biochem 1992;207:1035-
71 78.
79.
80.
81.
82. 83.
84.
85.
86.
87.
88. p9. 90. 91. 92. 93. 94.
95.
96. 97.
1044. Peek K, Veitch DP, Prescott M, Daniel RM, MacIver B, Bergquist PL. Some characteristics of a proteinase from a thermophilic Bacillus sp. expressed in Escherichia coli: Comparison with the native enzyme and its processing in E. coli and in vitro. Appl Environ Microbiol 1993;59:11681175. Kuriki T, Park J-H, Okada S, Imanaka T. Purification and characterization of thermostable pullulanase from Bacillus stearothermophilus and molecular cloning and expression of the gene in Bacillus subtilis. Appl Environ Microbiol 1988;54:2881-2883. Shen G-J, Srivastava KC, Saha BC, Zeikus JG. Physiological and enzymatic characterization of a novel pullulan-degrading thermophilic Bacillus strain 3 183. Appl Microbiol Biotechnol 1990;33: 340-344. Nashihara N, Nakamura N, Nagayama H, Horikoshi K. Cloning and expression of thermostable pullulanase gene from Thermus sp. strain AMD-33 in Escherichia coli. FEMS Microbiol Lett 1988;49:385-388. Plant AR, Morgan HW, Daniel RM. A highly stable pullulanase from Thermus aquaticus YT-1. Enz Microbiol Techno1 1986;8:668-678. Richter 0-MH, Schafer G. Purification and enzymic characterization of the cytoplasmic pyrophosphatase from the thermoacidophilic archaebacterium Thermoplasma acidophilum. Eur J Biochem 1992;208:343-349. Richter 0-MH, Schafer G. Cloning and sequencing of the gene for the cytoplasmic inorganic pyrophosphatase from the thermoacidophilic archaebacterium Thermoplasma acidophilum. Eur J Biochem 1992;209:351-355. Shao W, DeBlois S, Wiegel J. A high-molecular-weight, cell-associated xylanase isolated from exponentially growing Thermoanaerobacterium sp. strain JW/SL-YS485. Appl Environ Microbiol 1995;61:937-940. Lee Y-E, Zeikus JG. Genetic organization, sequence and biochemical characterization 01 recombinant P-xylosidase from Thermoanaerobacrerium saccharolyticurn strain B6A-RI. J Gen Microbiol 1993;139:1235-1243. Luthi E, Bergquist PL. A P-D-xylosidase from the thermophile Caldocellurn saccharolyticum expressed in Escherichia coli. FEMS Microbiol Lett 1990;67:29 1-294. Barnes EM, Akagi JM, Himes RH. Properties of fructose-l,6-diphosphatealdolase from two thermophilic and a mesophilic clostridia. Biochim Biophys Acta 1971;227:199-203. Russell RJM, Hough DW, Danson MJ, Taylor GL. The crystal structure of citrate synthase from the thermophilic Archaeon, Thermoplasma acidophilum. Structure 1994;2:1157-1 167. Kenealy WR, Zeikus, JG. Characterization and function of phosphoenolpyruvate carboxylase in Methanobacterium thermoautotrophicum. FEMS Microbiol Lett 1982;14:7- 10. Shing YW, Akagi JM, Himes RH. Psychrophilic, mesophilic, and thermophilic triosephosphate isomerases from three clostridial species. J Bacteriol 1975;122:177-184. Tchernajenko V, Vieille, C, Burdette, DS, Zeikus JG. Physicobiochemical studies of xylose isomerases from thermophilic and mesophilic microorganisms. Submitted for publication. Lee C, Zeikus JG. Purification and characterization of thermostable glucose isomerase from Clostridium thermosulfurogenes and Thermoanuerobacter strain B6A. Biochem J 1991;273:565-57 1. Lee Y-E, Ramesh MV, Zeikus JG. Cloning, sequencing, and biochemical characterization of xylose isomerase from Thermoanaerobacterium saccharolyticum strain B6A-RI. J Gen Microbiol 1993;139:1227-1234. Lee C, Bhatnagar L, Saha BC, Lee Y-E, Takagi M, Imanaka T, Bagdasarian M, Zeikus JG. Cloning and expression of the Clostridium therrnosulfurogenes glucose isomerase gene in Escherichia coli and Bacillus subtilis. Appl Environ Microbiol 199056:2638-2643. Lehmacher A, Bisswanger H. Isolation and characterization of an extremely thermostable D-XylOSe isomerase from Thermus aquaticus HB 8. J Gen Microbiol 199@136:679-686. Dekker K, Sugiura A, Yamagata H, Sakaguchi K, Udaka S. Efficient production of thermostable Thermus thermophilus xylose isomerase in Escherichia coli and Bacillus brevis. Appl Microbiol
72 Biotechnol 1992;36:727-732. 98. Dekker K, Yamagata H, Sakaguchi K, Udaka S. Xylose (glucose) isomerase gene from the thermophile Thermus thermophilus: Cloning, sequencing, and comparison with other thermostable xylose isomerases. J Bacteriol 1991;173:3078-3083. 99. Rainey FA, Donnison AM, Janssen PH, Saul D, Rodrigo A, Bergquist PL, Daniel RM, Stackebrandt E, Morgan HW. Description of Caldicellulosiruptor saccharolyticus gen. nov, sp. nov.: An obligately anaerobic extremely thermophilic, cellulolyticbacterium. FEMS Microbiol Lett 1994;120:263-266. 100. Rella R, Raia CA, Pensa M, Pisani FM, Gambacorta A, De Rosa M, Rossi M. A novel archaebacterial NAD'-dependent alcohol dehydrogenase. Eur J Biochem 1987;107:475-479. 101. Ma K, Robb FT, Adams MWW. Purification and characterization of NADP-specific alcohol dehydrogenase and glutamate dehydrogenase from the hyperthermophilic archaeon Thermococcus liroralis. Appl Environ Microbiol 1994;60:562-568. 102. Mukund S, Adams MWW. Characterization of a tungsten-iron-sulfur protein exhibiting novel spectroscopicand redox propertiesfrom the hyperthermophilic archaebacterium Pyrococcusfuriosus. J Biol Chem 1990265:11508-11516. 103. Chan MK, Mukund S, Kletzin A, Adams MWW, Rees DC. Structure of a hyperthermophilic tungstopterin enzyme, aldehyde ferredoxin oxidoreductase. Science 1995;267:1463- 1469. 104. Aono S, Bryant FO, Adams MWW. A novel and remarkably thermostable ferredoxin from the hyperthermophilic archaebacterium Pyrococcus furiosus. J Bacteriol 1989;171:3433-3439. 105. DiRuggiero J, Robb FT, Jagus R, Klump HH, Borges KM, Kessel M, Mai X, Adams MWW. Characterization, cloning, and in vitro expression of the extremely thermostable glutamate dehydrogenase from the hyperthermophilic archaeon, ES4. J Biol Chem 1993;268:17767-17774. 106. Consalvi V, Chiaraluce R, Politi L, Vaccaro R, De Rosa M, Scandurra R. Extemely thermostable glutamate dehydrogenase from the hyperthermophilic archaebacterium Pyrococcus furiosus. Eur J Biochem 1991;202:1189-1196. 107. Robb FT, Park J-B, Adams MWW. Characterization of an extremely thermostable glutamate dehydrogenase: a key enzyme in a primary metabolism of the hyperthermophilic archaebacterium, Pyrococcus furiosus. Biochim Biophys Acta 1992;1 120:267-272. 108. Britton KL, Baker PJ, Borges KMM, Engel PC, Pasquo A, Rice DW, Robb FT,Scandurra R, Stillman TJ, Yip KSP. Insights into thermal stability from a comparison of the glutamate dehydrogenases from Pyrococcus furiosus and Thermococcus litoralis. Eur J Biochem 1995;229: 688-695. 109. Consalvi V, Chiaraluce R, Politi L, Gambacorta A, De Rosa M, Scandurra R. Glutamate dehydrogenase from the thermoacidophilic archaebacterium Sulfolohus solfataricus. Eur J Biochem 1991;196:459-467. 1 10. Wrba A, Schweiger A, Schultes V, Jaenicke R. Extremely thermostable ~-glyceraldehyde-3phosphate dehydrogenasefrom the eubacterium Thermotoga maritima. Biochemistry 1990;29:75847592. 111. Tomschy A, Glockshuber R, Jaenicke R. Functional expression of ~-glyceraldehyde-3-phosphate dehydrogenase from the hyperthermophilic eubacterium Thermotoga maritima in Escherichia coli. Authenticity and kinetic properties of the recombinant enzyme. Eur J Biochem 1993;21443-50. 112. Schultes V, Deutzmann R, Jaenicke R. Complete amino acid sequence of glyceraldehyde-3phosphate dehydrogenase from the hyperthermophilic eubacterium Thermotoga maritima. Eur J Biochem 1990;192:25-31. 113. Zwickl P, Fabry S, Bogedain C, Haas A, Hensel R. Glyceraldehyde-3-phosphatedehydrogenase from the hyperthermophilic archaebacterium Pyrococcus woesei: Characterization of the enzyme, cloning, and sequencing of the gene, and expression in Escherichia coli. J Bacteriol 1990 172343294338. 114. Hensel R, Jakob I, Scheer H, Lottspeich F. Proteins from hyperthermophilic archaea: stability towards covalent modification of the peptide chain. Biochem SOCSymp 1992;58:127-133. 115. Hensel R, Laumann S, Lang J, Heumann H, Lottspeich F. Characterizationof two D-glyceraldehyde3-phosphatedehydrogenasesfrom the extremely thermophilic archaebacterium Thermoproreus tenax.
73 Eur J Biochem 1987;170:325-333. 116. Shah NN, Clark DS. Partial purification and characterizationof two hydrogenases from the extreme thermophile Merhanococcus jannaschii. Appl Environ Microbiol 1990;56:858-863. 117. Bryant FO, Adams MWW. Characterization of hydrogenase from the hyperthermophilic archaebacterium,Pyrococcus furiosus. J Biol Chem 1989;264:5070-5079. 118. Pihl TD, Maier RJ. Purification and characterization of the hydrogen uptake hydrogenase from the hyperthermophilic archaebacterium Pyrodicrium brockii. J Bacteriol 1991;173:1839-1844. 119. Juszczak A, Aono S, Adams MWW. The extremely thermophilic eubacterium, Thermofoga maritima, contains a novel iron-hydrogenase whose cellular activity is dependent upon tungsten. J Biol Chem 1991;266:13834-13841. 120. Wrba A, Jaenicke R, Huber R, Stetter KO. Lactate dehydrogenase from the extreme thermophile Thermotoga maritima. Eur J Biochem 1990; 188:195-201. 121. Ostendorp R, Liebl W, Schurig H, Jaenicke R. The L-lactate dehydrogenase gene of the hyperthermophilic bacterium Thermotoga maririma cloned by complementation in Escherichia coli. Eur J Biochem 1993;216:709-715. 122. Honka E, Fabry S, Niermann T, Palm P, Hensel R. Properties and primary structure of the L-malate dehydrogenase from the extremely thermophilic archaebacterium Mefhanorhermus fervidus. Eur J Biochem 1990; 188:623-632. 123. Klein A, Koch J, Stetter KO, Thauer RF. Two ~JV1O-methylenetetrahydromethanopterin dehydrogenases in the extreme thermophile Methanopyrus kandleri: characterization of the coenzyme F,,-dependent enzyme. Arch Microbiol 1993;160:186-192. 124. Ma K, Zirngibl C, Linder D, Stetter KO, Thauer RK. N5,N10-Methylenetetrahydromethanopterin dehydrogenase (H,-forming) from the extreme thermophile Merhunopyrus kandleri. Arch Microbiol 1991;156:43-48. 125. Ma K, Linder D, Stetter KO, Thauer RK. Purification and properties of N5,”o-methylenetetrahydromethanopterin reductase (coenzyme F4,-dependent) from the extreme thermophile Methanopyrus kundleri. Arch Microbiol 1991;155:593-600. 126. Kunow J, Schworer B, Stetter KO, Thauer RK. An F,,,-dependent NADP reductase in the extremely thermophilic sulfate-reducing Archaeoglobus fulgidus. Arch Microbiol 1993;160:19+205. 127. Kunow J, Linder D, Thauer RK. Pyruvate: ferredoxin oxidoreductase from the sulfate-reducing Archaeoglobus fulgidus: molecular composition,catalytic properties, and sequence alignments. Arch Microbiol 1995;163:21-28. 128. Blamey JM, Adam MWW. Purification and characterization of pyruvate ferredoxin oxidoreductase from the hyperthermophilic archaeon Pyrococcusfuriosus. Biochim Biophys Acta 1993;1161:19-27. 129. Blarney JM, Adams MWW. Characterization of an ancestral type of pyruvate ferredoxin oxidoreductase from the hyperthermophilic bacterium Thermoroga maririma. Biochemistry 1994;33: 11007. 130. Blake PR, Park J-B, Bryant FO, Aono S, Magnuson JK, Eccleston E, Howard JB, Summers MF, Adams MWW. Determinants of protein hyperthermostability: purification and amino acid sequence of rubredoxin from the hyperthermophilic Archaebacterium Pyrococcus furiosus and secondary structure of the zinc adduct by NMR. Biochemistry 1991;30:10885-10895. 131. Dahl C, Koch H-G, Keuken 0, Triiper HG. Purification and characterization of ATP sulfurylase from the extremely thermophilic archaebacterial sulfate-reducer, Archaeoglobus fulgidus. FEMS Microbiol Lett 1990;67:27-32. 132. Perler FB, Comb DG, Jack WE, Moran LS, Qiang B, Kucera RB, Benner J, Slatko BE, Nwankwo DO, Hempstead SK, Carlow CKS, Jannasch H. Intervening sequences in an Archaea DNA polymerase gene. Proc Natl Acad Sci USA 1992;89:5577-5581. 133. Uemori T, Ishino Y, Toh H, Asada K, Kato I. Organization and nucleotide sequence of the DNA polymerase gene from the archaeon Pyrococcus furiosus. Nucleic Acids Res 1993;21:259-265. 134. Breitung J, Borner G, Scholz S, Linder D, Stetter KO, Thauer RK. Salt dependence, kinetic properties and catalytic mechanism of N-formylmethanofuran: tetrahydromethanopterin formyltransferase from the extreme thermophile Methanopyrus kandleri. Eur J Biochem 1992;210971-
74 981. 135. Lehmacher A, Hensel R. Cloning, sequencing and expression of the gene encoding 2-phosphoglycerate kinase from Merhanothermus fervidus. Molec Gen Genet 1994;242: 163-1 68. 136. Koch R, Zablowski P, Spreinat A, Antranikian G. Extremely thermostable amylolytic enzyme from the archaebacterium Pyrococcus furiosus. FEMS Microbiol Lett 1990;71:21-26. 137. Laderman KA, Davis BR, Krutzsch HC, Lewis MS, Griko YV, Privalov PL, Anfinsen CB. The purification and characterization of an extremely thermostable a-amylase from the hyperthermophilic archaebacterium Pyrococcus furiosus. J Biol Chem 1993;268:24394-24401. 138. Laderman KA, Asada K, Uemori T, Mukai H, Taguchi Y, Kato I, Anfinsen CB. a-Amylase from the hyperthermophilic archaebacterium Pyrococcus furiosus. Cloning and sequencing of the gene and expression in Escherichiu coli. J Biol Chem 1993;268:24402-24407. 139. Koch R, Spreinat A, Lemke K, Antranikian G. Purification and properties of a hyperthermoactive a-amylase from the archaeobacterium Pyrococcus woesei. Arch Microbiol 1991;155572-578. 140. Chung YC, Kobayashi T, Kanai H, Akiba T, Kudo T. Purification and properties of extracellular amylase from the hyperthermophilic archaeon Therrnococcus profundus DT5432. Appl Environ Microbiol 199561:1502-1506. 141. Brown SH, Kelly RM. Characterization of amylolytic enzymes, having both a-1,4 and a-1,6 hydrolytic activity, from the thermophilic archaea Pyrococcusfuriosus and Thermococcus lirorulis. Appl Environ Microbiol 1993;59:2614-2621. 142. Schuliger JW, Brown SH, Baross JA, Kelly RM. Purification and characterization of a novel amylolytic enzyme from ES4, a marine hyperthermophilic archaeum. Molec Marine Biol Biotechnol 1993;2:76-87. 143. Phipps BM, Hoffman A, Stetter KO, Baumeister W. A novel ATPase complex selectively accumulated upon heat shock is a major cellular component of thermophilic archaebacteria. EMBO J 1991;10:1711-1722. 144. Villa A, Zecca L, Fusi P, Columbo S, Tedeschi G, Tortora P. Structural features responsible for kinetic thermal stability of a carboxypeptidase from the archaebacterium Sulfolobus solfataricus. Biochem J 1993;295:827-831. 145. Colombo S, D'Auria S, Fusi P, Zecca L, Raia CA, Tortora P. Purification and characterization of a thermostable carboxypeptidase from the extreme thermophilic archaebacterium Sulfolobus solfataricus. Eur J Biochem 1992;206:349-357. 146. Bronnenmeier K, Kern A, Liebl W, Staudenbauer WL. Purification of Thermoroga maritimu enzymes for the degradation of cellulosic materials. Appl Environ Microbiol 1995;61:1 3 9 F1407. 147. Simpson HD, Haufler UR, Daniel RM. An extremely thermostable xylanase from the thermophilic eubacterium Thermotoga. Biochem J 1991;277:4 13-4 17. 148. Ruttersmith LD, Daniel RM. Thermostable cellobiohydrolase from the thermophilic eubacterium Thermotoga sp. strain FjSS3-B.I. Biochem J 1991;277:887-890. 149. Ruttersmith LD, Daniel RM. Thermostable P-glucosidase and P-xylosidase from Thermotoga sp. strain FjSS3-B.l. Biochim Biophys Acta 1993;1156:167-172. 150. Costantino HR, Brown SH, Kelly RM. Purification and characterization of an a-glucosidase from a hyperthermophilic archaebacterium, Pyrococcusfuriosus, exhibiting a temperature optimum of 105 to 115°C. J Bacteriol 1990,172:3654-3660. 15 1. Klein AR, Breitung J, Linder D, Stetter KO, Thauer RK. Ns,N1O-methenyltetrahydromethanopterin cyclohydrolase from the extremely thermophilic sulfate reducing Archaeoglobus fulgidus: comparison of its properties with those of the cyclohydrolase from the extremely thermophilic Merhanopyrus kandleri. Arch Microbiol 1993;159:213-219. 152. Breitung J, Schmitz RA, Stetter KO, Thauer RK. NS,N"-methenyltetrahydromethanopterin cyclohydrolase from the extreme thermophile Methanopyrus kandleri: increase of catalytic efficiency (kca,/KM) and thermostability in the presence of salts. Arch Microbiol 1991;156:517-524. 153. Eggen R, Geerling A, Watts J, deVos WM. Characterization of pyrolysin, a hyperthermoactive serine protease from the archaebacterium Pyrococcusfuriosus. FEMS Microbiol Lett 19907I :17-20. 154. Morikawa M, Izawa Y, Rashid N, Hoaki T, Imanaka T. Purification and characterization of a
75 thermostable thiol protease from a newly isolated hyperthermophilic Pyrococcus sp. Appl Environ Microbiol 1994;60:455%4566. 155. Ruttersmith LD, Daniel RM, Simpson HD. Cellulolytic and hemicellulolytic enzymes functional above 100°C. Ann NY Acad Sci 1992;672:137-141. 156. Slesarev AI, Lake JA, Stetter KO, Gellert M, Kozyavkin SA. Purification and characterization of DNA topoisomerase V. J Biol Chem 1994;269:3295-3303. 157. Robb FT,Masuchi Y, Park J-B, Adams MWW. Key enzymes in the primary nitrogen metabolism of a hyperthermophile. In: Adams MWW, Kelly RM (eds) Biocatalysis at Extreme Temperatures. Washington, DC: American Chemical Society, 1992;74-85. 158. Tiboni 0,Cammarano P, Sanangelantoni AM. Cloning and sequencing of the gene encoding glutamine synthetase I from the archaeum Pyrococcus woesei: Anomalous phylogenies inferred from analysis of archaeal and bacterial glutamine synthetase I sequences. J Bacteriol 1993;175:2961-
2969. 159. Brown SH, Sjaholm C, Kelly RM. Purification and characterization of a highly thermostable glucose isomerase produced by the extremely thermophilic eubacterium, Thermotoga maritima. Biotechnol Bioeng 1993;41:878-886. 160. Vieille C, Hess JM, Kelly RM, Zeikus JG. xylA cloning and sequencing and biochemical characterization of xylose isomerase from Thermotoga neapolitana. Appl Environ Microbiol
1995;61:1867-1 875. 161. Aubert J-P, BCguin P, Millet J. Genes and proteins involved in cellulose and xylan degradation by Clostridium thermocellurn. In: Sebald M (ed) Genetics and Molecular Biology of Anaerobic Bacteria. New York: Springer-Verlag, 1993:412-422. 162. Hecht K, Wrba A, Jaenicke R. Catalytic properties of thermophilic lactate dehydrogenase and halophilic malate dehydrogenase at high temperature and low water activity. Eur J Biochem
1989;183:6%74. 163. Taguchi H, Yamashita M, Matsuzawa H, Ohta T. Heat-stable and fructose 1,6-bisphosphate-activated L-lactate dehydrogenase from an extremely thermophilic bacterium. J Biochem 1982;91:1343-1348. 164. Sundaram TK. Physiology and growth of thermophilic bacteria. In: Brock T (ed) Thermophiles: General, Molecular, and Applied Microbiology. New York: Wiley and Sons, 1986;75-100. 165. Sakamoto N, Kotre AM, Savageau MA. Glutamate dehydrogenase from Escherichia coli: Purification and properties. J Bacteriol 1975;124:775-783. 166. Coulton JW,Kapoor M. Studies on the kinetics and regulation of glutamate dehydrogenase of Salmonella typhimurium. Can J Microbiol 1973;19:43%450. 167. Hooper AB, Hansen J, Bell R. Characterization of glutamate dehydrogenase from the ammoniaoxidizing chemoautotroph Nitrosomonas europaea. J Biol Chem 1967;242:288-296. 168. =John HB, Suzuki I, Wright JA. Glutamate dehydrogenases of Thiobacillus novellus. J Biol Chem 1968;243:1 18- 128. 169. Hudson RC, Ruttersmith LD, Daniel RM. Glutamate dehydrogenase from the extremely thermophilic archaebacterial isolate ANl. Biochim Biophys Acta 1993;1202:244-250. 170. Batt CA, Jamieson AC, Vandeyar MA. Identification of essential histidine residues in the active site of Escherichia coli xylose (glucose) isomerase. Proc Natl Acad Sci USA 199087:618-622. 171. Smith CA, Rangarajan M, Hartley BS. D-xylose (D-glucose) isomerase from Arthrobacter strain N.R.R.L. B3728. Biochem J 1991;277:255-261. 172. Glaser P, Presecan E, Delepierre M, Surewicz WK, Mantsch HH, BLrzu 0, Gilles A-M. Zinc, a novel structure element found in the family of bacterial adenylate kinases. Biochemistry 1992;31 :3038-3043. 173. Baba T,Shinke R, Nanmori T. Identification and characterization of clustered genes for thermostable xylan-degrading enzymes, P-xylosidase and xylanase, of Bacillus stearothermophilus 21. Appl Environ Microbiol 1994;602252-2258. 174. BCguin P, Aubert J-P. The biological degradation of cellulose. E M S Microbiol Rev 1994;13:25-58. 175. Koyama Y, Furukawa K. Cloning and sequence analysis of tryptophan synthase genes of an extreme thermophile, Thermus thermophilus HB27: Plasmid transfer from replica-plated Escherichia coli
76 recombinant colonies to competent T. thermophilus cells. J Bacteriol 1990;172:3490-3495. 176. DiMarco AA,Sment KA, Konisky J, Wolfe RS. The formylmethanofuran: tetrahydromethanopterin formyltransferase from Methanobacterium thermoautotrophicum 6H. J Biol Chem 199@265:472476. 177. Xu M-Q, Southworth MW, Mersha FB, Homstra LJ, Perler FB. In vitro protein splicing of purified precursor and the identification of a branched intermediate. Cell 1993;75:1371-1377. 178. Holden JF, Baross JA. Enhanced thermotolerance and temperature-induced changes in protein composition in the hyperthermophilic archaeon ES4. J Bacteriol 1993;175:2839-2843. 179. Roberts DL, James-Hagstrom JE, Garvin DK, Gorst CM, Runquist JA, Baur JR,Haase FC, Ragsdale SW. Cloning and expression of the gene cluster encoding key proteins involved in acetyl-CoA synthesis in Clostridium thermoaceticum: CO dehydrogenase, the comnoid/Fe-S protein, and methyltransferase. Proc Natl Acad Sci USA 1989;86:32-36. 180. MenCndez-Arias L, Argos P. Engineering protein thermal stability: Sequence statistics point to residue substitutions in a-helices. J Molec Biol 1989;206:397-406. 181. Watanabe K, Chishiro K, Kitamura K, Suzuki Y. Proline residues responsible for thermostability occur with high frequency in the loop regions of an extremely thermostable oligo-l,6-glucosidase from Bacillus thermoglucosidasius KP1006.J Biol Chem 1991;266:2428?-24294. 182. Love11 CR, Przybyla A, Ljungdahl LG. Primary structure of the thermostable formyltetrahydrofolate synthetase from Clostridium thermoaceticum. Biochemistry 1990;29:5687-5694. 183. Bohm G, Jaenicke R. Relevance of sequence statistics for the properties of extremophilic proteins. Int J Pept Protein Res 1994;43:97-106. 184. Moll R, Schafer G. Chemiosmotic H' cycling across the plasma membrane of the thermoacidophilic archaebacterium Suljolobus acidocaldarius. FEBS Lett 1988;232:359-363. 185. Volkin DB, Middaugh CR. The effect of temperature on protein structure. In: Ahem TJ, Manning MC (eds) Stability of Protein Pharmaceuticals,Part A: Chemical and Physical Pathways of Protein Degradation. New York: Plenum Press, 1992;215-247. 186. Fontana A. How nature engineers protein (thermo) stability. In: di Prisco G (ed) Life Under Extreme Conditions: Biochemical Adaptation. Heidelberg: Springer-Verlag, 1990;89-I 13. 187. Alonso DOV, Dill KA, Stigter D. The three states of globular proteins: acid denaturation. Biopolymers 1991;31: 1631-1649. 188. Haynie DT, Freire E. Structural energetics of the molten globule state. Proteins Struct Funct Genet
1993;16:115-140. 189. Tomazic SJ, Klibanov AM. Mechanisms of irreversible thermal inactivation of Bacillus a-amylases. J Biol Chem 1988;263:308&3091. 190. Meldgaard M, Svendsen I. Different effects of N-glycosylation on the thermostability of highly homologous bacterial (1,3-1,4)-P-glucanases secreted from yeast. Microbiol 1994;140:159-166. 191. Olsen 0,Thomsen KK. Improvement of bacterial P-glucanase thermostability by glycosylation. J Gen Microbiol 1991;137:579-585. 192. Mirsky AE, Pauling L. On the structure of native, denatured, and coagulated proteins. Proc Natl Acad Sci USA 1936;22:439-447. 193. Pfeil W. Unfolding of proteins. In: Hinz H-J (ed) Thermodynamic Data for Biochemistry and Biotechnology. Berlin: Springer-Verlag. 1986;349-376. 194. Jaenicke R. Protein stability and molecular adaptation to extreme conditions. Eur J Biochem 1991;202:715-728. 195. Ahem TJ, Klibanov AM. Analysis of processes causing thermal inactivation of enzymes. Meth Biochem Anal 1988;33:91-127. 196. Fabry S,Hensel R. Purificationand characterizationof ~-glyceraldehyde-3-phosphate dehydrogenase from the thermophilic archaebacteriumMethanothermusfervidus. Eur J Biochem 1987;165:147-155. 197. Londesborough J. The causes of sharply bent or discontinuousArrhenius plots for enzyme-catalysed reactions. Eur J Biochem 1980;105:211-215. 198. Kauzmann W. Some factors in the interpretation of protein denaturation. Adv Protein Chem 1959;14:1-47.
77 199. Freire E, Haynie DT, Xie D. Molecular basis of cooperativity in protein folding. IV. CORE: a general cooperative folding model. Proteins Struct Funct Genet 1993;17:111-123. 200. Haynie DT, Freire E. Thermodynamic strategies for stabilizing intermediate states of proteins. Biopolymers 1994;34:261-271. 201. Privalov PL, Gill SJ. Stability of protein structure and hydrophobic interaction. Adv Protein Chem 1988;39:191-234. 202. Dill KA. Dominant forces in protein folding. Biochemistry 1990;29:7133-7155. 203. Dill KA, Alonso DOV, Hutchinson K. Thermal stabilities of globular proteins. Biochemistry 1989;28:543+5449. 204. Creighton TE. Protein Folding. Biochem J 1990;270:1-16. 205. Creighton TE. Proteins in solution and in membranes: stability of the folded conformation. In: Creighton TE (ed) Proteins: Structures and Molecular Properties, 2nd edn. New York W.H. Freeman and Company, 1993;287-309. 206. Tsuboi M, Ohta S , Nakanishi M. Structural fluctuation of protein and thermophily. In: Friedman M (ed) Biochemistry of Thermophily. New York: Academic Press, 1978;251-266. 207. Wutrich K, Roder H, Wagner G. Internal mobility and unfolding of globular proteins. In: Jaenicke R (ed) Protein Folding. Amsterdam: Elsevier/North Holland Biomedical Press, 1980549-564. 208. Argos P, Rossmann MG, Grau UM, Zuber H, Frank G, Tratschin JD. Thermal stability and protein structure. Biochemistry 1979;18:5698-5703. 209. Matthews BW. Structural and genetic analysis of protein stability. Ann Rev Biochem 1993;62:13+ 160. 210. Zhang X-J, Baase WA, Matthews BW. Multiple alanine replacements within a-helix 126-134 of T4 lysozyme have independent, additive effects on both structure and stability. Protein Sci 1992;1:761776. 21 1. Herning T, Yutani K, Inaka K, Kuroki R, Matsushima M, Kikuchi M. Role of proline residues in human lysozyme stability: A scanning calorimetric study combined with X-ray structure analysis of proline mutants. Biochemistry 1992;31:7077-7085. 212. Hurley JH, Baase WA, Matthews BW. Design and structural analysis of alternative hydrophobic core packing arrangements in bacteriophage T4 lysozyme. J Molec Biol 1992;224:1143-1159. 21 3. Watanabe K, Masuda T, Ohashi H, Mihara H, Suzuki Y. Multiple proline substitutions cumulatively thermostabilize Bacillus cereus ATCC7064 oligo-l,6-glucosidase. Eur J Biochem 1994;226:277-283. 214. Eriksson AE, Baase WA, Zhang X-J, Heinz DW, Blaber M, Baldwin EP, Matthews BW. Response of a protein structure to cavity-creating mutations and its relation to the hydrophobic effect. Science 1992;255:178-183. 215. Biro J, Fabry S, Dietmaier W, Bogedain C, Hensel R. Engineering thermostability in archaebacterial glyceraldehyde-3-phosphatedehydrogenase. FEBS Lett 1990275:1 3 s134. 216. Koizumi J-I, Zhang M, Imanaka T, Aiba S. Does single-amino-acid replacement work in favor of or against improvement of the thermostability of immobilized enzyme? Appl Environ Microbiol 199056:3612-36 14. 217. Matsumura M, Yasumura S , Aiba S. Cumulative effect of intragenic amino acid feplacements on the thermostability of a protein. Nature 1986;323:356358. 218. Kuroki R, Inaka K, Taniyama Y, Kidokoro S-I, Matsushima M, Kikuchi M, Yutani K. Enthalpic destabilization of a mutant human lysozyme lacking a disulfide bridge between cysteine-77 and cysteine-95. Biochemistry 1992;31 :8323-8328. 219. Herning T, Yutani K, Taniyama Y, Kikuchi M. Effects of proline mutations on the unfolding and refolding of human lysozyme. The slow refolding kinetic phase does not result from proline cistrans isomerization. Biochemistry 1991;30:9882-9891. 220. Bell JA, Becktel WJ, Sauer U, Baase WA, Matthews BW. Dissection of helix capping in T4 lysozyme by structural and thermodynamic analysis of six amino acid substitutions at Thr 59. Biochemistry 1992;31:3590-3596. 221. Pjura P, Matthews BW. Structures of randomly generated mutants of T4 lysozyme show that protein stability can be enhanced by relaxation of strain and by improved hydrogen bonding via bound
78 solvent. Protein Sci 1993;2:22262232. 222. Matsumura M, Signor G, Matthews BW. Substantial increase of protein stability by multiple disulphide bonds. Nature 1989;342:291-293. 223. Eriksson AE, Baase WA, Matthews BW. Similar hydrophobic replacements of Leu99 and Phe153 within the core of T4 lysozyme have different structural and thermodynamic consequences. J Molec Biol 1993;229:747-769. 224. Matthews BW, Nicholson H, Becktel W. Enhanced protein thermostability from site-directed mutations that decrease the entropy of unfolding. Proc Natl Acad Sci USA 1987;84:6663-6667. 225. Pjura P, McIntosh LP, Wozniak JA, Matthews BW. Perturbation of Trp 138 in T4 lysozyme by mutations at Gln 105 used to correlate changes in structure, stability, solvation, and spectroscopic properties. Protein Struct Funct Genet 1993;15:401-412. 226. Kuroki R, Taniyama Y,Seko C, Nakamura H, Kikuchi M, Ikehara M. Design and creation of a Ca2' binding site in human lysozyme to enhance structural stability. Proc Natl Acad Sci USA 1989;86:6903-6907. 227. Blaber M, Lindstrom JD, Gassner N, Xu J, Heinz DW, Matthews BW. Energetic cost and structural consequences of burying a hydroxyl group within the core of a protein determined from Ala --f Ser and Val + Thr substitutions in T4 lysozyme. Biochemistry 1993;32:11363-11373. 228. Anderson DE, Hurley JH, Nicholson H, Baase WA, Matthews BW. Hydrophobic core repacking and aromatic-aromatic interaction in the thermostable mutant of T4 lysozyme Ser 117 + Phe. Protein Sci 1993;2:1285-1290. 229. Eriksson AE, Baase WA, Wozniak JA, Matthews BW. A cavity-containing mutant of T4 lysozyme is stabilized by buried benzene. Nature 1992;355:371-373. 230. Hardy F, Vriend G, Veltman OR, van der Vinne B, Venema G, Eijsink VGH. Stabilization of Bacillus stearothermophilusneutral protease by introduction of prolines. FEBS Lett 1993;317:8%92. 231. Vriend G, Berendsen HJC, van der Zee JR, van den Burg B, Venema G, Eijsink VGH. Stabilization of the neutral protease of Bacillus stearothermophilusby removal of buried water molecule. Protein Eng I99 1;4:941-945. 232. Van den Burg B, Dijkstra BW, Vriend G, van der Vinne B, Venema G, Eijsink VGH. Protein stabilization by hydrophobic interactions at the surface. Eur J Biochem 1994;220:981-985. 233. Ishikawa K, Nakamura H, Morikawa K, Kanaya S. Stabilization of Escherichia coli Ribonuclease HI by cavity-filling mutations within a hydrophobic core. Biochemistry 1993;32:6171-6178. 234. Kimura S, Nakamura H, Hashimoto T, Oobatake M, Kanaya S. Stabilization of Escherichia coli ribonuclease HI by strategic replacement of amino acid residues with those from the thermophilic counterpart. J Biol Chem 1992;267:21535-21542. 235. Kimura S, Kanaya S, Nakamura H. Thermostabilization of Escherichiu coli ribonuclease HI by replacing left-handed helical Lys95with Gly or Asn. J Biol Chem 1992;267:22014-22017. 236. Fabian H, Schultz C, Backmann J, Hahn U, Saenger W, Mantsch HH, Naumann D. Impact of point mutations on the structure and thermal stability of ribonuclease TI in aqueous solution probed by Fourier transform infrared spectroscopy. Biochemistry 1994;33:10725-10730. 237. Mitchinson C, Wells JA. Protein engineering of disulfide bonds in subtilisin BPN'. Biochemistry 1989;28:4807-48 15. 238. Wells JA, Powers DB. In v.ivo formation and stability of engineered disulfide bonds in subtilisin. J Biol Chem 1986;261:6564-6570. 239. Takagi H, Takahashi T, Momose H, Inouye M, Maeda Y,Matsuzawa H, Ohta T. Enhancement of the thermostability of subtilisin E by introduction of a disulfide bond engineered on the basis of structural comparison with a thermophilic serine protease. J Biol Chem 1990;265:6874-6878. 240. Quax WJ, Mrabet NT, Luiten RGM, Schuurhuizen PW, Stanssens P, Lasters I. Enhancing the thermostability of glucose isomerase by protein engineering. Biotechnology 1991;9:738-742. 241. Mrabet NT, Van den Broeck A, Van den brande I, Stanssens P, Laroche Y, Lambeir A-M, Matthijssens G, Jenkins J, Chiadmi M, van Tilbeurgh H, Rey F, Janin J, Q u a WJ, Lasters I, De Maeyer M, Wodak SJ. Arginine residues as stabilizing elements in proteins. Biochemistry 1992;31:223%2253.
79 242. Meng M, Bagdasarian M, Zeikus JG. Thermal stabilization of xylose isomerase from Thermoanaerobacterium thermosulfurigenes. Biotechnology 1993;11:1 157- 1161. 243. Volkin DB, Klibanov AM. Thermal destruction processes in proteins involving cystine residues. J Biol Chem 1987;262:2945-2950. 244. Alber T, Dao-pin S, Nye JA, Muchrnore DC, Matthews BW. Temperature-sensitive mutations of bacteriophage T4 lysozyme occur at sites with low mobility and low solvent accessibility in the folded protein. Biochemistry 1987;26:3754-3758. 245. MacArthur MW, Thomton JM. Influence of proline residues on protein conformation. J Molec Biol 1991;218:397-412. 246. Dixon MM, Nicholson H, Shewchuk L, Baase WA, Matthews BW. Structure of a hinge-bending bacteriophage T4 lysozyme mutant, Ile3 + Pro. J Molec Biol 1992;227:917-933. 247. Perutz MF, Kendrew JC, Watson HC. Structure and function of haemoglobin. J Molec Biol 1965;13:66!&678. 248. El Hawrani AS, Moreton KM, Sessions RB, Clarke AR, Holbrook JJ. Engineering surface loops of proteins - a preferred strategy for obtaining new enzyme function. Trends Biotechnol 1994;12:207211. 249. Creighton TE. Proteins: Structures and Molecular Properties, 2nd edn. New York: W.H. Freeman and Company, 1993. 250. Davies GJ, Gamblin SJ, Littlechild JA, Watson HC. The structure of a thermally stable 3phosphoglycerate kinase and a comparison with its mesophilic equivalent. Proteins Struct Funct Genet 1993;15:283289. 251. Ishikawa K, Okumura M, Katayanagi K, Kimura S, Kanaya S, Nakamura H, Morikawa K. Crystal structure of ribonuclease H from Thermus thermophilus HB8 refined at 2.8.A resolution. J Molec Biol 1993;230:529-542. 252. Tomazic SJ, Klibanov AM. Why is one Bacillus a-amylase more resistant against irreversible thermoinactivation than another? J Biol Chem 1988;263:3092-3096. 253. Baker EN, Hubbard RE. Hydrogen bonding in globular proteins. Prog Biophys Molec Biol 1984W97- 179. 254. Cleland WW, Kreevoy MM. Low-barrier hydrogen bonds and enzymic catalysis. Science 1994;264:1887-1890. 255. Hibbert F, Emsley J. Hydrogen bonding and chemical reactivity. In: Bethell D (ed) Advances in Physical Organic Chemistry. London: Academic Press Ltd., 199Q255-379. 256. Matsumura M, Becktel WJ, Matthews BW. Hydrophobic stabilization in T4 lysozyme determined directly by multiple substitutions of Ile 3. Nature 1988;334:406410. 257. Yutani K, Ogasahara K, Tsujita T, Sugino Y. Dependence of conformational stability on hydrophobicity of the amino acid residue in a series of variant proteins substituted at a unique position of tryptophan synthase a subunit. Proc Natl Acad Sci USA 1987;84:4441-4444. 258. Rashin AA, Iofin M, Honig B. Internal cavities and buried waters in globular proteins. Biochemistry 1986;25:361!&3625. 259. Wilson KP, Malcolm BA, Matthews BW. Structural and thermodynamic analysis of compensating mutations within the core of chicken egg white lysozyme. J Biol Chem 1992;267:10842-10849. 260. McRee DE, Redford SM, Getzoff ED, Lepock JR, Hallewell RA, Tainer JA. Changes in crystallographic structure and thermostability of a Cu, Zn superoxide dismutase mutant resulting from the removal of a buried cysteine. J Biol Chem 1990;265:14234-14241. 261. Yamada H, Kanaya E, Inaka K, Ueno Y, Ikehara M, Kikuchi M. Stabilization and enhanced enzymatic activities of a mutant human lysozyme C77/95A with a cavity space by amino acid substitution. Biol Pharm Bull 1994;17:192-196. 262. Sandberg WS, Terwilliger TC. Influence of interior packing and hydrophobicity on the stability of a protein. Science 1989;245:54-57. 263. Kanaya S, Itaya M. Expression, purification, and characterization of a recombinant ribonuclease H from Thermus thermophilus HB8. J Biol Chem 1992;267:10184-10192. 264. Baldwin EP, Hajiseyedjavadi 0, Baase WA, Matthews BW. The role of backbone flexibility in the
accommodation of variants that repack the core of T4 lysozyme. Science 1993;262:1715- 1718. 265. Kwon K-S, Kim J, Shin HS, Yu M-H. Single amino acid substitutions of a,-antitrypsin that confer enhancement in thermal stability. J Biol Chem 199+269:9627-963 1. 266. Heinz DW, Baase WA, Matthews BW. Folding and function of a T4 lysozyme containing 10 consecutive alanines illustrate the redundancy of information in an amino acid sequence. Proc Natl Acad Sci USA 1992;89:3751-3755. 267. Zale SE, Klibanov AM. Why does ribonuclease irreversibly inactivate at high temperatures? Biochemistry 1986;25:5432-5444. 268. Ahem T, Klibanov AM. The mechanism of irreversible enzyme inactivation at 100°C. Science 1985;228: 128+1284. 269. Sicard PJ, Leleu J-B, Duflot P, Drocourt D, Martin F, Tiraby G, Petsko G, Glasfeld A. Site-directed mutagenesis applied to glucose isomerase from Streptomyces violaceoniger and Streptomyces olivochromogenes. Ann NY Acad Sci 1990613:371-375. 270. Hallewell RA, Imlay KC, Lee P, Fong NM, Gallegos C, Getzoff ED, Tainer JA, Cabelli DE, Tekamp-Olson P, Mullenbach GT, Cousens LS. Thermostabilization of recombinant human and bovine CuZn superoxide dismutases by replacement of free cysteines. Biochem Biophys Res Comm 1991;181:474-480. 271. Lepock JR, Frey HE, Hallewell RA. Contribution of conformational stability and reversibility of unfolding to the increased thermostability of human and bovine superoxide dismutase mutated at free cysteines. J Biol Chem 1990;265:21612-21618. 272. Volkin DB, Klibanov AM. Mechanism of thermoinactivation of immobilized glucose isomerase. Biotechnol Bioeng 1989;33:1104-1111. 273. Kuroki R, Kawakita S, Nakamura H, Yutani K. Entropic stabilization of a mutant human lysozyme induced by calcium binding. Proc Natl Acad Sci USA 1992;89:6803-6807. 274. Imanaka T. Enhancement of thermostability of neutral proteases. Ann NY Acad Sci 1990613:347351. 275. Wilson S-A, Peek K, Daniel RM. Immobilization of a proteinase from the extremely thermophilic organism Thermus Rt41A. Biotechnol Bioeng 1994;43:225-23 1. 276. Jensen VJ, Rugh S. Industrial scale application of immobilized glucose isomerase. Methods Enzymol 1987;136:356-370. 277. Genvig GJ, Kamerling JP, Vliegenthart JFG, Morag E, Lamed R, Bayer EA. Primary structure of 0-linked carbohydrate chains in the cellulosome of different Clostridium thermocellum strains. Eur J Biochem 1991;196:115--122. 278. Lee Y-E, Lowe SE, Zeikus JG. Regulation and characterization of xylanolytic enzymes of Thermoanaerobacterium saccharolyticum B6A-RI. Appl Environ Microbiol 1993;59:763-77 1. 279. Ng TK, Zeikus JG. Purification and characterization of an endoglucanase (1,4-p-D-glucan glucanohydrolase) from Clostridium thermocellum. Biochem J 1981;199:341-350. 280. Chen B-I, King J. Pathway for the thermal unfolding of wild-type and mutant forms of the thermostable P22 tailspike endorhamnosidase. In: Georgiou G, De Bemardez-Clark E (eds) Protein Refolding. Washington, DC: American Chemical Society, 1991;I I F 1 32. 281. Boel E, Brady L, Brzozowksi AM, Derewenda Z, Dodson GG, Jensen VJ, Petersen SB, Swift H, Thim L, Woldike HF. Calcium binding in a-amylases: an X-ray diffraction study at 2.1-A resolution of two enzymes from Aspergillus. Biochemistry 1990;29:624&6249. 282. Marg GA, Clark DS. Activation of the glucose isomerase by divalent cations: evidence for two distinct metal-binding sites. Enzyme Microb Techno1 1990 12367-373. 283. Whitlow M, Howard AJ, Finzel BC, Poulos TL, Winbome E, Gillialand GL. A metal-mediated hydride shift mechanism for xylose isomerase based on the 1.6A Streptomyces rubiginosus structures with xylitol and D-XylOSe. Proteins Struct Funct Genet 1991;9:153-173. 284. Kasumi T, Hayashai K, Tsumura N. Roles of magnesium and cobalt in the reaction of glucose isomerase from Streptomyces griseofuscus S-41. Agric Biol Chem 1982;46:21-30. 285. Coolbear T, Whittaker JM, Daniel RM. The effect of metal ions on the activity and thermostability of the extracellular proteinase from a thermophilic Bacillus, strain EA.1. Biochem J 1992;287:367-
81 374. 286. Stigter D, Alonso DOV, Dill KA.Protein stability: electrostatics and compact denatured states. Proc Natl Acad Sci USA 1991;88:41764180. 287. Hei DJ, Clark DS. Pressure stabilization of proteins from extreme thermophiles. Appl Environ Microbiol 1994;60:932-939. 288. Miller JF, Nelson CM, Ludlow JM, Shah NN, Clark DS. High pressure-temperature bioreactor: assays of thermostable hydrogenase with fiber optics. Biotechnol Bioeng 1989;34:101>1021. 289. Aaltonen 0, Rantakyla M. Biocatalysis in supercritical CO,. Chemtech 1991;1:24(t248. 290. Segel IH. Enzyme Kinetics. New York: John Wiley & Sons, 1975. 291. Luthi E, Reif K, Jasmat NB, Bergquist PL. In vitro mutagenesis of xylanase from the extreme thermophile Caldocellum saccharolyticum. Appl Microbiol Biotechnol 1992;36:503-506. 292. Glaser L, Gajewski G. 1994. Industrial uses of agricultural materials: Situation and outlook report. (50 pages). SOP U.S. Department of Agriculture, Washington DC. 293. Saha B, Zeikus JG. Biotechnology of maltose production. Proc Biochem 1987;22:78-82. 294. Ramesh MV, Saha B, Mathupala S, Podkovyrov SM, Zeikus JG. Biodegradation of starch and aglycan polymers. In: Ratledge C (ed) Biochemistry of Microbial Degradation. Amsterdam: Kluwer Academic Publishers, 1994;31>346. 295. Norman BE. The application of polysaccharide degrading enzymes in starch industry. In: Gooday GW, Ellwood DC (eds) Microbial Polysaccharidesand Polysaccharases. New York: Academic Press, 19793339-376. 296. Reilly PJ. Starch hydrolysis with soluble immobilized glucoamylase. In: Wingard LB, KatchalskiKatzir E, Goldstein L (eds) Applied Biochemistry and Bioengineering. New York: Academic Press, 1979;185-205. 297. Coleman RD, Yang S-S, McAlister MP. Cloning of the debranching-enzyme gene from Thermoanaerobium brockii into Escherichia coli and Bacillus subtilis. J Bacteriol 1987;169:43024307. 298. Spreinat A, Antranikian G. Analysis of the amylolytic enzyme system of Clostridium thermosulfurogenes EM1. Starch 1992;44:305-312. 299. Mathupala SP, Lowe SE, Podkovyrov SM, Zeikus JG. Substrate competition and specificity at the active site of amylopullulanase from Clostridium thermohydrosulfuricum. Biochem Biophys Res Comm 1993;166:1 2 6132. 300. Saha B, Zeikus JG. Novel highly thermostable pullulanase from thermophiles. Trends Biotechnol 1989;7:234-239. 301. Walsh G, Headon DR. Protein Biotechnology. Chichester, England: John Wiley & Sons, 1994. 302. Winter J, Zellner G. Thermophilic anaerobic degradation of carbohydrates - metabolic properties of microorganisms from the different phases. FEMS Microbiol Rev 199075: 13F154. 303. Lovitt RW, Longin R, Zeikus JG. Ethanol production by thermophilic bacteria: physiological comparison of solvent effects on parent and alcohol-tolerantstrains of Clostridium thermohydrosulfuricum. Appl Environ Microbiol 1984;48:171-177. 304. Ng TK, Ben-Bassat A, Zeikus JG. Ethanol production by thermophilic bacteria: fermentation of cellulosic substrates by cocultures of Clostridium thermocellum and Clostridium thermohydrosulfuricum. Appl Environ Microbiol 1981;41:1337-1343. 305. Winterhalter C, Liebl W. Two extremely thermostable xylanases of the hyperthermophilic bacterium Thermotoga maritima MSB8. Appl Environ Microbiol 1995;61:1810-1 8 15. 306. Outtrup H, Boyce COL. Microbial proteinases and biotechnology. In: Fogarty WM, Kelly CT (eds) Microbial Enzymes and Biotechnology, 2nd edn. New York: Elsevier Sciences Publisher, 1990227-254. 307. Coolbear T, Monk C, Peek K, Morgan HW, Daniel RM. Laboratory-scale investigations into the use of extremely thermophilic proteinases for cleaning ultrafiltration membranes fouled during whey processing. J Membr Sci 1992;67:93-101. 308. Crueger A, Crueger W. Glucose transforming enzymes. In: Fogarty WM, Kelly CT (eds) Microbial Enzymes and Biotechnology, 2nd edn. New York Elsevier Science Publishers, 1990; 177-226.
82 309. Jenkins J, Janin J, Rey F, Chiadmi M, van Tilbeurgh H, Lasters I, De Maeyer M, Van Belle D, Wodak SJ, Lauwereys M, Stanssens P, Mrabet NT, Snauwaert J, Matthyssens G, Lambeir A-M. Protein engineering of xylose (glucose) isomerase from Actinoplanes rnissouriensis. Biochemistry 199231:544%5458. 310. Carrel1 HL, Glusker JP, Burger V, Manfre F, Tritsch D, Biellmann J-F. X-ray analysis of D-xylose isomerase at 1.9 A: Native enzyme in complex with substrate and with a mechanism-designed inactivator. Proc Natl Acad Sci USA 1989;86:4440-4444. 31 1. Henrick K, Collyer CA, Blow DM. Structures of D-XylOSe isomerase from Arrhrobacter strain B3728 containing the inhibitors xylitol and D-sorbitol at 2.5 A and 2.3 8, resolution, respectively. J Molec Biol 198%208:129-157. 312. Meng M, Lee C, Badasarian M, Zeikus JG. Switching substrate preference of thermophilic xylose isomerase from D-xylose to D-glucose by redesigning the substrate binding pocket. Proc Natl Acad Sci USA 1991;88:4015-4019. 3 13. Allen KN, Lavie A, Glasfeld A, Tanada TN, Gemty DP, Carlson SC, Farber GK, Petsko GA, Ringe D. Role of the divalent metal ion in sugar binding, ring opening, and isomerization by D-xylose isomerase: Replacement of catalytic metal by an amino acid. Biochemistry 1994;33: 1488-1494. 314. Tainer JA, Roberts VA, Getzoff ED. Metal-binding sites in proteins. Curr Opinion Biotechnol 1991;2:582-591. 315. Tainer JA, Roberts VA, Getzoff ED. Protein metal-binding sites. Curr Opinion Biotechnol 1992;3:378- 387. 316. Glusker JP. Structural aspects of metal liganding to functional groups in proteins. Adv Protein Chem 1991;42:1-76. 317. Innis MA, Gelfand DH, Snisky JJ, White TJ. PCR Protocols: a Guide to Methods and Applications. San Diego: Academic Press, Inc., 1990. 3 18. Prangishvili DA, Vashakidze RP, Chelidze MG, Gabriadze IY. A restriction endonucleaseSuaI from the thermoacidophilic archaebacterium Sulfolobus acidocaldarius. FEBS Lett 1985;19257-60. 319. McConnell DJ, Searcy DG, Sutcliffe JG. A restriction enzyme Tha I from the thermophilic mycoplasma Thermoplasma acidophilum. Nucleic Acids Res 1978;5:172%1739. 320. Cowan DA. Biotechnology of the archaea. Trends Biotechnol 1992;10:31 5 3 2 3 . 321. Kullmann W. Enzymatic Peptide Synthesis. Boca Raton, Florida: CRC Press Inc., 1987. 322. Cheetham PSJ. The use of biotransformations for the production of flavors and fragrances. Trends Biotechnol 1993;11:478-488. 323. Dordick JS. Enzymatic and chemoenzymatic approaches to polymer synthesis. Trends Biotechnol 1992;10:287-293. 324. Faber K, Franssen MCR. Prospects for the increased application of biocatalysts in organic transformations. Trends Biotechnol 1993;11:461-470. 325. Sarney DB, Vulfson EN. Application of enzymes to the synthesis of surfactants. Trends Biotechnol 1995;13:164-172. 326. Fruton JS. Proteinase-catalyzed synthesis of peptide bonds. Adv Enzymol 1982;53:239-306. 327. Flegmann AW, Tattersall R. Energetics of peptide bond formation at elevated temperatures. J Molec Evol 1979;12:34+355. 328. Peek K, Wilson S-A, Prescott M, Daniel RM. Some characteristics of a serine proteinase isolated from an extreme thermophile for use in kinetically controlled peptide bond synthesis. Ann NY Acad Sci 1992;672:471-477. 329. Wilson S-A, Daniel RM, Peek K. Peptide synthesis with a proteinase from the extremely thermophilic organism Thermus Rt41A. Biotechnol Bioeng 1994;44:1- 10. 330. Katchalski-Katzir E. Immobilized enzymes - learning from past successes and failures. Trends Biotechnol 1993;11:471-478. 331. Cowan D, Daniel R, Morgan H. Thermophilic proteases: properties and potential applications. Trends Biotechnol 1985;3:68-72. 332. Kazlauskas RJ. Elucidating structure-mechanism relationships in lipases: Prospects for predicting and engineering catalytic properties. Trends Biotechnol 1994;12:464-472.
83 333. Zamost BL, Nielsen HK, Stames RL. Thermostable enzymes for industrial applications. J Ind Microbiol 1991;8:71-82. 334. May SW, Padgette SR. Oxidoreductase enzymes in biotechnology: current status and future potential. Biotechnology 1983;1:677-686. 335. Klibanov M. Biotechnological potential of the enzyme hydrogenase. Proc Biochem 1983;18:13-23. 336. Burdette D, Zeikus JG. Purification of acetaldehyde dehydrogenase and alcohol dehydrogenases from Thermoanaerobacrer ethanolicus 39E and characterization of the secondary-alcohol dehydrogenae (2O Adh) as a bifunctional alcohol dehydrogenase-acetyl-CoA reductive thioesterase. Biochem J 1994;302:163-170. 337. Keinan E, Seth KK, Lamed R. Organic synthesis with enzymes. 3. TBADH-catalyzed reduction of chloro ketones. Total synthesis of (+)-(S,S)-(cis-6-methyltetrahydropyran-2-y1)acetic acid: a civet constituent. J Am Chem Soc 1986;108:3474-3480. 338. Keinan E, Seth KK, Lamed R, Ghirlando R, Singh SP. Thermostable enzyme in organic synthesis. 4. TBADH-catalyzed preparation of bifunctional chirons. Total synthesis of S-(+)-Z = tetradec-5-en13-olide. Biocatalysis 1990;4:1-15. 339. Lamed RJ, Keinan E, Zeikus JG. Potential applications of an alcohol-aldehydeketon oxidoreductase from thermophilic bacteria. E m Microbiol Techno1 1981;3:144-148. 340. Wilks HM, Hart KW, Feeney R, DUM CR, Muirhead H, Chia WN, Barstow DA, Atkinson T, Clarke AR, Holbrook JJ. A specific, highly active malate dehydrogenase by redesign of a lactate dehydrogenase framework. Science 1988;242:1541-1544. 341. Wilks HM, Halsall DJ, Atkinson T, Chia WN, Clarke AR, Holbrook JJ. Designs for a broad substrate specificity keto acid dehydrogenase. Biochemistry 1990;29:8587-8591. 342. Chen K, Arnold FH.Enzyme engineering for nonaqueous solvents: random mutagenesis to enhance activity of subtilisin E in polar organic media. Biotechnology 1991;9:1073-1077. 343. Arnold FH. Engineering proteins for nonnatural environments. FASEB J 1993;7:744-749. 344. Matsumura M, Aiba S. Screening for thermostable mutant of kanamycin nucleotidyltransferae by the use of a transformation system for a thermophile, Bacillus stearorhermophilus. J Biol Chem 1985;260:15298-15303. 345. Liao H, McKenzie T, Hageman R. Isolation of a thermostable enzyme variant by cloning and selection in a thermophile. Proc Natl Sci Acad USA 1986;83:576-580. 346. Bai Y, Milne JS, Mayne L, Englander SW. Protein stability parameters measured by hydrogen exchange. Proteins Struct Funct Genet 1994;20:4-14. 347. Muheim A, Todd RJ, Casimiro DR, Gray HB, Arnold FH. Ruthenium-mediated protein cross-linking and stabilization. J Am Chem SOC1993;115:5312-5313.
This Page Intentionally Left Blank
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2.
85
M.R. El-Gewely, editor.
Production of secondary metabolites by solid-state fermentation Javier Barrios-GonzBlez and Armando Mejia Departmento de Biotecnologia, Universidad Autbnoma Metropolitana-lztapalapa.Apdo., Mexico
Abstract. Microbial secondary metabolites are useful high value products that are normally produced by liquid culture; but could be advantageously produced by solid-state fermentation (SSF). Particularly if SSF could benefit from a deeper understanding of microbial physiology in a solid environment. Recent research indicates that different kind of secondary metabolites can be produced by SSF: antibiotics, phytohormones, food grade pigments, alkaloids, etc. Physiology in SSF shows several similarities with physiology in liquid medium, so similar strategies must be adapted for efficient processes. However, there are certain particularities of idiophase in solid medium which dictate the need for special strains.
Key words: alkaloids, antibiotics, basic principles, control of metabolism, effect of environmental and nutritional factors, key variables, phytohormones, pigments, production, secondary metabolites, solid-state fermentation (SSF), strain improvement.
Introduction Solid-state fermentation (SSF) is an ancient microbial culture system that is being transformed for new purposes, using new approaches of microbiology, biochemistry and biochemical engineering. There are advantages in employing SSF processes, at least for certain applications, over the conventional submerged fermentation (SmF) ones. However, the latter is usually chosen, owing to the great success of large-scale commercial installations of this kind in many fields of biotechnology. Such units are excellent examples of how fundamental microbiological and biochemical knowledge has been applied to bioengineering principles which have set the procedures to design, control and optimize them. Most of the processes using the SSF technique are commercialized in oriental countries, mainly in Japan. Nevertheless, a resurgence of interest has occurred in Western countries over the last 1 e 1 5 years [1,2] in response to the ever rising demand for economy in processes. This is a well-justified trend which ultimately may lead to an extensive industrialization of SSF throughout the world [3]. Mitchell and Lonsane [3] have divided SSF processes into socioeconomic and profit-economic applications. Socioeconomic applications are either small scale processes performed to contribute something essential for the community (like ensiling of grasses and upgrading lignocellulosic products or staple foods) or involve the disposal of wastes (e.g., composting). A profit-driven application on the other
Address for correspondence: Javier Barrios-Gonzilez, Departmento de Biotecnologia, Universidad Aut6noma Metropolitana-Iztapalapa. Apdo. Postal 55-535, 09340 Mexico.
86 hand will seek commercial profit. In the short term, new profit-economic SSF processes are most likely to come from Japan, which is not only highly industrialized but also has significant experience with large scale SSF because of koji industry. In the Western society traditional applications of SSF are very scarce and SSF has been largely neglected since the Second World War. Soy sauce production, which involves the SSF step of koji fermentation, is an example of an originally small-scale traditional process which has become highly industrialized [3]. The advancement of SSF technology has permitted a higher degree of control over the process and hence, the production of high value products like antibiotics and other secondary metabolites. These compounds are characterized by the great variety of biological activities that confer their actual or potential use in different industrial fields e.g., pharmacy, agriculture and livestock. Normally these sophisticated metabolites would be produced by SmF, but the research performed in the last 10 years indicates that these compounds could be produced in great amounts, and probably with advantages, by SSF. Secondary metabolites are generally reported to be produced in higher concentrations in solid culture, often in shorter times and without the need of aseptic conditions [4-81. Capital costs are claimed to be significantly less [9]. In some instances, economic performance could be even better if the final product is required in solid form, for example antibiotics in animal feeds. However, if the use of SSF is going to generalize, several problems have to be solved. A serious problem with- SSF is the inexperience of engineers in Western nations in the design and scale-up of solid-state fermenters; and the nonexistence of standard fermenters for these processes.- Nevertheless, there have been noticeable advances in this field, which have been reviewed by several authors [lo-121. A very important disadvantage of SSF is the relatively little data on the physiology and genetics of SSF strains in solid medium, to enable optimization of production. In consequence SSF processes can be underestimated or their full potential not exploited. The aim of this work is to analyze the literature on production of secondary metabolites in SSF, and to start drawing conclusions about fundamental principles of microbial secondary metabolism in solid medium. As in the case of SmF, this basic principle will eventually be the basis for efficient process development, control strategies and reactor design, as well as for the methodology to generate hyperoducing strains, particularly suited for solid conditions. The first two sections describe general features of secondary metabolism and SSF. Later sections discuss the control of secondary metabolism in solid medium by environmental and nutritional factors (including the role of regulatory mechanisms), while the last section deals with the relation between the strain and the production level obtained in SSF.
Secondary metabolites The early microbial physiologists observed that in the logarithmic phase of growth
87 there is an intense metabolism, with microbes replicating rapidly their cellular components, as a prerequisite for growth and cell division. It was assumed initially, that stationary phase represented a complete metabolic inactivity of the microorganisms. It was the chemists of natural products who realized the falsehood of that concept. Between 1920 and 1930, organic chemists found that fungal cultures in stationary phase were an almost inexhaustible source of complex organic compounds. As the structures of these molecules were described, it became evident that these compounds did not play a role during exponential phase of growth. Some years before, plant physiologists had recognized two similar classes of compounds produced by plants. There were compounds like chlorophylls that were synthesized by all plants; these were named primary products of metabolism. Conversely, there were compounds like camphor and tannins (or carbinol), which were only produced by particular species of plants, and no metabolic function could be assigned to them. These were called secondary products of metabolism. A little more than 2 decades ago the term was used to describe a wide range of compounds produced by microorganisms, and as their plant counterparts, are not directly related to compounds that form the cell. Today the term secondary metabolite is more often associated to microbial products than plant products [13]. There is still a certain amount of controversy about the lack of function of secondary metabolites, since in some cases specific functions have been attributed to individual secondary metabolites, (mostly related to processes of differentiation) or, in other cases, an antagonistic role in natural environment has been proposed [14]. Hence an adequate definition is the one proposed by Demain et al. [ 151: metabolites that are not essential for vegetative growth of the producing organism in pure culture. They are usually formed as a mixture of closely related members of a chemical family. For example, there are at least eight aflatoxins, 10 polimixins over 20 penicillins and 20 actinomycines. Each is produced by a narrow taxonomic range of organisms and production ability is easily lost by mutation (strain degeneration). These compounds have been described as secondary metabolites in opposition to primary metabolites like amino acids, nucleotides, lipids and carbohydrates that are essential for growth. Despite the enormous diversity of chemical structures found in microbial secondary metabolites, most of these compounds can be grouped in a few classes depending on their biosynthetic origin. A convenient classification is the one proposed by Rose [13]. This author classifies these metabolites in only four groups or families, and enphasizes that this reflects the reduced number of groups in which low molecular weight compounds, precursor to of the cell constituents, can be classified. In this way, secondary metabolites derive from: a) amino acids; b) sugars; c) acetyl-CoA (and related compounds, including Krebs cycle intermediates); and d) terpenes.
Growth and secondary metabolism The relationship between growth and secondary metabolisms has been extensively studied [16-18]. It has been found that production of secondary metabolites starts when growth is limited by the exhaustion of one (or more) key nutrient. Bu’Lock introduced the term trophophase to name the phase of primary metabolism; and idiophase for secondary metabolism. Although these terms have a descriptive value, they are not absolute or mutually exclusive. It is considered that growth in the nonlimited phase (trophophase) is balanced and all the biosynthetic capacity of the cell is required to cope with a high growth rate. Growth is unbalanced when growth is limited by the shortage of a nutrient. Those cell processes that require the limiting nutrient are restricted, while those which do not require it are not. Since growth depends of a wide range of biosynthetic activities, the absence of one (or several nutrients) e.g., carbon, nitrogen, phosphate or trace elements, can restrict it. Not limiting nutrients can then be diverted towards biosynthesis not related to growth. Biosynthesis of secondary metabolites or specific compounds required for differentiation can be stimulated in this way. Consequently, a popular hypothesis proposes that the specific products of secondary metabolism are not important but the process of secondary metabolism itself is of selective advantage for the microorganism. It is considered to provide a mechanism by which excess intermediates can be metabolized in an adverse environment. Such a mechanism would serve to maintain the cell in a functional state during conditions which prevented growth [ 191. It is not possible to define a specific growth rate below which growth can be said to be of the idiophase type. However, such a growth rate (p) can be defined for a specific metabolite produced in the idiophase [20]. These ideas can be illustrated by mentioning the example of ergot alkaloids whose synthesis starts when phosphate source is exhausted from the medium. In gibberellic acid fermentation the growth rate is controlled by limiting the supply of nitrogen. Chemostat studies using glycine as the nitrogen source have shown that pmaxfor Gibberella fujikuroi is approximately 0.18/h and at growth rates approaching this value growth is balanced. At growth rates of O.lO/h carbohydrates accumulated but no bikaverin or gibberellic acid synthesis occurred. Growth rates of less than O.O5/h were required for bikaverin formation and of less than O.Ol/h for gibberellic acid synthesis [21]. In the case of penicillin, whose synthesis correlates with the exhaustion of the easily metabolized carbon source (glucose), lactose is used to reduce the growth rate of P. chrysogenum and stimulate penicillin biosynthesis. The molecular events that turn on secondary product formation at the end of the trophophase are unknown. It is considered that nutrients limitation cause an inducer (positive effector) of secondary metabolic enzymes to accumulate or release genes of secondary metabolism from catabolic repression (negative effector). In any case it is well known that at the end of the trophophase enzymes specifically related to formation of secondary products suddenly appear [22]. It has been demonstrated that the production of secondary metabolites is affected by regulatory mechanisms that appear to be similar to the ones that control primary metabolism, i.e., induction,
89 feedback regulation and catabolite regulation [ 2 3 ] . In this way, production of many secondary metabolites is negatively affected by a number of growth medium constituents, falling into three broad classes: carbon/energy sources, nitrogen sources, and phosphorus-containing compounds. Commercial importance Figure 1 shows the interrelation between the products of secondary metabolism and primary metabolism. It also gives an idea of the great variety of compounds produced. A consequence of the wide range of chemical structures of secondary metabolites is the enormous range of biological activities that can be found among these compounds. Hence, many secondary metabolites have great economic importance since they are applied to or have potential applications in different fields. This can be illustrated by the following examples: 1. Pharmaceutical industry: antibiotics, antitumorals, inmunodepressors, etc. 2. Agriculture: plant growth stimulants, antibiotics with herbicide and fungicide activities that would be more environmental friendly. 3. Cattle raising: growth promoters (antibiotics), antihelmints and coccidiostatics. 4. Food industry: food grade colorants and aromas. These compounds are or would normally be produced in liquid culture (submerged fermentation). However, at least some of them could be advantageously produced by SSF.
SSF SSF has been used since antiquity for the preparation of fermented foods, silages and composting. The use of koji for soy sauce in China, Japan and Southeast Asia goes back as far as 1,000 years BC, and can be considered as a prototype of SSF. It consists in the cultivation of Aspergillus oryzae on soybeans and other grains to produce proteases and amylases, which degrade proteins and transform starch into sugars. In this way, the fermented material is used for the production of soy sauce or, in a second stage, rice wine or sak6. Although the earliest commercial production of enzymes depended on SSF technology developed in China and Japan, the advent of sterile submerged culture techniques in the 40s displaced the solid-state methods in the Western countries [24]. Work performed by Hesseltine et al. [8,25,26] studied and described the traditional SSF processes of the East, and informed about the great technological importance of these culture systems. These reports are probably responsible, in great part, for the renewed interest in SSF observed during the last 10 or 15 years. As a consequence, SSF is being transformed for new purposes, using new approaches of microbiology, biochemistry and biochemical engineering, and often presents several advantages over SmF.
90 Centanycine
Kanvnycins DNA, RNA
phaphalr
c H-
1
Triov phorphate
Serine
- -+
Pe&
phosphate--
Tetfose phorphate Candiddin Chinramphenid
Shikimate
1
t
RrimFin
Phosphadmate
,
-
hidonate IMpIonyI CON
1
Fatty acas
Fig. 1. Relation between primary and secondary metabolism (modified from Rehacek and Sajdl, 1990) [102].
Modern SSF systems Besides the koji-type systems, SSF cultures are now performed on other starchy substrates like roots (cassava or potato flour), bananas, etc. or lignocellulosic material like straw or wood pulp. These SSF systems are referred to as nontraditional in this review. A new type of SSF uses an inert support with absorbed liquid medium. The support can be of natural origin like sugarcane bagasse, or artificial (or synthetic) like polyurethane, amberlite or vermiculite. This kind of SSF is very useful for basic studies since liquid media, with the desired composition can be used, and fermented broths can be extracted And analyzed at any time. Some of them, like sugarcane bagasse, also show great productivity (Fig. 2) [4,27,28]. SSF was defined by Hesseltine [25] as a fermentation in which the substrate is not liquid. However, a modem definition is the one proposed by Lonsane et al. [9], as a microbial culture that develops on the surface and at the interior of a solid matrix and in absence of free water. The porous matrix can be a moistened substrate or an inert support capable of absorbing nutrients dissolved in a solution. In this way two types of SSF can be distinguished, depending on the nature of the solid phase used:
91
Fig. 2. Laboratory scale column reactor packed with 12 g of sugarcane pith bagasse impregnated with liquid medium.
a) Solid culture of one support-substrate phase. Solid phase is constituted by a
material that assumes, simultaneously, the functions of support and of nutrients
Fig. 3. Pilot bioreactor for gibberellic acid production by SSF. This reactor works under sterile conditions and has a highly sophisticated control system (Solar et al. Depto. de Ing. Quimica y de Bioprocesos, Pontificia Universidad Cat6lica de Chile, Chile).
92
source. This material is generally of starchy or lignocellulosic nature. Most of the applications of SSF use this kind of system. b) SSF of two substrate-support phases. Solid culture with an inert support impregnated with a liquid medium. -In this type of fermentation, solid phase is considered as an inert support that serves as a reservoir for a nutritive solution. In this type of SSF water-holding capacity is of paramount importance in the selection of the support. New applications
SSF processes are used on a commercial scale for the production of different types of traditional fermented foods, fungal metabolites and for bioconversion of organic wastes in the East, mainly in Japan [3,29]. However, this culture system could have significative advantages over SmF methods for the manufacture of nontraditional products of interest. These advantages and the versatility of this culture system has driven the appearance of a great number of new application fields (Table 1) [29]. Table 1 . Examples of new applications of SSF. Application field
Product
Microorganisms
Ref.
Fermented foods
Cheeses Koji Pozol Bread Cacao
Penicillum spp. Aspergillus orizae Lactobacillus spp. and yeasts Yeasts and Acetobacter
[301 [311 [321 [331 1331
Protein enrichment of food stuff
Lignocellulosic fibers Starchy material
Aspergillus terreus Aspergillus niger
[341 [311
Enzymes production
Amylases Proteases Cellulases Pectinases
Aspergillus Aspergillus Trichoderma Aspergillus
[351 1361 [371 1381
Primary metabolites
Citric acid Galic acid Gluconic acid
A. niger A. niger A. niger
Secondary metabolites
Mycotoxins
A. flaws and other fungal species
181
Alcohol production
Ethanol
Saccharomyces spp.
[401
Spore production
Inoculum Biological control
Penicillium spp. Trichoderma hartzianum
[411 1371
Composting
Compost
Mixed flora
1421
Silage
Silage
Lactobacillus spp.
[431
Edible mushrooms
Pleurotus
Pleurotus spp.
[441
Biological filters
Treated water
Mixed flora
[421
93 Advantages and disadvantages SSF and SmF have been compared by several authors [9,12,26,45]. Advantages of SSF include: 1. Often sterile conditions are not required. The process proceeds as a pure culture since solid conditions (similar to their natural habitat) ecologically favor fungi and actinomycetes. 2. Mycelial morphology of fungi and actinomycetes (microorganisms most associated with secondary metabolite formation) is well suited to invasive growth on solid medium. This morphology is responsible for considerable difficulties in large-scale submerged culture. These include highly viscous, non-Newtonian broths and foam production. This results in very high power requirements for mixing and oxygen transfer efficiency and can lead to problems during product recovery ~461. 3. Conversely, energy requirements in SSF are relatively low since oxygen is transferred directly to the microorganisms. 4. Metabolites are often produced in much higher yields [4-81. This is an important point to counterbalance disadvantages described below. 5 . Quite often, the solid-state fermented product is used as such, requiring little downstream processing and, hence, improving the process economics relative to liquid-culture-based processes. In some instances the final product is required in solid form, e.g., antibiotics in animal feeds. 6. Extraction of soluble products from solid-state mash into a small volume of solvent can often yield a concentrated solution of the product, which once again, is amenable to purification by simpler means. This implies a reduction in liquid effluents. 7. Capital costs are claimed to be significantly less [9]. 8. Very concentrated media can be used without a deleterious effect on product formation. 9. Useful changes in enzyme characteristics when produced in SSF. Regarding secondary metabolites, secretion of pigments (intracellular in SmF) into the solid medium has been reported (see section on: Moisture content and water activity). Notwithstanding the advantages, large scale commercial use SSF requires solutions to several problems. Disadvantages of SSF include: 1. Heat dissipation problems. 2. Lack of sensors and efficient methods to handle solids. 3. Difficult to add nutrients and controlling agents. 4. The inexperience of engineers in Western nations in the design and scale-up of solid-state fermenters; and the nonexistence of standard fermenters for these processes. 5. A very important disadvantage is the relatively little data on the physiology and genetics of SSF strains to enable optimization of production.
94 Physiological aspects The physiological effect of the liquid environment on biomass and product formation by filamentous fungi has been extensively studied in submerged culture [20,48]. However, there is little information available of the effect of the solid environment on microorganisms in SSF. It can be thought that the effects are similar to the ones observed in liquid cultures, modified for the chemical composition, physical structure of the substrates and local variations of temperature, pH and nutrients concentration and dissolved gases [ 121. Nevertheless, only recently did it become apparent that the biochemical and physiological response of certain microorganisms in solid-state culture may differ greatly from those in SmF, leading to reduction in, or even total loss of, productivity of the desired enzyme or metabolite. Physiology in solid medium can be so different that microorganisms produce enzymes with different characteristics of size, Km, stability, and optimum pHs and temperatures, and a different sensitivity to substrate inhibition in SSF [49,50]. It has even been reported that intracellular enzymes become extra cellular when produced in SSF [51,52]. Process variables A key question for the industrial application of many of these processes is: what are the key process variables in SSF that control growth and microbial metabolism? In practice SSFs are controlled with the following process variables [12,53]: a) pretreatment; b) nutrients; c) particle size; d) moisture content; e) sterilization; f) temperature and pH; g) aeration and agitation; and h) inoculum size; In fact these are the same parameters that are controlled in SmF (except moisture content and particle size). This does not mean that the effect of these parameters is well understood in SSF, rather, that the concepts of SmF are applied directly (Fig. 3). Some theoretical considerations on the effect of these variables on microbial growth in SSF are commented below. 1. Pretreatment. Natural solid substrates generally require some kind of pretreatment to make their chemical constituents more accessible and their physical structure more susceptible to mycelial penetration and adhesion. 2. Nutrients. Although most traditional food fermentations do not require nutritional supplementation, it may be beneficial in nontraditional fermentations. Nutritional factors are usually limiting for microbial growth [54]. In solid materials this limitation is more severe due to the limited diffusion rates and the limited access of the fungus to it.
95
3.
4.
5. 6.
In fact the understanding of this subject is very poor, probably because most of the processes used are of one support-substrate phase. In this way the medium is already adequate for growth or is balanced with nitrogen and phosphorous sources. Particle size. It is generally considered that smaller sizes provide greater superficial area for transference of heat and gas exchange. It also results in higher superficial concentrations of nutrients and shorter pathways for their diffusion [12]. A smaller particle size results in a higher packing density and a concomitant reduction of the void space between particles. This tends to reduce the area of transference of heat and gas exchange. Moisture content. Water is a main component of microorganisms and in SSF has an important role in enzymes, nutrients and products diffusion through the solid matrix [35].It is noteworthy that free water must not be too abundant since it may reduce porosity and, in consequence, decrease the gases exchange. On the other hand, variations in moisture content are produced during the course of the fermentation. The cause of these changes are evaporation, water production by respiration of the culture and, possibly, moisture exchange between the air and the solid medium. Due to the importance of moisture content and water activity of the solid medium on microbial physiology, several authors have performed studies on this parameter. Oriol et al. [55] did interesting estimations of total water, consumed water and residual water in a SSF on cassava flour (one supportsubstrate phase). A theoretical calculation, based on the Ross equation, indicated that water activity of the solid medium descended to 0.85 at the end of the culture. The authors interpreted this as the cause of growth cessation. It is important to note that this theoretical Aw is different from the actual measurement, which is a mixture of Aw of the solid substrate and the Aw of the mycelium in it. Sterilization. Many SSF exclude or greatly reduce the problem of bacterial contamination, so many processes require no sterilization and/or nonsterile conditions during the process. Temperature and pH. In SSF metabolic heat generation can provoke an increase in temperature in the reactor and cause a serious problem [45]. This is due to the high local substrate concentration, the low water content and weak thermal conductivity of the biological materials [53,56,57]. With the aim of solving this problem several authors have established strategies for temperature regulation. Most of them use forced convection of air through the fermenter. Others have used the cooling properties of water evaporation to automatically control moisture and temperature [58-601. We consider that the solution involves the use of all these strategies plus water addition to compensate losses from evaporative cooling. Global pH of the liquid phase of a SSF can be considerably different than the local pH values on the solid surfaces where growth is taking place, due to the superficial charge effects and ionic equilibrium modified by the effect of solute
96 transport [12]. Although there is not a standard protocol for pH measurement, a general procedure is to determine global pH, after suspension of a sample in a 10 times greater volume of water. 7 . Aeration. Aeration fulfills four main functions in SSF, namely 1) maintain aerobic conditions; 2) desorption of CO,; 3) regulate temperature and 4) regulate the moisture level [61]. Partial pressures of CO, and 0, of the gaseous atmosphere of a SSF are important factors for growth and product formation. Oxygen should be enough not to limit growth, and is controlled by the air flow in the reactor. 8. Znoculum size. Generally, SSFs are inoculated with high spore concentrations. Mudgett [ 121 indicates that it is necessary to optimize this parameter, since a very low inoculum density can give insufficient biomass and allow growth of contaminants. Saucedo-Castafieda [62] studied the effect of inoculum size (5.7 x lo7 to 3.6 x lo3 cells/ml) on growth of yeast on SSF on sugarcane bagasse. Shorter lag phases were obtained with increased inoculum sizes. Specific growth rates of the culture also increased at increased inoculum sizes (it stabilized in 5.8 x lo5), while final biomass concentration seemed less sensitive to this factor. Experiments performed by Oriol et al. [63], with Aspergillus niger in the same SSF system, confirm these results (shorter lag phases), using inocula of 5.8 x lo5 to lo9 spores/g of dry medium. It is interesting to note that an inhibitory effect on germination has been observed at those spore densities in liquid culture [64]. Mycotoxin production by SSF Mycotoxins are dangerous metabolites of fungal origin, some of which can represent a health hazard in very small quantities. Although mycotoxins are secondary metabolites, its production in SSF is discussed in this previous section for historical reasons. The fact that mycotoxins are harmful metabolites is another reason to separate them from the useful high-valued compounds. In any case, knowledge of secondary metabolism in solid medium can also be applied to avoid mycotoxin contamination of stored agricultural goods, or even SSFs [65,66]. Research on SSF in the Western countries practically started with the studies on mycotoxin production. Hesseltine and co-workers, in the Northern Regional Research Center (Department of Agriculture of the USA) in Peoria, Illinois, adapted the koji SSF system to produce mycotoxins in much higher concentrations than the ones that could be obtained by liquid fermentation [8,26]. The authors explain that production of great amounts of aflatoxins was entrusted to them by the government for field trails. Initially they tried production by liquid fermentation with very unsatisfactory yields. After that they applied the methodology of rice koji with much higher yields. A very surprising increase in production came when a flask with the rice fermentation was agitated in shaker (200 rpm) as if it contained liquid medium. Very impressive concentrations of aflatoxins were obtained in this way (1.5 g/kg). The method was
97
adopted by the group to produce mycotoxins from different species with excellent results. Similar concentrations of ochratoxins were obtained with some species of Aspergillus cultivated on wheat (production was lower on corn) [25]. Ochratoxin was produced in concentrations up to 2.4 g/kg on wheat, by means of a rotating drum fermenter of four compartments, with deflectors that rise and drop the grains in each revolution [67]. The highest rotating speed tested (16 rpm) gave the highest production, but required 12 to 19 days to reach it. At 1 rpm, usual yields of 2.3-2.5 g/kg were obtained in 8-9 days, while only 0.1-0.2 g/kg were produced in static conditions. Hesseltine [26] summarizes the advantages of agitation on mycotoxin production as: a) effective distribution of spore inoculum; b) maintenance of homogeneity and prevention of pellet formation; c) improves aeration; and d) facilitates heat transfer. Two stages are distinguished on research on mycotoxin production by SSF. The first one was developed by Hesseltine’s group from the late 60s through to the mid-70s. They studied production of several mycotoxins using different grains and forages like alfalfa. This method became classic and the main process variables controlled were: 1. Selection of the grains (substrate), and recommended whole or dehulled rice, barley, wheat, corn and soybean. 2. Pretreatment, which includes soak and cooking. Also, light abrasion of the kernel surface (pearl) for barley and wheat and fractionation of the grains of corn and soybean into five or six pieces. 3. Particle size, kept within a limited range to prevent grain agglomeration. 4. Moisture content is important and should be kept low to prevent contamination. 5. Grains should occupy a relatively small volume of the vessel. A second phase was developed during the second half of the 70s, with the work of Demain and co-workers in Boston, USA. This group studied the production of other mycotoxins in the same SSF system [68]. In general terms it can be said that lower yields were obtained and that it became clear that agitation did not favor the synthesis of some toxins [69,70].
Secondary metabolite production in SSF ’
Although the works referred to in the preceding section suggested that high concentrations of secondary metabolites could be produced in solid cultures, there were no further attempts to produce useful secondary metabolites by SSF (with the exception of pigments), until the end of the 80s. In 1987 Kumar and Lonsane [7] reported the production of gibberellic acid by SSF on wheat bran in flasks. A year later our group published the first paper on penicillin production by SSF on inert support, using Raimbault-type column fermenters with forced aeration [4]. This marked the beginning of a new phase of research on secondary metabolites production by SSF. After this several authors have studied production of antibiotics
and other useful secondary metabolites by different SSF systems (Table 2). These works have constituted a second stage of study and interest on the subject, that is characterized by the diversity of secondary metabolites produced and by the use of nontraditional and novel SSF systems. Also, because some of these studies not only optimize the most usual process variables, but perform basic and deeper studies. The analysis of this literature is beginning to form a view of the nature and particularities of idiophase in solid medium, which is described in the next sections.
The SSF system As was previously mentioned, traditional-type SSFs are performed on grains. Our work on secondary metabolites production in solid culture started using SSF on cassava flour (constituted mainly by starch), and studied the synthesis of aflatoxins and gibberellic acid. In that period it was observed that when the carbon source constitutes part of the structure, the physical characteristics of the solid medium Table 2. Recent applications of SSF for the production of antibiotics and other secondary metabolites.
SSF system
Microorganism
Product
Conc.
(Pdd
Time (days)
Ref.
Wheat bran
Gibberellic acid
1217
7
[7,711
Gibberella fujikuroi
Wheat bran
Gibberellic acid
6800
8
[721
Penicillium chrysogenum
Impregnated support Penicillin (sugar cane bagasse)
5-6
[5]
Streptomyces viridifaciens
Sweet potato residues Tetracycline
4720
5
[731
Streptomyces clavuligerus
Barley
Cefamicine
300
10
[741
Achremonium chrysogenum
Barley
Cephalosporine
950
10
[741
Aspergillus parasiticus
Cassava flour
Aflatoxins
Bacillus subtilis
Wheat bran
Iturine
Claviceps purpurea C . fusiformis
Rye support support Rye
Ergot alcaloids
Monascus kaoling
Mantou meal
Pigments
Streptomyces cinnamonensis
Barley/oats
Monensine
Aspergillus oryzae
Wheat and soybean
Pyrazines
Streptomyces aureofaciens
Wheat bran
Aureomycin
Gibberellla fujikuroi, Fusarium moniliformis c
“Production in units of absorbency per ml at 500 nm.
10500
1.2
[66]
3660
2
1751
690 960 2080 0
11
[761
60
8.5
5430 ODU“
[61
250
[771 Rolz et al. [47] 1781
4800
3
[791
99 deteriorate progressively during the culture, reducing mass and energy transfer. Although metabolite synthesis was very fast, cultures could not proceed for long periods (Tomasini, Fajardo, Barrios-Gonzdez, unpublished results). It was thought that the use of an inert support with impregnated liquid medium would solve the problem, since the physical structure would be more or less constant throughout the culture. In practice this SSF system presented additional advantages for basic studies since the same media used in SmF can be used in SSF, the effect of the presence of different compounds can be determined, and the liquid-fermented broth can be extracted (by pressing) and analyzed at any time. Several supports were tested and sugarcane bagasse (pith) stood out, since good growth and production could be obtained, not only for secondary metabolites but also for different enzymes [4,27,80]. Studies on secondary metabolism in SSF were also impulsed by the use of penicillin as a convenient model secondary metabolite. As can be observed in Table 2, studies on secondary products formation, published during the last 8 years, have used different SSF systems; from the adaptation of the koji system used by Hesseltine and co-workers to the inert support with absorbed liquid medium system. Production levels suggest that lower titres are obtained in kojitype SSF systems, and in longer periods. It is possible that extraction processes are more complicated in those systems. High production is observed in impregnated support and in wheat bran systems. However, it is interesting tonote the high titres obtained in sweet potato residues. The culture period reported (5 days) seems very long for a one-support-substrate phase system. It is also convenient to keep in mind the high yields of mycotoxins obtained by the Peoria (Hesseltine) group using agitated grains SSFs (Fig. 4). The microorganisms and the products SSF is defined as a general method for the production not only of mycotoxins, but of antibiotics and other useful secondary metabolites, like plant growth stimulants, alkaloids and pigments. Panorama is also optimistic due to the range of microorganisms that can be used successfully for the production of these compounds. Besides different fungal species, processes using actinomycetes and at least one using a nonfilamentous bacteria (Bacillus) have been reported.
'
Control of secondary metabolism in SSF: environmental factors A careful analysis of the literature mentioned above shows evidence that basic principles, on the factors that govern secondary metabolite production in SSF, are beginning to emerge. Moisture content and water activity Like in the early works on mycotoxins, the new processes confirm the importance of
100
Fig. 4. Electron microphotograph showing the adherence of yeast cells to the sugarcane bagasse during a solid culture (Saucedo-Castafiedaet al., Depto. de Biotecnologia, Universidad Aut6noma Metropolitana, MBxico).
moisture content as a fundamental parameter. There is not a study in which this factor has not been optimized, and its importance becomes evident when production changes, under different initial moisture values, are observed. This is an expected fact since water has an important role in enzymes, nutrients and products diffusion through the solid matrix. However, the importance of this variable appears to be more critical for secondary metabolites formation, probably since initial conditions must generate an adequate environment for production, after the rapid growth phase. Water activity (Aw) is related to the concentration of solutes in a liquid medium and represents the unbounded, and therefore available, water. In SSF determination of available water during the culture is an important problem. Moisture content of a solid culture, at any time, is constituted by: 1) available water, 2) bound water, and 3) mycelial constitutive water. As the culture develops this latter fraction becomes greater, while available water decreases. However, available water is an unknown quantity, since there is no quantifying method. Through indirect calculations (estimations and balances) it has been proposed that, in SSFs of one substrate-support phase, growth stops because of exhaustion of the available water, i.e., Aw = 0.85. It is interesting to note that real Aw determination was much higher since it represents the global Aw, including mycelium [63].
101 Initial moisture in different SSF systems Optimal initial moisture contents for growth, depend on the water-holding capacity of the solid medium, and are lowest in grain SSF systems, with values around 35%. In starchy flour systems this parameter is near 50% and in wheat bran between 50 and 60%. The impregnated support system allows the usage of the highest initial moisture contents (up to 78%), with optimum values around 70%. Lotong and Suwanarit [81] have produced red pigments in plastic bags containing rice grains. They observed that pigmentation occurred only at relatively low initial moisture contents ( 2 6 3 2 % ) . It is possible that at these moisture levels nutrients transport or supply during idiophase is slower, giving rise to a slower and more convenient growth rate. In fact the authors report that at higher initial moisture levels higher activities of glucoamylase were obtained. Glucose was rapidly liberated in concentrations of 120 g/l, presumably inhibiting pigmentation. This information suggests that moisture level could be used to control growth rate (or catabolites concentration) in these culture systems. Johns and Stuart [82] confirmed solid culture was superior to liquid fermentation for red pigment production by Monascus purpureus. This result has been attributed to the derepression of pigment synthesis in SSF, due to the diffusion of intracellular pigments into the surrounding solid matrix. In submerged culture, the pigments normally remain in the mycelium due to their low solubility in the usually acidic medium. However, a word of caution is necessary for pigments production after the work of Blanc et al. [83]. The authors demonstrated that, besides the pigments, strains of M. purpureus and M. ruber produce the mycotoxin citrinin in liquid and solid cultures. Moisture content for different microbial groups Different SSF systems have distinct water retention capacity, therefore showing diverse optimal values of initial moisture content (IMC). However, as can be observed in Table 2, in the same SSF system, bacteria (actinomycetes and Bacillus) have a higher requirement for water than fungi. In the wheat bran SSF system, optimal IMC for gibberellic acid production by G. fujikuroi was 60%, while the optimal value for iturine production by Bacillus was 68%. In the case of cephalosporine production optimal IMC in a barley SSF system was 39.5% for Streptomyces clavuligerus, while for the fungus Achremoium chrysogenum was 33%. With tetracyclines the high moisture content that the material (sweet potato residue) can hold is surprising: 68% with an Aw of 0.995. Although there are no reports of an antibiotic of an actinomycete produced in SSF on support, the abovementioned Aw value can be compared with the penicillin production of a solid medium with an Aw of 0.967. Moisture content and water activity: experiments in SSF on support An interesting insight into the nature of secondary metabolism in solid culture is arising from the research performed in SSF on support. Up to what point these
102
conclusions can be extrapolated to other SSF systems remains to be studied (Fig. 5). Sato et al. [84]developed a mathematical model of microbial growth in SSF using wood pulp with impregnated liquid medium. Results with Cundidu lipoliticu showed that an increase in specific growth rate (p) is directly related to an increase in moisture of the solid medium, as well as with the relative humidity of the air flow. Later, On01 et al. [63]carried out studies with A. niger in SSF using sugarcane bagasse with absorbed liquid medium. The authors found that, when initial moisture is varied between 40 and 75%, but keeping medium concentration constant (Aw of 0.97),growth rate was not modified (p = 0.4/h).Nevertheless, when Aw (and initial moisture) was modified, between 0.9 and 0.986,with different glucose concentrations, p varied between 0.19 and 0.54/h.The total amount of biomass generated was greater in the more concentrated media, although the yield Yx/s did not vary greatly. The most important impact was on germination time, which varied from 5 to 20 h. It can be said thus, that Aw controls growth in SSF (at least in SSF on support), having a direct effect on growth rate and an inverse effect on germination. Also, that initial moisture content does not have any effect on the growth phase. Although IMC did not have an important effect on trophophase, experiments performed in our laboratory showed that this parameter has an important impact in idiophase. Barrios-Gonzillez et al. [4]showed that IMC has an important influence on the penicillin production level in SSF. The authors modified initial moisture (between 60 and 78%) keeping Aw constant, in a similar way as done by Oriol et al. [63].In cultures with IMCs of 70 and 73%, high production rates were observed during the last part of the fermentations, which allowed the cultures to reach penicillin titres of
Fig. 5. Automatic on-line measuring system of oxygen uptake and carbon dioxide production.This system was designed in Depto. de Biotecnologia, Universidad Aut6noma Metropolitana, MBxico (BaniosGonzilez and Mejia).
103 1,120 pg/g of dry matter, while cultures with lower or higher IMCs only reached 280 pg/g. In the same work, experiments were performed in which nutrients concentration was increased (decreasing Aw), keeping IMC constant in 70%. It was found that unlike trophophase, the use of very concentrated media favors the antibiotic production in SSF. This was an important effect since production increased 5-fold in 2x medium (twice the concentration recommended for SmF). Conversely, the use of concentrated media negatively affected production in SmF. This represents a marked difference between the effect of moisture content (and Aw) in trophophase and idiophase in SSF. The need of concentrated media in SSF for high secondary metabolite production also represents an outstanding difference between physiology in solid and liquid mediums. Very recent studies performed in our laboratory (M. Dominguez, A. Mejia, J. Barrios-Gonzlilez, 1995;unpublished results) widen the view in this field, by demonstrating that IMC and medium concentrations do not control production levels directly. At this point it is important to remember that the SSF on support system is constituted (initially) by three main components: support, water and nutrients. Support and nutrients represent the solids, so to increase the nutrients concentration of the solid medium, keeping a constant moisture value, the support content has to be decreased. The experiments on moisture content performed by Oriol and co-workers were made keeping medium concentration constant (Aw = 0.977) by decreasing bagasse content of the solid medium. To increase the concentration of the liquid medium, the authors decreased support and water content. In this work, it was found that a particular value of IMC (or nutrient content) can be adjusted by varying the other two components (nutrients and support content) in different ways, which resulted in very different penicillin yields. In other words, the same initial moisture (or concentration) value can create different conditions for idiophase. In fact, the different ways in which these two parameters can be adjusted represent different nutrients/bagasse/water combinations. It was observed that when nutrients or moisture content are increased, production can increase or decrease depending on how this variation is compensated with the other two components. However, when bagasse content was decreased, penicillin yields always increased. It was established then that penicillin synthesis in this system is controlled by the proportion of support and the other two components. At this point it is important to note that not all medium components are completely soluble. Solid media with high nutrients concentration and low bagasse content has a very different appearance (dense, humid, soggy) to a solid medium with higher support content (fluffy, dry, low density). Combinations used in these experiments were plotted in a triangle of combinations of nutrient support and water. Conditions of high production were localized in a narrow fringe of low support content (l(t12.5%). In this fringe, maximal titres were detected in a zone low moisture (62%) and high nutrients content (25.5%), as well as in a zone moisture (73-75%) and a medium level of nutrients concentration
104 (12.3-16.25%), which are equivalent to 1.5 and twice the recommended concentration for SmF. Although these initial conditions have low bagasse content in common, it is not clear why such different combinations can create a favorable environment (maybe similar) for penicillin production during idiophase. However, respiration measurements have been discovered to be a very sensitive method for detecting subtle changes in metabolism, and are beginning to throw some light on this and other basic aspects of secondary metabolism in SSF. Respiration studies of the cultures with diflerent components combinations Respiration studies were performed on these fermentations (J. Barrios-Gonzilez, M. Dominguez, A. Mejia, 1995;unpublished results) by means of an automated respirometer [85]. Two kinds of curves were obtained: the derivative (ml CO,/g/min vs. time) and the integrated or accumulative form (ml total COJg) which was calculated from the latter and resembles the growth curve. The accumulative respiration curve (Fig. 6) can be divided into a sharp peak that corresponds to trophophase. When this peak falls, penicillin synthesis starts in all cases. After this, a second peak is formed, which is lower and much wider, presenting a negative slope that falls steadily until, near the end, slopes down until it almost reaches zero at the end of the culture. Penicillin production continues during the second peak. In a typical integrated curve (Fig. 7), the culture starts with a short period of high slope (QC0,-t), followed by a long phase with a lower slope (QC0,-i), which is stable for a period and then decreases steadily. This stage correlates with penicillin production so is identified as idiophase (Fig. 8). The first conclusion drawn from the respiratory experiments is that, like in SmF, production starts when growth (respiratory activity) is limited. This observation agrees with the growth kinetic patterns observed by Kumar and Lonsane [71] on gibberellic acid production on wheat bran SSF. As previously mentioned, penicillin production increases when support content of the solid medium decreases. When respiratory rates (QCO, in ml/min*gdry medium)
f
6 w-
Q
1800 1500
1200
900
CI)
N
8 E
600
300
o0
20
40
00
80
100
120
140
Time (hl Fig. 6 . Respiration kinetics (derivative form) of penicillin SSF. Comparison of high (solid line) and low production conditions (thin line).
105
0
30
60 90 Time (h)
120
Fig. 7. Time course of C 0 2 production (accumulative form) in SSF by P. chrysogenum with different support (pith bagasse) content: 23,9%(thin line) and 10.3 (solid line).
for tropho- and idiophase were calculated, it became evident that decreasing bagasse content caused a decrease in respiratory rate. This suggests that in solid culture, like in SmF, optimum penicillin production (highest specific production rate) is obtained when the culture grows within a range of low specific growth rates (not lower than O.O15/h) [14]; and that low support content is the way to achieve this in SSF. Conditions of highest penicillin production presented certain common respiratory patterns: 1. The slope of phase 2 (derivative curve) was lower, i.e., tended to form a leveled plateau. 2. CO, evolution rates during idiophase (QC0,i) were always lower than the QC0,i under conditions of lower production. Although numeric values varied among the high production cultures, these QC0,i were always smaller (between 15 and 88% lower) than the ones determined in low production conditions. These results indicate that high production conditions generate metabolic stability during idiophase, with slower respiration rates (and hence growth rates). It appears thus that, under these conditions, nutrients are less exposed and mass transfer
0
30
60 90 Time (h)
120
Fig. 8. Time course of penicillin production by P. chrysogenum in SSF with different support (pith bagasse) content: 23.9% (thin line) and 10.3% (solid line).
declines, causing slower nutrient uptake. These initial combinations probably permit an adequate and constant nutrient supply during idiophase, supporting slower but constant growth rates that are more adequate for product biosynthesis. Particle size and packing density Oriol et al. [55] studied growth kinetics of A. niger in SSF on impregnated bagasse. It was found that increases in support particle size caused decreases in growth rate at the end of this growth phase and suggested that this depressive effect could be caused by limitations caused by the intraparticular nutrients diffusion. This effect could favor secondary metabolism since nutrients limitation triggers and maintains idiophase. Barrios-GonzBlez et al. [86] determined the effect of this parameter on penicillin production in a similar SSF system. Although highest titres were reached using the biggest particle size (14 x 1.7 mm; retained in 10 mesh), experiments using washed bagasse showed that the increase in production was due to higher sugar content in that fraction. However, the use of coarse wheat bran (particle size of 0.3-0.4 cm) resulted in an increase of 2.5 times in the yield of gibberellic acid, in relation with smaller particles [7]. The authors indicate that the increase could have been due to the availability of larger void fraction, noncompactness of the sterilized medium, better oxygen transfer and efficient removal of CO, and other volatile products. Packing density of the solid medium is a parameter that is seldom studied in SSF. However, this parameter will be increased in most static reactors that can be used for production scale-up. Barrios-GonzBlez et al. [83] reported a moderate increase in penicillin production (1,25W 1,680 pg/g) by P . chrysogenum Wisconsin 54- 1255 in a densely packed solid-state cultures (0.35 g/ml), using the impregnated bagasse system. The reason for this increment is not clear, but it should be related to the reduction of the interparticular space. Under these conditions, an inhibition of conidiation was also observed. Since sporulation occurs during secondary metabolites formation, it is possible that those two effects could be related. Inhibition of morphogenesis implicates that more intermediates and substrate might be available for secondary metabolites biosynthesis. This hypothesis has also been used to explain yield improvements by agitation in the early mycotoxin work [26]. Another possibility is that contact between mycelia inhibits growth but stimulates secondary metabolism. Aeration and agitation The gas environment may significantly affect the relative levels of biomass and enzyme production [87]. Han and Mudgettt [88] found that levels of 0, and CO, in the gas environment influence pigment production significantly and growth to a lesser extent in SSF. Maximum pigment yields, with Monascus purpurea on rice, were observed at 0.5 atm. of 0, partial pressure in closed pressure vessels. However, high CO, partial pressure progressively inhibited pigment production, with complete
107 inhibition at 1.0 atm. In a closed aeration system with a packed-bed fermenter, oxygen partial pressures ranging from 0.05 to 0.5 atm., at constant CO, partial pressures of 0.02 atm., gave high pigment yields with a maximum at 0.5 atm. of O,, whereas lower CO, partial pressures at constant 0, partial pressure of 0.21 atm. gave higher yields. Experiments performed in our laboratory also indicate that growth is less sensitive to high CO, partial pressures, than penicillin production. In fact these conditions seemed to stimulate germination (unpublished results). Earlier work in our laboratory studied the effect of different air flow rates on aflatoxin production by A. parasiticus in SSF on cassava flour [66]. A moderate positive effect was observed when increasing aeration rates between 0 and 0.3 l/h*g of moistened medium, while no effect was noted on growth of final pH. It was also found that high aflatoxin concentrations were still reached at low aeration rates. The form of the curve was similar to the one obtained by Silman et al. [89], studying aflatoxin production in corn (grains). An important difference is that these authors studied a very different aeration range: 0-0.0024 1h.g of humid corn. It is not clear if different ranges were studied in these two works or if the SSF systems show different aeration needs. In contrast to these findings, ochratoxin A yields in a rotating drum bioreactor were reduced by aeration [67]. It is considered that agitation prevents heterogeneity of the solid medium composition and in mycelial age. Mixing breaks long mycelial nets, generating shorter mycelia in similar physiological stages (this could be useful for secondary metabolite production). Agitation can be used to improve gas exchange and to remove heat from the solid medium. This operation is also a prerequisite for mixing water or other type of additions. As previously mentioned, this can be an essential part of a cooling strategy. Water additions can also be a key operation in prolonging production phase whether the limiting factor is available water or a nutrient depletion. However, it has been reported that some fungi do not tolerate agitation in SSF [12]. Barrios-Gonzilez et al. [86] determined the effect of agitation on penicillin production on impregnated bagasse. The authors did not find a negative effect when P. chrysogenum Wisconsin 54-1255 was used. Similar experiments with the higher producer P. chrysogenum P-2 showed a slight positive effect on product formation (4,000 vs. 5,700 pgJg dry medium) and an important increase in metablic activity: 189 vs. 261 ml CO,/g dry medium, at 96 h. Image analysis indicated that the microbial population was formed by shorter mycelia (approximately 200 vs. 500 p), presenting a higher branching frequency (branching every 67 f 25 vs. 246 f 70 p) [go]. In other words, a higher number of growing points, giving rise to a higher metabolic activity (and probably to a greater secretion surface), as well as greater physiological homogeneity of the culture (Fig. 9). Kumar and Lonsane studied, in a different way, the effect of the degree of aeration on gibberellic acid production in wheat bran SSF. The authors varied the ratio of the volume of the solid medium to the total capacity of the flask. As the ratio increased from 0.024 to 0.1 the yields of gibberellic acid decreased from 1,116 to 916 m a g of dry medium. The 0.024 ratio corresponds to a 1.0 cm depth of the moist medium.
108
Fig. 9. View of a pilot reactor (1.6 m3) for SSF; with screws and carriage. This reactor has a maximum working capacity of 1 ton (200 kg dry matter) and can be scaled up to production level (Durand and Cherau, INRA, Dijon, France).
Fig. 10. Pilot fermenter for solid-state cultures. This reactor has been used for the production of penicillin by P. chrysogenum; and for protein enrichment of cassava flour by A. niger (GutiCrrez-Rojas et al. Depto. de Biotecnologia, Universidad Authoma Metropolitana).
109 An important physical factor is pretreatment of the substrate or support. Autoclaving the moist solid medium can cause changes in the substrate and make it more amenable to degradation during fermentation. In the case of inert supports, pretreatment can bring out its full water-holding capacity. It has been reported that gibberellic acid production increased slightly (from 1,023 to 1,032 m a g ) with increased autoclaving time up to 45 min at 121°C. The yield was lower (910 mgkg) with 15 min autoclaving time, indicating insufficient modification of the solid substrate (Fig. 10). Inoculum size
In SSF inoculum must be distributed homogeneously and must be high enough to assure predominance of the strain. In recent work on secondary metabolites production two tendencies can be noted: some inoculate with spores and some inoculate with mycelium from a liquid seed culture. Among the studies in which a spore inoculum was used, the work on alkaloid production on rye grains is interesting in this regard since the effect of inoculum size was determined [76]. Inoculum size varied between 2 x lo6 and 2 x 10' spores/g of dry medium. Results showed that by increasing inoculum size, lag phase (germination) was slightly reduced, while growth rate was increased. However, the highest alkaloid production was obtained with 2 x lo7 spores/g. In the work of tetracycline production with Streptomyces vridifQciens, inoculum size was also optimized, varying this parameter between lo7 and lo9 spores/g of dry medium. In this case too, the highest titres were obtained using lo8 spores/g [73]. It seems that with an actinomycete, with less capacity to colonize the solid medium than fungi, a greater inoculum size is required. The effect of the inoculum type and size on gibberellic acid production on wheat bran was studied by Kumar and Lonsane [46]. Results showed that gibberellic acid production was more or less the same when an inoculum, grown for 7 days in liquid or solid media, was used. However, the yields of this metabolite were reduced by about 33% when inoculum grown in liquid medium was used for 20 h. The maximum production was obtained when 15 or 20% (w/v) liquid inoculum was used, although lower inoculum sizes caused a reduction in yields. The effect of inoculating with different mycelium concentrations (from 0.8 to 2.5 mg of dry biomass/ml) was also tested in cefamycine and cephalosporine production on barley SSF. It was found that antibiotic production increased proportionally to the inoculum size, so 2.5 mg/ml were used throughout the work [74]. In the case of iturine, SSF on wheat bran was inoculated with 3 ml of liquid culture for 15 g of moistened medium, which appears to be a similar proportion to the one used for gibberellic acid. Very recent results in our laboratory indicate that penicillin SSF can also be inoculated with mycelium from liquid culture. Preliminary results indicate that this can even have a positive effect on production and shorten production time.
110
Control of secondary metabolism in SSF: nutritional factors (and pathway regulation) In traditional SSFs, modification of the type and concentration of nutrients is done indirectly by changing the grain used. In early work on mycotoxin production on grains, fermentations were not supplemented with additional carbon source. It is possible that a convenient balance of nutrients was obtained by an adequate selection of the grain.
Carbon source In recent research on secondary metabolites production in solid culture, an additional carbon source was often used. The impact of this procedure on production indicates that this is a key parameter. Production of gibberellic acid on wheat bran increased 3.5 times when the solid medium was supplemented with 20% of soluble starch, although higher concentrations had a deleterious effect [71]. The use of a fed-batch SSF with intermittent feeding of soluble starch powder had a moderate positive effect of 18% on production [7]. A subsequent report [91] described a 47% increase in product yield over batch culture using feeds of corn starch. The reason for this increase was that production continued for another day. This is a very important result since high productivity process requires high production rates for as long as possible. Tetracycline production on SSF with sweet potato residues showed a 270% increase when the medium was enriched with 10% soluble starch [73]. It is important to note that the addition of 10% glucose caused a 60% decrease in the antibiotic titres. Supplementation with 5% sucrose had a positive effect on alkaloid production in SSF on rye grains. To increase alkaloid production on SSF on support (sugarcane bagasse) a careful optimization of the medium was done by statistical response surface methods. The results of the analysis identified sucrose (250 g/l) as one of the key factors, together with triptophan (precursor) and phosphate (limiting nutrient; phosphate regulation) [76]. Conversely, supplementation with glucose, KH,PO, and MgSO, did not affect iturin production on wheat bran SSF with a strain of Bacillus subtilis [75].
Pathway regulation Studies in liquid medium have shown that carbon source can also have negative effects on the synthesis of secondary metabolites. In the case of penicillin, glucose depresses the synthesis of the antibiotic, repressing transcription of enzymes of the pathway (ACV synthetase and isopenicillin N synthetase) [92]. The use of SSF on support has permitted the initiation of studies on how regulatory mechanisms operate in solid medium. This information is not only important from the basic point of view, it could find applications in the design of production processes as well as in the
111
development of overproducing strains. Initially it seemed likely that secondary metabolite synthesis in solid medium could be subregulated due to limitations in mass transfer in this system. However, our studies on aflatoxin production in cassava SSF showed that when the concentration of phosphate and ammonium in the solid medium were reduced, aflatoxin production increased proportionally [66]. Since the synthesis of these metabolites is regulated by phosphate and ammonium, these results insinuated that the biosynthesis of aflatoxins was regulated, in solid medium, in a similar way as in liquid medium. On the other hand, Ramesh and Lonsane [93] reported what they called the ability of SSF (wheat bran) to minimize catabolic repression on a-amylase of Bacillus lichenformis. The authors found that enzyme production in SmF decreased drastically when soluble starch concentration of the medium was increased from 0.2 to 1%. In comparison, enzyme production in SSF increased 29-fold when starch concentration was increased in wheat bran SSF. Later, the authors [94] observed that the addition of 1% glucose to a liquid culture completely suppressed production, while in SSF 15% glucose slightly stimulated production. Similar results were obtained by Solis et al. [28] on pectinases production by A. niger using SSF on impregnated bagasse. In SmF an initial glucose concentration of 3% caused a strong decrease of enzymatic activity, while in SSF, concentrations up to 10% had a stimulatory effect on production. Although these results give the impression of a much greater regulatory threshold in solid culture, sugar consumption kinetics suggest that these differences are more related to the capacity to consume sugars in solid culture (very rapid uptake rate), and not to differences in regulatory thresholds. In SSF 90% of the sugar was consumed in 24 h, while in SmF this proportion of the sugars was consumed in 96 h. Taking advantage of the facility of SSF on support to use exactly the same medium composition as in liquid culture, and to analyze the fermentation broth, our group has compared the effect of glucose on penicillin biosynthesis in SSF and in liquid culture. Regulatory thresholds were estimated by carefully correlating glucose uptake kinetics with the initiation of penicillin biosynthesis. Results (SmF: 28-20 gll; SSF 3G14 g/l) indicated that penicillin synthesis is also regulated by carbon catabolite regulation in SSF and at similar thresholds as the ones observed in SmF [951. On the other hand, the high concentrations of penicillin that can be obtained in solid medium (see section on: The strain and the production level) ih this culture system must represent a problem to the fungus. The relatively low diffusion rates in the solid medium can cause an accumulation gradient of antibiotics near the mycelium. Feedback regulation of penicillin G biosynthesis (by penicillin V) in SSF was recently studied in our laboratory. Results indicated that this mechanism is also active in SSF [96], and suggested that, as was found in liquid medium [23], the regulatory threshold of a strain is directly related to its productivity. However, the thresholds estimated in SSF appear to be higher than in liquid medium (Fig. 11). These results strongly suggest that mechanisms which regulate the biosynthetic pathway of secondary metabolites in SSF also play a role in solid culture.
112
Fig. 11. Laboratory scale rotatory reactor for SSF, designed and built in Depto. de Biotecnologia, Universidad Autonoma Metropolitana, MCxico. This fermenter is used to study the effect of mixing and water and/or nutrients additions on penicillin fermentation (Barrios-Gonzdez, Mejia and Miranda).
Fig. 12. General view of a pilot scale reactor (50 I total volume) for asceptic SSF. This reactor has been used for gibberellic acid production by fed-batch.
113 Nitrogen source Concentration and type of nitrogen source is important in growth and production media in SmF. Although during the early work on mycotoxin production no nitrogen source was added to the medium, recent work on secondary metabolites production in different SSF systems has shown the importance of doing so. Yang and Ling [73] optimized tetracycline production in SSF on sweet potato residues. One of the parameters that contributed most to increased production was nitrogen source. The authors started by decreasing C/N ratio from 65 to 20 with (NH,),SO,, increasing production from 48 to 322 pdg. They fixed the concentration of this compound in 0.5% (1% had a negative effect) and combined it with 20% of an organic nitrogen source (peanut flour or wheat bran). This had a synergic effect since tetracycline production increased to 2,000 pdg. It is also interesting to note that tetracyclines spectrum varied when the nitrogen source was changed. When NH,NO, was used, tetracycline and chlortetracycline were produced in a 59.4 and 28.1% proportion. When the combined nitrogen source was used these percentages changed to 18.3-38.6%. For alkaloid production in SSF on rye, a mixture of urea and ammonium sulfate was used to control pB during the culture. The studies on SSF on inert support found ammonium sulfate as the best nitrogen source, and was an important factor for production increase (2,500 pdml) [76]. Limiting nutrient In liquid medium secondary metabolism starts when growth is limited by the exhaustion of a key nutrient. The studies on respiration kinetics and penicillin regulation described above, together with the observations made in the works of aflatoxins (phopsphate and/or ammonium concentration caused a decrease in yields); and alkaloids (low phosphate, a key factor) formation, indicate that secondary metabolism is triggered, in SSF, by the same stimuli as in liquid medium. Precursor effect In liquid culture, the addition of a precursor (biosynthetic pathway intermediate) that is rate limiting, during antibiotic and other secondary metabolites formation, brings about an increase in production and/or directs the synthesis preferentially towards one of the metabolites of the family. No specific studies on this subject have been performed in solid culture, however, the effect of triptophane on alkaloid synthesis mentioned before, as well as the increase in tetracycline production by the addition of methionine indicates a precursor effect. In our research on penicillin production, phenylacetic acid has been used as a precursor for penicillin G. The HPLC analysis have not shown evidence that related penicillins are being synthesized. All this indirect evidence suggests that precursor effect is also manifesting in solid
114 culture. This can have practical importance in a commercial process since the formation of other penicillins, or related compounds in the case of other secondary metabolite production, is negative for its economy.
The strain and production level The development of highly productive microbial strains is a prerequisite for efficient biotechnological processes. Up to the present time, strain improvement was mainly based on induced mutagenesis. Mutation and selection has been responsible for the impressive increases in penicillin titres (several 100-fold), while only during the last few years has genetic recombination based on parasexual crosses and protoplast fusion became an additional practicable strategy [97]. Although recombinant DNA techniques (genetic engineering) provide an enormous potential for strain improvement in industrial microorganisms, its impact is just beginning to be felt. Basic cloning systems have now been developed for various microorganisms used in industrial processes [98]. Special strain for SSF
One of the major positive aspects of SSF is that metabolites are, in many cases, produced at much higher yields than by liquid fermentation (SmF) [4,5,25,99]. However, there are only a few reports on comparative studies, and these works have been carried out with low-yielding strains, which are closer to the wild-types than to the hyperproducing mutants used in modem SmF industry. In fact there is a lack of information regarding what kind of strains are needed for SSF processes. Overproducing strains, developed for liquid medium, might not have the expected performance in a SSF process. This can cause not only an underestimation of the potential of SSF for certain applications, but could be a serious handicap for this culture method, since most or all of the overproducing strains have been developed for SmF. Shankaranand et al. [lo01 observed discrepancies in the levels of a-amylase produced by strains of Bacillus in SSF in relation with SmF. The authors concluded that cultures that are good producers in SmF cannot be relied upon to perform well in SSF. A recent study of our group [5] showed that, as in SmF, the strain is a fundamental element in the production level that can be obtained in SSF. It was found that higher yielding strains, developed for SmF, also tend to be the highest producers in SSF. This agrees with our results on regulation of penicillin biosynthesis in SSF, since it implicates that mutations that allow high production in liquid medium are also useful to overproduce in solid medium. Strains efSiciency in SSF
In the same work [5] parameters were defined which allow quantitative evaluation of
115 the efficiency of different strains to produce in SSF. Relative production was defined as the ratio = peak production in solid divided by peak production in liquid. This parameter can be roughly interpreted as the number of times production in solid is greater that production in liquid culture (if >l). Relative productivity was defined in a similar way. With this tool it was possible to determine that higher yielding strains (for SmF) tend to be less efficient in SSF, that is these strains show relative productions of 0.1-4, while lower yielding strains can show relative productions of 17. This implicates that there is one or several characteristics, probably not related to the biosynthetic pathway, that allow high production in SSF (besides the ones needed to produce in liquid medium), which are more frequently found in low producing strains. These characteristic(s) are unknown but are expressed as high relative production. Since this value was higher in lower yielding strains, it is possible that these characteristic(s) have been lost during the genetic improvement programs.
Strain improvement for SSF Although higher yielding strains (for SmF) tend to be less efficient in solid culture, some mutants derived from these strains displayed very good performance in solid medium (high relative production). This is the basis of the method described to select high producing mutants particularly suited for SSF. In this work mutant strains were generated that produced up to 10,500 pg of penicillin per g of dry medium, representing production increases, in relation to parental strains, of between 500 and 640% [ 5 ] . This means that isolating mutants from (SmF) high-yielding strains, and selecting as performed in this work, is an efficient and rapid way of generating penicillin (and probably other secondary metabolites) overproducers particularly suited for SSF. The characteristics that allow a high relative production do not seem to be related with antibiotic biosynthetic pathway since are only expressed in SSF. Neither do these characteristics seem to be related to fast and abundant growth in solid medium since P. chrysogenum P2-4 (the highest producing mutant obtained) reached peak production in 130 h, while lower producing mutants like ASW8-20 (8,750 pg/g) in 85 h and Wisconsin 54-1255-6 (4,530 pg/g) in only 60 h, with much greater apparent growth. This can also be noted in the study on alkaloid production by SSF. Six strains of Cluviceps were evaluate in SmF and production of the two best was optimized in SSF on impregnated support. Alkaloid production of C. fusiformis (ATCC 26019) was high in both culture systems (solid and liquid), even though growth of this strain in SSF was very poor (4.7 g of dry biomassfl vs. 17.5 g/l) as compared with the other strain (C. purpurea 1029~).It is noteworthy that mentioning that C. fusiformis practically did not grow on rye grains and displayed the slowest apical growth rate in petri dish [76]. On the other hand it is surprising that Ohno et al. [75] produced 3,660 pg of ituridg in 2 days, in a wheat bran SSF system, using a wild strain of B. subtilis
116 isolated from compost. More surprising is the comparison of that titre with production reached in SmF: 150 pdml in 5 days (even though a different medium was used since this SSF system is not suitable for comparative experiments). From these figures a relative production of 24.4 can be calculated. This results support our earlier conclusion: that strains closer to the wild-type tend to show a higher relative production. The characteristics that allow good performance in SSF are not known, but this information would provide a deep understanding of microbial adaptation to solid medium. From an applied point of view, this knowledge would be very helpful to design more direct rational selection methods as well as genetic engineering strategies for genetic improvement of these strains. In studies on pectinases production by SSF on coffee pulp, Antier an co-workers [loll found that the use of low water activity as a selective factor, for mutants resistant to 2-deoxiglucose (2DG), seems to select a special kind of mutants named dgrAw96. These are pectinase hyperproducing mutants for SSF (228 U/g dry pulp). Mutant strains that overproduced pectinases in SmF were also obtained using a similar selective medium (pectin + 2DG) but with high Aw (0.999). The fact that the hyperproducing mutants for SSF show a rapid and vigorous growth in SSF attracts our attention. As mentioned before, in secondary metabolism this characteristic, although desirable, is not usually associated to high production. At the present time it is not clear if this difference is due to the fact that it is a catabolic enzyme or because the SSF system used is a one phase support-substrate. Since the support is also the substrate, during the course of the fermentation Aw decreases sharply, since available water is decreasing while sugars are constantly dissolving in it. In the system of support with absorbed medium, available water is probably increasing while sugars and other nutrients concentration decreases. In this way the characteristics needed in a strain could be different for different SSF systems. It can be speculated that in SSF the microorganism senses the solid environment and reacts producing special enzymes that are more efficient in a medium with restricted diffusion. It is possible that membrane structure varies also to adapt to a medium with an air-water-support interface, with different conditions for nutrient uptake and product secretion. Constant protection against osmotic pressure and ionic toxicity might be another particular feature of SSF. All this suggests the existence of some regulatory mechanisms of solid medium adaptive genes (Fig. 12).
Concluding remarks SSF could become an alternative production method for antibiotics, growth promoters and other secondary metabolites. It is in the production of such high value products that advances in SSF are most likely to be made. Only from the economic returns from such products will incentives and resources be available to carry out research and development to solve problems associated with SSF (Mitchell and Lonsane, 1992).
117
However, without a deeper knowledge of the nature of secondary metabolism in solid medium it will not be possible to obtain the systems full production potential. Moreover, there will be little foundation from which process engineers can design and build large-scale systems. Research performed over the last 8 years indicates that different kind of secondary metabolites can be produced by SSF: antibiotics, phytohormones, food grade pigments, alkaloids, etc., using fungi, actinomycetes and Bacillus. Very high concentrations have been obtained in SSF on inert support and on wheat bran, although other agricultural products or byproducts can also be used. Moisture content of the solid medium stands out as a key factor in secondary metabolites production. Experiments in SSF on support indicate that control of idiophase can be achieved by manipulating the support content of the solid medium. This operation appears to control metabolic activity (growth rate) during production phase. Hopefully, findings in this system will eventually be applied in other SSF systems. Basic studies on nutritional factors have shown that physiology in SSF has several similarities with the known behavior in liquid medium. Hence, similar strategies must be adapted to avoid regulation by carbon, nitrogen or phosphate. High production in SSF also requires supplementation with an adequate carbon source as well as with inducers and precursors. It has been found that special strains are needed for SSF processes, but that these strains can be obtained by modifying the overproducing mutants used in SmF. Moreover, the characterization of these strains could provide valuable information about the particularities of the microbial adaptation to a solid environment. Further research is needed to asses ways in which the productivity of SSF may be increased. Areas deserving attention include: the use of fed-batch or continuous plugflow modes of operation. Analysis of performance of SSF bioreactors to identify criteria for successful scale-up and to permit effective process control. The latter is particularly important given the lengthy nature of these cultures (Johns, 1992).
References 1. 2. 3. 4. 5. 6. 7. 8.
Hahn-Hagerdal B. Water activity: A possible external regulator in biotechnical processes. Enzyme Microb 1986;8:322-327. Steinkraus KH. Solid-state (solid substrate) food/beverage fermentations involving fungi. Acta Biotechnol 1984;4:83-88. Mitchell DA, Lonsane BK. In:Doelle HW, Mitchell DA, Rolz CE (eds) Solid Substrate Cultivation. London and New York: Elsevier, 1992;l-16. Barrios-GonzAlez J, Tomasini A, Viniegra-GonzAlezG , Lopez L. Penicillin production by solid-state fermentation. Biotechnol Lett 1988;10:793-798. Barrios-Gonziilez J, Castillo TE, Mejia A. Development of high penicillin producing strains for solid-state fermentation. Biotechnol Adv 1993;11:53+547. Lin TF, Iizuka H. Production of extracellular pigment by a mutant of Monascus kaoliung spp. Nov Appl Environ Microbiol 1982;43:671-676. Kumar PKR, Lonsane BK. Gibberellic acid by solid-state fermentation: consistent and improved yields. Biotechnol Bioeng 1987;30:267-27 1. Hesseltine CW. Solid-state fermentation, Part 2. Proc Biochem 1977;12:2%32.
118 9. Lonsane BK, Ghildyal NP, Budiatman S, Ramakrishna SV. Engineering aspects of solid-state fermentation. Enzyme Microbiol Techno1 1985;7:258-265. 10. Mitchell DA. In: Doelle HW, Mitchell DA, Rolz CE (eds) Solid Substrate Cultivation. London and New York: Elsevier, 1992;87-114. 11. Mitchell DA, Lonsane BK, Durand A, Renaud R, Almanza S, Maratray J, Desgranges C, Crooke K, Hong K, Tanner RD, Malaney GW. In: Doelle HW, Mitchell DA, Rolz CE (eds) Solid Substrate Cultivation. London and New York: Elsevier, 1992;115-140. 12. Mudgett RE. In: Demain AL, Solomon NA (eds) Manual of Industrial Microbiology and Biotechnology. Washington: American Society For Microbiology, 1986;66-84. 13. Rose AH. In: Rose AH (ed) Economic Microbiology Secondary Products of Metabolism. New York: Academic Press, 1979;2-32. 14. Trilli A. In: Poole R, Bazin M, Keevil W (eds) Microbial Growth Dynamics 28. London: Irl. Press, 1990;103-1 26. 15. Demain AL. In: Demain AL, Solomon NA (eds) Antibiotics Containing the b-Lactam Structure. Berlin: Springer Verlag, 1983;189-228. 16. Bu’Lock JD. Essays in Biosynthesis and Microbial Development. “Squibb Lectures”. New York: John Wiley and Sons, 1967. 17. Demain AL. Regulatory mechanisms and the industrial production of microbial metabolites. Lloydia 1968;31 :395-418. 18. Demain AL. Cellular and environmentalfactors affecting the synthesis of metabolites. J Appl Chem Biotech 1972;22:345-362. 19. Bu’Lock JD. Intermediary metabolism and antibiotic synthesis. Adv Appl Microbiol 1961;3:293342. 20. Smith JE, Berry DR. An Introduction to Biochemistry of Fungal Development. London, New York: Academic Press, 1976. 21. Bu’Lock JD. In: Smith JE, Berry DR (eds) “The Filamentous Fungi” 1. Industrial Mycology. London: Edward Arnold, 1974. 22. Demain AL. In: Wang DIC, Cooney CL, Demain AL, Dunnill P, Humphrey AE, Lilly MD (eds) Fermentation and Enzyme Technology. New York, Chichester, Brisbane, Toronto: John Wily & Sons, 1978;2636. 23. Martin JF, Demain AL. Control of antibiotic biosynthesis. Microbiol Rev 1980;44:230-251. 24. Underkofler LA, Severson GM, Goering KJ, Christensen LM. Commercial production and use of mold bran. Cereal Chem 1947;24:1-22. 25. Hesseltine CW. Solid-state fermentation. Biotechnol Bioeng 1972;14517-532. 26. Hesseltine CW. Solid-state fermentation, Part 1. Proc Biochem 1977;12:24-27. 27. Raimbault M, Roussos S, Barrios-Gonzikz J, Gutitrerez M, Viniegra G. Procedet de culture de microorganisms sur un milleu solide constitute de d’un support solide absorbant, compresibleet non fermentable. French patent No. 26471.19. 1989. 28. Solis-Pereira S, Favela-Torres E, Viniegra-Gonzalez G, Gutierrez-Rojas M. Appl Microbiol Biotechnol 1993;39:3641. . 29. Pandey A. Recent process developments in solid-state fermentation. Proc Biochem 1992;27:109117. 30. Larroche C, Gros JB. Strategies for spore production by Penicillium roquefortii using solid-state fermentation techniques. Proc Biochem 1989;97- 103. 31. Raimbault M. Fermentation en milieu solide, croissance de champignons filamenteux sur substrat amylact. These doct. es Sci., Universitt Paul Sabatier, Toulouse, France, 1980291. 32. Saucedo-Castaiieda G. Contribuci6n al estudio de la fermentacibn s6lida: enriquecimientoproteico y conservaci6n por ensilaje, Tesis Maestria en Ciencias Ingenieria Quimica, Universidad Aut6noma Metropolitana, Iztapalapa, Mtxico, 1987;169. 33. Shchez JE.Etude de la fermentation du cacao “essai de deux techniques nouvelles” thbse ing Doct., Ecole National Supereure Agronomique de Montpellier France, 1983;163. 34. Gonzhlez-Blanco P, Saucedo-Castaiieda G, Viniegra-Gonzhlez G. Protein enrichment of sugar cane
119 35.
36.
37.
38.
39. 40. 41. 42. 43.
44. 45. 46. 47. 48. 49. 50.
51. 52. 53. 54. 55.
56. 57.
by-products using solid-state cultures of Aspergillus terreus. J Ferment Technol 199O;70:351-354. Oriol E. Croissance de Aspergillus niger sur milieu solide: importance de I’eau et de I’activit6 de I’eau. These de Doctorat Microbiologie, Institut National Sciences AppliquCes de Toulouse, France, 1987;133. Fukushima D. Koji as an important source of enzymes in the orient and its unique composite systems of proteinases and peptidases. Use of enzymes in food technology. Versailles: C.N.E.R.N.A. (C.N.R.S.), 1982. Roussos S. Croissance de Trichoderma harzianurn par fermentation en milieu solide: physiologie, sporulation et production de cellulases. These Doct. es Sci., Universitd de Provence, France, 1985;193. Dufour D. Production de pectinases dans un milieu adsorb6 sur bagasse, contribution a l’btude de la physiologie des champignons pectolytiques cultiv6s en milieu solide, en relation avec la respiration et la synthese de pectinases. These Doctorat, Universit6 De Technologie De Compiegne, 199Q262. Hang YD, Woodams EE. Effect of substrate moisture content on fungal production of citric acid in a solid-state fermentation system. Biotechnol Lett 1987;9:183-186. Gibbons WR, Wetsby CA, Demain AE. Technology and economics of ethanol production from fodder beets via solid phase fermentation. Biotechnol Lett 1988;10:665-670. Larroche C, Desfarges C, Gros JB.Spore production of Penicilium roquefortii by simulated solidstate fermentation. Biotechnol. Letters 1986;8:453-456. Aidoo KE, Hendry R, Wood JB. Solid substrate fermentation. Adv Appl Microbiol 1982;28:201237. Saucedo-Castaiieda JG, Gonzilez P, Revah S , Viniegra-Gonzilez G, Raimbault M. Effect of lactobacilli inoculation on cassava (Manihor esculenta) silage: fermentation pattern and kinetic analysis. J Sci Food Agricul 1990;50:467-477. Zadrazil F. Influence of CO, concentration the mycelium growth of three Pleurotus species. Eur J Appl Microbiol Biotechnol 1975;1:327-335. Hesseltine CW. Solid-state fermentation: an overview. Biodeterioration 1987;23:79-89. Kumar PKR, Lonsane BK. Solid-state fermentation: physical and nutritional factors influencing gibberellic acid production. Appl Microbiol Biotechnol 1990,34:145-148. Johns MR. In: Doelle HW, Mitchell DA, Rolz CE (eds) Solid Substrate Cultivation. London and New York: Elsevier, 1992341-352. Dumenil GS. In: Larpent JP, Sanglier JJ (eds) Biotechnologie des Antibiotiques. Paris: Masson, 1989;195-2 17. Alazard D, Raimbault M. Comparative study of amylolytic enzymes production by Aspergillus niger in liquid and solid-state cultivation. Euro J Appl Microbiol 1981;12:113-117. AcuAa-Arguelles ME, Guti6rrez-Rojas M, Viniegra-Gonzilez G, Favela-Torres E. Production and properties of three pectinolytic activities produced by A. niger by submerged and solid-state fermentation. 1995;(In press). Lehka PK, Lonsane BK. Comparative titres location and properties of tannin acyl hydrolase produced by PKL 104 in solid-state surface and sumerged cultures. Proc Biochem 1994;29:497-503. Smiley KL, Hensley DE, Gasdorf HJ. Alpha-galactosidase production and use in a hollow-fibre reactor. Appl Environ Microbiol 1976;31:615. Moo-Young M, Moreira AR, Tengerdy RP. In: Smith JFi, Beny DR, Kristiansen B (eds) The Filamentous Fungi. Fungal Technology. London: Edward Arnold Publishers, 1983;4 117-144. Righelato RC. In: Smith JE, David DR (eds) Filamentous Fungi 1. London: Edward Arnold, 197379- 103. Oriol E, Raimbault M, Roussos S, Viniegra-GonzBlez G. Water and water activity in the solid-state fermentation of cassava starch by Aspergillus niger. Appl Microbiol Biotechnol 1988;27:498-503. Grajek W. Cooling aspects of solid-state cultures of mesophilic and thermophilic fungi. J Ferment Technol 1988;66:675-679. Saucedo-Castaiieda G, Guterrez-Rojas M, Bacquet G, Raimbault M, Viniegra-GonzBlez G. Heat
120 58. 59.
60. 61. 62. 63.
64. 65. 66. 67. 68.
69. 70. 71. 12.
73. 74. 75. 76. 77. 78. 79. 80.
81.
transfer simulation in solid substrate fermentation. Biotechnol Bioeng 1990;35:802--808. Gervais P, Bazeline C. Development of a solid substrate fermentor allowing the control of the substrate water activity. Biotechnol Lett 1986;8:191-196. Bastrow LM, Dale BE, Tengerdy RP. Evaporative moisture and temperature control in solid substrate fermentation. Biotechnol Tech 1988;2:233-236. Durand A, Chereau D. A new pilot reactor for solid-state fermentation: Application to the protein enrichment of sugar beet pulp. Biotechnol Bioeng 1988;31:476-486. Durand A, de la Broise D, Blachkre H. Laboratory scale bioreactor for solid-state processes. J Biotechnol 1988;8:5+66. Saucedo-Castaiieda G, Dufour D, Trejo M, Soccol RC, Roussos S, Raimbault M. On line gaz analysis in solid-state fermentation at fenabio biolatina 91. Sao Paulo, Brazil, 6 1 0 July, 1991. Oriol E, Schettino B, Viniegra-Gonzhlez G, Raimbault M. Solid-state culture of Aspergillus niger on support. J Ferment Techno1 1988;66:57-62. Barrios-Gonzilez J, Martinez C, Aguilera A, Raimbault M. Germinationof concentrated suspensions of spores from Aspergillus niger. Biotechnol Lett 1989;11551-554. Barrios-Gonzllez J, Tomasini A. Production of aflatoxins by solid-state fermentation. J Sci Indust Res 1995;(In press). Barrios-Gonzhlez J, Rodriguez GM, Tomasini A. Enviromental and nutritional factors controlling aflatoxin production in cassava solid-state fermentation. J Ferment Bioeng 1990;70:32+333. Lindenfelser LA, Ciegler A. Solid-substratefermenter for ochratoxin A production. Appl Microbiol 1975;29:323-327. Demain AL, Hunt NA, Malik V, Kobbe B, Hawkins H, Matsuo K, Wogan GN. Improved procedure for production of cytochalasin E and tremorgenic toxins by Aspergillus clavatus. Appl Environ Microbiol 1977;31: 138-140. Luk, Kobbe B, Townsend JM. Production of cyclopiazonic acid by Aspergillus flavus link. Appl Environ Microbiol 1977;33:211-212. Kobbe B, Cushman M, Wogan GN, Demain AL. Production and antibacterial activity of malformin C, a toxic metabolite of Aspergillus niger. Appl Environ Microbiol 1977;33:996997. Kumar PKR, Lonsane BK. Potential of fed-batch culture in solid-state fermentation for production of gibberellic acid. Biotechnol Lett 1987;9:179-182. Agosin E, Maureira M, Biffani V, PCrez F. Production of gibberellins by solid-substratecultivation of Gibberella fujikuroi. I1 International Symposium on Solid-state Fermentation. February 2&27, Montpellier, France, 1995;28. Yang SS, Ling MY. Tetracycline production with sweet potato residue by solid-state fermentation. Biotechnol Bioeng 1989;33:1021-1028. Jermini MFG, Demain AL. Solid-state fermentation for cephalosporin production by Streptomyces clavuligerus and cephalosporium acremonium. Experientia 1989;45:1061-1065. Ohno A, Ano T, Shoda M. Production of antifungal antibiotic, iturin in a solid-state fermentation by bacillus subtilis nb22 using wheat bran as a substrate. Biotechnol Lett 199214:817-822. Trejo MR. These de Doctorat Biologie Cellulaire-Microbiologie, UniversitC de Provence, AixMarseille I, France, 1992. Le6n R. In: Raimbault M (ed) Solid-state Fermentation in Bioconversion of Agricultural Raw Materials. France: Orstom, '1989; 139-143. Serrano-Carre6n L, Hathout Y, Bensousan M, Belin JM. A simple protocol for extraction and detection of pyrazines produced by fungi in solid-state fermentation.Biotechnol Tech 19926:1+22. Segura M, Bautista ME, Lug0 G, Garcia 0. Producci6n de aureomicina por fermentacidn s6lida con Streptomyces aureofaciens. Biotecnologia 1993;FSeFS12. Acuiia-Arguelles M, Gutierrez-Rojas M, Viniegra-Gonzilez G, Favela-Torres E. Effect of water activity on exopectinaseproduction by Aspergillus niger CH4 on solid-statefermentation.Biotechnol Lett 1993;16:23-28. Lotong N, Suwanarit P. Fermentation of ang-kak in plastic bags and regulation of pigmentation by initial moisture content. J Appl Bacteriol 1990;68:565-570.
121 82. Johns MR, Stuart DM. Production of pigments by monascus purpureus in solid culture. J Indust Microbiol 1991;8:23-28. 83. Blanc PJ, Loret MO, Goma G. Pigments and citrinin production during cultures of Monascus on rice and liquid media. n International Symposium on Solid-state Fermentation. Febrary 26-27, Montpellier, France, 1995;26. 84. Stato K, Nagatani M, Nakamura K, Sat0 S. Growth estimation of Candida lipolifica from oxigen uptake in solid-state fermentation with forced aeration. J Ferment Technol 1983;61:623-629. 85. Mejia A, Tomasini A, Barrios-Gonzhlez J. On-line measurement of oxygen uptake and carbon dioxide evolution in solid-state fermentation: Application in the study of secondary metabolism. I1 International Symposium on Solid-state Fermentation. Febrary 26-27, Montpellier, France, 1995. 86. Barrios-Gonzilez J, Gonzilez H, Mejia A. Effect of particle size, packing density and agitation on penicillin production in solid-state fermentation. Biotechnol Adv 1993;11525-537. 87. Mudget RE, Nash J, Rufner R. Controlled gas environments in solid substrate fermentation. Devel Indust Microbiol 1982;23:397-405. 88. Han YW, Mudgett RE. Effects of oxygen and carbon dioxide partial pressures on Monascus growth and pigment production in solid-state fermentations. Biotechnol Prog 1992;8:5-10. 89. Silman RW, Conway HF, Anderson RA, Bagley EB. Production of aflatoxin in corn by a large scale solid substrate fermentation process. Biotechnol Bioeng 1979;21:1799-1808. 90. D u r h J, Mejia A, Barrios-Gonzilez J. Efecto del mezclado sobre el creciemiento y la produccibn de penicilina de Pxhrysogenum P-2. Biotechnol 1993;FS2&FS28. 91. Kumar PKR, Lonsane BK. Batch and feed-batch solid-state fermentations: kinetics of cell growth, hydrolytic enzyme production and gibberellic acid production. Proc Biochem 1988;23(2):43-47. 92. Revilla G, Ramos FR, Lbpez-Nieto JM, Alvarez E, Martin JF. Glucose represses formation of d-(La-aminoadipy1)-L-cysteinyl-D-valineand isopenicillin N synthetase but not of penicillin acyltransferase in P. chrysogenum. J Bacteriol 1986;168:947-952. 93. Ramesh MV, Lonsane BK. Regulation of alpha-amylase production in Bacillus licheniformis M27 by enzyme endproducts in submerged fermentation and its overcoming in solid-state system. Biotechnol Lett 1991;13(5):355-360. 94. Ramesh MV, Lonsane BK. Ability of a solid-state fermentation technique to significantly minimize catabolic repression of a-amylase production by Bacillus licheniformis M27. Appl Microbiol Biotechnol 1991;607-609. 95. Garcia BE, Bamos-GonzAlez J, Mejia A. Regulation of penicillin biosynthesis by glucose and ammonium in solid-state fermentation.9th International Biotechnology Symposium, August 18-20, Crystal City, VA, USA, 1993. 96. Flores V, Mejia A, Barrios-Gonzilez J. Feedback regulation of penicillin biosynthesis in solid-state fermentation: thresholds in high and low producing mutants. 7th International Symposium on the Genetics of Industrial Microorganisms. June 27 - July 1, Montreal, Canada, 1994. 97. Ball C. In: Huter R, Leisinger T, Nuesch J, Wehrli W (eds) Antibiotics and Other Secondary Metabolites; Biosynthesis and Production. FEMS Symp. No. 5. London: Academic Press, 1987;165- 176. 98. Schwab H. Strain improvement in industrial microorganisms by recombinant dna techniques. Adv Biochem Eng Biotechnol 1988;37:129-164. 99. Ghildyal NP,Lonsane BK, Sreekantiah KR, Murty VS. Economics of submerged and solid-state fermentation for the production of amyloglucosidase. J Food Sci Technol 198522: 171-176. loo. ShankaranandaVS, Ramesh MV, Lonsane BK. Idiosyncrasiesof solid-state fermentation system in the biosynthesis of metabolites by some bacterial and fungal cultures. Proc Biochem 1992;27:33-36. 101 Antier P, Minjares A, Roussos S, Raimbault M, Viniegra-GonzBlez G. Pectinase-hyperproducing mutants of Aspergillus niger C28B25 for solid-state fermentation of coffee pulp. Enzyme Microbiol Technol 1993;15:254-260. 102. Rehacek Z, Sad1 P. Ergot Alkaloids. Amsterdam: Elsevier, 1990.
This Page Intentionally Left Blank
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R.El-Gewely, editor.
123
Genetics of lactobacilli in food fermentations Rudi F. Vogel and Matthias Ehrmann Lehrstuhlfur Technische Mikrobiologie, Technische Universitat Munchen, 85350 Freising-Weihenstephan, Germany
Abstract. Lactobacilli play a substantial role in food biotechnology and influence our quality of life by their fermentative and probiotic properties. Despite their obvious importance in fermentation ecology and biotechnology only recent years have brought some insight into the genetics of lactobacilli. These genetic investigations allow the elucidation of traits determinative for competitiveness and ecology and thus product safety and quality. They have concentrated only on a small selection of lactobacilli whereas others are hardly touched or remained recalcitrant to genetic analysis and manipulation. The knowledge gained on the biochemistry, physiology, ecology and especially genetics is a prerequisite for the deliberate application and improved handling of lactobacilli in traditional and novel applications. In this review, the achievements in the genetics of lactobacilli are described including detection systems, genetic elements, host vector systems, gene cloning and expression and risk assessment of genetically engineered lactobacilli.
Key words: biosafety expression, food fermentation, gene cloning, genetics, identification, lactic acid bacteria, Lactobacillus, plasmid vector, starter organism, transformation.
Introduction
'
The genus Lactobacillus comprises 56 species which are associated with plants or material of plant origin and often are abundant in man-made habitats [l]. They are predominant whenever rich, carbohydrate containing substrates are available as in mucosal membranes of man and animal (oral cavity, intestine and vagina), in manure sewage and fermenting or spoiling food [2]. They influence our quality of life through their probiotic and metabolic activities which are exploited in their deliberate use in the biotechnical processes of food fermentation. Lactobacilli are intrinsically safe and none of the species described exhibits a toxigenic or pathogenic potential, although selected strains of some species were found to produce biogenic amines [3] or were isolated from deceased individuals without being able to reinfect other individuals [4].Application of lactobacilli since ancient times of food biotechnology and their association with man and animal ensures that they still represent a pool of safe organisms for biotechnical purposes. Research on the genetics of lactic acid bacteria is most developed with dairy organisms namely Lactococcus lactis [5]. On the other hand the application of lactococci is restricted to the dairy field whereas all other lactic food fermentations are dominated by lactobacilli. Despite this substantial role in ecology and biotechnology, only recent years have brought some insight into the genetics of lactobacilli. These genetic investigations have concentrated only on
124
a small selection of lactobacilli whereas others are hardly touched or remained recalcitrant to genetic analysis and manipulation. The variety of lactobacilli adapted to different habitats represent a pool for comparative studies which allow the elucidation of traits determinative for competitiveness and ecology, and thus product safety and quality. The knowledge gained in the studies reviewed here and in forthcoming developments on the biochemistry, physiology, ecology and especially genetics is a prerequisite for the deliberate application and improved handling of lactobacilli in traditional and novel applications.
Lactobacilli in food fermentations Lactobacilli are involved in many food fermentations which are listed in Table 1. Some species are frequently found and may be predominant in selected fermentations, e.g., L. curvatus and L. sake in sausage fermentation, L. sanfrancisco in sourdough or L. kefir and L. kejiranofaciens in Kefn. Others like L. plantarum have enhanced metabolic capabilities and may participate or even dominate fermentations with different ecological measures, e.g., sausage vs. olive fermentation. L. brevis seems to be a nearly ubiquitous contaminant although recent use of molecular taxonomy has demonstrated that this may be partly due to misidentification [6]. r
Prospects in the use of Lactobacillus genetics Industrial scale production of fermented foods which are hygienically safe and of reproducible quality while maintaining a big product variety, requires knowledge on the metabolism, ecology and genetics of the lactobacilli involved. Up to now the development of starter organisms was merely achieved by screening followed by trail and error pilot scale experiments. Figure 1 demonstrates the use of metabolic studies and genetics in an iterative process for the improvement of lactobacilli for food fermentations. For such a strategy, which in principle can be applied to other microorganisms in biotechnology, basic knowledge on the genetic elements of lactobacilli, genes encoding special properties and tools for the detection and modification, are required. Genetical methods enable us to detect and characterize lactobacilli which are suitable and competitive in specific fermentation processes. Their intrinsic as well as special properties can be investigated on a molecular basis, and genes encoding specific traits can be transferred to starter lactobacilli which might exhibit improved performance or affect the quality of a product in a desired way. With such recombinant organisms furthermore the effect of a single trait on the behaviour of a starter organism in a fermentation can be investigated. This knowledge can be used to screen for suitable natural organisms or direct application of recombinant lactobacilli in food fermentation. Major targets of such a genetic modification are listed in Table 2. In the following, the achievements made in basic understanding of Lactobacillus genetics
125 Table 1. Lactobacilli in food fermentations. Food
Raw material
Lactobacillus (others)”
Acidophilus milk
milk
L. acidophilus
Cheese
milk curd
Lactobacillus spp. (other lactic acid bacteria namely Lactococcus)
Coffee
coffee beans
Lactobacillus spp. (other lactic bacteria, yeasts)
Fermented sausage
meat
L. sake, L. curvatus, L. plantarum
Izushi
fresh fish, rice, vegetables
Lactobacillus spp.
Kefir
milk
L. kefr, L. kefiranofaciens (yeasts)
Kumiss
raw mare’s milk
L. delbrueckii subsp. bulgaricus, L. “leichmiannii”
Olives
olives
L. plantarum (Leuconostoc spp., Pediococcus spp., yeasts)
Sauerkraut
cabbage
L. curvatus, L. sake (Leuconostoc mesenteroides)
Soja sauce
rice, wheat, soja beans
Lactobacillus spp. (Aspergillus oryzae, yeasts)
Sourdough
wheat or rye flour
L. sanfrancisco, L. pontis, L. fructivorans, L. fermentum (Candida milleri)
Sour wort for beer production
wort
L. amylovorus
Vegetable juice
carrots, cabbage
L. casei, L. sake
Yakult
milk
L. casei strain Shirota
Yoghurt
milk
L. delbrueckii subsp. bulgaricus (Streptococcus thermophilus)
“Species other than Lactobacillus are given in parentheses and may even be predominant in the respective fermentation process.
and regulation, the developments of tools for identification and targeted genetic alterations, are discussed. The application of this knowledge and tools for the improvement of lactobacilli and understanding of their behaviour and metabolism in food fermentations is highlighted.
Genetic tools for identification and classification of food lactobacilli Modem food technology requires fast detection and specific identification systems for food-bome pathogens or spoilage organisms, as well as for rapid monitoring of starter organisms in fermentation processes. Furthermore, there is a strong demand of
126
- removal of undesired
- hygiene baderiocins - sensorial properties
properties
catalase lipase, proteinase amylase, pektinase - competitiveness bacteriocins - probiosis colonisation acid tolerance
I
improvement of technology use of new strains modelling and prediction Fig. I. Development of starter organisms for food fermentations.
companies to identify their strains in fermentations and follow the development of microbial flora, especially in nondairy fermentation where no pasteurization of the raw materials can be performed, and the producer only relies on the competitiveness of the starter organisms. Due to their need of cultivation steps, conventional methods based on biochemical and physiological tests are often time consuming and give ambiguous results. This is also hampering to the identification and evaluation of new suitable starters (from different environments) with the required key metabolic features, which is time consuming when using conventional biochemical screening methods (compare Fig. 1). In addition to identification and monitoring, the phylogenetic relationship of lactobacilli is hardly characterized by biochemical data alone, nor necessarily reflected by their metabolic traits or traditional subgrouping [l]. The expression of genes can be easily affected by environmental factors, and so genetic methods have an inherent advantage not 'only for reliable identification systems, but also comprehensive taxonomic grouping. In food microbiology genetic methods are no more restricted to diagnostics concerned with hygiene. Modem taxonomic tools based on PCR techniques and DNA hybridization methods, are finding increasing applications in detection of foodspoiling and fermentation organisms [7,8]. A variety of genetic methods are successfully applied for detection and differentiation of lactobacilli.
127 Table 2. Discernible aims in the genetic engineering of starter organisms for food fermentations. 1. Reduction of hygienical risks e.g., antagonism to food poisoning micro-organisms, removal of toxins originating from microorganisms or raw materials
2. Enhancement of the nutritional value e.g., enrichment in amino acids and vitamins
3. Production following ecological requirements e.g., energy savings, exploitation of new resources including residual wastes 4. Enhancement of process safety e.g., resistance to bacteriophage, stabilisation of plasmid encoded traits by chromosomal integration, enhancement of the competitiveness
5. Simplification of the microbial action e.g., combination of traits in one organisms as degradation of malate in yeasts, reduction of nitrate in lactic acid bacteria
6. Improvement of the ecological adaptation e.g., killer factors, bacteriocins 7. Economical production e.g., reduction of process time
8. Improvement of efficiency e.g., introduction of new traits including enhanced sugar fermentation capability, prototrophy, stronger or new flavour compounds, improved colour 9. New products e.g., lite beer, fruit and vegetable juices, metabolites and enzymes produced by intrinsically safe organisms
Hybridization techniques A nucleic acid probe used for hybridization is a fragment of a single-stranded nucleic acid, which specifically hybridizes to a complementary region of a target nucleic acid. The target can be DNA or RNA whereby randomly selected fragments or specific genes are selected. Also total DNA can be extracted and applied as gene probe as demonstrated for the identification of lactobacilli, commonly found in musts and wines [9]. In the last years rRNA genes proved to be one of the most suitable target genes for bacterial taxonomy and identification because of their incontestable advantages regarding their ubiquitous distribution and very broad specificity ranging [ 101. Moreover, the sensitivity of the test systems is considerably increased because in exponentially growing cells, at least 10.000 more rRNA molecules are present. Because of these advantages, rRNA targeted oligonucleotide probes were developed during the last few years for a couple of different Lactobacillus species. These probes are summarized in Table 3. They are targeted against 16s or 23s rRNA and usually possess species or group specificity. Using these probes it is also possible to identify many organisms present in one sample simultaneously by using reverse dot blot
128 Table 3. rRNA-targeted oligonucleotide probes useful for identification of LAB at the species level. probe
Target
Specificity
Lba
23s
L. acidophilus
Lbam
16s
L. amylovorus
Lbb
16s
L. brevis
Lbc
23s
L. curvatus
Lbco
16s
L. collinoides
Lbcp
23s
L. crispatus
Lbcr
23s
L. caseilrhamnosus
Lbd
23s
L. delbrueckii
Lbfe
16s
L. fermentum
Lbfr
16s
L. fructivoranslhomohiochii
Lbg
23s
L. gasseri
Lbh
23s
L. helveticus
Lbhi
16s
L. hilgardii
Lbj
23s
L. johnsonii
Lbk
16s
L. kefir
Lbkf
23s
L. kefranofaciens
Lbl
16s
L. lindneri
LbP
23s
L. pentosuslplantarum
Lbpa
23s
L. paracasei
Lbpb
23s
L. parabuchneri
LbPe
16s
L. pentosuslplantarum
Lbpo
16s
L. pontis
Lbre
16s
L. reuteri
Lbru
16s
L. ruminis
Lbs
23s
L. sake
Lbsa
16s
L. sanfrancisco
Reference
hybridization [ 111. In one of our studies we were able to identify misclassified strains and show a taxonomically intermixed situation of L. brevis, L. buchneri, L. parubuchneri and L. hilgurdii strains (unpublished data). The use of rRNA sequencing and development of species-specific probes, may also be useful in the identification and characterization of new taxa, as demonstrated for lactobacilli prevailing in sourdough preparations and description of Lactobacillus pontis [6].
129 PCR analysis Alternatively to their use in hybridization, specific probes can be used as PCR primers. Systems involving PCR-amplification exist for the detection of various food spoiling organisms. As lactobacilli can also be spoiling organisms in (non)-alcoholic beverages as fruit juice, beer or wine, this sensitive method may be helpful for their detection even at low numbers. Nakagawa et al. [ 171 reported the detection of 1 1 alcohol tolerant Lactobacillus strains by amplifying spacer region between 16s and 23s rRNA genes. In this investigation the detection limit was about one cell in 50 ml of artificially contaminated Sake.
Reverse dot blots and capture probes Despite the advantages of the probe and PCR technologies over the conventional differentiation methods, the application of probes for the analysis of food samples is still laborious when individual hybridization experiments must be performed sequentially or in parallel for each probe or primer-pair. This disadvantage is solved by application of the reverse dot blot hybridization technique, which combines both techniques and allows analysis of even complex samples in one single experiment. In this approach a large number of probes with different specificities, is bound to a membrane and used as a capture probe targeted to PCR-amplified rDNA from crude nucleic acid preparations. Ehrmann et al. [13] demonstrated the reliable and simultaneous identification of different lactobacilli in fermented food by using this technique. A typical result of a reverse dot blot analysis is shown in Fig. 2.
Other techniques The largely conserved character of rRNA genes and their spacer regions, extremely seldom allows the design of strain specific probes. However, identification of lactobacilli at the strain level is very important for patent purposes, in the identification of starter strains, and the ecological analysis of artisanal or traditionally fermented food, where no starter is applied. Conventional typing such as serological studies, phage sensitivity or antibiotic resistance, have the disadvantages previously described. Therefore a variety of molecular techniques including plasmid profiles, restriction fragment length polymorphism (RFLP) and random amplified polymorphic DNA (RAPD) have proved as useful tools. Since plasmid profiles have firstly been evaluated for taxonomic purposes within lactobacilli by Nes [18], they were used to monitor the succession of lactobacilli during the fermentation process [19] and in the piglet digestive tract [20]. A limitation of plasmid profiling may be their change upon time as discussed below. For RFLP analysis chromosomal DNA is cleaved with restriction enzymes, subjected to electrophoresis, and the resulting patterns are hybridized with a gene probe. The reduced number of distinct hybridization signals as compared to restriction endonuclease digestion analysis (REA), is easily analyzed by using statistical methods
130
Fig. 2. Identification of lactic acid bacteria in food samples by reverse dot blot hybridization. PCR products derived from total DNA were visualized after hybridization to membrane-boundoligonucleotide probes. (Abbreviations: Lb = Lactobacillus; Stc = Streptococcus)
like Jaccard coefficients and the unweighted pair group algorithm with arithmetic averages (UPGMA) or principal-component analysis (PCA) and soft independent modelling of class analogy (SIMCA). The probes may either be random DNA fragments or targeted against specific genes like rFWA genes (ribotyping). A few examples are given in the following. On the basis of their restriction patterns, the genetically closely related strains of L. helveticus and L. jugurti, exhibiting DNA-DNA hybridization values ranging from 89 to loo%, could be differentiated from one another [21]. Using DNA of Phage M13 as a probe, (M13 DNA fingerprinting) Mitewa et al. [22] were able to show that this method might be a powerful way for taxonomic studies of Lactobacillus subsp. Jeune et al. [23] achieved differentiation of L. brevis and L. hilgardii by ribotyping. Comprehensive studies using REA and ribotyping of many strains of L. plantarum, L. reuteri, L. fermentum were performed in order to clarify their taxonomic position
131 [24-261. For examination of the strain heterogeneity in grass silage plasmid profiling, REA and ribotyping were applied [27]. During the last few years a new taxonomic method called random amplified polymorphic DNA (RAPD) was established and has found broad application in fungal and bacterial taxonomy [28]. For this investigation a PCR experiment with unspecific primers is set up under very relaxed conditions, resulting in characteristic DNA fragment patterns. In a comparative analysis using RAPD and oligonucleotide probing of numerous strains L. brevis, L. buchneri, L. parabuchneri and L. hilgardii, we have evaluated this method regarding its potential for strain and species differentiation. These experiments revealed that differentiation of species and even strains can be possible, whenever these type of results are useful only to one laboratory under specific experimental conditions (unpublished observations). At present a powerful selection of techniques are available for the identification, characterization and monitoring of the microbial flora in fermenting or spoiling food. For choosing the right method it is necessary to realize that there are different potentials of these techniques either for detection and identification on the one hand and differentiation with elucidation of phylogenetic relation of the organisms on the other. The potential of these techniques is summarized in Table 4.
Genetic elements in lactobacilli Plasmids Within Lactobacillus plasmids were found as extra chromosomal elements first in L. casei [29] and subsequently in all lactobacilli investigated. Many strains carry several plasmids ranging in size from 2 to >lo0 kbp [3&33]. The electrophoretic profiles obtained with such plasmids can be used for the identification of starter strains and detection of genetic variations [34]. Most of these plasmids are cryptic without any
Table 4. Genetic methods and their potential for the identification and differentiation in molecular taxonomy of lactobacilli. Method
RNA sequencing RFLP/REA rRNA targeted probes and PCR primers" RAPD
Potential Differentiation
Identification
strain
swcies
strain
-
-k
-
+
-
+ +
+
+
species
Phylogenetic grouping
Detection
+
-
-
-
-
-
+
+
+
+
-b
-b
-
+
+
The potential of other probes and primers depends on the target sequence; bthe majority of strains of one species may, however, fall into one cluster under specific experimental conditions.
132 known function. This is especially true for the small plasmids of 2-4 kbp which might represent suitable candidates for vector construction. Their strong segregational stability which is desired in the latter respect, hampers the construction of plasmidfree host organisms. However, such plasmid free variants can be obtained upon prolonged growth at sublethal temperatures [35]. The small cryptic plasmids exhibit a varying extent of homology within different species as demonstrated for a cryptic plasmid of L. plantarum [36]. The respective investigation of pLc2, a small cryptic plasmid from L. curvatus LTH683 which is used as starter for sausage fermentation, revealed the presence of homologous plasmids in L. subsp. pseudoplantarum, L. curvatus, L. sake, L. alimentarius, L. farciminis and L. halotolerans from meat and L. curvatus and L. sake from sauerkraut. These small plasmids can be used for vector construction (discussed below) and merely promote their own replication [37-391 which was characterized for selected plasmids as a rolling circle mechanism [40], which can be controlled by transcriptional attenuation [41]. Despite their cryptic nature, these small plasmids may encode important functions, e.g., a trans-acting protein which allows replication of truncated plasmids [37]. The finding of mob and pre genes in some of these plasmids could allow plasmid exchange and explain the presence of homologous plasmids in Lactobacillus strains sharing a habitat [31,32]. Large plasmids in lactobacilli were found to carry genes encoding a variety of metabolic traits which are listed in Table 5 . Nevertheless, most of the properties
Table 5 . Plasmid encoded properties of lactobacilli.
Property
Lactobacillus species
Assimilation of cysteine
L. sake
Bacteriocin production
L. acidophilus L. curvatus L. sake
Fermentation of lactose
L. casei L. plantarum
Fermentation of maltose
Lactobacillus spp.
Fermentation of N-acetyl-glucosamine
L. helveticus
Formation of dextran
Lactobacillus spp.
Formation of protease
L. helveticus
Resistance to antibiotics
L. acidophilus L. bulgaricus L. fermenturn L. helveticus L. reuteri L. plantarum
Slow acid formation in milk
L. helveticus
Reference
133 encoded by such plasmids remain to be elucidated. In many cases the properties encoded include properties of practical interest, e.g., bacteriocin production which may be transferred to other lactobacilli to construct new starter strains with the desired properties [42,43]. Some plasmids can be transferred between lactobacilli and also from other organisms into lactobacilli by conjugation. This has been used to investigate and develop natural gene transfer systems which are discussed below. Bacteriophages Bacteriophages are known as potent tools for genetic studies and modification of bacteria. On the other hand, they are known to attack lactococci and acetic acid bacteria, deliberately used in cheese and vinegar fermentations, respectively, and even induce complete failure of such processes. For lactobacilli, such severe implications have not been proved in pilot scale fermentations or challenge tests. Nevertheless, bacteriophages were described which are specific for lactobacilli in food fermentations. For a review see [44]. More recently, bacteriophages were also described for lactobacilli from meat [45] and sourdough fermentations [46], as well as the rumen of cows [47]. The genetic organization of these bacteriophages is hardly investigated whenever some genes encoding structural proteins have been characterized [48,49]. Some attempts have been made to use these bacteriophages in gene transfer systems. i.e., transduction [50] or transfection [51]. At present there are no effective systems for the use of these bacteriophages in gene transfer systems in lactobacilli. Insertion elements Insertion elements may cause genetic instability in bacteria leading to metabolic switches and (in)activation of genes upon their move and integration at a new site in the bacterial genome. On the other hand, they might be used to deliver genes into bacterial chromosomes or construct mutants and subsequently isolate the respective genes. IS-elements have been characterized for L. casei [68] and L. delbrueckii subsp. bulgaricus [69]. The latter IS-element caused spontaneous deletion formation within the P-gal gene of L. delbrueckii subsp. bulgaricus NCD01489. Similarly, the production of lactocin S, a bacteriocin from L. sake, was abolished by insertion of IS1 163, a member of the IS3 family [70]. In genes of L. sake encoding the catalase and P-galactosidase, we have detected the presence of multiple direct and inverted repeats of 9 bp or more and observed homologous silent P-galactosidase genes. A speculation that this could be the result of transposition would be corroborated by the observation of genetical instability of traits found in fresh isolates but lost upon propagation under laboratory conditions (unpublished observations). The use of transposons from Enterococcus faecalis to induce mutations in L. curvatus and L. sake [71], revealed that insertion of the transposons used was random, however, the frequency of integration was too low for practical application.
134
Development of host vector systems One of the basic requirements for the transfer of genetic material is the availability of suitable vectors as well as efficient in vitro transfer systems. Because major efforts in genetic research has been done for lactococci [5], many in vitro transfer systems currently applied for lactobacilli are vectors originally constructed for the use in lactococci. Conjugation Intergeneric and intrageneric conjugation has been found in lactobacilli. For intergeneric plasmid transfer, mostly broad host range plasmids like pAMP1 [66] and pIP501 [72] have been used. Intrageneric exchange has been described for lactose plasmids or bacteriocin encoding plasmids [42,73]. We have demonstrated the conjugal transfer of plasmid pAMP1 from Enterococcus faecalis to L. curvatus, L. pentosus and L. plantarum (unpublished). Whenever the transfer rates were low with all lactobacilli the potential of natural gene transfer with lactobacilli was demonstrated. In addition to the results obtained in in vitro experiments, it was demonstrated that conjugal plasmid transfer among lactic acid bacteria is possible in natural ecosystems, e.g., in fermenting sausages [74] or during cheese making [75]. Mobilization Mobilization is a method for gene transfer which, to date, has not been thoroughly examined. Under special conditions conjugative plasmids have the ability to transfer (mobilize) other resident non conjugative plasmids. After successful transfer omission of selective pressure causes the loss of the conjugative plasmid. Kozlowa et al. [76] demonstrated the successful mobilization of plasmid PUB 102-4 encoding endoglucanase into Lactobacillus ‘‘jiermenti” using Bacillus thuringiensis carrying the mobilizing plasmid pAMP1 and the donor strain Bacillus subtilis. Furthermore, a sequence coding for a putative mobilization protein and its corresponding RSA site, has been found in plasmid pLABlOOO of L. hilgardii [32]. These findings demonstrate that mobilization and conjugation provide the possibility for genetic modification of lactobacilli by using natural gene transfer systems. Transduction A drawback for the use of conjugation and mobilization is the requirement of special plasmids fitting to given strains. Similar limitation hold true for other methods like protoplast fusion, transformation and transduction, which are of minor importance because of their restriction to few species and their low efficiency [77-801. Transducting L. gasseri with phage phi.adh, Raya and Klaenhammer [80] were able to increase the transduction frequencies of a recombinant plasmid 102-105-fold as compared to the native plasmid. The increase in frequency generally corresponded with the extent of DNA-DNA homology between plasmid and phage DNAs [go].
135 Electroporation An essential improvement for the genetics of lactobacilli was the transfer of foreign DNA into these organisms by electroporation. The possibility of using intact cells avoids the problems arising from protoplast fusion and regeneration. In this procedure the cells are exposed to a current impulse of very defined parameters which permeabilizes the cell membrane allowing DNA uptake. Successful transfer depends on factors like restriction/modification systems, plasmid incompatibility and the ability to replicate and express the foreign DNA. Since the first electroporation of lactic acid bacteria by Harlander [81], numerous other lactic acid bacteria including lactobacilli have been successfully transformed [82-861. Obtainable transfer rates range from I to lo7 transformants/pg DNA. Transformation frequencies of lo7 transformants per pg DNA were routinely obtained only with L. reuteri using a native L. reuteri plasmid and its derivatives [87]. Typical transformation rates are rather lo2 to lo3, unless highly transformable strains of a species are available. A remarkable fact is that the optimal conditions depending on numerous parameters, are highly strain specific and have to be defined case by case. We have developed electroporation for typical sausage starter organisms like L. curvatus and L. sake [88], which was adopted for the use in different strains by these species [89]. Plasmid vectors These electroporation-based transformation systems can be used to introduce genes encoding new properties into lactobacilli with any vector naturally exhibiting an appropriate host range, or with vectors constructed on the basis of Lactobacillus plasmids which are used to transform the respective species. In addition to Lactobacillus vectors replicating only in few species, it has been proved that some of these plasmids possess a wide host range within lactobacilli and therefore they can be applied to different species [32,90]. Usually a vector consists of a small cryptic plasmid carrying a Lactobacillus replicon and a reporter gene typically encoding resistance to antibiotics e.g., erythromycin or chloramphenicol. The minimal size of Lactobacillus replicons needed for a stable maintenance was determined as 2.4 to 3,1 kb for an L. reuteri plasmid [87], 1,6 kb in pLAB 1000 of L. hilgardii [32] and even as small as 1,34 kb in pBul from L. delbrueckii [9 11. A shuttle vector additionally contains sequences responsible for replication in other hosts usually provided by a part of pUC or pBR derivatives or an origin of replication, which is characterized by a broad host range. A disadvantage of vectors based on small cryptic Lactobacillus plasmids is their structural instability which is probably caused by their mode of replication, which follows the rolling circle mechanism. During this process single stranded DNA intermediates are formed [40]. Integration of foreign DNA into these plasmids seems to interfere with this replication mode by accumulation of single stranded DNA, resulting in complete loss of plasmids [92]. Segregative instability and partial deletions were also observed upon subcloning of the L. curvatus plasmid pLc2 [93].
136 On the other hand, Posno et al. [90] and Shimidzu-Kadota et al. [94] have been able to construct vectors of improved stability although following the rolling circle mode of replication. At present, only few host vector systems were derived from Lactobacillus plasmids in which transformants had not undergone detectable rearrangements, or deletions, even if restriction and modification systems in hosts [95,96]. The construction of plasmid vectors based on the replicon of %replicating (large) plasmids may solve some of these problems. The transfer of vectors into host strains often shows their incompatibility with resident plasmids of the recipient organisms. This results in the loss of either the vector, or the resident plasmid, demonstrating the limited host range of the most vectors. On the other hand, incompatibility can be used to cure specific plasmids encoding special properties. While curing of plasmid of lactococci is relatively easy done by applying stress conditions, it is much harder to achieve in lactobacilli [35]. Examples for plasmid curing by using incompatibility have been described for L. acidophilus [42], L. plantarum [36] L. pentosus [90] and L. sake [97]. Food grade vectors and chromosomal integration Whereas these vector systems have been successfully used in the laboratory, most of them are not suitable for the construction of genetically engineered lactobacilli which will be applied in food production or the agro industry. The requirements include absence of antibiotic resistance markers, stable integration of foreign genes into the host chromosome, sufficient expression of heterologous genes as well as a secretion in adequate quantities, that until now are mostly too low for industrial exploitation. These problems can be solved by the construction of food-grade vectors, integrative suicide vectors and expression vectors carrying the required signals. Examples are: 1) the development of a food grade host vector for L. helveticus by cloning a pgalactosidase gene from L. delbrueckii subsp. bulgaricus for the use as a nonantibiotic reporter gene in a Lac negative mutant strain [96]; 2) stable chromosomal integration in L. plantarum [98-1001 and 3 ) expression of the a-amylase gene from Bacillus licheniformis (amylL) in L. plantarum by replacement of the amylL promotor by a strong L. plantarum promotor. The expression and secretion signals used in this latter investigation, led to efficient gene expression and secretion of more than 90% of the recombinant protein into the culture supernatant. [loll. In conclusion, the development of host vector systems has tremendously advanced during the last years. The current limit is caused by the restriction of these systems to a limited number of Lactobacillus species. Therefore, a major task will be adaptation and development of new host vector systems for specific species and applications.
Gene cloning in lactobacilli Cloning of genes from Lactobacillus species not only serves basic interest in their
137 genetic structure and regulation, but selection of genes for investigation was predominantly guided by their impact on practical applications, namely food fermentations. In addition, some genes have been cloned and analysed because of their potential usefulness as nonantibiotic reporter genes for the construction of food grade vector systems. An overview on the genes cloned and characterized from lactobacilli is provided in Table 6. Most of the work has concentrated on few Lactobacillus species namely L. casei, L. delbrueckii, and more recently, L. plantarum and L. sake. Genes of the sugar metabolism and proteolytic systems are major targets of investigation. Their analysis provides insight into gene regulation, transport and metabolism within lactobacilli. Carbohydrate metabolism The analysis of lactose metabolism in lactobacilli revealed the presence of two systems involving either P-galactosidase [140] or a PTS system [57,109]. In L. sake the P-galactosidase gene consists of two open reading frames, which are transcribed into a single mRNA and finally form a heterodimer of the active enzyme [140]. This has also been reported for L. casei and Leuconostoc luctis [141] and L. plantarum [58]. However, the L. sake gene is chromosomally encoded as compared to the plasmid borne L. casei and Le. lactis genes, and shares only 61% homology with them. Such dissimilarity of genes encoding the same property in different lactobacilli may be caused by preferential use of codons. Analysis of the codon usage patterns in 70 cloned Lactobacillus genes from different species revealed a high bias [143]. Nevertheless, preferential use of A or T in L. sake codons can not solely explain the difference of the P-galactosidase gene. Furthermore, there is no indication of deletion formation as observed with the L. bulgaricus lacZ gene [142], although the L. sake gene is scattered with direct and inverted repeats [144]. This indicates that the pgalactosidase genes have diverged from a common ancestor. Within L. sake all strains investigated appear to contain a p-galactosidase gene, some of which do not express detectable P-galactosidase activity. The presence of such silent genes was also observed for genes involved in the biosynthesis of amino acids. These genes may be activated upon treatment with mutagenic agents [ 145; unpublished observations]. Nevertheless, the distribution and regulation of silent genes in lactobacilli and their putative role in ecological adaptation remains unclear. Special properties Following the practically oriented approach, genes encoding special properties which are found in selected strains of some species, were only characterized. These properties may be useful in food fermentations, e.g., the formation of bacteriocins [53,117,119,120] or catalase [ 1361, but also undesired as histidine decarboxylase whose presence may lead to the formation of biogenic amines in food fermentations [ 133,1341. The ability of lactobacilli to produce a bacteriocin may be used to develop starter
138 Table 6. Genes cloned from Lactobacillus species. Cloned gene
Species
Reference
Peptidases aminopeptidase
L. delbrueckii subsp. lactis
aminopeptidase C
L. helveticus
lysyl aminopeptidase
L. delbrueckii subsp. lactis
peptidases
L. helveticus
Carbohydrate utilization factor IIIlac
L. casei
lactose permease
L. delbrueckii subsp. bulgaricus
lactose-specific enzyme I1
L. casei
Phospho-R-galactosidase
L. casei
phosphofructokinase
L. bulgaricus
8-D-phosphogalactosid-galactoh ydrolase
L. casei
8-galactosidase
L. delbrueckii
B-galactosidase
L. delbrueckii subsp. bulgaricus
8-galactosidase
L. sake
galactokinase
L. helveticus
galactose- 1-phosphate-uridyl-transferase
L. helveticus
malolactic enzyme
L. delbrueckii
xylose operon
L. pentosus
Bacteriocins curvacin A
L. curvatus
helveticin J
L. helveticus
lactacin F
L. acidophilus
plantaricin A
L. plantarum
sakacin 674
L. sake
sakacin P
L. sake
sakacin A immunity
L. sake
Basic metabolism D-lactate dehydrogenase
L. plantarum
D-lactate-dehydrogenase
L. delbrueckii subsp. bulgaricus
L-lactate dehydrogenase . -
L. plantarum (continued)
139 Table 6. Continued.
Cloned gene
Species
Reference
lactate dehydrogenase
L. casei
[ 1241
~2-h ydrox y isocaproate-dehydrogenase
L. casei
11251
dihydrofolate reductase
L. casei
[ 1261
folylpoly -g-glutamate-synthetase
L. casei
1271
~2-h ydrox y isocaproate-deh ydrogenase
L. confusus
[I281
orotic acid-phosphoribosyl-transferase
L. plantarum
[1291
ribonucleotide reductase
L. leichmnnii
[ 1301
valyl-tRNA synthetase
L. casei
[I311
glutamate racemase
Lactobacillus spp.
~321
histidine decarboxylase A
Lactobacillus 30A
[I331
histidine decarboxylase B
Lactobacillus 30A
[I341
tryptophan operon
L. casei
[ 1351
catalase
L. sake
[I361
conjugated bile acid hydrolase
L. plantarum
[ 1371
erythromycin resistance
L. reuteri
[201
S-layer protein
L. acidophilus
U381
S-layer protein
L. brevis
[I391
Amino acid metabolism
Special properties
organisms with an enhanced capability to suppress the competing fortuitous flora in fermentations, where no pasteurization of the raw materials is possible, e.g., sausage fermentation [97]. In addition, growth of opportunistically pathogenic Listeriu monocytogenes may be reduced in the presence of a bacteriocin producing Lactobacillus as starter organisms [146]. The analysis of the bacteriocin genes of lactobacilli and their comparison with bacteriocins of other lactic acid bacteria, revealed the presence of homologous parts especially in the N-terminal region, which seems to be responsible for secretion. This sequence, however, has no homology to known transport signals [120]. It may, however, be used to develop novel secretion systems for peptides or proteins in lactic acid bacteria. Catalase is also a rare property in lactobacilli and is only formed if an exogenous heme source is available [147]. Expression of the L. sake catalase in L. casei led to an increased tolerance to hydrogen peroxide, which is formed by many lactobacilli [ 1361. The enzyme removes hydrogen peroxide from fermented foods, preventing premature occurrence of rancidity and discoloration of the product.
140 The study of adhesion of lactobacilli to mucosal membranes helps in understanding probiotic effects often claimed to be inherent with the consumption of fermented milk products, preferably those containing live cells of L. acidophilus. Boot et al. [ 1381 and Vidgren et al. [ 1391 reported on the cloning and characterization of S-layer proteins of L. brevis and L. acidophilus, respectively, providing first insight in the genetics of colonizing factors which may play a role in probiosis. Although there is only little genetic work within this field, many publications on the biochemistry and physiology of adhesion within lactobacilli associated with man and animals indicate forthcoming research on the genetics of these properties. Heterologous gene expression Some of these genes and genes from microorganisms other than lactobacilli were heterologously expressed in lactobacilli to study expression, and also construct strains with enhanced capabilities for studies in model food fermentation processes (Table 2). Table 7 provides an overview on the heterologous expression of genes in lactobacilli. These reports demonstrate that it is possible to enhance capability of L. plantarum to ferment polymeric substrates as cellulose, starch, xylans or glucans and thus use substrates which are abundant in plant material. This may be used to construct improved starter organisms for silage fermentation, but also to use cheap substrates for biotechnical fermentations with L. plantarum. As the respective enzymes are released into the medium, secretion mechanisms can be elucidated and modules for the construction of expression and secretion vectors are available. The heterologous expression of bacteriocin genes in lactobacilli provides an opportunity to identify genes involved in the regulation, secretion, modification of, and immunity against bacteriocins. Furthermore, strains adapted to specific environments and therefore suitable as starter organisms in specific fermentations, can be provided with the ability to produce bacteriocin in situ. This can be of crucial importance, as an added bacteriocin may be inactivated in proteinase-rich habitats or adsorbed by phospholipids if not continuously delivered by the starter culture. Such environments are present e.g:, in meat with its endogenous cathepsins, a putative load with proteinases from Pseudomonas or Micrococcaceae, or the surface of a red-smear cheese foods covered with proteolytic bacteria or fungi. On the other hand, these foods are major targets of bacteriocin application, as they provide habitats for opportunistic pathogens such as Enterococcus faecalis and Listeria monocytogenes. In addition to bacteriocins, specific enzymes with the ability to attack and induce lysis of selected bacteria, may be used to prevent pathogen growth in foods. Lysostaphin is an endopeptidase specifically recognizing pentaglycyl-peptides which are present as interpeptide bridges in the staphylococcal peptidoglycan [ 1581. Therefore, starter cultures expressing lysostaphin might reduce the risk of Staphylococcal enterotoxicosis [ 1541. Most of the genes listed in Table 7 were introduced into their recipient strains via plasmid shuttle vectors. Therefore, they are subjected to segregation, especially in the absence of a selective agent. The segregational stability of a new gene introduced into
141 Table 7. Heterologous expression of genes in Lactobacillus species (without anticiotic resistance genes).
Source
Expressed in
Acidocin B
L. acidophilus
L. plantarum
Helveticin J
L. helveticus
L. acidophilus
Cellulase
C. acetobutylicum
L. plantarum
a-Amylase
B . stearothermophilus C . thermocellum B. amyloliquefaciens L. amylovorus
L. casei, Lc. lactis L. plantarum L. plantarum L. plantarum. L. casei
Xylanase
C . thermocellum C . acetobutylicum
L. plantarum
Endoglucanase
C . thermocellum B . subtilis
L. plantarum L. plantarum
L. pentosus L. sake
L. plantarum, L. casei L. casei
catalase
L. sake
L. casei
lysostaphin
Staphylococcus simulans
L. casei
lipase
Staphylococcus hyicus
L. casei
manganese-superoxidedismutase
E. coli
L. gasseri
lux-genes
Vibrio fischeri
L. casei
Pverty
Reference
Bacteriocin
Polysaccaride cleavage
Carbohydrate utilization Xy lose-Operon B-galactosidase Other
a host Lactobacillus can be enhanced by chromosomal integration. For a comprehensive review on Lactobacillus vectors and gene expression in lactobacilli, see Pouwels and Leer [ 1591. Scheirlinck et al. [ 1481 introduced an a-amylase and an endoglucanase into the chromosome of L. plantarum by homologous recombination. This was achieved with a plasmid vector containing sequences of the L. plantarum chromosome. Chromosomal integration may be desirable from many points of view, also including reduced transferability to other strains and stable expression of a gene. From a technological point of view chromosomal integration may not always be necessary or even useful. Strong expression of a gene is often desired only during the fermentation, and plasmids present at high copy numbers are suitable vehicles making use of a desired gene-dosis effect. When plasmids are subsequently lost this does not affect the process. It may even be of an advantage from aspects of biosafety, when genetically engineered organisms are applied which will loose their new properties
142 upon time after the release into the environment. However, the mobility of extrachromosomal elements, e.g., plasmids, is determinative for the distribution of properties among organisms sharing the habitat, and therefore crucial in the assessment of biological safety when using genetically engineered microorganisms. Chromosomal integration systems in Lactobacillus have not only been used to introduce and stabilize new traits under nonselective conditions, but were also used for gene disruption in Lactobacillus plantarum [ 1001 and L. helveticus [ 1601. In these experiments strains were constructed deficient in bile salt hydrolase or an X-prolyl dipeptidyl aminopeptidase, respectively. This demonstrates that chromosomal integration systems may be used to inactivate genes encoding undesired properties of otherwise useful starter lactobacilli.
Behaviour of recombinant lactobacilli in food and potentially emerging risks The application of recombinant microbes which may even remain living in food, requires special attention to the construction of microorganisms for the use as starter cultures of the production of compounds used in food manufacture. The use of genetically engineered lactobacilli as starters for food fermentations, is generally combined with their intimate contact with man and its intestinal microflora, and the release into the environment. For the construction of a genetically engineered (Lactobacillus) starter, the nature of the newly introduced gene(s), the donor and the recipient of the DNA and the following parameters may be important: 1) low or absent transmissibility by conjugation or mobilization; 2) absence of antibiotic resistance reporter genes; 3) known nucleotide sequence, e.g., to prove the absence of known toxin genes; 4) source and host range of the replicon used for transfer of the DNA; 5 ) survival, behaviour and genetic stability of the new strain; and 6) identification of genetically engineered strains. When genetical engineering is performed within the intrinsically safe lactobacilli, using their genetic elements and genes, or even within strains of one species, this may be regarded as safe to our current knowledge. Nevertheless, currently a case by case study should be made with a new strain and process. We have investigated behaviour and plasmid transfer between genetically engineered model starter lactobacilli during sausage fermentation [74]. Plasmids showed the same segregation as observed in vitro. Recent experiments with vectors derived from a cryptic plasmid of L. curvatus showed, that even minor changes may cause strong differences in segregational and structural plasmid stability in L. curvatus. Therefore, such studies should be performed with the final constructs case by case, and results on plasmid stability obtained previously, can not be extrapolated to other constructs. For Lactococcus, similar experiments were performed during cheese ripening [75]. Klijn et al. [161] furthermore followed the behaviour of lactococci in natural ecosystems using DNA probes and PCR techniques. These studies provide a basis for the biosafety assessment of genetically engineered microbes in food systems.
143
Perspectives Recent developments for lactic acid bacteria are most advanced for lactococci and include food grade, self contained vector systems [ 162,1631, high efficiency chromosomal integration [164] and monitoring of strains with DNA probes [161]. Research on lactobacilli currently enters a similarly productive phase with some breakthroughs in genetic investigations which have already been demonstrated. The knowledge obtained during the investigations discussed here, has already provided basic understanding of the genetic organization of lactobacilli and it can be used in classical genetics for the improvement of starter organisms. The methods and constructs, however, are designed for targeted engineering of starter organisms. They may soon be available for more lactobacilli opening a big field of fascinating and safe novel applications.
References 1.
2.
3. 4. 5.
6.
7.
8. 9.
10.
11. 12.
13. 14.
Hammes WP, Vogel RF. The genus Lactobacillus. In: Wood BJB, Holzapfel WH (eds) The Genera of Lactic Acid Bacteria. London: Blackie Academic & Professional, 1995;(In press). Hammes WP,Weiss N, Holzapfel WH. The genera Lactobacillus and Camobacterium. In: Balows A, Triiper HG, Dworkin M, Harder W, Schleifer KH (eds) The Prokaryotes Handbook on the Biology of Bacteria: Ecophysiology, Isolation Identification, Applications. New York: Springer Verlag, 1991;1535-1594. Joosten HMU, Northolt MD. Detection, growth and amine-producing capacity of lactobacilli in cheese. Appl Environ Microbiol 1989;55:2356-2359. Bourne KA, Beebe JL, Lue YA, Ellner PD. Bacteremia due to Bifidobacterium, Eubacterium or Lactobacillus. Twenty-one cases and review of the literature. Yale J Biol Med 1978;51:505-512. Gasson MJ, de Vos WM (eds). Genetics and Biotechnology of Lactic Acid Bacteria. London: Blackie Academic and Professional, 1994. Vogel RF, Bocker G, Stolz P, Ehrmann M, Fanta D, Ludwig W, Pot B, Kersters K, Schleifer KH, Hammes WP. Identification of lactobacilli from sourdough and description of Lactobacillus ponfis sp. nov. Int J System Bacteriol 1994;44:223-229. Blackbum C de W. Rapid and alternative methods for the detection of salmonellas in food. J Appl Bacteriol 1993;75:19F214. Schleifer KH. DNA probes in food microbiology. Food Biotechnol 1990;4:585-598. Lonvaud-Funel A, Joyeux A, Ledoux 0. Specific enumeration of lactic acid bacteria in fermenting grape must and wine by colony hybridization with nonisotopic DNA probes. J Appl Bacteriol 1991;71501-508. Woese CR. Bacterial evolution. Microbiol Rev 1987;51:221-271. Ehrmann M, Ludwig W, Schleifer KH. Reverse dot blot hybridization: a useful method for the direct identification of lactic acid bacteria in fermented food. FEMS Microbiol Lett 1994117:143-150. Pot B, Hertel C, Ludwig W, Descheemaeker P, Kersters K, Schleifer KH. Identification and classification of Lactobacillus acidophilus, L. gasseri and L. johnsonii strains by SDS-PAGE and rRNA-targeted oligonucleotide probe hybridization. J Gen Microbiol 1993;139513-5 17. Ehrmann M. Khsifizierung und Identifizierung von Milchsaurebakterien mit Hilfe molekularbiologischer Methoden. PhD Thesis, 1994. Technische Universitat Miinchen, Germany. Hertel C, Ludwig W, Obst M, Vogel R, Hammes WP, Schleifer KH. 23s rRNA-targeted oligonucleotide probes for rapid identification of meat lactobacilli. Syst Appl Microbiol 1991;14: 173- 177.
144 15. Hertel C, Ludwig W, Pot B, Kersters K, Schleifer KH. Differentiation of lactobacilli occurring in fermented milk products by using oligonucleotide probes and electrophoretic protein profiles. Syst Appl Microbiol 1993;16:453-467. 16. Hensiek R, Krupp G, Stackebrandt E. Development of diagnostic oligonucleotide probes for four Lactobacillus species occurring in the intestinal tract. Syst Appl Microbiol 1992;15:123-128. 17. Nakagawa T, Shimada M, Mukai H, Asada K, Kato I, Fujino K, Sat0 T. Detection of alcohol tolerant hiochi bacteria by PCR. Appl Environ Microbiol 1994;60:637-640. 18. Nes IF. Plasmid profiles of ten strains of Lactobacillus plantarurn. FEMS Microbiol Lett 1984;21:35*361. 19. Hill HA, Elliot J. The value of plasmid profiling in monitoring Lactobacillus plantarum in silage fermentations. Curr Microbiol 1986;13:91-94. 20. Tannock GW, Luchansky JB, Miller L, Hugh C, Thode-Andersen S, Mercer AA, Klaenhammer TR. Molecular characterisation of a plasmid borne (pGT633) erythromycin resistance determinant (ermGT) from Lactobacillus reuteri 100-63. Plasmid 1994;31:60-71. 21. Manachini PL, Parini C. DNA restriction endonuclease cleavage patterns, DNA sequence similarity, and phenotypical characteristicsin some Lactobacillus helveticus and Lactobacillusjugurti. Antonie van Leeuwenhoek 1983;49:143--152. 22. Mitewa VI, Abadjewva AN, Stefanova TT.MI3 DNA fingerprinting, a new tool for classification and identification of Lactobacillus spp. J Appl Bacteriol 1992;73:349-354. 23. Jeune C Le,Lonvaud-Funel A. Lactobacillus hilgardii and Lactobacillus brevis DNA analysis by restriction fragment length polymorphism (RFLP). Food Microbiol 1994;ll: 195-202. 24. Staahl M, Molin G, Persson A, Ahrne S, Staahl S. Restriction endonucleasepatterns and multivariate analysis as a classification tool for Lactobacillus spp. Int J Syst Bacteriol 1990;40:189-193. 25. Staahl M, Molin G. Classification of Lactobacillus reuteri by restriction endonuclease analysis of chromosomal DNA. Int J Syst Bacteriol 1994;44:9-14. 26. Staahl M, Petterson B, Molin G, Uhlen M, Ahrne S. Restriction fragment length polymorphism of Lactobacillus reuteri and Lactobacillusferrnentum, originating from intestinal mucosa, based on 16s rRNA gens. Syst Appl Microbiol 1994;17:108--115. 27. Duffner F, O’Connell M. Comparative evaluation of plasmid profiling and ribotyping in the analysis of Lactobacillus plantarum strain heterogeneity in silage. J Appl Bacteriol 1995;78:20-27. 28. Welsh J, McCleeland M. Fingerprinting genomes using PCR with arbitrary primers. Nucl Acid Res 1990 18:7213-7219. 29. Chassy B, Gibson M, Giuffrida E. Evidence for extrachromosomal elements in Lactobacillus. J Bacteriol 1976;12:1576- 1578. 30. AhmC S, Molin G, Stahl S. Plasmids in Lactobacillus strains isolated from meat and meat products. Syst Appl Microbiol 1989;11;32&325. 31. Bates EEM, Gilbert HJ. Characterization of a cryptic plasmid from Lactobacillus plantarum. Gene 1989;85:253-258. 32. Josson K, Scheirlinck T, Michiels F, Platteeuw C, Stanssens P, Joos H, Dhaese P, Zabeau M, Mahillon J. Characterizationof a gram-positive broad-host-range plasmid isolated from Lactobacillus hilgardii. Plasmid 1989;21:9-20. 33. Ruiz-Barba JL, Piard JC, Jipenez-Diaz R. Plasmid profiles and curing of plasmids in Lactobacillus plantarurn strains isolated from green olive fermentations. J Appl Bacteriol 1991;71:417-421. 34. von Husby KO, Nes IF. Changes in the plasmid profile of Lactobacillus plantarurn obtained from commercial meat starter cultures. J Appl Bacteriol 1986;60413-417. 35. Heidel M. Genetische Untersuchungen an Lactobacillus species mit Eignung fur die Rohwurstfermentation. PhD thesis 1989, Universitat Hohenheim, Germany. 36. Bringel F, Frey L, Hubert J-C. Characterization, cloning, curing and distribution in lactic acid Bacteria of pLPl a plasmid from Lactobacillus plantarurn CCM 1904 and its use in shuttle vector construction. Plasmid 1989;22:193-202. 37. Bouia A, Bringel F, Frey L, Kammerer B, Belarbi A, Guyonvarch A, HubertJ-C. Structural organization of pPLl a, cryptic plasmid from Lactobacillus plantarurn CCM 1904. Plasmid
145 38. 39.
40. 41.
42.
43.
44. 45.
46. 47. 48.
49.
50.
51.
52. 53. 54. 55.
56. 57.
58.
1989;22:185-192. Skaugen M. The complete nucleotide sequence of a small cryptic plasmid from Lactobacillus plantarum. Plasmid 1989;22:175-179. Takiguchi R, Hashiba H, Aoyama K, Ishi S. Complete nucleotide sequence and characterization of a cryptic plasmid from Lactobacillus helveticus subsp. jugurti. Appl Environ Microbiol 1989;55: 1653-1 655. Vujic M, Topisirovic L. Molecular analysis of the rolling-circle replicating plasmid pAl of Lactobacillus plantarum A1 12. Appl Environ Microbiol 1993;59:274-280. Pouwels PH, van Luijk N, Leer RJ, Posno M. Control of replication of the Lactobacillus pentosus plasmid p353-352: Evidence for a mechanism involving transcriptional attenuation of the gene coding for the replication protein. Molec Gen Genet 1994;242:614-622. Muriana PM, Klaenhammer TR. Conjugal transfer of plasrnid-encoded determinants for bacteriocin production and immunity in Lactobacillus acidophilus 88. Appl Environ Microbiol 1987;53:553560. van der Vossen JMBM, van Herwijnen MHM, Leer RJ, ten Brink B, Pouwels PH, Huis In’t Veld JHJ. Production of acidocin B, a bacteriocin of Lactobacillus acidophilus M46 is a plasmidencoded trait: Plasmid curing genetic marking by in vivo plasmid integrationand gene transfer. FEMS Microbiol Lett 1994;116:333-340. Lahbib-Mansais Y, Mata M, Ritzenthaler P. Molecular taxonomy of Lactobacillusphages. Biochimie 1988;70:429-435. Leuschner RGK, Arendt EK, Hammes WP.Characterizationof a virulent Lactobacillus sake phage PWH2. Appl Microbiol Biotechnol 1993;39:617-621. Foschino R, Galli A, Pagani MA. Lactobacillusfermentum bacteriophages isolated from sour dough. Annali Microbiol Enzimol 1993;43:225-234. Nemcova R, Styriak I, Stachova M, Kmet V. Isolation and partial characterization of three rumen Lactobacillus plantarum bacteriophages. Microbiologica (Pavia) 1993;16:177-180. Vasala A, Dupont L, Baumann M, Ritzenthaler P, Alatossava T. Molecular comparison of the three structural proteins encoding gene clusters of two related Lactobacillus delbrueckii bacteriophages. J Virol 1993;67:3061-3068. Forsman P, Tanskanen J, Alatossava T. Structural similarity and genetic homology between Lactobacillus casei bacteriophages isolated in Japan and in Finland. Biosci Biotechnol and Biochem 1993;57:2043-2048. Raya RR, Kleeman EG, Luchansky JB, Klaenhammer TR. Characterization of temperate bacteriophage phi-adh and plasmid transduction in Lactobacillus acidophilus ADH. Appl Environ Microbiol 1989;55:220&2213. Watanabe K, Kakita Y, Nakashima Y, Miake F. Calcium requirement for protoplast transfection mediated by polyethylene glycol of Lactobacillus casei PL- 1 phage DNA. Biosci Biotech Biochem 199236: 1859-1862. Shay BJ, Egan AF, Wright M, Rogers PJ. Cysteine metabolism in an isolate of Lactobacillus sake: plasmid composition and cysteine transport. E M S Microbiol Lett 1988;56:183-188. Tichaczek P, Vogel RF, Hammes WP. Cloning and sequencing of curA ancoding curvacin A, the bacteriocin produced by Lactobacillus curvatus LTHll74. Arch Microbiol 1993;160:27+283. Schillinger U, Liicke K-F. Antibacterial activity of Lactobacillus sake isolated from meat. Appl Environ Microbiol 1989;55:1901-1906. Mmvedt CI, Nes IF. Plasmid-associated bacteroicin production by a Lactobacillus sake strain. J Gen Microbiol 1990;136:1601- 1607. Hofer F. Involvement of plasmids in lactose metabolism in Lactobacillus casei suggested by genetic experiments. FEMS Microbiol Lett 1977;1:167-170. Lee L-J, Hansen JB, Jagusztyn-Krynicka E, Chassy BM. Cloning and expression of the p-Dphosphogalactoside-galactohydrolase gene of Lactobacillus casei in Escherichia coli K- 12. J Bacteriol 1982;152:1138-1146. Mayo B, Gonzalez B, Arca P, Suarez J.E Cloning and expression of the plasmid encoded p-D-
146
59.
60. 61.
62. 63.
64. 65.
66. 67. 68. 69. 70. 71. 72. 73.
74.
75. 76. 77. 78. 79.
80. 81.
galactosidase gene from a Lactobacillus plantarum strain of dairy origin. FEMS Microbiol Lett 199%122:145-15 1. Liu ML, Kondo JK, Barnes MB, Bartholomew DT. Plasmid linked maltose utilization in Lactobacillus ssp. Biochimie 1988;70:351-355. Smiley MB, Fryder V. Plasmids, lactic acid production and N-acetyl-glucosamine fermentation in Lactobacillus helveticus subsp. jugurti. Appl Environ Microbiol 1978;35:777-781. de Rossi E, Brigidi P, Riccardi G, Milano A, Matteuzzi D. Preliminary studies on the correlation between the plasmid plHJl and its proteolytic activity in Lactobacillus helveticus S36.2. Physical mapping and molecular cloning of the plasmid in Escherichia coli. Microbiologica 198%97:27+ 277. Vescovo M, Morelli L, Bottazzi V. Drug resistance plasmids in Lactobacillus acidophilus and Lactobacillus reuteri. Appl Environ Microbiol 1982;43:50-56. Morelli L, Vescovo M, Bottazzi V. Plasmids and antibiotic resistances in Lactobacillus helveticus and Lactobacillus bulgaricus isolated from natural whey culture. Microbiologica 1983;6:145- 154. Ishiwa H, Iwata S. Drug resistance plasmids in Lactobacillus fermenturn. J Gen Microbiol 198026:7 1-74. Tannock GW, Luchansky JB,Miller L, Connell H, Thode-Andersen S, Mercer AA, Klaenhammer TR. Molecular characterization of a plasmid-borne (pGT633) erythromycin resistance determinant (ermGT) from Lactobacillus reuteri 1 W 6 3 . Plasmid 1994;31:6+71. West CA, Warner PJ. Plasmid profiles and transfer of plasmid encoded antibiotic resistance in Lactobacillus plantarum. Appl Environ Microbiol 1985;50:1319-1321. Morelli L, Vescovo M, Cocconcelli P, Bottazzi V. Fast and slow milk coagulating variants of Lactobacillus helveticus HLM- 1 Can. J Microbiol 1986;32:758-760. Shimidzu-Kadota M, Flickinger JL, Chassy BM. Evidence that Lactobacillus casei insertion element ISLl has a narrow host range. J Bacteriol 1988;1704976-4978. Germond JE, Lapierre L, Delley M, Mollet B. Novel IS element (ISL3) located next to the Lactobacillus bulgaricus lacZ gene. FEMS Microbiol Rev 1993 12:A13. Skaugen M, Nes I. Transposition in Lactobacillus sake and its abolition of lactocin S production by insertion of IS1 163, a new member of the IS3 family. Appl Environ Microbiol 1994;60:281&2825. Knauf HJ, Vogel RF, Hammes WP. Introduction of the transposon Tn919 into Lactobacillus curvatus Lc2-c. FEMS Microbiol Lett 1989;65:101-104. Thompson JK, Collins MA. Evidence for conjugal transfer of the broad host range plasmid pIP5 into strains of Lactobacillus helveticus. J Appl Bacteriol 1988;65:30+3 19. Chassy BM, Rokaw E. Conjugal transfer of plasmid-associated lactose metabolism in Lactobacillus casei subsp. casei. In: Levy S, Clowes R, Koenig E (eds) Molecular Biology Pathogenesis and Ecology of Bacterial Plasmids. New York: Plenum Press, 1981;590. Vogel RF, Becke-Schmid M, Entgens P, Gaier W, Hammes WP.Plasmid transfer and segregation in Lactobacillus curvatus LTH1432 in vitro and during sausage fermentations. Syst Appl Microbiol 1992;15:129-1 36. Gabin-Gauthier K, Gratadoux J-J, Richard J. Conjugal transfer between lactococci on surface matings and during cheese making. FEMS Microbiol Ecol 1991;85:133-140. Kozlowa EV, Aminov RI,, Boronin AM. Molec Gen Microbiol Virol 1991;9:15-17. Yokokura T, Kodaira S, Ishiwa H, Sakurai T. Lysongeny in Lactobacilli. J Gen Microbiol 1974;84:277-284. Morelli L, Cocconelli PS, Bottazzi V, Damiani G, Ferretti L, SgaramellaV. Lactobacillus protoplast transformation. Plasmid 1987;17:73-75. Badii-Azandahi R, Jones S, Warner PJ. Sphaeroblast and electroporation-mediated transformation of Lactobacillus plantarum. Lett Appl Microbiol 1989;9:41-44. Raya RR, Klaenhammer TR. High frequency plasmid transduction by Lactobacillus gasseri bacteriophage phi adh. Appl Environ Microbiol 1992;58:187-193. Harlander SK. Transformation Streptococcus lactis by electroporation. In: Ferretti JJ, Curtiss R I11 (eds) Streptococcal genetics. Washington, DC:ASM, 1987;229-233.
147 82. Chassy BM, Flickinger JL. Transformation of Lactobacillus casei by electroporation. FEMS Microbiol Lett 1987;44173--177. 83. Leer RJ, Posno M, van Rijn JMM, Lockman BL, Pouwels P. Transformation of Lactobacillus by plasmid DNA. FEMS Microbiol Rev 1987;46:2*32. 84. Luchansky JB,Kleeman EG, Raya RR, Klaenhammer TR. Genetic transfer systems for delivery of plasmid deoxyribonucleic acid to Lactobacillus acidophilus ADH:conjugation electroporation and transduction. J Dairy Sci 1989;72:1408-1417. 85. Bringel F, Hubed-C. Optimized transformation by electroporation of Lactobacillus plantarum strains with plasmid vectors. Appl Microbiol Biotechnol 199033:664-670. 86. Posno M, Leer RJ, van Luijk N, Lokman BC, van Giezen JF,Heuvelmans PTHM, Pouwels PH. A host vector system for Lactobacillus. 6th International Symposium on Genetics of industrial microorganisms, Strasbourg 1990 189:D76. 87. Ahrne S, Molin G, Axelsson L. Transformationof Lactobacillus reuteri with electroporation: studies on the erythromycin resistance plasmid pLUL63 1. Curr Microbiol 1992;24:19F205. 88. Gaier W, Vogel RF, Hammes WP.Genetic transformation of intact cells of Lactobacillus curvatus Lc2-c and Lact sake Ls2 by electroporation. Lett Appl Microbiol 199011:81-83. 89. Aymerich MT, Hugas M, Garriga M, Vogel RF, Montfort JM. Electrotransformation of meat lactobacilli. Effect of several parameters on their efficiency of transformation. J Appl Bacteriol 1993;75:320-325. 90. Posno M, Leer RJ, van Luijk N, van Giezen MJF, Heuvelmans PTHM, Lokman BC, Pouwels PH. Incompatibility of Lactobacillus vectors with replicons derived from small cryptic Lactobacillus plasmids and segregational instability of the introduced vectors. Appl Environ Microbiol 199137: 1822-1 828. 91. Ito Y, Sasaki Y, Sasaki T. Novel plasmid pBULl from Lactobacillus and the preparation of derivatives. PCT Int Appl 1992;34. 92. te Riele H, Michel B, Ehrlich SD. Single-stranded plasmid-DNA in Bacillus subtilis and Staphylococcus aureus. Proc Natl Acad Sci 1986;83:2541-2545. 93. Vogel RF, Lohmann M, Weller AN, Hugas M, Hammes WP. Structural similarity and distribution of small cryptic plasmids of Lactobacillus curvatus and L. sake. FEMS Microbiol Lett 1991;84:183190. 94. Shimizu-Kadota M, Shibahara-Sone H, Ishiwa H. Shuttle plasmid vectors for Lactobacillus casei and Escherichia coli with a minus origin. Appl Environ Microbiol 1991;57:3292-3300. 95. Hashiba H, Takiguchi R, Ishii S, Aoyama K. Transformation of Lactobacillus helveticus subsp. jugurti with plasmid pLHR by electroporation. Agric Biol Chem 199054:1537-1541. 96. Hashiba H. Plasmid vectors for gene cloning and expression in Lactobacillus helveticus and molecular breeding. Snow Brand R&D Rep. 1994;101:93-143. 97. Vogel RF, Pohle S, Tichaczek PS, Hammes WP. The competitive advantage of Lactobacillus curvatus LTH1174 in sausage fermentations is caused by formation of curvacin A. Syst Appl Microbiol 1993;16:457-462. 98. Cosby WM, Axelsson LT, Dobrogosz WJ. Tn917 transposition in Lactobacillus plantarum using the highly temperature-sensitive plasmid pTV 1Ts as a vector. Plasmid 1989;22:236-243. 99. Rixon JE, Geoffrey GP, Gilbert HJ. Integration of an unstable plasmid into the chromosome of Lactobacillus plantarum. FEMS Microbiol Lett 19907I :105- 109. 100. Leer RJ, Christiaens H, Verstraete W, Peters L, Posno M, Pouwels PH. Gene disruption in Lactobacillus plantarum strain 80 by site-specific recombination: Isolation of a mutant strain deficient in conjugated bile salt hydrolase activity. Molec Gen Genet 1993;239:26% 272. 101. Hols P, Ferain T, Garmyn D, Bernard N, Delcour J. Use of homologous expression-secretion signals and vector-free stable chromosomal integration in engineering of Lactobacillusplantarum for alphaamylase and levanase expression. Appl Environ Microbiol 1994;60:1401-1413. 102. Meyer-Barton EC, Klein JR, Imam M, Plapp R. Cloning and sequence analysis of the X-prolyldipeptidyl-aminopeptidasegene (pepX) from Lactobacillus delbrueckii ssp. lactis DSM7290. Appl Microbiol Biotechnol 1993;40:82-89.
148 103. Fernandez L, Bhowmik T, Steele JL. Characterization of the Lactobacillus helveticus CNRZ32 pepC gene. Appl Environ Microbiol 1994;60:333-336. 104. Klein JR, Klein U, Schad M, Plapp R. Cloning DNA sequence analysis and partial characterization of pepNa, lysyl aminopeptidase from Lactobacillus delbrueckii ssp lactis DSM7290. Eur J Biochem 1993;217:105-114. 105. Nowakowski CM, Bhowmik TK, Steele JL. Cloning of peptidase genes from Lactobacillus helveticus CNRZ32. Appl Microbiol Biotechnol 1993;39:204-210. 106. Alpert C-A, Chassy BM. Molecular cloning and nucleotide sequence of the factor IIIlac gene of Lactobacillus casei. Gene 1988;62:277-288. 107. Leong-Morgenthaler P, Zwahlen MC, Hottinger H. Lactose metabolism in Lactobacillus bulguricus - analysis of the primary structure and expression of the genes involved. J Bacteriol 1991;173: 1951-1957. 108. Alpert CA, Chassy BM. Molecular cloning and DNA sequence of lacE the gene encoding the lactose-specific enzyme I1 of the phosphotransferase system of Lactobacillus casei - evidence that a cysteine residue is essential for sugar phosphorylation. J Biol Chem 1990;265:22561-22568. 109. Shimizu-Kadota M. Cloning and expression of the phospho-P-galactoside genes on the lactose plasmid and the chromosome of Lactobacillus C257 in Escherichia coli. Biochimie 1989;70:523529. 110. Branny P, de la Torre F, Garel JR. Cloning, sequencing, and expression in Escherichia coli of the gene coding for phosphofructokinase in Lactobacillus bulgaricus. J Bacteriol 1993;1755344-5349. 1 11. Schmidt BF, Adams RM, Requadt C, Power S, Mainzer SE. Expression and nucleotide sequence of the Lactobacillus bulgaricus P-galactosidase gene cloned in Escherichia coli. J Bacteriol 1989;171:625-635. 112. Obst M, Vogel RF, Hammes WP. Cloning of the P-galactosidase gene of Lactobacillus sake DSM20017. BioBngineering 1992;1:P406 1 13. Mollet B, Pilloud N. Galactose utilization in Lactobacihs helveticus - Isolation and characterization of the galactokinase (galK) and galactose- 1-phosphate uridyl transferase (galT) genes. J Bacteriol 1991;173:4464-4473. 114. Williams SA, Hodges RA, Strike TI, Snow R, Kunkee RE. Cloning the for the malolactic fermentation of wine from Lactobacillus delbrueckii gene in Escherichia coli and yeast. Appl Environ Microbiol 1989;42:288-293. 115. Lokman BC, Vansanten P, Verdoes JC, Kruse J, Leer RJ, Posno M, Pouwels PH. Organization and characterization of three genes involved in D-xylose catabolism in Lactobacillus pentosus. Molec Gen Genet 1991;230:161-169. 116. Joerger MC, Klaenhamrner TK. Cloning expression, and nucleotide sequence of the Lactobacillus helveticus 48 1 gene encoding the bacteriocin helveticind. J Bacteriol 1990;172:6339-6347. 117. Muriana PM, Klaenhammer TR. Cloning, phenotypic expression, and DNA sequence of the gene for lactacin-F, an antimicrobial polypeptide produced by Lactobacillus spp. J Bacteriol 1991;173: 1779-1788. 118. Diep DB, Haverstein LS, Nissen-Meyer J, Nes IF. The gene encoding plantaricin A, a bacteriocin from Lactobacillusplantarum C11, is located on the same transcription unit as an agr-like regulatory system. Appl Environ Microbiol 1994;60:1-166. 119. Holck AL, Axelsson L, Huhne K, Krockel L. Purification and cloning of sakacin 674a bacteriocin from Lactobacillus sake Lb674. FEMS Microbiol Lett 1994;115:143-149. 120. Tichaczek PS, Vogel RF, Hammes WP. Cloning and sequencing of sakP encoding sakacin P, the bacteriocin produced by Lactobacillus sake LTH 673. Microbiology 1994;140:361-367. 121. Axelsson L, Holck A, Birkeland SE, Aukrust T, Blom H. Cloning and nucleotide sequence of a gene from Lactobacillus sake Lb706 necessary for sakacin A production and immunity. Appl Environ Microbiol 1993;59:2868-2875. 122. Taguchi H, Ohta T. D-Lactate dehydrogenase is a member of the D-isomer-specific 2-hydroxyacid dehydrogenase family - Cloning, sequencing and expression in Escherichia coli of the D-hCtate dehydrogenase gene of Lactobacillus plantarum. J Biol Chem 1991;266:2588-2594.
149 123. Bernard N, Ferayn T, Garmyn D, Hols P, Delcour J. Cloning of the D-laCtate dehydrogenase from Lactobacillus delbrueckii subsp. bulgaricus by complementation in Escherichia coli. FEBS Lett 1991;290:61-65. 124. Kim SF, Baik SJ, Pack MY. Cloning and nucleotide sequence of the Lactobacillus casei lactate dehydrogenase gene. Appl Environ Microbiol 1991;57:2413-2417. 125. Lerch H-P, Blocker H, Kallwass H, Hoppe J, Tsai H, Collins J. Cloning, sequencing and expression in Escherichia coli of the D-2-hydroxyisocaproatedehydrogenase gene of Lactobacillus casei. Gene 1989;78:47-57. 126. Andrews RE, Clore GM, Davies RW, Gronenborn AM, Gronenborn B, Kalderon D, Papadopoulos RC, Schiifer S , Sims PFG, Stancombe R. Nucleotide sequence of the dihydrofolate reductase gene of methotrexate-resistant Lactobacillus casei. Gene 1985;35:217-222. 127. Toy 3, Bognar AL. Cloning and expression of the gene encoding Lactobacillus casei folylpolygamma-glutamate synthetase in Escherichia coli and determination of its primary structure. J Biol Chem 1990;265:2492-2499. 128. Lerch H-P, Frank R, Collins J. Cloning, sequencing and expression of the L-2-hydroxycaproate dehydrogenase-encodinggene of Lactobacillus confusus in Escherichia coli. Gene 1989;83:263-270. 129. Bouia A, Bringel F, Frey L, Belarbi A, Guyonvarch A, Kammerer B, Hubert J-C. Cloning and structure of the pyrE gene of Lactobacillus plantarum CCM1904. FEMS Microbiol Lett 199e69: 233-23 8. 130. Booker S , Stubbe J. Cloning, sequencing, and expression of the adenosylcobalamin-dependent ribonucleotide reductase from Lactobacillus leichmannii. Proc Natl Acad Sci 1993;90:8352-8356. 131. Taylor BV, Toy J, Sit T, Bognar AL. Cloning and sequence determination of the valS gene, encoding valyl-tRNA synthetase in Lactobacillus casei. J Bacteriol 1993; 175:2475-2478. 132. Gallo KA, Knowles JR. Purification, cloning and cofactor independence of glutamate racemase from Lactobacillus. Biochemistry 1993;32:3981-3990. 133. Vanderslice P, Copeland WC, Robertus JD.Cloning and nucleotide sequence of wild type and a mutant histidine decarboxylase from Lactobacillus 30a. J Biol Chem 1986;261:15186-15191. 134. Copeland WC, Domena JD, Robertus JD. The molecular cloning, sequence and expression of the hdcB gene from Lactobacillus 30A. Gene 1989;85:25+265. 135. Natori Y, Kano Y, Imamoto F. Nucleotide sequences and genomic constitution of five tryptophan genes of Lactobacillus casei. J Biochem 1990;107:248-255. 136. Knauf HJ,Vogel RF, Hammes WP. Cloning, sequence and phenotypic expression of katA encoding the catalase of Lactobacillus sake LTH677. Appl Environ Microbiol 1992;58;832-839. 137. Christiaens H, Leer R, Pouwels PH, Vaerstraete W. Cloning and expression of a conjugated bile acid hydrolase gene from Lactobacillus plantarum by using a direct plate assay. Appl Environ Microbiol 1992;58:3792-3798. 138. Boot H, Kolen CPAM, van Noort JM, Pouwels PH. S-layer protein of Lactobacillus acidophilus ATCC4356: Purification, expression in Escherichia coli and nucleotide sequence of the corresponding gene. J Bacteriol 1993;175:608+6096. 139. Vidgren G, Palva J, Pakkanen R, Lounatmaa K, Palva A. S-layer protein gene of Lactobacillus brevis: Cloning by polymerase chain reaction and determination of the nucleotide sequence. J Bacteriol 1992;174:7419-7427. 140. Obst M. Genetische und physiologische Untersuchungen des Lactosernetabolismusvon Lactobacillus sake und Lactobacillus curvatus. PhD thesis, 1994, Universittit Hohenheim, Germany. 141. David S, Stevens H, van Riel M, Simons G, de Vos WM. Leuconostoc lactis p-galactosidase is encoded by two overlapping genes. J Bacteriol 1992;174475-448 1. 142. Mollet B, Delley M. Spontaneous deletion formation within the P-galactosidase gene of Lactobacillus bulgaricus. J Bacteriol 1990;172:567@5676. 143. Pouwels PH, Leunissen JAM. Divergence in codon usage of Lactobacillus species. Nucleic Acids Res 199+22:929-936. 144. Obst M, Meding ER, Vogel RF, Hammes WP. Two genes encoding the beta-galactosidase of Lactobacillus sake. Microbiology 1996;(In press).
145. Batt CA. Genetic engineering of Lactobacillus. Food Techno1 1986;4:95-98. 146. Schillinger U, Kaya M, Liicke FK. Behaviour of Listeria monocytogenes in meat and its control by a bacteriocin producing strain of Lactobacillus sake. J Appl Bacteriol 1991;70473-478. 147. Wolf G, Strahl A, Schmitz J, Hammes WP.Heme-dependent catalase activity of lactobacilli. Int J Food Microbiol 1991;12:133-140. 148. Scheirlinck T, de Meutter J, Arnaut G, Joos H, Classens M, Michiels F. Cloning and expression of the cellulase and xylanase genes in Lactobacillus plantarum. Appl Microbiol Biotechnol 1990;33:534-541. 149. Cocconcelli PS, Gasson MJ, Morelli L, Bottazzi V. Single-stranded DNA plasmid vector construction and cloning of Bacillus stearothennophilus alpha-amylase in Lactobacillus. Res Microbiol 1991;142:643-652. 150. Scheirlinck T, Mahillon J, Joos H, Dhease P, Michiels F. Integration and expression of alphaamylase and endoglucanase genes in the Lactobacillus plantarum chromosome. Appl Environ Microbiol 1989;55;2130-2 137. 151. Jones S, Warner PJ. Cloning and expression of alpha-amylase from Bacillus amyloliquefaciens in a stable plasmid vector in Lactobacillus plantarum. Lett Appl Microbiol 1990.1 1:214-219. 152. Baik BH, Pack MY. Expression of Bacillus subtilis endoglucanase gene in Lactobacillus acidophilus. Biotechnol Lett 1990; 12:919-924. 153. Posno M, Heuvelmans PTHM, van Giezen MJF, Lokman BC, Leer RJ, Pouwels PH. Complementation of the inability of Lactobacillus strains to utilize D-xylose with D-xylose catabolism-encoding genes of Lactobacillus pentosus. Appl Environ Microbiol 1991;57:2764-2766. 154. Gaier W, Vogel RF, Hammes WP. Cloning and expression of the lysostaphin gene in Bacillus subtilis and Lactobacillus casei. Lett Appl Microbiol 1992;1473-78. 155. Vogel RF, Gaier W, Hammes WP. Expression of the lipase gene from Staphylococcus hyicus in Lactobacillus curvatus Lc2c. FEMS Microbiol Lett 1990;69:289-292. 156. Roy DG, Klaenhammer TR, Hassan HM. Cloning and expression of the manganese superoxide dismutase gene of Escherichia coli in Lactococcus lactis and Lactobacillus gasseri. Molec Gen Genet 1993;239:33-40. 157. Ahmad KA, Stewart GSAB. Cloning of the lux genes into Lactobacillus casei and Streptococcus lactis: phosphate-dependent light production. Biochem Soc Trans 1988;16:1068. 158. Heinrich P, Rosenstein R, Bohmer M, Sonner S, Gotz F. The molecular organization of the lysostaphin gene and its sequences repeated in tandem. Molec Gen Genet 1987;209:563-569. 159. Pouwels PH, Leer RJ. Genetics of Lactobacilli: Plasmid and expression. Antonie van Leeuwenhoek 1993;64:85107. 160. Bhowmik T,Femandez L, Steele JL. Gene replacement in Lactobacillus helveticus. J Bacteriol 1993;175:6341-6344. 161. Klijn N, Weerkamp AH, de Vos WM. Detection and characterization of lactose-utilizing Lactococcus spp. in natural ecosystems. Appl Environ Microbiol 1995;61:788-792. 162. MacCormick CA, Griffin HG, Gasson MJ. Construction of a food grade host/vector system for Lactococcus lactis based on the lactose operon. FEMS Microbiol Lett 1995;l27:105-109. 163. Dickely F, Nilsson D, Hansen EB, Johansen E. Isolation of Lactococcus lactis nonsense suppressors and construction of a food-grade cloning vector. Molec Microbiol 1995;15:839-847. 164. Biswas P, Gruss A, Ehrlich SD, Maguin E. High-efficiency gene inactivation and replacement system for gram-positive bacteria. J Bacteriol 1993;1753628-3635.
01996 Elsevier Science B.V. All rights reserved.
Biotechnology Annual Review Volume 2. M.R. El-Gewely, editor.
15 1
Nitrogen fixing root nodule symbioses: legume nodules and actinorhizal nodules Katharina Pawlowski, Ana Ribeiro and Ton Bisseling Department of Molecular Biology, Agricultural University, Wageningen, The Netherlands
Abstract. Since decades, research has been performed to answer the question whether the ability to form an endophytic symbiosis with N,-fixing bacteria can be transferred to agriculturally important crops. Here, two root nodule symbioses between angiosperms and N,-fixing bacteria, Rhizobium/legumeand symbioses between the actinomycetous bacterium Frankia and actinorhizal plants, will be described. In contrast to Rhizobium, which with one exception, can only enter symbioses with plants of the legume family, Frankia can enter symbioses with plant species from eight different families, mostly perennial woody shrubs. While extensive research has been done on physiological, ecological molecular and genetic aspects of Rhizobium/legumesymbioses,molecular studies on actinorhizalsymbioseshave been started only recently. Nodule development, structure, and metabolism will be compared between both systems, indicating that actinorhizal symbioses represents a more primitive situation with a less sophisticated pattern of signal exchange. The developmental program of actinorhizal nodules shows less differences from the one of lateral roots than it is the case for the program of legume nodules. Also in contrast to legume nodules, there is a considerable diversity in actinorhizal symbioses regarding the differentiation of the endosymbiont in symbiosis and the oxygen protection systems provided by the plant. The implications of this comparison will be discussed. Key words: actinorhiza, Frankia, legume, nodulin, Rhizobium, root hair deformation, root nodule, symbiotic nitrogen fixation.
Symbiotic nitrogen fixation systems Biosphere nitrogen is subjected to a rapid turnover, and part of it is used as a terminal electron acceptor by bacteria, and thereby lost as N, into the atmosphere. A continuous supply with reduced nitrogen from atmospheric N,, is therefore required to maintain the biosphere balance. This can be provided by two processes: chemical reduction in the Haber-Bosch process, or biological N, fixation. However, while chemical nitrogen fixation is cost intensive and about 4CF50% of the nitrogen applied as fertiliser is lost via denitrification, runoff or leaching, only l(t20% of the biologically fixed nitrogen is lost that way [1,2]. Thus, there is a strong interest in a better understanding of biological N, fixation in order to increase agricultural productivity. Biological N, fixation can only be performed by certain prokaryotes which contain genes encoding nitrogenase. This enzyme catalyzes the reaction:
Address for correspondence; Katharina Pawlowski, Department of Molecular Biology, Agricultural University, Dreijenlaan 3, 6703 HA Wageningen, The Netherlands. Tel.: +31-3 17-483278. Fax: +31-317-483584.
152 N,
+ 8H' + 8e- + 16Mg-ATP + 2NH, + H, + 16Mg-ADP + 16Pi
Nitrogenase consists of the homodimeric Fe protein, encoded by the nitrogen fixation (niJ) gene nim, and the tetrameric MoFe protein, encoded by n i p and nijK, which contains the FeMo-cofactor [3]. Since nitrogenase is irreversibly denatured by 0,, the process of N, fixation is highly 0,-sensitive [4,5]. Because of this and the high amount of energy (ATP) necessary for the nitrogenase reaction, the expression of N, fixation systems is strictly regulated and takes place only under nitrogen starved conditions, either under low 0, tension or when special 0, protection systems are provided. Several nitrogen-fixing organisms can form endophytic symbioses with higher plants, where the energy for nitrogen fixation and in most cases the 0, protection system, is provided by the plant partner. Symbiotic N, fixation accounts for 70% of total biological nitrogen fixation [2]. In two groups of symbioses the prokaryotic partners are soil bacteria (rhizobia in legume symbioses and Frunkiu spp. strains in actinorhizal symbioses, respectively), while in the case of the Nostoc-Gunneru symbiosis [6], the cyanobacterium Nostoc is the N,-fixing partner. These systems share some common features: the prokaryotes fix N, living as endophytes inside the cells of special organs of their host, separated from the plant cytoplasm by membranes derived from the plant plasmalemma. In the case of Gunneru, these infected cells are located in specialized stem glands whose development does not depend on the symbiont, while in the case of legumes and actinorhizal plants, the symbionts are hosted in root nodules that are formed by the plant upon infection with the symbiont. Most agriculturally important plant species are belonging to the monocotyledonous plants, for example rice, corn, and wheat. To date, no monocotyledonous plants were known to form endophytic symbioses with N,-fixing bacteria, although nitrogen fixing bacteria like Azospirillum brusilense are associated with the roots of several grasses. However, in this association bacterial nitrogen fixation does not contribute to plant growth, i.e., it cannot substitute for nitrogen fertilizer [7,8]. In contrast, nitrogen-fixing root nodule symbioses lead to independence of nitrogen fertilizer for the plant and are hence of major importance to design strategies by which the ability to form an endophytic symbiosis with N,-fixing bacteria can be transferred to agriculturally important crops like rice. Recently, such a project has been initiated by the International Rice Research Institute in Manila, Philippines [9]. In this chapter we will give an overview of the Rhizobiumllegume symbiosis as well as actinorhizal symbioses. We will describe the Rhizobiumllegume symbiosis because the system is well studied at the molecular level and forms a paradigm for plant-microbe interactions. On the other hand, far less knowledge is available on actinorhizal symbioses. However, Frunkiu bacteria can interact with several plant families while Rhizobium only interacts with leguminosae. Moreover, Frunkiu-induced nodules are in fact modified lateral roots while legume nodules are in general, considered to be unique new organs. Due to the more promiscuous nature of Frunkiu as well as by the more root-like nature of actinorhizal nodules, this system might provide useful clues on how to transfer nodulation ability to other plant species.
153
Root nodule symbioses In this review, the two types of nitrogen-fixing root nodule symbioses, Rhizobiumlegume and actinorhizal symbioses, will be compared. Although the structures of the respective nodules are different, the process of nodule induction involves some steps similar in both types of symbioses. The Rhizobiumllegume symbiosis starts with an interaction between the bacteria and the root epidermis. In general, deformation and curling of root hairs is induced (Fig. lA,B,C). The bacteria become entrapped in the curl and there the host cell wall
Fig. 1. Signal exchange during legume nodule induction. A: a schematic picture of a legume plant; B: a closeup of a root tip. C: the fiist steps of the interaction between legume roots and rhizobia are shown. Flavonoids present in the plant root exudate are binding the rhizobial NodD protein [236] which in turn binds the nod gene promoters and induces the expression of the other nod genes by binding to their promoters (pro) [43]. The nod gene products catalyze the biosynthesis of the Nod factors, that induce the deformation and curling of root hairs on the host plant. Flavonoid structure (for review see (2371): NodD proteins from different rhizobia require specific flavonoids from their respective host plants for optimal activation. For example, hesperetin (R, = OH, R, = OCH,, R, = OH) activates the NodD protein of R. leguminosarum biovar viciae, but not that of R. leguminosarum biovar rrifolii, which is activated by 7-hydroxyflavone (R, = H, R, = H,R, = H). In addition to nod gene-inducing flavonoids, several flavonoids have been identified that inhibit nod gene activation, for example luteolin (R, = OH, R, = OH, R, = OH) inhibits nod gene induction by NodD from R. Ieguminosarum biovar phaseoli, but activates nod gene induction by NodD from R. leguminosarum biovars viciae and trifolii. The stippled double bound is present in luteolin and 7-hydroxyflavone, but not in hesperetin. Nod factor structure (for review see [26,238]): The number of the N-acetylglucosamine residues can vary between three and five. The following substitutions can be found: position R,, -H or methyl group; position R,, acyl group (C16:1, C16:2, C16:3, C18:l or C18:4); position R,, -H, acetyl(O-6), or carbamyl group; position R.,, -H, sulfate, acetyl, (2-0-methyl)fucosyl or D-arabhosyl group; position R,, -H or glyceryl group. A single Rhizobium strain can produce several Nod factors;. for example Rhizobium NGR234 which can nodulate various tropical legumes, synthesizes 18 different Nod factors [239].
154 is hydrolysed and a new tubular structure, the infection thread, is formed. The bacteria invade the root hair and then the root cortex with this infection thread. Meanwhile, cells of the cortex are mitotically activated and form the nodule primordium. Infection thread grows towards this primordium and there rhizobia are released from the tips of the infection threads into the cytoplasm of the plant cells. This is an endocytotic process by which the bacteria become surrounded by a membrane derived from the host plasma membrane. In some cases, the bacteria do not enter the plant via root hairs, but between epidermal cells (“crack entry”). Frankia bacteria induce nodulation in their host plants in a slightly different way. In some interactions, root hairs are invaded by the formation of tube-like structures that resemble the infection threads in legumes. In other cases, intercellular penetration of the root and colonization of the intercellular spaces takes place. After root hair infection, like Rhizobium, Frankia induces mitotic activity in the root cortex but additionally, cell divisions are induced in the pericycle. The latter center of mitotic activity develops into a root nodule, and like in legume nodules, specialized cells become fully packed with the microsymbiont that again is surrounded by a membrane derived from the plasma membrane of the host.
Rhizobium-legume symbioses Although leguminous plants have been used for soil enrichment by green manuring for centuries, it was first discovered in 1888, that bacteria living in symbiosis with the plant are responsible for the reduction of atmospheric N, to ammonium [10,11]. Gram-negative soil bacteria, members of the family Rhizobiaceae (including the genera Rhizobium, Bradyrhizobium and Azorhizobium), induce the formation of root nodules on their leguminous host plants. These symbioses show different degrees of host specificity, ranging from the stem-nodulatingAzorhizobium caulinodans ORS571 that can only interact with the tropical leguminous shrub Sesbaniu rostrutu [12], to the wide host range strain Rhizobium spp. NGR234 which can induce nodules on several different tropical and temperate legume species and even on one nonlegume, Parasponia (Ulmaceae) [ 131, In the last decades, considerable research has been devoted to the understanding of this symbiosis, not only because of the importance of biological nitrogen fixation to agriculture, but also because it provides insights in mechanisms controlling plant development.
Formation of legume nodules Plant genes involved in nodulation. During legume nodulation, plant genes play an important role and these genes have been studied by genetic and molecular approaches. Genetic studies have revealed that certain plant genes (sym genes) are required during all stages of nodulation (for reviews see [14-161). The phenotypes of the sym mutants show that the products of these genes are involved in, e.g., root hair deformation [17], infection [18], and bacterial release [191. To date, none of the
155 sym genes have been cloned, but several researchers have started programs to isolate sym genes by a positional cloning strategy [20-221. Molecular studies have concentrated on the plant genes whose expression is induced during the consecutive stages of nodulation (for review see [23-261). Here, in most cases, the genes have been cloned, but their functions in the nodulation process are still poorly understood. Some of these genes are not expressed in any plant organ other than nodules and are termed nodulin genes [27]. The recent use of more sensitive methods to detect gene expression, has shown that several genes which were thought to be nodule specific, are actually expressed in other organs also. For convenience, they are still being called nodulin genes in this review. Nodulin genes are thought to be derived, either from the duplication of genes involved in nonsymbiotic processes (nodulin genes sensu strictu) like in the case of leghemoglobins [28] and nodulin-26 [29], or to be genes recruited from other developmental programs, like the early nodulin genes ENODZ2 [30] and ENOD40 [31,32]. Nodulin genes expressed before the onset of nitrogen fixation are called early nodulin genes (ENOD) and are probably involved in building up the nodule structure and in the infection process. Nodulin genes expressed at or after the onset of nitrogen fixation, are termed late nodulin genes (NOD) [33]. In general, late nodulins are involved in the metabolic specialization of the nodule. For example, leghemoglobin is the most abundant late nodulin in legume nodules. It works as an 0, carrier in the central tissue of nodules, transporting 0, to the sites of respiration [34]. Bacterial Nod factors. The rhizobial signal molecules that induce the early steps of legume nodulation, are the so-called Nod factors which are lipchito-oligosaccharides containing a backbone of 4-5 N-acetylglucosamine residues and a fatty acid at the nonreducing terminal sugar residue (Fig. 1C)(for reviews see [35-371). All rhizobia secrete similar lipochito-oligosaccharides, but their host specificity is determined by substitutions at the terminal sugar residues. For example, in the case of R. meliloti, the major host determinant is a sulfate group at the reducing sugar residue [38,39]. The biosynthesis of Nod factors is mediated by enzymes encoded by the rhizobial nod genes (for reviews see [40,41]), whose expression is induced by flavonoids excreted by the plant roots (Fig. 1C)[42]. One of the nod genes, IzodD, is expressed constitutively, and upon binding of host flavonoids, the NodD protein activates the transcription of the other nod genes (Fig. 1C) [43]. Nod factors can induce several responses in the host plant [39,44-481, as will be described in the following parts of this review. In some cases, as in Medicago sativa and Glycine soja, purified Nod factors are even sufficient to induce the formation of bacteria-free nodules [35,37,49]. Root hair deformation. Basically, two different ways of rhizobial infection are known: infection through deformed root hairs, and infection via so-called crack entry, where the bacteria enter the plant root through gaps in the epidermis [50,51] or enter between intact epidermal cells [52]. The latter mode of infection only occurs in some tropical legumes. The infection through deformed root hairs is the most frequently
156 used way, and will be described in detail in this chapter. During root hair infection, the first microscopically visible response of the host plant on rhizobial infection, is the deformation and curling of root hairs (Figs. 1C and 2A) [53]. Microscopical studies have shown that root hair deformation is due to a new induction of root hair tip growth by the Nod factors [54]. This process is accompanied by the induction of several host genes. Examples are Mtripl [55], encoding a peroxidase and the early nodulin genes ENODS and ENODIZ that encode proline rich polypeptides, which probably represent cell wall components [30,56]. Purified Nod factors can induce root hair deformation, and also the expression of the above mentioned plant genes [39,47,57]. Also nodulation by crack entry depends on Nod factors [58]. Nod factors act in concentrations as low as lo-'* M, suggesting that they are recognized by a receptor in the root epidermis [54]. Studies on gene induction by Nod factors, have shown that the length of the N-acetylglucosamine backbone as well as the modifications at the terminal sugar residues, are crucial for
Fig.2. Induction of the nodule primordium via root hair deformation in legumes and actinorhizal plants. The different steps in nodule primordium induction are indicated in root cross sections. A: Induction of an indeterminate legume nodule. Stage I shows an uninfected root. Stage 11, Nod factors secreted by the bacteria induce deformation and curling of root hairs. Stage 111, after root hair deformation, an infection thread is formed in the curled root hair by which the bacteria enter the plant. At the same time, cell divisions are induced in the inner cortical layers. The outer cortical cells form preinfection thread structures preparing the passage,of the infection thread. Stage IV, the infection thread has reached the nodule primordium in the inner cortex and cells of the primordium become infected by Rhizobium. (Abbreviations: E = epidermis consisting of atrichoblasts and trichoblasts (forming root hairs); EN = endodermis; VB = vascular bundle consisting of pericycle (outer layer), phloem, cambium, and xylem.) The protoxylem cells are indicated. B: Induction of an actinorhizal nodule. Stage I shows an uninfected root. Stage 11, after root hair deformation, an infection thread-like structure is formed by which bacterial hyphae enter the plant, encapsulated in plant cell wall material. Concomitant with formation of the infection thread-like structure, cell divisions are induced in the outer cortical layers. Stage 111, the encapsulated hyphae have grown towards the dividing cortical cells and infected them, resulting in the formation of a prenodule. Cell divisions are induced in the pericycle of the nodule vascular bundle. Stage IV, the encapsulated hyphae grow from the prenodule to the nodule primordium and infect cells of the primordium.
157 the induction of responses in the root epidermis [39,47]. However, since the presence of a fatty acyl moiety is essential but its structure is not important, probably the receptor does not recognize this part of the molecule. Cytological studies have shown that the expression of ENODl2 and Mtripl is induced in all epidermal cells of a zone of the root, starting above the root tip even before root hairs have emerged, and extending to the region containing mature root hairs [55,59]. A direct contact between Nod factors and epidermal cells is required for the induction of such genes [39]. Thus, it is likely that within the zone of the root able to respond to Nod factors, they are recognized by all epidermal cells, not only by those containing root hairs. The function of the Nod factor induced plant genes is unclear. However, it is unlikely that all of these genes are essential for the infection process, since alfalfa plants lacking ENODl2 can form effective nodules when inoculated with Rhizobium [60].
Infection thread formation. When rhizobia induce root hair curling, they become entrapped in the curls. There they induce the formation of infection threads in the crooks of curled root hairs, beginning with a local hydrolysis of the plant cell wall (Fig. 2A) [61,62]. At the site of hydrolysis the plasma membrane grows inward and new cell wall material is deposited along the invaginating plasma membrane ([62,63]; for reviews see [64,65]). This way, a tubular structure, the so-called infection thread, is formed by which the bacteria enter the plant [66]. The mechanism by which this local hydrolysis of the plant cell wall is achieved is unclear, but it seems unlikely that hydrolytic enzymes secreted by the bacteria can establish such a localized effect. Hence, it has been suggested that the bacteria induce the local secretion of hydrolytic enzymes by the plant which also happens when a trichoblast forms a root hair (for review see [65]). Bacteria inside the infection thread are surrounded by a matrix, which contains (glyco-) proteins and other compounds of the plant as well as compounds secreted by the bacteria [67]. The infection thread wall is most likely of plant origin, and has an ultrastructure similar to that of the plant cell wall [68]. The products of the early nodulin genes ENODS and ENODZ2, have been suggested to be involved in infection thread formation, since ENODS and ENOD12 are transcribed in cells containing an infection thread tip, and ENOD12 expression is also induced in the dividing cortical cells in front of the infection thread [30,56]: It has been proposed that infection thread growth resembles the development of root hairs, but the direction of growth is inverted [46]. Thus, the mechanisms controlling initiation as well as growth of the infection thread, might be derived from root hair development. During infection thread formation, root cortical cells are mitotically activated and form nodule primordia (see below). The infection thread grows toward the base of the root hair and subsequently toward the nodule primordium. If the primordium is formed by inner cortical cells (see below), the infection thread has to cross the outer cortex to reach the nodule primordium. Prior to infection thread penetration, the cortical cells between the infected root hair and the nodule primordium are activated and form radial tracks of cytoplasm (Fig. 2A) [62]. Such cytoplasmic structures are
158 called “preinfection threads” and resemble phragmosomes [46]. Therefore, it was postulated that the cells forming a preinfection thread enter the cell cycle, although they do not divide [46]. Studies on expression of cell cycle specific genes have proven that the cortical cells forming a preinfection thread indeed enter the cell cycle and become arrested in the G , phase [69]. Thus, preinfection thread formation is derived from the cell cycle machinery. The infection thread penetrates root cortical cells by local hydrolysis of the cell wall and grows through the preinfection thread structures to the nodule primordium, where bacteria are endocytotically released into the plant cells and differentiate into their symbiotic form, the bacteroids [70,7 11. When the plants are not infected through deformed root hairs, but by crack entry, the situation is less uniform with regard to infection thread formation. In case of Neptunia and during stem nodule’ induction on Sesbania rostrata, infection threads are formed when the bacteria have entered the plant [72-741, while in roots of Arachis hypogaea, intercellular infection centers develop and bacteria enter the cells of the nodule primordium directly via invagination of the plant plasma membrane, without infection thread formation [75]. In Stylosanthes and Aeschynomene, a similar process takes place without the formation of infection centers [50,76]. Cortical cell divisions and nodule meristem formation. Concomitant with infection of root hairs, root cortical cells, mostly opposite a protoxylem pole of the root stele [77], are activated and start dividing (Fig. 2A). Several nodulin genes are expressed in the dividing cortical cells. Examples are ENODI2 [30], ENOD40 [31,32,78,79] and GmN93 [31]. Which of the root cortical cells divide, is determined by the plant [80,81]. In temperate legumes such as pea, vetch and alfalfa, inner cortical cells divide and form the nodule primordium [7 1,821. When the infection threads reach the primordia, they ramify, and cells at the base of the primordium are infected. At the same time, a meristem is formed at the distal part of the primordium, consisting of small cells with dense cytoplasm [82]. The nodule meristem differentiates during the complete nodule life time into the different cell types that build up the nodule. Consequently, these nodules have an indeterminate development like lateral roots. Nod factors have the ability to induce the formation of nodule primordia [48]. For this action, they have to be present in higher concentrations, and to fulfill more stringent structural requirements than for the induction of responses, in the root epidermis [83]. Thus, it has been postulated that at least two different Nod factor receptors are present in the root epidermis: a “signaling receptor” involved in the induction of reactions in the epidermis, and an “uptake receptor” that initiates the infection process and is activated only by a very specific structure [83]. However, the mechanism by which they induce mitotic activity in the cortex, is not completely understood. Cytokinin and compounds that block the polar transport of auxin, phenocopy the Nod factors, since they can cause the formation of nodule-like structures [84,85]. Therefore it is assumed that Nod factors cause a change in the cytokinin/auxin balance which subsequently results in the mitotic reactivation of cortical cells. Before cell division occurs, expression of the early nodulin ENOD40
159 is induced in the regions of the root pericycle opposite to a protoxylem pole (W.-C. Yang and T. Bisseling, unpublished results). Thus, a tissue even deeper inside the root responds faster to Nod factors than the cortex. It is possible that ENOD40 expression in the pericycle is involved in a process which leads to a change in hormone balance or perception in the cortex, which finally causes cortical cells to divide. This hypothesis is based on the observation that ENOD40 is affecting the response to auxin when expressed in tobacco protoplasts (K. Pawlowski, K. van de Sande, R. Walden and T. Bisseling, unpublished results). In tropical legumes such as soybean, outer cortical cells of the roots divide to form the nodule primordium, while the inner cortical cells between the primordium and the stele are activated to divide and will in turn form the connection with the vascular bundle of the root. The growing infection threads directly invade primordium cells after penetrating the root hair (for review see [sl]). Cells at the periphery of the primordium remain mitotically active and form a spherical meristem which loses its activity at an early stage of development. Thus, these nodules have a determinate growth pattern and are called determinate nodules. Infection of cells by Rhizobium. When the infection thread has reached the nodule primordium, bacteria are released from the infection thread into the cytoplasm. During the release, the bacteria become surrounded by a plant-derived membrane, the peribacteroid membrane (PBM; Fig. 3A,B). The bacteria, together with the space within the PBM (peribacteroid space, PBS), and the PBM, form a functional structure called symbiosome [86]. The PBM works as an interface between both symbiotic partners, controlling the metabolite exchange. In accordance with its specialized function, it is different from the plasma membrane, from which it is derived, in phospholipid and protein composition [24,87]. It has been suggested that the PBM has obtained some properties of the membrane of the vacuole [24,88,89] since within the PBS, hydrolytic enzymes have been found which are also present in vacuoles [90,91]. An integral PBM protein, nodulin-26, is targeted to the vacuolar membrane when expressed in tobacco, supporting the hypothesis of the similarity between PBM and vacuolar membrane [92]. In the same line of argument, it was also proposed that the symbiosome resembles a lytic compartment, similar to the vacuole, that the bacteroides continuously have to neutralize by exporting ammonia, be a product of nitrogen fixation, in order to avoid being degraded by the plant [93]. Thus, rather than a symbiosis, the interaction between rhizobia and legumes would constitute a case of parasitism of the plant on the bacteria. This hypothesis is supported by the fact that for Rhizobium mutants unable to fix nitrogen, premature degradation of bacteroides can be detected in the infected cells of the nodules [94]. Legume nodule structure Determinate and indeterminate legume nodules have a similar tissue organization, a central tissue where bacteria are hosted, surrounded by several peripheral tissues (Fig. 4A) (for review see [25,64,71]). The peripheral tissues comprise the nodule cortex, the endodermis and the nodule parenchyma [95]. The latter tissue harbors the
160
Fig. 3. Nitrogen-fixing endosymbionts in Rhizobiumllegume and actinorhizal symbioses. A: Intracellular rhizobia in a nodule formed on clover by R. trifolii. This region of the indeterminate clover nodule shows the transition of the prefixation zone II to the interzone II-III. In the upper cell (11, prefixation zone), intracellular bacteria (b) have not yet differentiated into their nitrogen-fixing form. The bottom cell (11-III, interzone) contains amyloplasts (a), and nitrogen-fixing bacteroides (ba) have differentiated. In both cells, intracellular bacteria are surrounded by a peribacteroid membrane. Bar = 1 pm. B: Detail - the bacterial membrane (b) and the peribacteroid membrane (p) which separate the bacteroides (ba) from the cytoplasm (cy) can be clearly distinguished. Bar = 500 nm.The photographs were kindly provided by U. Bialek and A. van Lammeren, Department of Cytology, Agricultural University Wageningen, The Netherlands. C: Intracellular Frunkiu in a nodule formed on Alnus serrulutu. Vegetative hyphae (h) and nitrogen fixing vesicles (v) can be seen. Arrows point at the lipid envelope of a vesicle (e) and at a sept in a vesicle (s). Bar = 1 pm. D: Detail - a vesicle is separated from the plant cytoplasm (cy) by its own membrane (b), the lipid envelope (e) and the invaginated plant plasmamembrane (p). Arrowheads point at the plant cell wall-like encapsulation material between plant plasmamembrane and vesicle lipid envelope. Bar = 500 nm.Photographs were kindly provided by H.M. Berg, Biology Department, University of Memphis, Tennessee, USA.
161
Fig. 4. Structure of indeterminate root nodules. A: Scheme of an indeterminate legume nodule. Zonation: I, nodule meristem; 11, prefixation zone; 11-HI, interzone; III, nitrogen fixation zone; IV, senescence zone ([93]; see text). B: Scheme of an actinorhizal nodule from Ahus glutinosa. Zonation: 1, nodule meristem; 2, infection zone; 3, nitrogen fixation zone; 4, senescence zone ([144]; see text). The central vascular bundle contains a multilayered pericycle [136,219]. Due to the activity of the apical nodule meristem, a developmental gradient of infected cells forms in the central tissue (A) or in the cortex (B), respectively. The zones in which bacterial nitrogen fixation takes place are indicated for both types of nodules.
nodule vascular bundles. The central tissue consists of two cell types, infected and uninfected cells. The infected cells are fully packed with bacteria. A few cell layers of uninfected cells, the boundary layers, separate the central tissue from the nodule parenchyma [25,80]. Meristems of indeterminate nodules go on differentiating into the different nodule tissues. The effect is that the central tissue can be divided into several zones representing successive stages of development (Fig. 4A). A nomenclature has been developed for the successive zones of indeterminate nodules [96]. The meristem at the apex is designated as zone I. It consists of small cells with dense cytoplasm that are not infected by rhizobia. This zone is immediately followed by & prefixation zone II. In the distal part of this zone 11, infection threads penetrate meristematic cells and bacteria are released into the plant cytoplasm. In the proximal part of the prefixation zone 11, plant cells elongate and symbiosomes proliferate. The interzone 11-111 is characterized by the start of starch accumulation in infected cells, and the presence of differentiated bacteroides (Fig. 3A,B) [96]. It is also marked by dramatic changes in gene expression by both plant and bacteria. For example, the bacteria induce expression of the N,-fixation genes (Fig. 5A,B) [97], while the expression of bacterial ropA encoding an outer membrane protein is switched off [98]. The expression of the plant nodulin genes ENODS and ENOD40 [30,79] is strongly reduced at this transition, whereas several other nodulin genes like NOD6
163 Fig. 5. In situ localization of gene expression in indeterminate legume nodules. In A and C, bright field microscopy was used; silver grains denoting hybridization appear in black. In B and D,darkfield microscopy was used; silver grains are visible as white dots. Due to the very high density of silver grains in some infected cells of both sections, the light scattering by darkfield illumination is impaired. A and B:Expression of a Rhizobium leguminosarum nitrogenase structural gene, n i f l , in a longitudinal section of a 15-day-old pea nodule. The different zones of the developmental gradient are indicated: m, nodule meristem; p. prefixation zone; i, interzone; f, fixation zone. A senescent zone has not yet developed. R. Ieguminosarum n z f l expression starts in the first cell layer of the interzone. A nodule vascular bundle (v) is indicated. An arrow points at a protoxylem pole of the root vascular bundle. The root cortex (rc) is labeled. C and D: Expression of a pea leghemoglobin (Ib) gene in an adjacent section of the same nodule lobe. lb gene expression starts in the prefixation zone. The beginning of the interzone, i.e., the zone of R. leguminosarum nijW induction, is indicated by arrowheads. Bar = 500 pm.
are induced [99].The signal or mechanism that controls this developmental switch is not yet understood, although there is evidence that the 0, concentration is involved in the induction of bacterial nif genes ([ 1001;see below). In the nitrogen fixation zone 111, the plant cells have reached their maximal size and bacteroides are fixing nitrogen. In older nodules a senescent zone IV is present. Senescence of nodule tissues has hardly been studied at the molecular level. Based on analogy to other senescent organs, it is likely that the expression of genes encoding hydrolytic enzymes like proteases and RNases will be induced in this zone. Indeed proteases, e.g., thiol proteases, have been found to be active in senescent nodules ([ 101,1021 and references contained therein). A nodulin gene specifically expressed in senescent nodules, has been isolated from winged bean and found to encode a proteinase inhibitor [ 1031.Protease-inhibiting activity has also been found in the peribacteroid space of soybean nodules [ 1041.These data suggest that the plant has developed a system to control bacteroid senescence.
Actinorhizal symbioses A rather diverse group of plants from eight different families have the ability to establish a symbiosis with Frankia bacteria resulting in actinorhizal root nodules. Up to now, about 194 actinorhizal plant species from 24 genera have been identified (for review see [ 1051). Frankia is a filamentous gram-positive actinomycetms bacterium (reviewed in [106,107]).In contrast to Rhizobium, Frankia normally grows in hyphal form, being able to form also two other specialized cell types, namely vesicles, the sites of N, fixation (see below), and sporangia. Although actinorhizal nodules were f i s t described in 1829 [log],only in 1895 it was shown that they contributed to the nitrogen nutrition of the plant [109].The identification of the microsymbiont as an actinomycete finally took place in the 1930s [110,111]. Due to their symbiosis with Frankia, actinorhizal plants can grow on marginal soils. They are used in soil reclamation and reforestation, for timber-, fuelwood-, and pulp production, as windbreak plants in desert agroforestry systems and also to stabilize coastal sand dunes in tropical and subtropical countries [ 112-1201.
164 Frankiae have not been classed into species thus far. Physiological criteria could only be used to define two broad groups of strains [121]. As in the case of Rhizobium-legume symbioses, Frankiae show different degrees of host specificity, but here the attempt to use the host specificity as a taxonomic criterion has proven impracticable [ 1221. Meanwhile, Frankia strains have been isolated from several host plants and can be grown in culture (for review see [lOS]). During isolation, some strains require the addition of a root steroid, dipterocarpol, for initial growth stimulation [ 123,1241. The inability to isolate Frankia strains from some actinorhizal plant families, for example Datiscaceae and Coriariaceae, might reflect special requirements of these strains for growth stimulation. For the strains cultured thus far, four major host specificity groups have been defined [125]: group 1 includes those strains capable of nodulating Alms, Comptonia, Myrica and Gymnostoma, group 2 includes strains inducing nodules on Casuarina, Allocasuarina, Myrica and Gymnostoma, group 3 includes Frankia strains able to nodulate Elaeagnus, Hippophae, Shepherdia, Myrica and Gymnostoma, and group 4 includes strains that are able to nodulate either Alnus and Comptonia, or Casuarina and Allocasuarina, or Elaeagnus, Hippophae, and Shepherdia, but not Myrica or Gymnostoma. Strains which do not fit into this scheme are referred to as atypical, an expression also used for strains that are not able to reinfect their host plant from whose nodules they were isolated (for review see [lOS]). Induction of actinorhizal nodules As in Rhizobiumllegume symbioses, there are two ways known by which Frankia can initiate a symbiotic relationship with a compatible host plant, namely root hair infection, observed in Alnus, Casuarina, Comptonia, and Myrica [ 126-1281 and intercellular penetration which has been reported for Elaeagnus and Ceanothus [ 129,1301. Also in actinorhizal symbioses, the mode of infection is plant-determined [ 129,1311. Root hair infection starts with the deformation of the root hairs (Fig. 2B) [132]. There is no need for direct contact between the host plant cells and Frankia in order to induce root hair deformation [133,134]. Therefore it has been suggested that, in analogy to the signalling between Rhizobium and legumes, plant root exudates may stimulate the synthesis and/or release of a diffusible “Nod” factor by Frankia that in turn causes root hair deformation [ 1341. To date, the characterization of Frankia “nod genes” has not been reported, but it has been observed that (a) factor@)present in the supernatant of a Frankia culture can cause root hair deformation on Alnus glutinosa (M. van Ghelue, E. Lfivaas,E. Ring@and B. Solheim, personal communication). This suggests that like in the Rhizobiumllegume symbiosis, the interaction indeed is initiated by an exchange of signals between the two symbionts. Upon root hair deformation, Frankia hyphae associated with deformed root hairs, initiate digestion of the primary root hair cell wall and as a response, the host plant starts to build up a cell wall-like matrix around the microsymbiont (Fig. 2B) [135]. In this way, a tubular ingrowth, termed encapsulation, is created which functionally resembles the infection thread observed in the Rhizobium-legume symbiosis, and like the latter, grows through cortical cells (Fig. 2B) [132]. However, no equivalent of the
165 infection thread matrix exists in actinorhizal symbioses, but the hyphae are surrounded by the cell wall-like material of the encapsulation, equivalent of the infection thread wall in Rhizobiumbegume symbioses. In response to the invading microsymbiont, root cortical cells proximal to the infected root hairs start to divide and enlarge, giving rise to the so-called prenodule whose cells enlarge further upon infection by encapsulated hyphae (Fig. 2B) [ 1361. Thereupon, cell divisions are induced in the pericycle resulting in the formation of a nodule lobe primordium, that upon infection develops into a nodule lobe. Thus, while initially, Frunkiu, like Rhizobium, induces cell divisions in the nodule cortex, the final nodule primordium is formed in the root pericycle like a lateral root primordium. Like in Rhizobiuml legume symbioses, the primordia of actinorhizal root nodules are formed mostly opposite to a protoxylem pole of the root stele [137]. Encapsulated hyphae grow from the prenodule towards the nodule primordium, thereby again crossing cortical cells [ 1361. After entering the nodule primordium, Frunkiu hyphae infect part of the primordium cells. During infection, the plant plasma membrane invaginates and encapsulating material is continuously deposited around the growing hyphae (for review see [ 1321). Thus, like Rhizobium, Frunkiu is surrounded by a membrane derived from the plant plasma membrane when it is present in the host cell (Fig. 3C,D). However, in contrast to Rhizobium, Frunkiu bacteria remain in the infection thread-like structures and are not released endocytotically. After infection, Frunkiu hyphae grow until they occupy most of the volume of the infected cell. Then, specialized vesicles are formed in which nitrogenase is expressed (Fig. 3C,D) [138,139]. From now on, new cortical cells are formed from the nodule meristem and these become infected by hyphae progressively. In this way a nodule with an indeterminate growth pattern is formed (Fig. 4B). The process of infection by intercellular penetration is more primitive. Frunkia hyphae enter the root by partial digestion of the middle lamella between adjacent epidermal cells and move on strictly intercellularly [ 129,1311, while epidermal and cortical cells secrete some pectinaceous and proteinaceous material into the intercellular space [ 1401. No prenodule is formed, but immediately upon intercellular colonization of the root cortex, cell divisions are induced in the root pericycle resulting in the formation of the nodule primordium. While in Rhizobiumbegume symbioses, only a few cases of infection by crack entry are known, in actinorhizal symbioses infection by intercellular penetration seems to take place in most actinorhizal plant families except for Betulaceae, Myricaceae and Casuarinaceae. In legumes it has been shown that before an infection thread traverses a cortical cell, a dramatic rearrangement of the cytoplasm occurs. In a normal cortical cell the cytoplasm including the nucleus, is located at the periphery of the cell. Before a cortical cell is penetrated by the infection thread, the nucleus moves to the center of the cell and the cytoplasm obtains a radial polar organization which is named preinfection thread. The preinfection thread forms the pathway that the infection thread follow on their way to the nodule primordium, and the polar organization of the cytoplasm seems to be essential to support the polar growth of the infection thread [46]. Preinfection threads are reminiscent of phragmosomes, suggesting that
166 the corresponding cells enter the cell cycle and become arrested in the G, phase. This hypothesis was confirmed by analysing the expression of cell cycle specific genes [69]. Reinfection thread formation has not been studied in actinorhizal nodulation but it seems very likely that in case of infection via root hair deformation, preinfection threads have to be formed also here. Therefore it is striking that infection via root hairs is correlated with the induction of cell divisions in the cortex, although no specific function has been assigned to the dividing cortical cells forming the prenodule. We hypothesize that in actinorhizal plants, when infection threads traverse cortical cells, preinfection thread structures have to be formed, implying that cortical cells enter the cell cycle and become arrested in the G, phase. However, in some cortical cells this arrest is not established, leading to cell division, although this is not functional in the infection process. This hypothesis is supported by the fact that after infection of Alnus by Frankia, irregular undulated cell walls in combination with bundled arrays of microtubules, were found in postmeristematic cells proximal to the root meristem and the nodule meristem [ 1411, indicating nonfunctional activation of the cell cycle machinery.
Structure of actinorhizal nodules Actinorhizal nodules are perennial structures consisting of multiple lobes [ 1421. By ontogeny, as well as by tissue organization, each nodule lobe represents a modified lateral root with a central vascular cylinder. However, actinorhizal nodule lobes differ from lateral roots, in that they lack a root cap, have a superficial periderm, and contain both infected and noninfected cortical cells (Fig. 4B) (for review see [132,143]). Like in the case of temperate legumes, actinorhizal nodule lobes have an indeterminate growth pattern due to the presence of an apical meristem that differentiates continuously in a proximal direction [ 1321. Depending on the developmental stage of the infected cortical cells, a zonation of the nodule lobe can be observed (Fig. 4B) [144]. Thus, starting from the distal end of the lobe, four zones can be distinguished. Zone 1, the meristematic zone, consists of small dividing cells that do not contain bacteria. Zone 2, the infection zone, corresponding to the prefixation zone in legume nodules, contains enlarging cortical cells, some of which Fig. 6. In situ localization of gene expression in actinorhizal nodules of Alnus glufinosu. In A and C, bright field microscopy was used; silver grains denoting hybridization appear in black. In B and C, darkfield microscopy and epipolarized light were used; silver grains are visible as white dots. A and B: Expression of a Frunkia nitrogenase structural gene, nijH, in a longitudinal section of a nodule lobe. The different zones of the developmental gradient are indicated: 1, meristematic zone; 2, infection zone; 3, fixation zone; 4, senescence zone. Arrowheads point at infected cells of zone 2 that are not yet completely filled with hyphae and do not contain vesicles. These cells show no Frunkia nijH expression. The central vascular bundle (v) and the periderm (p) are indicated. C and D: Expression of A. glufinosu ug12 in an adjacent longitudinal section of the same nodule lobe. Arrowheads point to infected cells of zone 2 that are not yet completely filled with hyphae; these cells show high ug12 expression levels. An infected cell of zone 3 showing ug12 expression at a high level is marked by a star. An arrow points at an adjacent cell which shows little ag12 expression. Bar = 500 pm. This is a modified version of Figure 3C-3F of Ribeiro et al. [144], reprinted with the permission of the American Society of Plant Physiologists.
167
168 are infected and in turn, enlarge more than uninfected cells while being gradually filled with hyphae from the center outward [ 139,1451. Once the infected cells are completely filled with hyphae, provesicles are formed as terminal swellings on hyphae'or on short side branches [146]. In zone 3, the fixation zone, provesicles differentiate into vesicles. During this step of differentiation, the synthesis of nitrogenase, the enzyme responsible for the reduction of atmospheric nitrogen in ammonia, is induced [147]. The expression of the structural nif genes encoding nitrogenase is a marker for the shift from zone 2 to zone 3 (Fig. 6A) [144,148]. In zone 4, the senescence zone, cortical cells become senescent and the microsymbiont as well as the host cytoplasm is degraded. This zonation of the nodule cortex has been found to be applicable to nodules of Alnus glutinosu, Casuarinu gluucu and Ceunothus griseus, where infected cells are distributed over the cortex. However, in Dutiscu and Coriariu, where only a defined area of the cortex can be infected (see below), the developmental pattern is more complicated ([144]; K. Pawlowski and A.M. Berry, unpublished results). Due to the structural similarities between actinorhizal nodules and lateral roots, the products of nodule-specific genes can be expected to be involved in either of three processes. First, genes whose products are involved in the developmental shift from lateral root to nodule development, would determine the difference between lateral root- and nodule meristems and therefore should be differentially expressed in the respective meristems. Second, genes whose products are involved in the infection process should be expressed in the young infected cells. Third, there could be nodulespecific genes whose products are involved in the metabolic specialization of the nodule, i.e., in the assimilation of the ammonium exported by symbiotic Frunkiu, or in the transport and synthesis of carbon sources for the bacteria. Since actinorhizal plants mostly represent woody shrubs or trees, recalcitrant to molecular biological analysis [ 1491, their nodule-specific genes have not been examined as thoroughly as those of legumes. Only recently, actinorhiza-specific genes have been cloned from Cusuarina [ 150,1511 and A h u s [ 144,152,1531. One nodule-specific gene from Alnus, ug12, was found to be expressed at the highest levels in the infected cells of the infection zone 2, i.e., in cells where nifgenes have not been induced yet (Fig. 6C,D) [144]. Ag12 encodes a serine protease which thus seems to be involved in the infection process. Another nodule-specific gene family has been found ([154]; K. Pawlowski, C. Guan and T. Bisseling, unpublished results) showing an expression pattern similar to that of ag12. These genes encode glycine-rich proteins with a signal peptide indicating that they might be localized in the cell wall. Thus, the infection process and the interaction with the bacterial symbiont, appear to involve sets of nodule-specific genes. Several other genes were found to be expressed at elevated levels in Alnus nodules compared to roots [1531. Their products mostly were involved in nodule nitrogen and carbon metabolism, i.e., in the metabolic specialization of nodules. No member of the putative group of genes important for the developmental shift from root to nodule development has been identified yet.
169 Actinorhizal and legume nodule metabolism Legume nodules as well as actinorhizal nodules have to provide a suitable environment for nitrogen fixation by the endosymbiont, i.e., they have to protect bacterial nitrogenase from O,, supply the intracellular bacteria with carbon sources and assimilate the product of N, fixation, ammonium, which is exported by the bacteria. Furthermore, in the context of the complete plant, nodules represent carbon sinks and nitrogen sources: efficient transport of carbon sources to the nodules and of nitrogen sources from the nodules has to be provided. Different strategies have been adopted to fulfill these requirements that will be discussed in the following paragraphs. Oxygen protection of bacterial nitrogenase While nitrogenase itself is 0, sensitive, the high amount of energy required for the nitrogenase reaction has to be generated by oxidative processes, leading to a high demand for 0, in nodules. To meet these conflicting demands, different strategies have been developed. For legume nodules, physiological studies have shown that the nodule parenchyma forms an 0, diffusion barrier (Fig. 4A) [ 1551. This, together with the high 0, consumption rate of Rhizobium, leads to a low 0, concentration in the central tissue of the nodule, while in the infected cells, high levels of the 0,-carrier protein leghemoglobin facilitate 0, diffusion to the sites of respiration (Fig. 5C,D) [ 156,1571. Since in indeterminate legume nodules, the nodule parenchyma is interrupted by the meristem at the distal end of the nodule, an 0, gradient is formed (Fig. 4A). Data on bacterial nitrogenase gene expression in the free-living state have shown that nitrogenase expression is induced by low 0, tension [158]. It has been suggested that this type of regulation may also play a role in symbiosis, where nitrogenase gene expression starts in the first layer of interzone II--III (Fig. 5A,B). In fact, when overall 0, concentration in alfalfa nodules was reduced by submerging the nodules in agar, the nitrogenase structural gene nifi was expressed also in the prefixation zone [loll, confirming the role of 0, in nifregulation. 0, is generally assumed to diffuse via the intercellular spaces, because its diffusion is about lo4 times faster in air than in water. The 0, diffusion barrier in the nodule parenchyma is established by cell layers in which the size of the intercellular spaces can be controlled [157], presumably by the release and uptake of intercellular water in the nodule parenchyma [ 15!&161]. Furthermore, nodulins like ENODZ which are specifically expressed in the nodule parenchyma might contribute to the formation of the 0, diffusion barrier [95]. Actinorhizal nodules are structurally rather diverse (for review see [143]). An example of this diversity are the 0, diffusion pathways. In order to provide 0, access to the sites of respiration, i.e., N,-fixing Frankia vesicles [162] and plant mitochrondria, 0, has to pass through the nodule periderm and reach the infected cells via intercellular spaces. To provide 0, access through the periderm, two strategies have been developed: either the periderm is disrupted by lenticels like in some legume nodules, or agraviotropically growing nodule roots containing large air spaces, are
170 protruding from the lobes [163]. The 0, concentration can affect nodule anatomy, such as causing changes in the size of lenticels in Alnus and Coriaria, in the thickness of the periderm in Coriaria or in the length of nodule roots in Myrica [ 164-1661. There is also variability in the arrangement of infected cells in the cortex. While the infected cells are distributed over the nodule cortex, interspersed with uninfected cells, in nodules formed by Alnus, Casuarina, Ceanothus and Myrica, in nodules formed by Coriaria or Datisca they are arranged in a continuous kidneyshaped patch at one side of the acentric stele, not interspersed with uninfected cells [ 166,1671. The mechanisms of 0, protection among actinorhizal plants diverge considerably as well. In contrast to Rhizobium, Frankia can fix nitrogen also in the free-living state at atmospheric 0, concentrations [168]. This is achieved by the location of the 0, sensitive nitrogenase, in special vesicles which provide 0, protection by their outer envelopes consisting of multilayered hopanoid membranes (Fig. 3C,D) [ 169,1701. In symbiosis the shape and position of the vesicles in the infected cells is determined by the host plant [131]. While vesicles formed in culture are spherical, in symbiosis their shape [107], envelope morphology [171], and internal structure (septate or nonseptate) (for review see [107,172]) depends on the host plant. Vesicles can also contribute to oxygen protection in symbiosis. The symbiotic vesicles have a high respiratory capacity [162], thereby further decreasing the amount of 0, in the direct neighbourhood of nitrogenase. A different situation is found in Casuarina- and Allocasuarina symbioses. Here, Frankia forms atypical hyphae instead of vesicles, for nitrogen fixation in the infected cells [173]. In these symbioses, but also in nodules of Myrica where Frunkia forms vesicles [174], the infected cells are surrounded by an 0, diffusion barrier, achieved by lignification of the cell walls of the infected cells and of the adjacent uninfected cells [175-1791. Furthermore, Casuarina, Allocasuarinu, and Myrica synthesize high amounts of hemoglobin in the infected cells [ 151,18(tl82]. Hemoglobin is homologous to leghemoglobin, and like in the legume nodules it facilitates 0, diffusion toward the sites of respiration. Hence, in some actinorhizal symbioses, like in legumes, the plants seem to be mainly responsible for providing 0, protection to bacterial nitrogenase. However, in contrast to legumes, in actinorhiza both partners can contribute to 0, protection, as signified by the formation of Frankia vesicles in nodules.
Hydrogen metabolism As shown above, hydrogen' (H,) production is an obligatory part of the nitrogenase mechanism; furthermore, in the absence of other reducible substrates, the total electron flux through nitrogenase is funnelled into H, production [183]. H, is a competitive inhibitor of N, fixation [ 1841. Consequently, nitrogen fixers tend to express an uptake hydrogenase to oxidize H, to H,O, resulting in 0, consumption and energy (ATP) generation. In free-living Frankia as well as in cultures of some rhizobia, activation of hydrogenase expression by H, results in an increased nitrogenase activity [ 185- 1891. Research on the benefits of uptake hydrogenase activity for symbiotic nitrogen
171 fixation, has yielded inconclusive data. For legume symbioses, where the effects of isogenic strains differing only in hydrogenase activity could be examined, contradictory results were obtained. While in some cases, hydrogenase activity was beneficial for the symbiotic performance of a rhizobial strain [190-1921 in others it was detrimental [ 193,1941. At any rate, no selection pressure favouring rhizobia which can express hydrogenase seems to exist, as signified by the fact that many rhizobial strains have been isolated which do not contain uptake hydrogenase [195-1971. Therefore, it seems likely that uptake hydrogenase activity is not important under conditions of sufficient carbon supply and 0, protection. This hypothesis is supported by physiological studies on free-living Azorhizobium caulinodans ORS571, showing that hydrogenase activity is a disadvantage under conditions of 0, limitation [198]. For Frankia, no isogenic strains are available, hampering studies on the role of hydrogenase activity for symbiotic nitrogen fixation (for review see [ 1991). However, the vast majority of Frankia strains isolated thus far shows hydrogenase activity ([200]; F. Tavares, U. Mattsson and A. Sellstedt, personal communication). Thus, in actinorhizal nodules where the bacteria have to contribute to 0, protection themselves, bacterial uptake hydrogenase activity may be more important for symbiotic efficiency.
Nitrogen metabolism In both legume and actinorhizal symbioses, ammonium, the product of nitrogen fixation, is exported by the bacteria and assimilated in the plant cytoplasm via the glutamine synthetase (GS)/glutamate synthase (GOGAT) pathway (for review see [201]). High levels of plant GS activity were found both in legume and in A. glutinosa nodules [201]. After ammonium assimilation, glutamate has to be metabolized into nitrogen transport forms, which depend on the plant species. In temperate legumes as well as in most actinorhizal plants examined thus far, the major nitrogen transport forms are amides, namely, glutamine and asparagine [201,202]. Tropical legumes with the exception of some trees [203] transport ureides [201]. Most Alnus species and Casuarina equisetijolia use an ureide, citrulline, as nitrogen transport form [201,202]. In indeterminate legume nodules as well as in actinorhizal nodules of Alnus, ammonium assimilation as signified by GS expression seems to be confined to the infected cells and to the nodule vascular system [153,204,205]. In determinate nodules, however, where ammonium assimilation and the synthesis of ureide for nitrogen transport are spatially separated, the situation is more diverse: GS activity in soybean nodules was found both in the infected and in the uninfected cells [206], while GS expression in nodules of Phaseolus vulgaris was confined to the infected cells and to the nodule vascular system [207]. However, uricase (nodulin-35), which catalyzes the oxidation of uric acid to allantoin, in both soybean and Phaseolus nodules is localized in the specialized peroxisomes of uninfected cells only [208-2101, and the activity of the enzyme catalyzing the next step in purine oxidation, allantoinase, was also confined to the uninfected cells [208]. The uninfected cells of determinate nodules, contain specialized peroxisomes where these
172 enzymes are localized and ureide biosynthesis takes place [211]. They form a network and are involved in transport of nitrogenous compounds to and carbon sources from the nodule vascular bundles [212]. Connected by an elaborate tubular endoplasmic reticulum system which is appressed to the specialized peroxisomes, the places of ureide biosynthesis, and continues through plasmodesmata, they constitute a more or less continuous network throughout the central tissue, facilitating the transport of nitrogenous compounds to the nodule vascular bundle [212]. In indeterminate legume nodules, however, where amides serve as nitrogen transport form, whose biosynthesis takes place in the cytoplasm, no specialized function could be assigned to the uninfected cells in the central tissue, which are fewer than in determinate nodules and do not form a network [73]. Here, efficient transport of nitrogenous compounds seems to be achieved by the presence of transfer cells in the pericycle of the nodule vascular bundles, providing an abundant surface area across which the transport can occur [213]. In actinorhizal nodules of Afnus gfutinosa, citrulline biosynthesis seems to take place in the infected cells, since acetyl omithine transaminase, an enzyme involved in citrulline biosynthesis, has been found to be expressed in these cells [153]. Although Alnus is a ureide transporter, there is no homology with determinate legume nodules, because citrulline is not synthesized via de novo ureide biosynthesis as in the case of the tropical legumes [214,215], but via omithine [216]. Thus, the biosynthesis of citrulline does not require peroxisomal enzymes, but seems to take place at least partially in the mitochondria where omithine carbamoyl transferase was detected [217]. No metabolic specialization of the uninfected cells of Alnus nodules, except for starch storage [218], could be found up to now. Transport functions in Afnus nodules seem to be fulfilled by the pericycle of the central vascular bundle of the nodule lobes, which consists of several layers of small cells with a dense cytoplasm, but without the cell wall structures typical for transfer cells [136,219]. Considering that in Afnus nodule lobes, the transport function which in legume nodules is carried out by several vascular bundles, is concentrated on the central stele, it seems likely that the proliferation of the pericycle serves to improve the transport capacities. Carbon metabolism Nodules need to be supplied with carbon sources for maintenance and growth, energy for N, fixation and for supply of acceptor molecules for assimilation of the fixed nitrogen. Shoot carbohydrate pools have been identified as the primary source for the maintenance of nodule N, fixation activity, also during darkness ([218,220]; reviewed for legumes in [221]), in spite of the presence of starch grains in legume [96,210] and actinorhizal nodules [218,220,222]. Assimilates are transported in the form of sucrose from source to sink tissues [223] where they are introduced into metabolism by the action of symplastic sucrose synthase, or apoplastic invertase. In mature legume nodules, high sucrose synthase activities have been detected [224], and sucrose synthase transcription has been shown to be induced in legume as well as actinorhizal nodules [ 153,225,2261.
173 Analysis of nodule enzyme activities has shown that malate is the primary product of glycolysis in legume nodules, and in turn seems to be exported to the bacteroides as an energy source or to serve as an ammonium acceptor and to be metabolized to aspartate [221]. This is achieved by high activities of phosphoenolpyruvate (PEP) carboxylase, malate dehydrogenase (MDH) and aspartate transaminase (AAT) in legume nodules [221,227]. Nodulin-26 which is located in the PBM, has been suggested to mediate the transport of malate into the PBS [29], although its low substrate specificity in vitro indicates that it is more likely to form a pore responsible for the uptake of ions or small metabolites in general [228]. In actinorhizal nodules of A. glutinosa, high activities of PEP carboxylase [229] and MDH activity were found [230]. However, the identity of the carbon source provided by the plant to endosymbiotic Frankia is not clear yet, although a malate-aspartate-shuttle has been suggested [230]. Furthermore, it remains to be examined how far nodule carbon transport and metabolism differs between different actinorhizal symbioses.
Conclusions and future prospects In nodules, specific needs have to be fulfilled to allow nitrogen fixation. The comparison between legume- and actinorhizal nodules shows that these requirements can be met in a variety of ways. This implies that there will be multiple possibilities to solve these problems in new nitrogen fixing systems. The comparison of structure and development of Rhizobiumllegume and actinorhizal nodules has revealed several differences. First, Rhizobiumllegume nodules have a stem-like morphology with peripheral vascular system and infected cells in the central tissue, while actinorhizal nodule lobes represent modified lateral roots with a central vascular cylinder and infected cells in the cortex. Second, the two types of nodules are also developmentally different, legume nodule primordia being induced in the root cortex and actinorhizal nodule primordia in the root pericycle. Third, while rhizobia in symbiosis are released into the plant cytoplasm by a process resembling endocytosis, no such release is taking place in actinorhizal symbioses. Fourth, there is no compartment in actinorhizal nodules corresponding to the infection thread matrix in Rhizobiumllegume systems. Thus, both types of symbioses seem fundamentally dissimilar. However, in spite of these differences there is evidence for a phylogenetic relationship between both symbioses, since the comparison of sequences of the gene encoding the large subunit of ribulosebisphosphate carboxylase from different plant species has shown that there seems to be a single phylogenetic origin of susceptibility to nitrogen-fixing root nodule symbioses in angiosperms [2311. This is supported by the fact that nodules induced by Rhizobium on Parasponia (Ulmaceae), the only nonlegume being able to enter a symbiosis with rhizobia [232], structurally and developmentally resemble actinorhizal nodules. Furthermore, in Parasponia nodules, rhizobia are not released from the infection threads, and an infection thread matrix is not discernable [233,234]. Thus, the differences between both symbioses may simply be due to the variability of ways to meet the re-
174
quirements for symbiotic nitrogen fixation in legumes vs. other plant families. While previously legume nodules were considered unique organs and root nodule induction seemed to require a set of specific genes, new results have changed our view on nodule development. Proteins previously thought to be nodule-specific, have been shown to have counterparts in nonsymbiotic plant development, as it has been found for hemoglobin [28,235]. Infection thread growth during legume nodule induction has been related to a common developmental process, namely cell cycling [69]. Root hair deformation has been identified as newly induced tip growth [54]. The identification of the developmental pathways from which symbiotic processes are derived, will allow the development of strategies to engineer new systems. Hence, the possibility of transferring the ability to enter N,-fixing symbioses to other crop plants can be considered more optimistically nowadays because of the results in recent research on nodule development.
Acknowledgements We want to thank Alison M. Berry (University of California, Davis, USA) and Ab van Kammen for critical reading of the manuscript, R. Howard Berg (University of Memphis, Tennessee, USA), Ula Bialek and Andrt5 van Lammeren (Agricultural University Wageningen) and Wei-Cai Yang for providing photographs and Nicholas J. Brewin (John Innes Institute, Norwich, UK), Marijke van Ghelue and Bjam Solheim (University of Tromsg, Norway) and Anita Sellstedt (University of UmeA, Sweden) for providing unpublished information.
References 1. Heichel GH, Vance CP. Legume nitrogen: symbiotic fixation and recovery by subsequent crops. In: Helsel ZR (ed) Energy in Plant Nutrition and Pest Control. Amsterdam: Elsevier, 1987;63-80. 2. Paul EA. Towards the year 2000: Directions for future nitrogen research. In: Wilson JR (ed) Advances in Nitrogen Cycling in Agricultural Ecosystems. Wallingford, UK: CAB International, 1988;417-425. 3. Howard JB, Rees DC. Nitrogenase - a nucleotide-dependent molecular switch. Ann Rev Biochem 1994;63:235-264. 4. Shaw VK, Brill WJ. Isolation of an iron-molybdenum cofactor from nitrogenase. Proc Natl Acad Sci USA 1977;74:149+1497. 5 . Bums RH.Nitrogenases. J Biol Chem 1991;266:9339-9342. 6. Bergman B, Johansson C, Soderback E. The Nostoc-Gunneru symbiosis. New Phytologist 1992;122:379-400. 7. Elmerich C, Zimmer W, Vieille C. Associative nitrogen-fixing bacteria. In: Stacey G, Bums RH, Evans HJ (eds) Biological Nitrogen Fixation. New York Chapman and Hall, 1992;212-258. 8. Okon Y, Labandera-Gonzalez CA. Agronomic applications of Azospirillum - an evaluation of 20 years worldwide field inoculation. Soil Biol Biochem 1994;26:1591-1601. 9. Khush GF, Bennett J (eds). Nodulation and Nitrogen Fixation in Rice: Potentials and Prospects. Manila: International Rice Research Institute, 1992. 10. Hellriegl H. Welche Stickstoffquellen stehen der Pflanze zu Gebote? Zeitschrift des Verbands fur
175 die Riibenzucker-Industrie des Deutschen Reiches 1886;36:863-877. 11. Beyerinck M W . Die Bakterien der Papilionaceen-Knollchen.Bot ZRitung 1888;4&50:725-804. 12. De Bruijn FJ. The unusual symbiosis between the diazotrophic stem-nodulating bacterium Azorhizobium cuulinoduns ORS571 and its host, the tropical legume Sesbuniu rosfrufu: plant and bacterial aspects. In: Kosuge T, Nester EW (eds) Plant-MicrobeInteractions. New York: Macmillan, 1989;457-493. 13. Marvel DJ, Torrey JG, Ausubel FM. Rhizobium symbiotic genes required for nodulation of legume and nonlegume hosts. Proc Natl Acad Sci USA 1987;84:1319-1323. 14. Weeden N, Kneen BE, LaRue TA. Genetic analysis of Sym genes and other nodule-related genes in Pisum sutivum. In: Gresshoff PM (ed) Nitrogen Fixation: Achievement and Objectives. New York: Chapman and Hall, 1990323330. 15. Caetano-AnollCs G, Gresshoff PM. Plant genetic control of nodulation. Ann Rev Microbiol 1991;45:345-382. 16. Kolchinsky A, Funke R, Gresshoff PM. Dissecting molecular mechanisms of nodulation: taking a leaf from Arabidopsis. Plant Molec Biol 1994;26:54%-552. 17. Carroll BJ, McNeil DL, Gresshoff PM. Mutagenesis of soybean (Glycine mux (L.) Merr.) and the isolation of nonnodulating mutants. Plant Sci 1986;47:109-114. 18. Nutman PS. Improving nitrogen fixation in legumes by plant breeding; the relevance of host selectionexperimentsin red clover (Trifoliurn prufense L.) and subterranean clover (T. subterruneurn L.). Plant Soil 1984;82:285-301. 19. Hiiser A, Robinson DL, Duc G, Vance CP. A mutation in Viciufabu results in ineffective nodules with impaired bacteroid differentiation and reduced synthesis of late nodulins. J Exp Bot 1992;43:1397-1407. 20. Landau-Ellis D, Angermuller SA, Shoemaker RC, Gresshoff PM. The genetic locus controlling supernodulation in soybean (Glycine mux L.) co-segregates tightly with a cloned molecular marker. Molec Gen Genet 1991;228:221-226. 21. Landau-Ellis D, Gresshoff PM. Supernodulating soybean mutant alleles nfs382 and nfs1007 show no recombination with the same restriction length polymorphism marker. Molec Plant Microbe Interact 1992;5:428-429. 22. Kneen BE, Weeden NF, LaRue TA. Nonnodulating mutants of Pisum sutivum (L.) cv Sparkle. J Hered 199485: 12%-133. 23. Shchez F, Padilla JE, PCrfiz H, Lara M. Control of nodulin genes in root-nodule development and metabolism. Ann Rev Plant Physiol Plant Molec Biol 1991;42:507-528. 24. Verma DPS. Signals in root nodule organogenesis and endocytosis of Rhizobium. Plant Cell 1992;4:373-382. 25. Franssen HJ, Vijn I, Yang W-C, Bisseling T. Developmental aspects of the Rhizobium-legume symbiosis. Plant Molec Biol 1992;19:89-107. 26. Mylona P, Pawlowski K, Bisseling T. Symbiotic nitrogen fixation. Plant Cell 1995;7:869-885. 27. Van Kammen A. Suggested nomenclature for plant genes involved in nodulation and symbiosis. Plant Molec Biol Report 1984;2:43-45. 28. Taylor ER, Nie XZ,MacGregor AW, Hill RD. A cereal haemoglobin gene is expressed in seed and root tissues under anaerobic conditions. Plant Molec Biol 1994;24:853-862. 29. Miao G-H, Verma DPS. Soybean nodulin-26 gene encoding a channel protein is expressed only in the infected cells of nodules and is regulated differently in roots of homologous and heterologous plants. Plant Cell 1993;5:781-794. 30. Scheres B, Van de Wiel C, Zalensky A, Horvath B, Spaink H, Van Eck H, Zwartkruis F, Wolters AM, Gloudemans T, van Kammen A, Bisseling T. The ENODl2 product is involved in the infection process during pea-Rhizobium interaction. Cell 199060:28 1-294. 31. Kouchi H, Hata S. Isolation and characterization of novel cDNAs representing genes expressed at early stages of soybean nodule development. Molec Gen Genet 1993;238:106-119. 32. Yang W-C, Katinakis P, Hendriks P, Smolders A, de Vries F, Spee J, van Kammen A, Bisseling T, Franssen H. Characterization of GmENOD40, a gene showing novel patterns of cell-specific
176 33. 34. 35. 36. 37. 38.
39. 40. 41. 42. 43.
44.
45.
46.
47.
48. 49. 50. 51.
52. 53. 54. 55.
expression during soybean nodule development. Plant J 1993;3:573-585. Nap JP, Bisseling T. Developmental biology of a plant-prokaryote symbiosis: the legume root nodule. Science 1990;250948-954. Becana M, Klucas RV. Oxidation and reduction of leghemoglobin in root nodules of leguminous plants. Plant Physiol 1992;98:1217-1221. Fisher RF, Long SR. Rhizobium-plant signal exchange. Nature 1992;357:655-660. Spaink HP. Rhizobial lipopolysaccharides: Answers and questions. Plant Molec Biol 1992;20977986. Mnarie J, Cullimore J. Lipo-oligosaccharidenodulation factors: a new class of signalling molecules mediating recognition and morphogenesis. Cell 1993;7495 1-954. Roche P, DebellC F, Maillet F, Lerouge P, Truchet G, DBnari6 J, Prom6 J-C. Molecular basis of symbiotic host specificity in Rhizobium meliloti: nodH and nodPQ genes encode the sulfation of lipo-oligosaccharidesignals. Cell 1991;67:1131-1143. Journet EP, Pichon M, Dedieu A, de Billy F, Truchet G, Barker DG. Rhizobium meliloti Nod factors elicit cell-specific transcription of the ENODl2 gene in transgenic alfalfa. Plant J 1994;6:241-249. Okker RJH, Schlaman HRM, Spaink HP, Lugtenberg BJJ. Function of nodulation genes of Rhizobium. Symbiosis 1993;14:283-295. Van Rhijn P, Vanderleyden J. The Rhizobium-plant symbiosis. Microbiol Rev 199x59: 124-142. Peters NK, Frost JW, Long SR. A plant flavone, luteolin, induces expression of Rhizobium melilori nodulation genes. Science 1986;233:977-980. Goethals K, Van Montagu M, Holsters M. Conserved motifs in a divergent nod box of Azorhizobium cuulinoduns ORS571 reveals a common structure in promoters regulated by LysR-type proteins. Proc Natl Acad Sci USA 1992;89:16461650. Lerouge P, Roche P, Faucher C, Maillet F, Truchet G, Prom6 J-C, DdnariB J. Symbiotic host specificity of Rhizobium meliloti is determined by a sulphated and acylated glucosamine oligosaccharide signal. Nature 199@344:781-784. Spaink HP, Sheely DM, van Brussel AAN, Glushka J, York WS, Tak T, Geiger 0, Kennedy EP, Reinhold VN, Lugtenberg BJJ. A novel highly unsaturated fatty acid moiety of lipo-oligosaccharide signals determines host specificity of Rhizobium. Nature 1991;354:125-130. Van Brussel AAN, Bakhuizen R, Van Spronsen PC, Spaink HP, Tak T, Lugtenberg BJJ. Induction of preinfection thread structures in the leguminous host plant by mitogenic lipo-oligosaccharides of Rhizobium. Science 1992;257:70-71. Horvath B, Heidstra R, Lados M, Moerman M, Spaink HP, Prom6 J-C, Van Kammen A, Bisseling T. Induction of pea early nodulin gene expression by Nod factors of Rhizobium. Plant J 1993;4:727733. Vijn I, Das Neves L, Van Kammen A, Franssen H, Bisseling T. Nod factors and nodulation in plants. Science 1993;260:1764-1765. Stokkermans TJW, Peters NK. Brudyrhizobium elkunii lipooligosaccharidesignal induce complete nodule structures on Glycine soju Siebold et Zucc. Planta 1994;193:413-420. Chandler MR, Date RA, Roughley RJ. Infection and root nodule development in Stylosunrhes species by Rhizobium. J Exp Bot 1982;33:47-57. Dreyfus BL, Alazard D, Dommergues YR. Stem-nodulating rhizobia. In: Klug MC, Reddy CE (eds) Current Perspectives in Microbial Ecology. Washington, DC: American Society of Microbiology, 1986;161- 169. De Faria SM, Hay GT, Sprent JI. Entry of rhizobia into roots of Mimosa scubrellu Bentham occurs between epidermal cells. J Gen Microbiol 1988;134:2291-2296. Fahraeus G. The infection of clover root hairs by nodule bacteria studied by a simple glass slide technique. J Gen Microbiol 1957;16:374-381. Heidstra R, Geurts R, Franssen H, Spaink HP, Van Kammen A, Bisseling T. Root hair deformation activity of nodulation factors and their fate on Viciu sutivu. Plant Physiol 1994;105:787-797. Cook D, Dreyer D, Bonnet D, Howell M, Nony E, VandenBosch K. Transient induction of a perolidase gene in Medicugo truncutulu precedes infection by Rhizobium meliloti. Plant Cell
177 1995;7:43-55. 56. Scheres B, Van Engelen F, Van der Knaap E, Van de Wiel C, Van Kammen A, Bisseling T. Sequential induction of nodulin gene expression in the developing pea nodule. Plant Cell 1990;2:687-700. 57. Vijn I, Martinez-AbarcaF, Yang W-C, Das Neves L, Van Brussel A, van Kammen A, Bisseling T. Early nodulin gene expression during Nod factor-induced processes in Viciu sutivu. Plant J 1995;8:111-119. 58. Goethals K, Leyman B, Van den Eede G, Van Montagu M, Holsters M. An Azorhizobium cuulinoduns ORS57 1 locus involved in lipopolysaccharide production and nodule formation on Sesbuniu rosfrutu stems and roots. J Bacteriol 199%176:92-99. 59. Pichon M, Joumet E-P, de Billy F, Dedieu A, Huguet T, Truchet G, Barker DG. Rhizobium meliloti elicits transient expression of the early nodulin gene ENODIZ in the differentiating root epidermis of transgenic alfalfa. Plant Cell 1992;4:1199-1211. 60. Csanadi G, Szecsi J, Kalo P, Kiss P, Endre G, Kondorosi A, Kondorosi E, Kiss GB. ENODIZ, an early nodulin gene, is not required for nodule formation and efficient nitrogen fixation in alfalfa. Plant Cell 1994;6:201-213. 61. Callaham D, Torrey JG. The structural basis for infection of root hairs of Trifolium repens by Rhizobium. Can J Bot 1981;59:1647-1664. 62. Bakhuizen R. The Plant Cytoskeleton in the Rhizobium-Legume Symbiosis. PhD Thesis, Leiden University, Leiden, The Netherlands, 1988. 63. Turgeon BG, Bauer WD. Ultrastructure of infection-thread development during the infection of soybean by Rhizobium juponicum. Planta 1985;163:328-349. 64. Brewin NJ. Development of the legume root nodule. Ann Rev Cell Biol 1991;7:191-226. 65. Kijne JW.The Rhizobium infection process. In: Stacey G, Bums RH, Evans HJ (eds) Biological Nitrogen Fixation. New York: Chapman and Hall, 1992349-398. 66. Dart PJ. The infection process. In: Quispel A (ed) Biology of Nitrogen Fixation. Amsterdam: North Holland Publishing Company, 1974;381-429. 67. VandenBosch KA, Bradley DJ, Knox JP,Perotto S, Butcher GW, Brewin NJ. Common components of the infection thread matrix and the intercellular space identified by immunochemical analysis of pea nodules and uninfected roots. EMBO J 1989;8:335-342. 68. Rae AL, Bontante-Fasolo P, Brewin NJ. Structure and growth of infection threads in the legume symbiosis with Rhizobium leguminosurum. Plant J 1992;2:385-395. 69. Yang W-C, de Blank C, Meskiene I, Hirt H, Bakker J, van Kammen A, Franssen H, Bisseling T. Rhizobium Nod factors reactivate the cell cycle during infection and nodule primordium formation, but the cycle is only completed in primordium formation. Plant Cell 1994;6:1415-1426. 70. Newcomb W. A correlated light and electron microscopic study of symbiotic growth and differentiation in Pisum sutivum root nodules. Can J Bot 1976;54:2163&2186. 71. Newcomb W. Nodule morphogenesis and differentiation.Int Rev Cytol 1981;13(Suppl):S247-S297. 72. Tsien HC, Dreyfus BL, Schmidt EL. Initial stages in the morphogenesis of nitrogen-fixing stem nodules of Sesbuniu rostrutu. Molec Gen Genet 1983;206:291-299. 73. Schaede R. Die Knollchen der adventiven Wasserwurzeln von Neptuniu oleruceu und ihre Bakteriensymbiose. Planta 1940;31: 1-21. 74. James EK, Sprent JI, Sutherland JM, McInroy SG, Michin RT. The structure of nitrogen fixing root nodules on the aquatic mimosoid legume Neptuniu plena. Ann Bot 199269: 173-180. 75. Chandler MR. Some observations on infection of Aruchis hupogueu L. by Rhizobium. J Exp Bot 1978;29:74*755. 76. Alazard D, Duhoux E. Development of stem nodules in a tropical forage legume, Aeschynomene ufrusperu. J Exp Bot 1990;41:1199-1206. 77. Libbenga KR, Bogers RJ. Root-nodule morphogenesis. In: Quispel A (ed) The Biology of Nitrogen Fixation. Amsterdam: North-Holland Publishing Company, 1974;43&472. 78. Asad S, Fang Y, Wycoff KL, Hirsch AM. Isolation and characterization of cDNA and genomic clones of MsENOD40; transcripts are detected in meristematic cells of alfalfa. Protoplasma
178 1994;183:10-23. 79. Matvienko M, van de Sande K, Yang W-C, van Kammen A, Bisseling T, Franssen H. Comparison of soybean and pea ENOD40 cDNA clones representing genes expressed during both early and late stages of nodule development. Plant Molec Biol 1994;26:487-493. 80. Gresshoff PM, Delves AC. Plant genetic approaches to symbiotic nodulation and nitrogen fixation in legumes. In: Blonstein AD, King PJ (eds) A Genetic Approach to Plant Biochemistry. Heidelberg: Springer, 1986;15+206. 81. Rolfe BG, Gresshoff PM. Genetic analysis of legume nodule initiation. Ann Rev Plant Physiol Plant Molec Biol 1988;39:297-319. 82. Libbenga KR, Harkes PAA. Initial proliferation of cortical cells in the formation of root nodules in Pisum sutivum L. Planta 1973;114:17-28. 83. Ardourel M, Demont N, DebellC F, Maillet F, de Billy F, PromC J-C, D6nariC J, Truchet G. Rhizobium meliloti lipooligosaccharide nodulation factors: different structural requirements for bacterial entry into target root hair cells and induction of plant symbiotic developmental responses. Plant Cell 1994;6:1357-1374. 84. Hirsch AM, Bhuvaneswari TV,Torrey JG, Bisseling T. Early nodulin genes are induced in alfalfa mot outgrowthselicited by auxin transport inhibitors. Proc Natl Acad Sci USA 1989;86:1244--1248. 85. Cooper JB, Long SR. Morphogenetic rescue of Rhizobium melilori nodulation mutants by trunszeatin secretion. Plant Cell 1994;6:215-225. 86. Roth LE, Stacey G. Bacterium release into host cells of nitrogen-fixing soybean nodules: the symbiosome membrane comes from three sources. Euro J Cell Biol 1989;49:13-23. 87. Perotto S, Donovan N, Droback BK, Brewin NJ. Differential expression of a glycosyl inositol phospholipid antigen on the peribacteroid membrane during pea nodule development. Molec PlantMicrobe Interact 1995;8:560-568. 88. Mellor RB, Werner D. Peribacteroid membrane biogenesis in mature legume root nodules. Symbiosis 1987;3:89-114. 89. Mellor RB. Bacteroides in the Rhizobiurn-legume symbiosis inhabit a plant internal lytic compartment: implication for other microbial endosymbioses. J Exp Bot 1989;40831-839. 90. Kinnback A, Mellor RB, Werner D. a-Mannosidase I1 isoenzyme in the peribacteroid space of Glycine m a root nodules. J Exp Bot 1987;38:1373-1377. 91. Mellor RB, MSrschel E, Werner D. Legume root response to symbiotic infection: enzymes of the peribacteroid space. Z Naturforsch 1984;39:123-125. 92. Zhang Y, Roberts DM. Expression of soybean nodulin-26 in transgenic tobacco. Targeting to the vacuolar membrane and effects on floral and seed development. Molec Biol Cell 1995;6:10!&117. 93. Kannenberg EL, Brewin NJ. Expression of a cell surface antigen from Rhizobium leguminosarum 3841 is regulated by oxygen and pH. J Bacteriol 1989;171:4543-4548. 94. Hirsch AM, Smith CA. Effects of Rhizobium melilati nif and f i x mutants on alfalfa root nodule development. J Bacteriol 1987;169:1137-1146. 95. Van de Wiel C, Scheres B, Franssen H, van Lierop MJ, van Lammeren A, van Kammen A, Bisseling T. The early nodulin transcript ENOD2 is located in the nodule parenchyma (inner cortex) ’ of pea a d soybean root nodules. EMBO J 19909:l-7. 96. Vasse J, de Billy F, Camut S, Truchet G. Correlation between ultrastructural differentiation of bacteroides and nitrogen fixation in alfalfa nodules. J Bacteriol 1990172:4295-4306. 97. Yang W-C, Horvath B, Hontelez J, van Kammen A, Bisseling T. In situ localization of Rhizobium mRNAs in pea root nodules: nifA and n$Y localization. Molec Plant-Microbe Interact 1991;4:464468. 98. De Maagd RA, Yang W-C, Goosen-de Roo L, Mulders IHM,Roest HP, Spaink HP, Bisseling T, Lugtenberg BJJ. Downregulation of expression of the Rhizobium leguminosarum outer membrane protein gene ropA occurs abruptly in interzone 11-111 of pea nodules and can be uncoupled from nif gene activation. Molec Plant-Microbe Interact 1993;7:276-281. 99. Kardailsky I, Yang W-C, Zalensky A, van Kammen A, Bisseling T. The pea late nodulin gene PsNOD6 is homologous to the early nodulin genes PsENOD3114 and is expressed after
179 leghaemoglobin genes. Plant Molec Biol 1993;23:1029-1037. 100. Soupbne E, Foussard M, Boistard P, Truchet G, Batut J. Oxygen as a key developmental regulator of Rhizobium meliloti N, fixation: gene expression within the alfalfa root nodule. Pmc Natl Acad Sci USA 1995;92:375%3763. 101. Pladys D, Rigaud J. Senescence in French bean nodules: Occurrence of different proteolytic activities. Physiol Plant 1985;63:43-48. 102. Pladys D, Vance CP. Proteolysis during development and senescence of effective and plant genecontrolled ineffective alfalfa nodules. Plant Physiol 1993;103:37+384. 103. Manen JF, Simon P, van Slooten J-C, 0sterls M, Frutiger S, Hughes GJ. A nodulin specifically expressed in senescent nodules of winged bean is a protease inhibitor. Plant Cell 1991;3:259-170. 104. Garbers C, Meckbach R, Mellor RB, Werner D. Protease (themolysin) inhibition in the peribacteroid space of Glycine m a mot nodules. J Plant Physiol 1988;132:442-445. 105. Benson DR, Silvester WB. Biology of Frankia strains, actinomycete symbionts of actinorhizal plants. Microbiol Rev 1993;57:293-319. 106. Simonet P, Normand P, Hirsch AM, Akkemans ADL. The genetics of the Frankia actinorhizal symbiosis. In: Gresshoff PM (ed) The Molecular Biology of Symbiotic Nitrogen Fixation. Boco Raton, FL: CRC Press, 1990;77-109. 107. Baker DD, Mullin BC. Actinorhizal symbioses. In: Stacey G, Bums RH, Evans HJ (eds) Biological Nitrogen Fixation. New York Chapman and Hall, 1992;259-292. 108. Meyen J. Uber das Hervonvachsen parasitischer Gebilde aus den Wuneln anderer Pflanzen. Flora 1829;12:4%64. 109. Hiltner L. Uber die Bedeutung der Wurzelkniillchen von Alnus glutinosa fiir die Stickstoffernhung dieser Pflanze. Landwirschaft Versmd Stud 1985;46:153-161. 110. Krebber 0. Untersuchungen iiber die Wurzelknollchen der Erle. Arch Microbiol 1932;3:588-608. 11 1. Schaede R. Uber die Symbionten in den Knollchen der Erle und des Sanddornes und die cytologischen Verhaltnisse in ihnen. Planta 1933;19:389-416. 112. Dawson JO. Actinorhizal plants: their use in forestry and agriculture. Out1 Agric 1986;15:202-208. 113. Albouchi A. Comparative study of 2 young windbreaks of Cupressus sempervirens and Casuarina gluuca. Ann Instit Nat Rech Agron Tunisie 1989;74-83. 114. Adegbehin JO, Igboanugo ABI,Omijeh JE.Potentials of agroforestry for sustainablefood and wood production in the savanna areas of Nigeria. Savanna 1990;11:12-26. 115. Diem HG, Dommergues YR. Current and potential uses and management of Casuarinaceae in the tropics and subtropics. In: Schwintzer CR, Tjepkema JD (eds) The Biology of Frankia and Actinorhizal Plants. New York: Academic Press, 1990;317-342. 116. El-Lakany MH. Provenance trials of Casuarina gluuca and C . cunninghamiana in Egypt. In: ElLakany MH, Tumbull JW,Brewbaker JL (eds) Advances in Casuarina Research and Utilization. Cairo: American University in Cairo, Desert Development Center, 1990,12-22. 117. Skoupy J. The role of trees in agroforestry. Desert Cont Bull (UNEP) 1991;193845. 118. Sundararaju R, Chinnathurai AK. Technology packages for reclamation and development of wastelands. Indian Forester 1992;118:60M15. 119. Pun S, Singh S, Bhushan B. Evaluation of fuelwood quality of indigenous and exotic tree species of India’s semiarid region. Agroforest Syst 1994;26:123-130. 120. Sagwal S S . Casuarinu on farm: A plant of economic importance. Indian Farming 1993;43:12-14 121. Lechevalier MP, Lechevalier HA. Genus Frunkia Brunchorst 1886, 174AL.In: Williams ST, Sharpe ME, Holt HG (eds) Bergey’s Manual of Systematic Bacteriology, vol 4. Baltimore, MD: Williams and Wilkins, 1989;241*2417. 122. Lechevalier MP, Lechevalier HA. Systematics,isolation, and culture of Frankia. In: SchwintzerCR, Tjepkema JD (eds) The Biology of Frankia and Actinorhizal Plants. New York: Academic Press, 199Q35-60. 123. Quispel A, Burggraaf AJP, Borsje H, Tak T. The role of lipids in the growth of Frankia isolates. Can J Bot 1983;61:2801-2806. 124. Quispel A, Svendsen AB, Schripserna J, Baas WJ, Erkelens C, Lugtenberg J. Identification of
dipterocarpol as isolation factor for the induction of primary isolation of Frunkiu from root nodules of Alnus glutinosu (L.) Gaertner. Molec Plant-Microbe Interact 1989;2:107-1 12. 125. Baker DD. Relationships among pure cultured strains of Frunkiu based on host specificity. Physiol Plant 1987;70:245-248. 126. Torrey JG. Initiation and development of root nodules of Cusuurinu (Cuswrinuceue). Am J Bot 1976;63:335-344. 127. Berry AM, McIntyre L, McCully ME. Fine structure of root hair infection leading to nodulation in the Frunkiu-Alnus symbiosis. Can J Bot 1986;64:292-305. 128. Callaham D, Newcomb W, Torrey JG, Peterson RL. Root hair infection in actinomycete-induced root nodule initiation in Cusuurinu, Myricu, and Comptoniu. Bot Gaz 1979;14O(Suppl):S1 4 9 . 129. Miller IM, Baker DD. The initiation, development and structure of root nodules in Elueugnus ungustifoliu L. (Elaeagnaceae). Protoplasma 1985;128:107-1 19. 130. Liu Q, Berry AM. The infection process and nodule initiation in the Frunkiu-Ceunothus root nodule symbiosis: a structural and histochemical study. Protoplasma 1991;163:82-92. 131. Racette S, Torrey JG. Root nodule initiation in Gymnostom (Casuarinaceae) and Shepherdiu (Elaeagnaceae) induced by Frunkiu strain HFPGpIl. Can J Bot 1989;67:2873-2879. 132. Berry AM, Sunell LA. The infection process and nodule development. In: Schwintzer CR, Tjepkema JD (eds) The Biology of Frunkiu and Actinorhizal Plants. New York: Academic Press, 1990;61-81. 133. Burggraaf AJP, van der Linden J, Tak T. Studies on the localization of infectible cells of AInus glutinosu roots. Plant Soil 1983;74:175-188. 134. Prin Y, Rougier M. Preinfection events in the establishment of Alnus-Frunkiu symbiosis: Study of the root hair deformation step. Plant Physiol (Life Sci Adv) 1987;6:9+108. 135. Berry AM, Torrey JG. Root hair deformation in the infection process of Alnus rubru. Can J Microbiol 1983;61:2863-2976. 136. Burgess D, Peterson RL. Development of Alnus juponicu root nodules after inoculation with Frunkiu strain HFPArI3. Can J Bot 1987;65:1647-1657. 137. Callaham D, Torrey JG. Prenodule formation and primary nodule development in roots of Comptoniu (Myricaceae). Can J Microbiol 1977;51:2306-23 18. 138. Mian S, Bond G. The onset of nitrogen fixation in young alder plants and its relation to differentiation in the nodular endophyte. New Phytologist 1978;80:187-192. 139. Schwintzer CR, Berry AM, Disney LD. Seasonal patterns of root nodule growth, endophyte morphology, nitrogenase activity and shoot development in Myricu gale. Can J Bot 1982;60:746757. 140. Liu QQ, Berry AM. Localization and characterization of pectic polysaccharides in roots and root nodules of Ceunothus spp. during intercellular infection by Frunkiu.Protoplasma 1991;163:93-101. 141. Sunell LA, Berry AM. Preinfection cell wall formation in roots and developing nodules of Alnus rubru bong. Protoplasma 1992;168:87-93. 142. Bond G. Root-nodule symbiosis with actinomycete-like organisms. In: Quispel A (ed) The Biology of Nitrogen Fixation. Amsterdam: North-Holland Publishing Company, 1974342-378. 143. Silvester WB, Harris SL, Tjepkema JD. Oxygen regulation and hemoglobin. In: Schwintzer CR, Tjepkema JD (eds) The Biology of Frunkiu and Actinorhizal Plants. New York: Academic Press, 1990;157- 176. 144. Ribeiro A, Akkermans ADL, Van Kammen A, Bisseling T, Pawlowski K. A nodule-specific gene encoding a subtilisin-like protease is expressed in early stages of actinorhizal nodule development. Plant Cell 1995;7:785-794. 145. Lalonde M. Immunological and ultrastructural demonstration of nodulation of the European AInus glutinosu (L.) Gaertn. host plant by an actinomycetal isolate from the North American Comptoniu peregrinu (L.) Coult. root nodule. Bot Gaz 1979;140(S):35-43. 146. Fontaine MF, Lancelle SA, Torrey JG. Initiation and ontogeny of vesicles in cultured Frunkiu spp. strain HFPArI3. J Bacteriol 1984;160:921-927. 147. Huss-Dane11 K, Bergman B. Nitrogenase in Frunkiu from root nodules of Ahus incunu (L.) Moench: Immunolocalization of the Fe- and MoFe-proteins during vesicle differentiation. New
181 Phytologist 1990;116:443-455. 148. Pawlowski K, Akkermans ADL, van Kammen A, Bisseling T. Expression of Frankia nifgenes in actinorhizal nodules of AInus glutinosa. Plant Soil 1995;170:371-376. 149. SCguin A, Lalonde M. Expression of actinorhizins in the development of the Frankia-Alnus symbiosis. In: Polsinelli M, Materassi R, Vincenzini M (eds) Developments in Plant and Soil Sciences: Nitrogen Fixation. Dordrecht: Kluwer Academic Publishers, 1991;602-608. 150. Fleming AI, Wittenberg JB, Wittenberg BA, Dudman WF, Appleby CA. The purification, characterizationand ligand-bindingkinetics of hemoglobins from root nodules of the nonleguminous Casuurina glauca-Frunkia symbiosis. Biochim Biophys Acta 1987;911:209-220. 151. Jacobsen-Lyon K, Plstergaard-Jensen E, Jlrgensen J-E, Marcker KA, Peacock J, Dennis E. Symbiotic and nonsymbiotic hemoglobin genes of Casuarina glauca. Plant Cell 1995;7:213-223. 152. Goetting-Minesky MP, Mullin BC. Differential gene expression in an actinorhizal symbiosis: Evidence for a nodule-specific cysteine proteinase. Proc Natl Acad Sci USA 1994;91:9891-9895. 153. Pawlowski K, Guan C, Ribeiro A, van Kammen A, Akkermans ADL, Bisseling T. Genes involved in Alnus glutinosa nodule development. In: Kiss GB, Endre G (eds) Proceedings of the 1st European Nitrogen Fixation Conference. Szeged: Officina Press, 1994220-224. 154. Twigg PG.Isolation of a Nodule-Specific cDNA Encoding a Putative Glycine-Rich Protein from Alnus glutinosa. PhD Dissertation, University of Tennessee, Knoxville, TN, USA 1993. 155. Tjepkema JD, Yocum CS. Measurement of oxygen partial pressure within soybean nodules by oxygen microelectrodes. Planta 1974;119:351-360. 156. Appleby CA. Leghemoglobin and Rhizobium respiration. Ann Rev Plant Physiol1984;35:443-2478. 157. Witty JF, Minchin FR, Skot L, Sheely JE.Nitrogen fixation and oxygen in legume mot nodules. Oxford SUIVPlant Cell Biol 1986;3:275-315. 158. David M, Daveran ML, Batut J, Dedieu A, Domergue 0, Ghai J, Hertig C, Boistard P, Kahn D. Cascade regulation of nif gene expression in Rhizobium meliloti. Cell 1988;54671-683. 159. Parsons R, Day DA. Mechanism of soybean nodule adaptation to different oxygen pressures. Plant Cell Environ 1990;13:501-512. 160. Purcell LC, Sinclair TR. Soybean (Glycine max) nodule physical traits associated with permeability responses to oxygen. Plant Physiol 1993;103:149-156. 161. Denison RF, Kinraide TB. Oxygen-induced membrane depolarizations in legume root nodules possible evidence for an osmoelectrical mechanism controlling nodule gas permeability. Plant Physiol 1995;108:235-240. 162. Vikman P-A. The symbiotic vesicle is a major site for respiration in Frankia from Alnus incana root nodules. Can J Microbiol 1992;38:77+784. 163. Bond G. Some features of root growth in nodulated plants of Myricu gale L. Ann Bot 1952;16: 467-475. 164. Silvester WB, Silvester JK, Torrey JG. Adaptation of nitrogenase to varying oxygen tension and the role of the vesicle in root nodules of Alnus incana subsp rugosa. Can J Bot 1988;66:1772-1779. 165. Silvester WB, Whitbeck J, Silvester JK, Torrey JG. Growth, nodule morphology and nitrogenase activity of Myrica gale grown with roots at various oxygen levels. Can J Bot 1988661762-1771. 166. Silvester WB,Hanis SL. Nodule structure and nitrogenase activity of Coriaria arborea in response to varying PO,. Plant Soil 1989;118:97-109. 167. Hafeez F, Akkermans ADL, Chaudhary AH. Observations on the ultrastructure of Frankia spp. in root nodules of Datisca cunnabim L. Plant Soil 1984;79:383-402. 168. Parsons R, Silvester WB, Harris S , Gruijters WTM, Bullivant S. Frankia vesicles provide inducible and absolute oxygen protection for nitrogenase. Plant Physiol 1987;83:728-73 1. 169. Meesters TM, Van Vliet WM, Akkermans ADL. Nitrogenase is restricted to the vesicles in Frankia strain EANlpec. Physiol Plant 1987;70:267-271. 170. Berry AM, Haniot OT, Moreau RA, Osman SF, Benson DR, Jones AD. Hopanoid lipids compose the Frankia vesicle envelope, presumptive barrier of oxygen diffusion to nitrogenase. Proc Natl Acad Sci USA 1993;90:6091-6094. I7 1. Berg RH. Symbiotic vesicle ultrastructure in high pressure-frozen, freeze-substituted actinorhizae.
182 Protoplasma 1994;183:37-48. 172. Newcomb W, Wood SM. Morphogenesis and fine structure of Frunkiu (Actinomycetales): the microsymbiont of nitrogen-fixing actinorhizal root nodules. Int Rev Cytol 1987;109:1-88. 173. Berg RH, McDowell L. Endophyte differentiation in Cuswrinu actinorhizae. Protoplasma 1987;136:104-1 17. 174. Henry MF. h d e ultrastructurale de I’endophyte prksent dans les nodusitts radiculaires de Myricu gale L. Bull Soc Bot France 1977;124:291-300. 175. Berg RH, McDowell L. Cytochemistry of the wall of infected cells in Cuswrinu actinorhizae. Can J Bot 1987;66:2038-2047. 176. Zeng S , Tjepkema JD, Berg RH. Gas diffusion pathway in nodules of Cusuarinu cunninghumiunu. Plant Soil 1989;118:119-123. 177. Sellstedt A, Reddell P, Rosbrook PA, Ziehr A. The relations of haemoglobin and lignin-like compounds to acetylene reduction in symbiotic Cusuurinu. J Exp Bot 1991;42:1331-1337. 178. Zeng S, Tjepkema JD. The wall of the infected cell may be the major diffusion barrier in nodules of Myricu gale L. Soil Biol Biochem 1994;5:633-639. 179. Zeng S, Tjepkema JD. The resistance of the diffusion barrier in nodules of Myricu gale L. changes in response to temperature but not to partial pressure of 0,. Plant Physiol 1995;107:1269-1275. 180. Tjepkema JD, Asa DJ. Total and C0,-reactive heme content of actinorhizal nodules and the roots of some nonnodulating plants. Plant Soil 1987;100:225-236. 181. Sellstedt A, Reddell P, Rosbrook PA. The occurrence of haemoglobin and hydrogenase in nodules of 12 Cuswrinu-Frunkiu symbiotic associations. Physiol Plant 1991;82:458-464. 182. Pathirana SM, Tjepkema JD. Purification of hemoglobin from the actinorhizal root nodules of Myricu gale L. Plant Physiol 1995;107:827-831. 183. Hadfield KL, Bulen WA. Adenosine triphosphate requirement of nitrogenase from Azotobucrer vinelundii. Biochem 1969;8:5103-5108. 184. Wilson PW, Umbreit WW. Mechanism of symbiotic nitrogen fixation. In. Hydrogen as specific inhibitor. Arch Microbiol 1937;8:44*457. 185. Emerich DW, Ruiz-Argiieso T, Ching TM, Evans HJ. Hydrogen dependent nitrogenase activity and ATP-formation in Rhizobium juponicum bacteroides. J Bacteriol 1979;137:153-160. 186. Hanus FJ, Maier RJ, Evans HJ. Autotrophic growth of H,-uptake-positive strains of Rhizobium juponicum in an atmosphere supplied with hydrogen gas. Proc Natl Acad Sci USA 1979;76:17881792. 187. De Vries W, Stam H, Stouthamer AH. Hydrogen oxidation and nitrogen fixation in rhizobia, with special attention focused on strain ORS571. Antonia van Leeuwenhoek 1984;52:85-96. 188. Stam H, van Versefeld H, de Vries W, Stouthamer AH. Hydrogen oxidation and efficiency of nitrogen fixation in succinate-limited chemostate cultures of Rhizobium ORS57 1. Arch Microbiol 1984;139~53-60. 189. Murry MA, Lopez MF. Interaction between hydrogenase, nitrogenase and respiratory activities in a Frunkiu isolate from Ahus rubru. Can J Microbiol 1989;35:636-641. 190. Evans HJ, Hanus Ff, Russell SA, Harker m,Lambert GR, Dalton DA. Biochemical characerization and genetics of H, recycling in Rhizobium. In: Ludden PW, Bums JE (eds) Nitrogen Fixation and CO, Metabolism. New York: Elsevier, 1 9 8 5 ; F l l . 191. Garg N, Garg RP, Nainawatee HS. In pluntu comparison of Hup+ and isogenic Hup- Rhizobium leguminosurum. Indian J Exp Bot 199028:427-429. 192. Sajid GM, Campbell WF.Symbiotic activity in pigeon pea inoculated with wild-type Hup-, Hup+ and transconjugant Hup+Rhizobium. Tropical Agriculture 19947212:182-1 87. 193. Serensen GM, Wyndaele R. Effect of transfer of symbiotic plasmids and of hydrogenase genes (hup) on symbiotic efficiency of Rhizobium leguminosurum strains. J Gen Microbiol 1986132:3 17-324. 194. Hume DJ, Shelp BJ. Superior performance of the Hup- Brudyrhizobium juponicum strain 532C in Ontario soybean field trials. Can J Plant Sci 199070:661--666. 195. Bedmar EJ, Phillips DA. P isum surivum cultivar effects on hydrogen metabolism in Rhizobium. Can J Bot 1983;62:1682-1686.
183 196. Truelsen TA, Wyndale R. Recycling efficiency in hydrogenase uptake positive strains of Rhizobium leguminosarum. Physiol Plant 1984;62:45-50. 197. Saini I, Chander DR, Hagpal P. Uptake hydrogenase in fast-growing strains of Rhizobium spp. (Sesbania) in relation to nitrogen fixation. J Appl Bacteriol 1987;62:449-452. 198. Boogerd FC, Ferdmandy-Van Vlerken MMA, Mawadza C, Pronk AF, Stouthamer AH, Van Verseveld HW. Nitrogen fixation and hydrogen metabolism in relation to the dissolved oxygen tension in chemostat cultures of the wild-type and a hydrogenase-negative mutant of Azorhizobium caulinodans. Appl Environ Microbiol 1994;60 1859- 1866. 199. Huss-Dane11K. The physiology of actinorhizal nodules. In: Schwintzer CR, Tjepkema JD (eds) The Biology of Frankia and Actinorhizal Plants. New York: Academic Press, 1990; 129-156. 200. Sellstedt A, Lindblad P. Activities, occurrence, and localization of hydrogenase in free-living and symbiotic Frankia. Plant Physiol 199092:809-815. 201. Schubert KR. Products of biological nitrogen fixation in higher plants: synthesis, transport, and metabolism. Ann Rev Plant Physiol 198637539474. 202. Sellstedt A, Atkins CA. Composition of amino compounds transported in xylem of Casuarina spp. J Exp Bot 1991;42:1493-1497. 203. Yoneyama T, Kondo M. Sesbania spp., Aeschynomene indica and Crotaluria spp. are amideexporters. Soil Sci Plant Nutr 199036:689-693. 204. Hirel B, Perrot-Rechenmann C, Maudinas B, Gadal P. Glutamine synthetase in alder (Alnus glutinosa)root nodules. Purification,properties and cytoimmunochemical localization.Physiol Plant 1982;55:197-203. 205. Brears T, Walker EL, Coruzzi GM. A promoter sequence involved in cell-specific expression of the pea glutamine synthetase GS3A gene in organs of transgenic tobacco and alfalfa. Plant J 1991;1:235-244. 206. Miao G-H, Hirel B, Marsolier MC, Ridge RW, Verma DPS. Ammonia-regulated expression of a soybean gene encoding cytosolic glutamine synthetase in transgenic Lotus corniculatus. Plant Cell 1991;3:11-22. 207. Forde BG, Day HM, Turton JF, Wen-jun S , Cullimore JV,Oliver JE.Two glutamine synthetase genes from Phaseolus vulgaris L. display contrasting developmental and spatial patterns of expression in transgenic Lotus corniculatus plants. Plant Cell 1989;1:391-401. 208. Hanks IF,Tolbert NE, Schubert KR. Localization of enzymes of ureide biosynthesis in peroxisomes and microsomes of nodules. Plant Physiol 1981;68:65-69. 209. Nguyen T, Zelechowska M, Foster V, Bergmann H, Verma DPS. Primary structure of the soybean nodulin-35 gene encoding uricase I1 localized in the peroxisomes of uninfected cells of nodules. Proc Natl Acad Sci USA 1985;82:504(r5044. 210. Tat6 R, Pahiarca El, Riccio A, Defez R, Iaccarino M. Development of Phuseolus vulgaris root nodules. Molec Plant-Microbe Interact 1994;7:582-589. 21 1. Vaughn KC. Structural and cytochemical characterizationof three specialized peroxisome types in soybean. Physiol Plant 1985;64:1--12. 212. Selker JML. Three-dimensional organization of uninfected tissue in soybean root nodules and its relation to cell specialization in the central region. Protoplasma 1988;147:178-190. 213. Pate JS, Gunning GES, Briarty LG. Ultrastructure and functioning of the transport system of the leguminous root nodules. Planta 1969;85:1 1-34. 214. Schubert KR. Enzymes of purine biosynthesis and catabolism in Glycine mar. Plant Physiol 1981;86:1115--1122. 215. Atkins CA. Ammonia assimilation and export of nitrogen from the legume nodule. In: Dilworth M, Glenn A (eds) Biology and Biochemistry of Nitrogen Fixation. Amsterdam: Elsevier, 1991;293-319. 216. Martin F, Hirel B, Gadal P. Purification and properties of omithine carbamoyl transferase 1 from Alnus glutinosa root nodules. Z Pflanzenphysiol 1983;111:413-422. 217. BIom J, Roelofsen W, Akkermans ADL. Assimilation of nitrogen in root nodules of alder (Alnus glutinosa). New Phytologist 1981;89;321-326. 218. Wheeler CT, Lawrie AC. Nitrogen fixation in root nodules of alder and pea in relation to the supply
184 of photosynthetic assimilates. In: Nutman PS (ed) Symbiotic Nitrogen in Plants. Cambridge: Cambridge University Press, 1976497-509. 219. Burgess D, Peterson RL. Effect of nutrient conditions on root nodule development in Alnus juponicu. Can J Bot 1987;65:1658-1670. 220. Wheeler CT. The causation of the diurnal changes in nitrogen fixation in the nodules of Alnus glutinosu. New Phytologist 1971;70487-495. 221. Vance CP, Heichel GH. Carbon in N, fixation: Limitation or exquisite adaptation. Ann Rev Plant Physiol Plant Molec Biol 1991;42:373-392. 222. Wheeler CT, Watts SH, Hillman JR. Changes in carbohydrates and nitrogenase compounds in the root nodules of AInus glutinosu in relation to dormancy. New Phytologist 1983;95:20+218. 223. Pate JS. Root exudation studies on the exchange of ''C-labeled organic substances between the roots and shoot of the nodulated legume. Plant Soil 1962;17:333-356. 224. Morel1 M, Copeland L. Sucrose synthase of soybean nodules. Plant Physiology 1985;78:14+154. 225. Thummler F, Verma DPS. Nodulin-100 of soybean is the subunit of sucrose synthase regulated by the availibility of free heme in nodules. J Biologic Chem 1987;262:14730- 14736. 226. Kiister H, Friihling M, Perlick AM, Piihler A. The sucrose synthase gene is predominantly expressed in the root nodule tissue of Viciu fubu. Molec Plant-Microbe Interact 1993;6:507-5 14. 227. Vance CP, Gregerson RG, Robinson DL, Miller SS, Gantt JS. Primary assimilation of nitrogen in alfalfa nodules: molecular features of the enzymes involved. Plant Sci 1994;101:51-64. 228. Weaver CD, Shower NH, Louis CF, Roberts DM. Nodulin-26, a nodule specific symbiosome membrane protein from soybean, is an ion channel. J Biologic Chem 1994;269:17858-17862. 229. McClure PR, Coker GT, Schubert KR. Carbon dioxide fixation in roots and nodules of Alnus glutinosu. I. Role of phosphoenolpuryvate carboxylase and carbamyl phosphate synthetase in dark CO, fixation, citrulline synthesis, and N, fixation. Plant Physiol 1983;71:652-657. 230. Akkermans ADL, Huss-Dane11 K, Roelofsen W. Enzymes of the tricarboxylic acid cycle and the malate-aspartateshuttle in the N,-fixing endophyte of A h u s glutinosu. Physiol Plant 1981;53:289294. 23 1. Soltis DE, Soltis PS, Morgan DR, Swensen SM, Mullin BC, Dowd JM, Martin PG. Chloroplast gene sequence data suggest a single origin of the predisposition for symbiotic nitrogen fixation in angiosperms. Proc Natl Acad Sci USA 1995;92:2647-2651. 232. Trinick MJ. Structure of nitrogenrfixing nodules formed by Rhizobium on roots of Purusponiu undersonii Planch. Can J Microbiol 1979;25:565-578. 233. Lancelle SA, Torrey JG. Early development of Rhizobium-induced root nodules of Purusponiu rigidu. I. Infection and early nodule initiation. Protoplasma 1984;123:26-37. 234. Lancelle SA, Torrey JG. Early development of Rhizobium-induced root nodules of Purusponiu rigidu. 11. Nodule morphogenesis and symbiotic development. Can J Bot 1984;63:25-35. 235. Christensen T, Dennis ES, Peacock Landsmann J, Marcker KA. Hemoglobin genes in nonlegumes: cloning and characterizationof a Cusuurinu gluucu hemoglobin gene. Plant Molec Biol 1991;16:33+344. 236. Schlaman HRM, Spaink HP, Okker RJH, Lugtenberg BJJ. Subcellular localization of the nodD gene product in Rhizobium leguminosurum. J Bacteriol 1989;171:4686-4693. 237. Siqueira JO, Nair MG, Hammerschmidt R,Safir GR. Significance of phenolic compounds in plantsoil-microbial systems. Crit Rev Plant Sci 1991;10:63-121. 238. Carlson RW, Juan SJ, Bhat UR, Glushka J, Spaink HP, Wijfjes AHM, Van Brussel AAN, Stokkermans TJW, Peters NK, Stacey G. The structures and biological activities of the lipooligosaccharide nodulation signals produced by type-1 and type-2 strains of Brudyrhizobiurn juponicum. J Biologic Chem 1993;168:18372-18381. 239. Price NPJ, Relic B, Talmont F, Lewin A, Prom6 J-C, Broughton WJ. Broad-host-range Rhizobium species strain NGR234 secretes a family of carbamoylated, and fucosylated, nodulation signals that are 0-acetylated or sulphated. Molec Microbiol 1992;6:3575-3584.
w,
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R.El-Gewely, editor.
185
Using nonviral genes to engineer virus resistance in plants Christophe Robaglia and Mark Tepfer Laboratoire de Biologie Cellulaire, INRA-Versailles, Versailles Cedex, France
Abstract. In this review are presented some emerging methods to improve virus resistance in transgenic plants, based on the use of nonviral genes from various organisms. The genes discussed include ones coding for proteins involved in virus resistance: antibodies, PKR kinase and 2’-5’ oligoadenylate synthase from mammalian cells; the yeast SKI antiviral genes. Genes not clearly involved in resistance are also described, including ones encoding protease inhibitors or ribosome-inactivating proteins. The natural pathways for pathogen resistance in plants may also be enhanced, through modification of genes involved in the hypersensitive response, systemic acquired resistance, or nonhost and extreme resistance. Key words: 2’-5‘ oligoadenylate synthase, antiprotease, antibodies, dsRNA-activated protein kinase, hypersensitive reaction, resistance gene, ribosome-inactivating protein, SKI genes, systemic acquired resistance, transgenic plants, virus resistance.
Introduction The first report of virus-resistant transgenic plants appeared 10 years ago [ 11, and was in fact one of the very first of many others that have appeared since then describing transgenic plants of potential agronomic interest. That plants expressing a viral coat protein gene were resistant to the donor virus, was generally thought to be quite surprising, although this was in fact consistent with the concept of pathogen-derived resistance that had been formulated the year before [2]. Briefly stated, this concept postulates that the expression of a pathogen’s own genes in a host can interfere with host-pathogen interactions, thus rendering the host resistant. Over the last 10 years, numerous fragments of diverse viral genomes have been introduced into plants, and resistance obtained with a remarkable range of such genes, including ones that encode either coat protein or modified forms of movement protein or replicase. Genes whose transcripts include noncoding segments of viral genome can also confer resistance. Since pathogen-derived virus resistance genes have been reviewed regularly [3,4],our purpose here is to focus on other less well-known means of creating virus resistance genes, which are not virus derived, and which are in all cases either at the very beginning stages of evaluation or remain simply potential strategies for inducing resistance that may be explored in the future. Considering the remarkable successes of virus-derived resistance genes, one might
Address for correspondence: Christophe Robaglia, Laboratoire de Biologie Cellulaire, MA-Versailles, F-78026 Versailles Cedex, France. Tel.: +33- 1-30-83-30-29. Fax: +33-1-30-83-30-99.
186
wonder if it is in fact useful to continue seeking new strategies. In fact there are several excellent reasons for doing so. One is that, so far, we have no certainty concerning the durability of the resistance conferred by virus-derived genes. As is the case for natural resistance genes, resistance-breaking viral strains could appear quite rapidly in the field, in an extremely unpredictable manner. Thus, it would be wise to prepare replacement genes, ready for rapid deployment if necessary. A second point is that there is some concern that the expression of viral sequences by plants could lead to novel perturbations of the natural populations of plant viruses, which could give rise to negative ecological side effects (for review see [5,6]).If in fact serious problems do arise, here again it would be prudent to have ready other types of virus resistance genes. A third, quite positive potential of at least certain strategies for using nonviral genes to create resistance, is that many of them have the potential for conferring protection against a wide range of viruses, whereas virus-derived genes generally confer resistance only to closely related viruses.
Genes involved in virus resistance in organisms others than plants Viruses parasitize all living organisms. Like their hosts, although being extremely diversified, they can present certain common structural features that have been conserved through evolution. This presumably reflects their ability to use and divert to their profit conserved biochemical processes of the host cells. These characteristics can be the basis of virus classification in superfamilies transcending their hosts classification in kingdoms. An example is the grouping of some plant and animal viruses in sindbis-like and picoma-like superfamilies, based on protein sequence similarities and on overall genome structure and organization [7]. Given that among viral genomes there are conserved structural elements and genome expression strategies, it is conceivable that antiviral defense mechanisms developed by primitive organisms would have been conserved through evolution or that antiviral mechanisms developed independently by hosts could still be operational when transposed into an heterologous context. Since cultivated plants are the result of extremely long-term selection dating back to prehistory, they contain only a fraction of the genetic diversity present in their wild-type relatives. Wild plants are often more resistant to many plant viruses than cultivated species, and can often tolerate the multiplication of viruses without visible symptoms. It is thus probable that ancestral genes involved in coping with the existence of viruses were lost during the selection process. In this section, we will focus on the possible introduction into plants of genes involved in virus resistance in mammals and in yeasts.
Antiviral antibodies Many essential functions are carried out by similar proteins in plants and animals. Nonetheless, the well-accepted truism that plants are not simply green animals, is strongly supported by certain cases of radically different strategies in the different
187 kingdoms. One of the clearest cases of plant/animal differences is the absence of proteins homologous to antibodies in plants. Over the past several years, and for a variety of reasons, there has been considerable activity in the area of transformation of plants with genes encoding immunoglobulins (see [8] for review). Often, it has been preferred to use genes encoding only part of the complete antibody (Fig. 1). Among these is a single report concerning transgenic plants expressing a gene encoding a single-chain antibody (SCFV)directed against the coat protein of artichoke mottled crinkle tombusvirus (AMCV) [9]. One of the advantages of a gene encoding an scFv, in which the variable domains of the heavy and light immunoglobulin chains are synthesized as a single fusion protein, is that this avoids problems of subunit assembly, which often requires secretion into the ER of separate heavy and light chains, and which would thus favour accumulation of the antiviral antibodies in the wrong cell compartment for interaction with viral coat protein, which one would expect to be essentially cytosolic. The transgenic plants obtained presented good levels of resistance to the virus (Fig. 2), which was not overcome by increasing the inoculum level. Needless to say, it would be of great interest to know by what mechanism(s) the binding of coat protein by the recombinant antibody leads to resistance. The question is complex, however, since viral coat protein plays multiple roles in the infection cycle; in addition to protection of viral nucleic acids, coat protein is usually involved
Fig. I . Schematic representation of the protein structure of various forms of complete and deleted immunoglobulins. Oval shaded domains: hypervariable domains of light and heavy chains; Fab: antigenbinding fragment; scFv: single-chain antibody in which the hypervariable domains of light and heavy chains are joined by a linker peptide; VH: hypervariable domain of heavy chain.
188
0
10
20 Time after inoculation(days)
30
Fig. 2. Resistance of plants expressing an scFv directed against ACMV coat protein. Top: virus titer in inoculated and second upper leaves of control and transformed plants 14 days after inoculation with 40 ndml ACMV; bottom: percentage of nontransformed control plants (0)and two lines of transformed plants (0and 0) that were infected after inoculation with ACMV virions. (After Tavladoraki et al. [9].)
in short- and long-distance movement of virus in the plant, and may also play a direct role in regulation of replication of the viral genome. Some indications are given by Tavladoraki et al. [9], who carried out infection experiments in protoplasts (isolated cells from which the cell wall has been removed enzymatically). They observed a lower percentage of infection in protoplasts prepared from the transgenic plants, which suggests that at least part of the resistance observed is likely to occur at the
189 level of virus replication or decapsidation/encapsidation in initially infected cells. Further characterization of the protection observed would certainly cast more light on the mechanism(s) involved. For instance, is protection still observed if viral RNA rather than intact virus is used as an inoculum? And, how strain- or virus-specific is the protection? Further examination of the mechanisms could also be expected to be informative regarding the role of the coat protein in virus infection. More recently, Voss et al. [lo] have explored a different strategy for using antibody genes to confer virus resistance. They constructed genes allowing expression in plants of full-length heavy and light chain subunits of an antibody directed against an epitope specific to assembled particles of tobacco mosaic tobamovirus (TMV). In the transgenic plants, since the proteins include N-terminal signal peptides, the antibody is secreted into the extracellular space. The genes were introduced into a tobacco variety bearing the N TMV-resistance gene, in which the resistance phenotype is expressed as local necrotic lesions (see section below on natural hypersensitive resistance). Upon infection with TMV, fewer lesions were observed in the plants expressing the antibody genes, suggesting that they may interfere with the very initial stages of infection, involving penetration of virus particles into host cells. Unfortunately, the question that these results do not answer is whether such extracellular antibodies will confer protection in genotypes that are not already naturally resistant. Voss et al. [ 101 have also introduced the same genes into a TMVsensitive tobacco genotype, so clarification of this question should soon be available. Mammalian antiviral activities not mediated by the immune system As shown schematically in Fig. 3, mammals possess two types of virus resistance mechanisms activated by interferon and by the presence of double-stranded RNA (dsRNA). One is the 2’-5’ oligoadenylate (2-5A) synthetase pathway, and the other is based on the inhibition of translation by a dsRNA-activated protein kinase PKR (also called DAI for dsRNA-activated inhibitor), p68 (human) or p65 (murine) protein kinase [ll]. The 2 ‘-5‘oligoadenylate pathway Studies of cells infected with viruses have shown that interferons a,p and y induce the accumulation of both 2-5A synthetase and 2-5A-dependent RNase [12]. 2-5A oligonucleotides, ppp(A2’pS’),A, are produced from ATP following activation of the 2-5A synthethase by low concentrations of dsRNA of viral origin. In tum, 2-5A activates the 2-5A-dependent RNase, which has no detectable RNase activity until it binds to 2-5A. This RNase can cleave single-stranded RNA, of viral and cellular origin, 3‘ of UpNp dimers. The system can be regulated by the rapid degradation of 2-5A by 2’-5’ phosphodiesterase and 5’ phosphatase. Several studies have pointed out the implication of the 2-5A pathway in the inhibition of picomavirus replication, and overexpression of the 2-5A synthetase in animal cells was found to increase their resistance to picomaviruses [13]. It is, however, thought that the full biological significance of the system can go beyond virus resistance, possible other roles being
190
A
Degradatlon of singlastranded RNA
t
Q
activation of RNase L
@--
1
synthesis of 2’-5’Aoligomers
activation of 2-5’synthetase
Viral infection+replication+l
I
double-stranded RNA I I
B
‘’
PKR !tivat
self-phosphorylation
I
ion
phosphotylation translation initiation factor elF2 a
. )elF2 a-P
t t
sequestration of GEF Block of translation initiation
Fig. 3. The two mammalian antiviral activities activated by double-stranded RNA. A: The 2’-5’ oligoadenylate pathway. The 2’-5’ oligoadenylate synthase is activated by the presence of dsRNA and synthesizes 7 - 5 ’ oligoadenylates from ATP. These bind and activate RNAase L, which can degrade both cellular and viral RNA. The system is regulated by the rapid degradation of the 2’-5’ oligomers by the combined action of phosphodiesterasesand phosphatases (see [ 11,161). B: The PKR pathway. PKR is activated by binding of two molecules to a single molecule of dsRNA, followed by reciprocal phosphorylation. When activated, PKR can phosphorylate the a-subunit of the translation initiation factor eIF2. Phosphorylated eIF2-GDP cannot be recycled into active eIF2-GTP by the guanine nucleotide exchange cofactor (GEF) and instead sequesters it, leading to blockage of translation initiation (see [18]). 2’-5’ synthetase, RNase L and PKR accumulate in response to interferon.
the control of general RNA stability and cell proliferation [14]. This system has been found in reptiles, birds and mammals [15]. It was recently shown that plant extracts contain activities degrading 2-5A oligomers, and that these compounds decrease protein synthesis and RNA stability in translation extracts. However, no 2-5A synthetase activity was detected [16]. It was thus proposed that plants contain certain components of the 2-5A antiviral pathway and that the system could be reconstituted by the addition by genetic engineering of missing components, such as 2-5A synthetase.
191 Indeed, a gene encoding a rat 2-5A synthetase was found to increase the tolerance of transgenic tobacco and potato plants to different plant RNA viruses. In a first study [17], transgenic potato plants and control plants were challenged under field conditions with potato virus X, potexvirus (PVX) and potato virus Y, potyvirus (PVY). Transgenic plants expressing the mammalian 2-5A synthetase were found to contain significantly lower PVX concentrations than control plants, and one clone also displayed a slight tolerance to PVY infection. In further experiments ([ 161; Truve and Saarma, personal communication), transgenic tobacco plants expressing the rat 2-5A synthetase were challenged with PVX, TMV and potato virus S carlavirus (PVS). All transgenic clones were resistant to PVS infection, and in one clone the accumulation of all three viruses was reduced. These data suggest that the manipulation of the 2-5A pathway can be efficiently used to improve virus resistance in plants. Interestingly, the least efficient protection was observed against a potyvirus, PVY, which are among the closest plant relatives of the animal picomaviruses, against which the system is the most efficient. This suggests that these related viruses may somehow differ in their replication characteristics, perhaps in the intracellular localisation of replication sites or in other features that may affect RNA accessibility. The dsRNA-activated protein kinase (PKR) In addition to the 2-5A pathway, mammalian cells possess another antiviral mechanism mediated by the translation inhibition activity of the dsRNA-activated protein kinase PKR, also known as p68 or p65 protein kinase (reviewed in [ 181). This protein has two serine-threonine kinase activities, one for its own activation and the other for the phosphorylation of the a-subunit of the translation initiation factor, eIF2. The presence of low concentrations of dsRNA leads to autophosphorylation of the PKR protein and its activation, whereas high concentrations of dsRNA inactivate it. It is thought that activation results from the simultaneous binding of two molecules of the kinase on one molecule of dsRNA, followed by reciprocal protein phosphorylation. If dsRNA is in excess, activation is inhibited, perhaps because of the binding of most of the pre-existing b a s e to different RNA molecules. When activated, the enzyme can phosphorylate the a-subunit of the e F - 2 translation initiation factor, involved in the formation of the ternary initiation complex, eIF-2/GTP/Met-tRNA. When eIF-2 completes one round of initiation it is ejected from the ribosome as an eIF-2-GDP complex, which is then recycled into eIF2-GTP by the guanine nucleotide exchange cofactor eIF-2B. Phosphorylation of eIF2 a by the PKR kinase blocks initiation by sequestering eIF-2B. Two other proteins have been found to regulate translation initiation by phosphorylating eIF-201, one is the yeast GCN2 b a s e and the other is the hemin-regulated inhibitor (HRI) found in reticulocytes [ 181. The cDNA of the human p68 kinase has been constitutively expressed in NIH 3T3 mouse cells [19] and found to confer resistance to encephalomyocarditis virus (a picomavirus) in the absence of an inducing interferon treatment, but not to vesicular stomatitis virus (a rhabdovirus). Further proof that the PKR kinase plays an important role in cellular defences against viruses is found in the impressive array of strategies
192 developed by animal viruses to counteract its activity [20]. Viruses have been shown to induce PKR degradation (poliovirus), to inhibit its activity by the production of RNA decoys (adenovirus VA RNA), or to block eIF-2a phosphorylation (influenza virus). It is also likely that the PKR kinase, besides its antiviral function, is involved in the regulation of cellular gene expression. In yeast, the GCN2 kinase can sense the levels of uncharged tRNA in the cell. Under conditions of amino acid starvation, eIF-2a becomes phosphorylated and allows ribosomes to skip over the upstream pORFs of the GCN4 mRNA to reinitiate translation at the GCN4 ORF. GCN4, which is a transcription factor, can thus activate a set of genes involved in amino acid biosynthesis. The mammalian PKR kinase can complement yeast mutants devoid of the GCN2 kinase by phosphorylating yeast eIF2 a and allowing production of the GCN4 protein [21]. A slow growth phenotype has been associated with the presence of PKR in yeast, and has been attributed to a constant activation by endogenous yeast dsRNA viruses (see below) [22]. It is thus conceivable that mammalian PKR could also function in a plant context, and thus lead to virus'resistance. Recently, evidence suggesting the existence of a PKR analog in plants that is phosphorylated upon infection by viruses and viroid, has been reported [23,24]. Most plant viruses have single-stranded RNA genomes and produce dsRNA as replication intermediates. As this dsRNA would be an activator of the PKR kinase, it is anticipated that the expression of PKR in plant cells would lead to a form of resistance active against many virus species. To test this hypothesis, a cDNA coding for the human PKR (p68) has been engineered for proper expression in plants cells and has been transferred to tobacco and Arabidopsis thaliana plants (F. Vilaine and C. Robaglia, unpublished). The yeast antiviral system
Most laboratory strains of yeast contain several RNA viruses that can only be transmitted by cytoplasmic fusion. Some of these genetic elements are composed of dsRNA: the L-A, M, and L-BC viruses (Fig. 4). The M RNA appears to depend on L-A for its replication, and was found to code both for a toxin (the killer toxin) and for a protein conferring immunity to the toxin onto the yeast cells containing M. Thus, the production of the killer toxin by yeast cells (the killer phenotype) harboring M is lethal only to cells not containing M (reviewed in [25]). Mutations in the nuclear SKI genes lead to increased replication of the M RNA and to a superkiller phenotype. These genes also control the level of the L-A and LBC RNA viruses, and to an unrelated single-stranded RNA replicon called 20s RNA. Although most simple SKI mutants are not affected in any known biological function, ski2 and ski3 mutants were recently found to be lethal, but only if the cell is also mutated in another gene (SEPI) involved in RNA stability and DNA recombination [26], suggesting that some of the SKI genes might be involved in cell functions other than virus resistance. Six SKI genes have been identified, of which three have been cloned and sequenced. The SKI3 gene product is a 165 kD nuclear protein, and the SKI8 protein
193
killer protoxir
killer toxin I*
cytoplasm
Fig. 4. The yeast antiviral system. SKI genes regulate the copy number of double-stranded endogenous RNA viruses and of the 20s circular RNA replicon. Only the role of the SKI2 gene has been elucidated. The Ski2 protein inhibits specifically the translation of noncapped and nonpolyadenylated M and LA mRNA. The reduction of M mRNA translation prevents the accumulation of the killer toxin encoded by M (which is dependent on the presence of LA for replication). This toxin is activated by proteolytic cleavage, and is active only on cells not containing M.
bears a repeated amino acid pattern homologous to P-transducin, but their mode of action is still unresolved. The SKZ2 gene codes for a 145 kD protein, with motifs typical of the expanding helicase family, as well as ones found in certain nucleolar proteins. The SKZ2 gene product appears to act by reducing the level of translation of viral and other mRNAs devoid of the 5’ M7GpppN cap structure and/or of a polyadenylate tail at their 3‘ end [27,28]. Homologs of SKZ2 were also recently identified in animal cells, but their functions in animals are at this time unknown ~91. Many plant virus RNAs are uncapped and/or nonpolyadenylated [30]. For example, necrovirus genomic RNAs are neither capped nor polyadknylated. The genomic RNA of many viruses (e.g., tobamo-, cucumo- and bromovimses) is capped but nonpolyadenylated, and the RNA of viruses belonging to the como-, nepo- and potyviruses are polyadenylated but functionally uncapped, since they possess, instead, a protein (viral protein genome-linked, VPg) covalently linked to their 5’ end. Luteovirus and sobernovirus genomic RNAs also possess a VPg, but are nonpolyadenylated at their 3’ end. Experiments are under way to test whether the introduction of yeast SKI genes into transgenic plants can provide a new method to increase resistance to a large array of viruses.
194
Inhibitors of viral proteases The infection cycle of many plant viruses, as well as that of certain animal viruses such as HIV or poliovirus, requires precise cleavage of viral polyproteins to produce essential functional proteins. As a result, there has been considerable interest in using protease inhibitors as antiviral agents, particularly as a strategy for inhibition of HIV multiplication [313. In a preventive approach, plant scientists have proposed constitutive expression of peptide antiproteases in transgenic plant as a means of protection against attack by viruses of the picoma-associated supergroup, including the potyviruses and the comoviruses, which all synthesize polyprotein precursors [32,33]. There are, as yet, no published results showing the efficacy of this strategy in transgenic plants. However, the results of Garcia et al. [34] are promising, since they observed inhibition in vitro of two plum pox potyvirus proteases by cystatin C, an inhibitor of cysteine proteases. Although this section deals mainly with results obtained with proteinase inhibitors of animal origin, similar proteins have also been abundantly described in the plant kingdom. Their genes are being introduced for overexpression in transgenic plants in the perspective of improving not only virus but also insect resistance [32,33].
Plant genes potentially useful for creating virus resistance c
Many plant species display natural mechanisms for resistance to viruses (for reviews see [35,36]). Resistances governed by single genes are good candidates for being reintroduced by genetic transformation into elite cultivars without having to rely on the tedious process of sexual crossing followed by numerous rounds of backcrosses. It is anticipated that useful resistance genes of plant origin might be functional in other plant species, overcoming the barriers of sexual hybridization.
Genes encoding plant ribosome inactivating proteins The best-studied plant virus resistance genes provide relatively narrow strain- or virus-specific protection. In contrast, certain plant antiviral proteins would be expected to provide broad-range protection, as has been shown in the case of the ribosome inactivating protein (RIP) of Phytolacca americana, often referred to as pokeweed antiviral protein' (PAP), which is a potent inhibitor of infection of plants by a range of RNA and DNA viruses [37]. In fact, RIPS have attracted great interest in recent years for a range of potential therapeutic uses, including as anti-HIV therapeutic agents, or as toxic moieties fused to monoclonal antibodies to create cellspecific cytotoxins (see [38] for review). The toxic activity of RIPS is due to their RNA N-glycosidase activity, specific to a particular adenine residue in a highly conserved region near the 3' end of 28s ribosomal RNA (Fig. 5). RIP activity leads to complete inactivation of the 60s ribosomal subunit, which is no longer able to bind elongation factor 2 (EF-2). RIPS
195
A
U
G
G A
A
A C
C
C
U C - G
G - C G - A
C - G U C - G A -
4319 A
-
I
I
Ribose-Ribose-Ribose-Ribose-Ribose-Ribose
1
I
U
U 4350
Fig. 5. The mechanism of action of ribosome-inactivating proteins (RIPs). RIPS have a specific Nglycosidase activity that cleaves a single adenine base of the 28s RNA of the large ribosomal subunit in a highly conserved loop, resulting in the loss of ribosome function. (After [38].)
are extremely widespread in plants, having been reported in both monocot and dicot species. Surprisingly, certain bacterial toxins, such as those produced by Shigella dysenferiae and the Shiga-like toxins of certain Escherichia coli strains, also have an identical enzymatic activity [39]. Although the catalytic activity of all known RIPs is the same, there is great variability in their range of specificity. For instance, certain dicot RIPS, such as PAP, have similar levels of activity against all eucaryote ribosomes, whereas the RIP of barley seeds has essentially no activity against plant ribosomes [40]. Since RIPs are often accumulated to high levels, insensitivity of “self’ ribosomes is very likely to be an important mechanism for preventing autotoxic effects. Other RIPs, including several well-studied dicot RIPS such as ricin or trichoxanthin, are synthesized in the form of inactive preproproteins, which are proteolytically activated during sequestration, either in vacuoles or in the cell wall space [4 1,421. In 1993, Lodge et al. described the introduction of a PAP gene into tobacco (Nicotiana tabucum), N. bentharniana and potato [43]. A first point of interest is that they had difficulty in obtaining the transformed plants, presumably due to counter selection due to the toxic effects of high levels of PAP expression. Among the tobacco transformants obtained, there were also signs of toxicity, since those with the highest levels of expression showed showed stunting, leaf mottling, and were sterile.
196 However, when the plants were inoculated with PVX, PVY or cucumber mosaic cucumovirus (CMV), protection was observed, even in lines where the level of PAP was low enough for there to be no visible signs of toxicity. Thus, as was predicted from studies of inoculation with various viruses in the presence of PAP, they have indeed obtained broad-spectrum protection. The critical outstanding question is that of the mechanism by which PAP confers virus resistance. The obvious hypothesis is that resistance is mediated by the known RIP activity. How then can there be virus resistance without cytotoxicity? One could imagine that as yet unknown changes in translation in virus-infected cells would lead to an infection-specific cytotoxic effect, leading to the death of only the first infected cell(s). However, there is also evidence that RIPS can inhibit HIV replication in animal cells without cytotoxicity [MI. If this is the case, then it might be possible to reduce cytotoxicity without losing antiviral activity. If RIP genes are to find application for virus resistance in transgenic plants, it will certainly also be well worth while to test the activity of genes encoding other RIPS, since, as mentioned above there is great variation in the toxicity of the different RIPS to ribosomes of various organism. For instance, it has been shown that a barley RIP, which is active against fungal ribosomes and inactive against plant ones [40], confers resistance to fungi in transgenic plants [45]. It would be interesting to know if virus infection is inhibited in these plants. From experiments with inoculation of nontransgenic plants with TMV and with or without barley RIP, one would predict that the barley gene would have little or no effect in transgenic tobacco [40]. In particular, if the antiviral activities of RIPS are distinct from their cytotoxic effects, other dicot RIPS may naturally present different degrees of the two activities, and could be the source of antiviral RIPS that would have less toxic effects than PAP. Natural gene-for-gene resistance The concept of gene-for-gene resistance was first proposed by Flor in 1956 [46]. It postulates that a particular gene (the avirulence or incompatibility gene) in a pathogen, interacts with a corresponding gene (the resistance gene) in the host to determine the outcome of the infection. In molecular terms, it is generally thought that the product of the resistance gene can recognize a direct or indirect product of the avirulence gene, and thus induce the establishment of diverse types of resistance reactions [47]. Hypersensitive resistance Gene-for-gene resistance is often characterized by the rapid appearance of local necrosis, as shown in Fig. 6, and by the activation of an array of defense responses at the site of infection, which together consitute the hypersensitive reaction (HR). In the case of HR, it appears that a complex chain of transduction events is relaying the initial recognition signal. Both the death of host cells and the production of defense factors in surrounding cells can contribute to the localization of infection [48]. In the case of HR leading to virus resistance, it has been found that the virus can often
197
Fig. 6. Hypersensitive reaction of tobacco resistant to TMV due to the N gene. The lesions are approximately 2 mm in diameter.
replicate in host plant protoplasts, implying that the induction of HR depend on the presence of cell-to-cell contacts or is developmentally regulated [36]. The sequence analysis of viral strains overcoming resistance genes and the ease of manipulation of their genome through in vitro recombination and site-directed mutagenesis, allowed identification of viral gene products and amino acid sequences involved in gene-for-gene relationships. The coat protein of TMV has been found to be the avirulence determinant recognized by the N' hypersensitivity gene from Nicotiana sylvestris [49,50], and similarly, the hypersensitive response to TMV controlled by the N gene from N . glutinosa depends on an amino acid sequence of the 126-kD polymerase protein [51]. Only one plant gene leading to virus resistance through a gene-for-gene interaction has been isolated [52]. The N gene originating from N . glutinosa was cloned using a transposon tagging strategy and found to encode a 131.4 kDa protein with sequence similarities with the Drosophila Toll protein and with the mammalian interleukin-1 receptor, suggesting that the N gene product may play a role in an as yet unidentified signal transduction pathway. The introduction of the cloned gene into TMV-sensitive tobacco readily converted them to TMV resistance. It will undoubtedly be known in the near future whether the N gene can confer TMV resistance to other plant species, opening the way for the use of natural resistance genes as tools for improving crop resistance against pathogens. Interestingly, the N gene product also has structural features in common with genes leading to hypersensitive resistance against bacterial and fungal pathogens, in particular, a leucine-rich imperfectly repeated region in its C-terminus. These domains are generally found to be involved in protein/protein interactions [53], and are thus possible candidates for the recognition of the various pathogen avirulence gene products. It has been further shown that sequences homologous to the cloned resistance genes are located in their close vicinity, and it
198 has long been known that different resistance genes are often clustered within plant genomes (reviewed in [54]). This leads to the speculation that the generation of new resistance specificities can be generated by recombination events within and between homologous sequences. These observations, together with the presence of tandem repeats within a family of Xanthomonas avirulence genes [55], which can lead to the generation of new host specificities through inter- or intragenic recombination, give a conceptual framework for the molecular interpretation of the original Flor gene-forgene model. Most of the necessary tools are now available for a molecular dissection of the host and pathogen components of the recognition processes leading to the hypersensitive response. It is anticipated that this knowledge may ultimately lead to the artificial design and introduction into crops of resistance genes conferring resistance to new pathogens and to those for which no natural source of resistance is known. Extreme resistance A less common kind of resistance is the appearance of an immune (or extreme resistance) state that is also functional at the protoplast level and is controlled by dominant genes. This is the case of the Rx and Ry genes found in wild potato species, which confer resistance against PVX and PVY, respectively [35,56]. The coat protein of some PVX strains was found to trigger the resistance response controlled by the Nx (a hypersensitivity gene) and Rx genes from wild potato species [57]. Interestingly, the activation of the Rx gene by PVX appears also to confer resistance to the unrelated CMV at the protoplast level, demonstrating that although the recognition is specific, the activated resistance mechanism is not [58]. The existence of such types of extreme resistance genes acting nonspecifically at the cell level in plants, is not without parallels with the antiviral response existing in mammalian cells, namely the 2-5A and PKR pathways, which have been reviewed in a previous section. Biochemical studies further suggest that plant cells contain an analog of the mammalian dsRNA-activated PKR kinase, which is specifically phosphorylated during viroid and virus infections [23,24]. The molecular characterization of these genes may provide useful tools for engineering plants with broad resistance to viruses. It should be noted that some cases of apparent immunity were found to result, in fact, from limitation of infection to the initially infected cells, as demonstrated by the full replication of the virus in protoplasts of resistant plants. The resistance is thus acting by limitation of the virus movement from cell to cell. This is the case of the tomato Tm2 genes. This was clearly demonstrated when it was shown that mutations in strains of tomato mosaic tobamovirus that overcome Tm-2 and Tm22were localized in the viral 30-kD movement protein [59,60]. Modification of systemic acquired resistance
As shown schematically in Fig. 7, the initial attack of a plant by a pathogen can trigger a complex range of systemic responses leading to the establishment of a state
199
SAR PR genes expression
-
\
antivira~acthrities .%increase
oxidah'veburst
Fig. 7. Simplified scheme of the hypersensitive and systemic acquired resistances to virus infection ill plants. Following virus entry, the cell recognizes the pathogen in an unknown manner, through the action of resistance genes (R genes). The recognition signal leads to the accumulation of active oxygen species (AOS) and salicylic acid (SA), to the cross-linking of cell wall proteins and to localized cell death (HR). The accumulation of SA triggers the transcriptional activation of genes coding for pathogenesis-related proteins (PR-proteins), the synthesis of the antimicrobial compounds known as phytoalexins and of unknown antiviral activities. An unknown signal is translocated through the plant vascular system and leads to an accumulation of SA and to the synthesis of PR-proteins, phytoalexins and antiviral activities in parts of the plant distant from the site of infection, preventing infection by subsequently applied pathogens. (Adapted from Dempsey and Klessig [62].)
of resistance to subsequent infections by the same and by other pathogens, which is known as systemic acquired resistance (SAR). S A R is active against all types of pathogens, including viruses, and is characterized by an increase in salicylic acid, by the induction of an array of genes coding pathogenesis-related (PR) proteins and by the production of the antimicrobial compounds known as phytoalexins (reviewed in [61,62]). Genes coding for PR proteins have been manipulated and introduced into transgenic plants [63,64], in some cases resistance to fungal pathogen was obtained, but no increase in virus resistance has been reported. Studies are now focusing on the isolation of genes involved in the induction or maintenance of SAR. Arabidopsis thaliana mutants affected in the transduction pathways leading to SAR have already been isolated [65-671. Transgenic plants with a modified S A R response and an increase in TMV resistance have been obtained by expression of a gene coding for a small GTF' binding protein from rice [68]. Interestingly, these plants also contained
200 higher cytokinin levels, suggesting the involvement of this plant hormone in SAR. This is supported by another type of experiment, where the expression of the tobacco gene coding for S-adenosylhomocysteine hydrolase (SAHH) was reduced using antisense RNA [69]. The resulting plants also had higher cytokinin levels, which was attributed to a reduction in SAHH cytokinin-binding activity, and also displayed resistance to TMV, CMV, PVX, and PVY. Nonhost resistance
Another level of resistance is known as nonhost resistance, where a plant cannot be infected by a virus by any means. In fact, this situation is very common, since most plants are nonhost to most viruses. It is generally thought that this results from an inadequation of host cell components to virus replication or movement. The operational basis for the possibility of transferring a resistance character from one plant to another is the evidence that it is genetically controlled in a dominant or semidominant manner. Since nonhost resistance appears generally to be a recessive and multigenic trait, one would predict that it would be difficult to transfer by genetic engineering. However, the development of techniques for site-directed recombination in the plant genome [70], might nevertheless allow the replacement of a gene essential in some step of a virus life cycle by an inadequate counterpart from a related plant species (or by an allele mutated in vitro). In the same manner, as plant breeders often introgress recessive resistance genes into new varieties, these could possibly be introduced by transformation to replace the pre-existing alleles that lead to virus sensitivity.
Conclusions Effective genetic engineering strategies to control certain virus diseases in plants already exist, (e.g., coat-protein-mediated protection) and it is expected that plants incorporating these new traits will be soon launched on the market. Large-scale field cultivation of such plants will also open important new research areas, such as evaluation of the durability in the field of the protection conferred by the new resistance genes. Since viruses are very adaptable biological entities, it is probable that singly applied strategies may ultimately be overcome, as can be natural resistance genes. This strengthens the' need for alternative strategies. A second area of future research is to explore possibilities for designing genes that can improve the resistance of plants to a large range of viruses, and even to other pathogens. For this goal the manipulation of SAR offers promising perspectives. However, since S A R is a complex physiological response, its constitutive activation may lead to alterations of the plants agronomical potential. For instance, plants expressing PAP, the rice rgp 1 protein, or SAHH antisense RNA, are reported to display morphological alterations. Further detailed molecular analysis is necessary to allow optimization of natural pathways for virus resistance in plants. The efficiency of the 2-5A system in plant
20 1 cells is further evidence that viruses have been a constant problem in the evolution of cells and organisms. The apparent involvement in the basal cell machinery of genes also involved in virus resistance (2-5A, PKR, SKI) further suggests that related systems may be identified in distantly related organisms such as plants. As discussed in the sections Mammalian antiviral activities not mediated by the immune system and The yeast antiviral system, dsRNA is an essential intermediate of RNA virus replication, and often plays a role in the induction and/or regulation of virus resistance mechanisms. Watanabe et al. (1995) [71] expressed in transgenic plants a gene (pacl) encoding a dsRNA-specific RNase from the yeast Schizosaccharomyces pombe. They showed that the plants were partially protected against infection by CMV and PVY, and developed fewer lesions when infected by TMV. Expression of the pacl gene had no apparent deleterious effect on plant development.
Acknowledgements We thank M. Saarma, M. Hartley and N. Tumer for communicating unpublished results, and E. Benvenuto, D.A. Lappi and D.F. Kessig for allowing us to reproduce or readapt their published results.
References 1. Powell-Abel P, Nelson RS, Barun De B, Hoffman N, Rogers SG, Fraley RT, Beachy RN. Delay in disease development in transgenic plants that express the tobacco mosaic virus coat protein gene. Science 1986;232:738-743. 2. Sanford JC, Johnston SA. The concept of parasite-derived resistance-deriving resistance genes from the parasite’s own genome. J Theor Biol 1985;113:395-405. 3. Wilson TMA. Strategiesto protect crop plants against viruses: pathogen derived resistanceblossoms. Proc Natl Acad Sci USA 1993;90:3134-3141. 4. Beachy RN. Transgenic resistance to plant viruses. Sem Virol 1993. 5. Tepfer M. Viral genes and transgenic plants: what are the potential environmental risks? Bio/Technology 1993;11:1125-1132. 6. Palukaitis P. Virus-mediated genetic transfer in plants. In: Levin M, Strauss H (eds) Risk Assessment in Genetic Engineering. New York: McGraw-Hill, 1991;140-162. 7. Goldbach R. Genome similarities between plant and animal viruses. Microbiol Sci 1987;4:197-202. 8. Conrad U, Fiedler U. Expression of engineered antibodies in plant cells. Plant Molec Biol 1994;26:1023-1030. 9. Tavladoraki P, Benvenuto E, Trinca S, De Martinis D, Cattaneo A, Galeffi P. Transgenic plants expressing a functional single-chain Fv antibody are specifically protected from virus attack. Nature 1993;366:469-472. 10. Voss A, Niersbach M, Hain R, Hirsch HJ, Liao YC, Kreuzaler F, Fischer R. Reduced virus infectivity in N. tabacurn secreting a TMV-specific full-size antibody. Molec Breeding 1995;1:3950. 11. Hovanessian AG. Interferon-inducedand double-stranded RNA-activated enzymes: a specific protein kinase and 2’,5’-oligoadenylate synthetases. J Interfer Res 1991;11:199-205. 12. Sen GC, Lengyel P. The interferon system - a bird’s eye view on its biochemistry. J Biol Chem 1992;267:5017-5020.
13. Chebath J, Benech P, Revel M, Vigneron M. Constitutive expression of (2'-5') oligo A synthase confers resistance to picomavirus infection. Nature 1987;330:587-588. 14. Rysiecki G, Gewerts DR, Williams BRG. Constitutive expression of a 2',5'-oligoadenylate synthetase cDNA results in increased antiviral activity and growth suppression. J Interfer Res 1989;9:649-657. 15. Cayley PJ, White RF, Antoniw JF, Walesby NJ, Kerr IM. Distribution of the ppp(A2'p)"A-binding protein and interferon-related enzymes in animals, plants, and lower organisms. Biochem Biophys Res Commun 1982;108:1243-1250. 16. Truve E, Kelve M, AaspSllu A, Kuusksalu A, Sepphen P, Saarma M. Principles and background for the construction of transgenic plants displaying multiple virus resistance. Arch Virol 1994;9:41-50. 17. Truve E, AaspSllu A, Honkanen J, Puska R, Mehto M, Hassi A, Teen TH, Kelve M, Sepphen P Saarma M. Transgenic potato plants expressing mammalian 2'-5' oligoadenylate synthase are protected from potato virus X infection under field conditions. Bio/Technology 1993;ll: 1048-1052. 18. Hovanessian AG. Interferon-induced dsRNA-activated protein kinase (PKR): antiproliferative, antiviral and antitumoral functions. Sem Virol 1993;4:237-245. 19. Meurs EF, Watanabe Y, Kadereit S, Barber GN, Katze MG, Chong K, Williams BRG, Hovanessian AG. Constitutive expression of human double-stranded RNA activated p68 kinase in murine cells mediated phosphorylation of eukaryotic initiation factor 2 and partial resistance to encephalomyocarditis virus growth. J Virol 1992;66:5805-5814. 20. Katze MG. Games viruses play: a strategic initiative against the interferon-induced ds-RNA activated 6 8 , 0 0 0 Mr protein kinase. Sem Virol 1993;4:259-268. 21. Dever TE, Chen JJ, Barber GN, Cigan AM, Feng L, Donahue TF, London IM, Katze MG, Hinnebusch AG. Mammalian eukaryotic initiation factor 2 a kinases functionally substitute for GCN2 protein kinase in the GCN4 translational control mechanism of yeast. Proc Natl Acad Sci USA 1993;90:461&4620. 22. Chong KL, Feng L, Schappert K, Meurs E, Donahue TF, Friesen JD, Hovanessian AG, Williams BRG. Human p68 kinase exhibits growth suppression in yeast and homology to the translational regulator GCN2. EMBO J 1992;11:1553-1562. 23. Hiddinga HJ, Crum CJ, Hu J, Roth DA. Viroid-induced phosphorylation of a host protein related to a dsRNA-dependent protein kinase. Science 1988;241:451-453. 24. Langland JO, Jin S, Jacobs BL, Roth DA. Identification of a plant encoded analog of PKR, the mammalian double-stranded RNA-dependent protein kinase. Plant Physiol 1995;108:1259-1 267. 25. Wickner RB. Yeast RNA virology: the killer system. In: Broach JR,Pringle JR,Jones EW (eds) The molecular and cellular biology of the yeast Saccharomyces: genome dynamics, protein synthesis, and energetics. Plainview, NY: Cold Spring Harbor Laboratory Press, 1991;263-296. 26. Johnson AW, Kolodner RD. Synthetic lethality of sepl (xrnl) ski2 and sepl (xrnl) ski3 mutants of Saccharomyces cerevisiae is independent of killer viruses and suggests a general role for these genes in translation control. Molec Cell Biol 1995;15:2719-2727. 27. Widner WR, Wickner RB. Evidence that the SKI antiviral system of Saccharomyces cerevisiae acts by blocking expression of viral mRNA. Molec Cell Biol 1993;13:4331-4341. 28. Masison DC, Blanc A, Ribas JC, Carroll K, Sonenberg N, Wickner RW. Decoying the cap- mRNA degradation system by a double-stranded RNA virus and poly (A)- mRNA surveillance by a yeast antiviral system. Molec Cell Biol 1995;15:2763-2771. 29. Lee SG, Lee I, Park SH, Kang C, Song K. Identification and characterization of a human cDNA homologous to yeast SKI2. Genomics 1995;25:660-666. 30. Zaccomer B, Haenni AL, Macaya G. The remarkable variety of plant RNA virus genomes. Virology 1995;76:231-247. 31. Roberts NA, Martin JA, Kinchington D, Broadhurst AV, Craig JC, Duncan IB, Galpin SA, Handa BK, Kay J, Krohn A, Lambert RW, Memet JH,Mills JS, Parkes KEB, Redshaw S, Ritchie AJ, Taylor DL, Thomas GJ, Machin PJ. Rational design of peptide-based HIV proteinase inhibitors. Science 1990;248:358-361.
203 32. 33. 34. 35. 36. 37. 38. 39.
40.
41. 42. 43. 44.
45.
46. 47. 48. 49. 50.
51. 52. 53. 54. 55.
56. 57.
Ryan CA. Protease inhibitors in plants: genes for improving defenses against insects and pathogens. Ann Rev Phytopathol 1990;28:425-449. Masoud S, Johnson LB, White FF, Reeck GR. Expression of a cysteine proteinase inhibitor (oryzacystatin-I) in transgenic tobacco plants. Plant Molec Biol 1993;21:655-663. Garcia JA, Cervera MT, Riechmann JL, Lopez-0th C. Inhibitory effects of human cystatin C on plum pox potyvirus proteases. Plant Molec Biol 1993;22:697-701. Fraser RSS. The genetic of resistance to plant viruses. AM Rev Phytopathol 1990;28:179-200. Dawson WO, Hilf ME. Host range determinants of plant viruses. Ann Rev Plant Physiol Plant Molec Biol 1992;43:527-555. Chen ZC, White RF, Antoniw JF, Lin Q. Effect of pokeweed antiviral protein (PAP) on the infection of plant viruses. Plant Pathol 1991;40:612-620. Stirpe F, Barbieri L, Giulia Battelli M, Soria M, Lappi DA. Ribosome-inactivating proteins from plants: present status and future prospects. Bioflechnology 1992;10:405-412. Endo Y, Tsurugi K, Yutsudo T, Ogasawara T, Igarashi K. Site of action of a Vero toxin (VT2) from Escherichiu coli 0157:H7 and of Shiga toxin on eukaryotic ribosomes. RNA glycosidase activity of the toxin. Eur J Biochem 1988;171:45-50. Taylor S, Massiah A, Lomonosoff G, Roberts LM, Lord JM, Hartley M. Correlation between the activities of five ribosome-inactivating proteins in depurination of tobacco ribosomes and inhibition of tobacco mosaic virus infection. Plant J 1994;5:827-835. Lamb FI, Roberts LM, Lord JM. Nucleotide sequence of cloned cDNA coding for preproricin. Eur J Biochem 1985;148:265-270. Chow TP, Feldmann RA, Lovett M, Piatak M. Isolation and DNA sequence of a gene encoding atrichosanthin, a type 1 ribosome-inactivating protein. J Biol Chem 1990;265:8670-8674. Lodge JK, Kaniewski WK, Tumer NE. Broad-spectrum virus resistance in transgenic plants expressing pokeweed antiviral protein. Proc Natl Acad Sci USA 1993;90:7089-7093. Lee-Huang S, Huang PL, Kung HF, Li BQ, Huang P, Huang HI, Chen HC. TAP-29: an anti-human immunodeficiency virus protein from Trichosanrhes kirilowi that is nontoxic to intact cells. Proc Natl Acad Sci USA 1991;88:6570-6574. Logemann J, Jach G, Tommerup H, Mundy J, Schell J. Expression of a barley ribosome-inactivating protein leads to increased fungal protection in transgenic tobacco plants. Bionechnology 1992;101305-308. Flor HH. The complementary genetic system in flax and flax rust. Adv Genet 1956;8:2!&54. Keen N. Gene-for-gene complementarity in plant-pathogen interactions. AM Rev Genet 199&24:447-463. Keen N. The molecular biology of disease resistance. Plant Molec Biol 1992;19:109-122. Saito T, Meshi T, Takamatsu N, Okada Y. Coat protein gene sequence of tobacco mosaic virus encodes a host response determinant. Proc Natl Acad Sci USA 1987;84:6074-6077. Knorr DA, Dawson WO. A point mutation in the tobacco mosaic virus capsid protein gene induces hypersensitivity in Nicotiana sylvesrris. Proc Natl Acad Sci USA 1988;85:170-174. Padgett HS, Beachy RN. Analysis of a tobacco mosaic virus strain capable of overcoming the N gene-mediated resistance. Plant Cell 19935577-586. Witham S, Dinesh-Kumar SP, Doil C, Hehl R, Corr C, Baker B. The product of the tobacco mosaic virus resistance gene N: similarity to Toll and the Interleukin-I receptor. Cell 1995;78:1101-1115. Kobe B, Deisenhofer J. The leucine rich repeat: a versatile binding motif. Trends Biochem Sci 1994;19:415-42 1. Michelmore R. Isolation of disease resistance genes from crop plants. Curr Opin Biotechnol 1995;6:145-152. Herbers K, Conrad-Strauch J, Bonas U. Race-specificity of plant resistance to bacterial spot disease determined by repetitive motifs in a bacterial avirulence protein. Nature 1992;356:172-174. Valkonen JPT. Natural genes and mechanisms for resistance to viruses in cultivated and wild potato species (Solanurn spp.) Plant Breed 1993;112:1-16. Kavanagh T, Goulden M, Santa Cruz S, Chapman S, Barker I, Baulcombe DC. Molecular analysis
204 58. 59.
60. 61. 62. 63.
64.
65. 66. 67. 68.
69.
70. 71.
of a resistance-breaking strain of potato virus X. Virology 1992;189:60+617. Kohm BA, Goulden MG, Gilbert JE,Kavanagh TA, Baulcombe DC. A potato virus X resistance gene mediates an induced nonspecific resistance in protoplasts. Plant Cell 19935913-920. Meshi T, Motoyoshi F, Maeda T, Yohiwoka S, Watanabe H, Okada Y. Mutations in the tobacco mosaic virus 30-kD protein gene overcome Tm-2 resistance gene in tomato. Plant Cell 1989;1:5 15522. Calder VL, Palukaitis P. Nucleotide sequence analysis of the movement genes of resistance breaking strains of tomato mosaic virus. J Gen Virol 1992;73:165--168. Ryals J, Uknes S, Ward E. Systemic acquired resistance. Plant Physiol 1994;104:1109--1112. Dempsey DA, Klessig DF. Signals in plant disease resistance. Bull Inst Pasteur 1995;933:167-186. Broglie K, Chet I, Holliday M, Cressman R, Biddle P, Knowlton S, Mauvais CJ, Broglie R. Transgenic plants with increased resistance to the fungal pathogen Rhizoctonia solani. Science 1991;254:1194-1197. Alexander D, Goodman Rh4, Gut-Rella M, Glascock C, Weymann K, Friedrich L, Maddox D, AhlGoy P, Luntz T, Ward E, Ryals J. Increased tolerance to two oomycete pathogens in transgenic tobacco expressing pathogenesis-related protein la. Proc Natl Acad Sci USA 1993;90:7327-733 1. Cao Hui, Bowling SA, Gordon AS, Dong X. Characterization of an Arabidopsis mutant that is nonresponsive to inducers of systemic acquired resistance. Plant Cell 1994;6: 1583-1592. Bowling SA, Guo A, Cao Hui, Gordon AS, Klessig DF, Dong X. A mutation of Arabidopsis that leads to constitutive expression of systemic acquired resistance. Plant Cell 1994;6:1 8 4 51857. Delaney TP, Friedrich L, Ryals JA. Arabidopsis signal transduction mutant defective in chemically and biologically induced disease resistance. Proc Natl Acad Sci USA 1995;92:6602-6606. Sano H, Seo S, Orudgev E, Youssefian S, Ishizuka K, Ohashi Y. Expression of the gene for a small GTP binding protein in transgenic tobacco elevates endogenouscytokinin levels, abnormally induces salicylic acid in response to wounding, and increases resistance to tobacco mosaic virus infection. Proc Natl Acad Sci USA 1994;91:10556-10560. Masuta N, Tanaka H, Uehara K, Kuwata S, Koiwai A, Noma M. Broad resistance to plant viruses in transgenic plants conferred by antisense inhibition of a host gene essential in S-adenosylmethionine dependent transmethylation reactions. Proc Natl Acad Sci USA 1995;92:6117-6121. Morton R, Hooykaas PJJ. Gene replacement. Molec Breed 1995;1: 123-132. Watanabe Y, Ogawa T, Takahashi H, Ishida I, Takeuchi Y, Yamamoto M, Okada Y. Resistance against multiple plant viruses in plants mediated by a double-stranded RNA-specific ribonuclease. FEBS Lett 1995;372:165-168.
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R. El-Gewely, editor.
205
Transgenic fish and its application in basic and applied research Thomas T. Chen', Nicholas H. Vrolijk', Jenn-Kan Lu'.', Chun-Mean Lid, Renate Reimsch~essel~ and Rex A. Dunham4 'Biotechnology Center, University of Connecticut, Storrs, Connecticut; 2Departmentof Biological Sciences, University of Maryland at Baltimore County, Baltimore, Maryland; 'Aquatic Pathobiology Group, Department of Pathology, University of Maryland at Baltimore, Baltimore, Maryland; and 4Department of Fisheries and Allied Aquacultures, Auburn University, Auburn, Alabama, USA
Abstract. Since 1985, transgenic fish have been successfully produced by microinjecting or electroporating desired foreign DNA into unfertilized or newly fertilized eggs using many different fish species. More recently, transgenic fish have also been produced by infecting newly fertilized eggs with pantropic, defective retroviral vectors carrying desired foreign DNA. These transgenic fish can serve as excellent experimental models for basic scientific investigations as well as in biotechnological applications. In this paper, we will review the current status of the transgenic fish research and its potential application in basic and applied research.
Key words: CYPl Al, CYPlA2, electroporation, growth hormone, insulin-like growth factor, microinjection, P450, pantropic viral vector, transgene integration, transgene expression, transgene inheritance, transgenes, transgenic fish.
Introduction Animals or plants into which heterologous DNA (transgene) has been artificially introduced and integrated in their genomes are called transgenics. Since the early 1980s, transgenic plants [l], nematodes [2], fruit flies [3], sea urchins [4,5], frogs [6], laboratory mice [7,8] and farm animals such as cows, pigs and sheep [9] have been successfully produced. In plants, transgenes are introduced into cells by infection with Agrobucterium tumefuciens or by physical means. In animals, transgenes are introduced into the pronuclei of fertilized eggs by injection and the injected embryos are incubated in vitro or implanted into the uterus of a pseudopregnant female for subsequent development. In these studies, multiple copies of transgenes are integrated at random locations in the genome of the transgenic individuals. If the transgenes are linked with functional promoters, expression of transgenes as well as display of change in phenotype is expected in some of the transgenic individuals. Furthermore, the transgenes in many transgenic individuals are also transmitted through the germline to subsequent generations. These transgenic animals play important roles in
Address for correspondence: Dr Thomas T. Chen, Biotechnology Center, University of Connecticut, 184 Auditorium Road, U-149, Storrs, CT 06269-3149, USA. Tel.: +I-860-486-5011/5012. Fax: +I-860-4865005.
206 basic research as well as applied biotechnology. In basic research, transgenic animals provide excellent models for studying molecular genetics of early vertebrate development, actions of oncogenes, and the biological functions of hormones at different stages of development. In applied biotechnology, transgenic animals offer unique opportunities for producing animal models for biomedical research, improving the genetic background of broodstock for animal husbandry or aquaculture, and designing bioreactors for producing valuable proteins for pharmaceutical or industrial purposes. Since 1985, a wide range of transgenic fish species have been produced [1@12] by microinjecting or electroporating homologous or heterologous transgenes into newly fertilized or unfertilized eggs. Several important steps are routinely taken to produce a desired transgenic fish. First, an appropriate fish species must be chosen, depending on the nature of the studies and the availability of the fish holding facility. Second, a specific gene construct must be prepared. The gene construct contains the structural gene encoding a gene product of interest and the regulatory elements that regulate the expression of the gene in a temporal, spatial and developmental manner. Third, the gene construct has to be introduced into the developing embryos in order for the transgene to be integrated stably into the genome of every cell. Fourth, since not all instances of gene transfer are efficient, a screening method must be adopted for identifying transgenic individuals. Although remarkable progress has been made in producing transgenic fish by gene transfer technology, a critical review of the published results has shown that a majority of the research effort has been devoted to confirming the phenomenon of foreign gene transfer into various fish species. Very few attempts have been made to explore the application of transgenic fish technology in basic as well as applied research. Recently, we have devoted a substantial amount of our research effort to this problem with promising results. In this paper, we will discuss the potential application of this technology using results generated in our laboratories as examples.
Production of transgenic fish Selection of fish species Gene transfer studies have been conducted on several different fish species including: channel catfish, common carp, goldfish, Japanese medaka, loach, northern pike, rainbow trout, salmon, tilapia, walleye, and zebrafish (for review; [ 10,121).Depending on the purpose of the transgenic fish studies, the embryos of some fish species are more suited for gene transfer studies than the others. For example, Japanese medaka (Oryzius lutipes) and zebrafish (Burchydunio rerio) have short life cycles (3 months from hatching to mature adults), produce hundreds of eggs on a regular basis without exhibiting a seasonal breeding cycle, and can be maintained easily in the laboratory for 2 to 3 years. Eggs from these two fish species are relatively large (diameter: 0.7-1.5 mm) and possess very thin, semitransparent chorions, features that permit
207 easy microinjection of DNA into the eggs if appropriate glass needles are used. Furthermore, inbred lines and various morphological mutants of both fish species are available. These fish species are thus suitable candidates for conducting gene transfer experiments for: 1) studying developmental regulation of gene expression and gene action; 2) identifying regulatory elements that regulate the expression of a gene; 3) measuring the activities of promoters; and 4) producing transgenic models for environmental toxicology. However, a major drawback of these two fish species is their small body size which makes them unsuitable for some endocrinological or biochemical analyses. Channel catfish, common carp, rainbow trout and salmon are commonly used large body size model fish species in transgenic fish studies. Since the endocrinology, reproductive biology, and basic physiology of these fish species have been well worked out, they are well suited for conducting studies on comparative endocrinology and aquaculture applications. However, the long maturation time of these fish species and a single spawning cycle per year will limit research progress in this field. Loach, killifish, goldfish and tilapia are the third group of model fish species suitable for conducting gene transfer studies since their body sizes are large enough for most biochemical and endocrinological studies. Furthermore, shorter maturation times, as compared to catfish, rainbow trout or salmon, allow easier manipulation of transgenic progeny. Unfortunately, the lack of a well-defined genetic background and asynchronous reproductive behavior of these fish species render them less amenable to gene transfer studies. Transgene constructs
A transgene used in producing transgenic fish for basic research or biotechnological applications should be a recombinant gene construct that produces a gene product at an appropriate level in the desired tissue(s) at the desired time(s). Therefore, the prototype of a transgene is usually constructed in a plasmid to contain an appropriate promoter/enhancer element and the structural gene. Depending on the purpose of the gene transfer studies, transgenes can be grouped into three main types: 1) gain-of-function, 2) reporter function, and 3) loss-offunction. The gain-of-function transgenes are designed to add new functions to the transgenic individuals or to facilitate the identification of the transgenic individuals if the genes are expressed properly in the transgenic individuals. Transgenes 'containing the structural genes of mammalian and fish growth hormones (GH, or their cDNAs) fused to functional promoters such as chicken and fish p-actin gene promoters, are examples of the gain-of-function transgene constructs. Expression of the GH transgenes in transgenic individuals will result in increased production of growth hormone and ultimate growth enhancement [ 13-16]. Bacterial chloramphenicol acetyl transferase (CAT), P-galactosidase or luciferase genes fused to functional promoters are examples of transgenes with reporter function. These reporter function transgenes are commonly used to identify the success of gene transfer effort. A more important function of a reporter gene is used to identify and
measure the strength of a promoter/enhancer element. In this case, the structural gene of the CAT, P-galactosidase or luciferase gene is fused to a promoter/enhancer element in question. Following gene transfer, the expression of the reporter gene activity is used to determine the transcriptional regulatory sequence of a gene or the strength of a promoter [17]. The “loss-of-function’’ transgenes are constructed for interfering with the expression of host genes. These genes might encode an antisense RNA to interfere with the posttranscriptional process or translation of endogenous mRNAs. Alternatively, these genes might encode a catalytic RNA (a ribozyme) that can cleave specific mRNAs and thereby cancel the production of the normal gene product [18]. Although these genes have not yet been introduced into a fish model, they could potentially be employed to produce disease resistant transgenic broodstocks for aquaculture or transgenic model fish defective in a particular gene product for basic research. Methods of gene transfer
Techniques such as calcium phosphate precipitation, direct microinjection, lipofection, retrovirus infection, electroporation, and particle gun bombardment have been widely used to introduce foreign DNA into animal cells, plant cells, and germlines of mammals and other vertebrates. Among these methods, direct microinjection and electroporation of DNA into newly fertilized eggs have been proven to be the most reliable methods of gene transfer in fish systems. Microinjection of eggs or embryos Microinjection of foreign DNA into newly fertilized eggs was first developed for the production of transgenic mice in the early 1980s. Since 1985, the technique of microinjection has also been adopted for introducing transgenes into Atlantic salmon, common carp, catfish, goldfish, loach, medaka, rainbow trout, tilapia, and zebrafish [ 10,111. The gene constructs that were used in these studies include human or rat growth hormone (GH) gene, rainbow trout or salmon GH cDNA, chicken 6crystalline protein gene, winter flounder antifreeze protein gene, E. coli P-galactosidase gene, and E. coli hygromycine resistance gene [10,11]. In general, gene transfer in fish by direct microinjection is conducted as follows. Eggs and sperm are collected in separate, dry containers. Fertilization is initiated by adding water and sperm to the eggs, with gentle stirring to enhance fertilization. Eggs are microinjected within the first few hours .after fertilization. The injection apparatus consists of a dissecting stereo microscope and two micromanipulators, one with a micro-glassneedle for delivering transgenes and the other with a micropipette for holding fish embryos in place. Routinely, about lo6-lo8 molecules of a linearized transgene in about 20 nl is injected into the egg cytoplasm. Following injection, the embryos are incubated in water until hatching. Since natural spawning in zebrafish or medaka can be induced by adjusting photoperiod and water temperature, precisely staged newly fertilized eggs can be collected from the aquaria for gene transfer. If the medaka eggs are maintained at 4°C immediately after fertilization, the micropyle on the fertilized
209 eggs will remain visible for at least 2 h. The DNA solution can be easily delivered into the embryos by injection through this opening. Depending on the fish species, the survival rate of injected fish embryos ranges from 35 to 80% while the rate of DNA integration ranges from 10 to 70% in the survivors (Table 1; [ 10,111). The tough chorions of the fertilized eggs in some fish species, e.g., rainbow trout and Atlantic salmon, can frequently make insertion of glass needles difficult. This difficulty can be overcome by any one of the following methods: 1) inserting the injection needles through the micropyle, 2) making an opening on the egg chorions by microsurgery, 3) removing the chorion by mechanical or enzymatic means, 4) reducing chorion hardening by initiating fertilization in a solution containing 1 mM glutathione, or 5) injecting the unfertilized eggs directly. Electroporation Electroporation is a successful method for transferring foreign DNA into bacteria, yeast, and plant and animal cells in culture. This method has become popular for transferring transgenes into fish embryos in the past 3 years [ 15,991. Electroporation utilizes a series of short electrical pulses to permeate cell membranes, thereby permitting the entry of DNA molecules into embryos. The patterns of electrical pulses can be emitted in a single pulse of exponential decay form (i.e., exponential decay generator) or high frequencies multiple peaks of square waves (i.e., square wave generator). Studies conducted in our laboratory [ 15,991 and those of others [ 191 have shown that the rate of DNA integration in electroporated embryos is on the order of 20% or higher in the survivors (Table 1). Although the overall rate of DNA integration in transgenic fish produced by electroporation was equal to or slightly higher than that of microinjection, the actual amount of time required for handling a large number of embryos by electroporation is orders of magnitude less than the time required for microinjection. Recently, several research groups have also reported successful transfer of transgenes into fish by electroporating sperm instead of embryos [20,2 11. Electroporation is therefore considered as an efficient and versatile massive gene transfer technology. Table 1. Transfer of foreign DNA into medaka embryos by different gene transfer methods. Microinjection"(%)
Electroporation (%) Ib
Viability (at hatching) Integration rate' Transgene expression Efficiency (eggs/min)
50 20
70 15
yes 1-2
200 yes
11'
90
Pantropic retroviral vector (%) Electroporationd
Incubation"
70
Yes
50 50 Yes
Yes
200
200
200
25
70
'Injecting is carried out via micropyle prior to blastodisc formation. bExponential-decay impulse mode. 'Square wave impulse mode. dElectroporation with square wave mode at 3.5 kV. 'Fertilized eggs are exposed to a mixture of medaka hatching enzyme and pancreatin for 2 h. The dechorinated embryos are incubated with the pantropic pseudotyped retrovirus overnight at room temperature. 'Integration rate is calculated from the surviving embryos after gene transfer.
210 Transfer of transgenes by infection with pantropic retroviral vectors Although transgenes can be reproducibly introduced into various fish species by microinjection or electroporation, the resulting P, transgenic individuals possess mosaics germlines as a result of delayed transgene integration. Furthermore, these two gene transfer methods are not effective or successful in producing transgenics in marine fish and invertebrates. Recently a new gene transfer vector, a defective pantropic retroviral vector, has been developed [22]. This vector contains the long terminal repeat (LTR) sequence of Moloney murine leukemia virus (MoMLV) and transgenes packaged in a viral envelop with the G-protein of vesicular stomatitis virus (VSV). Since the entry of VSV into cells is mediated by interaction of the VSV-G protein with a phospholipid component of the cell, this pseudotyped retroviral vector has a very broad host range and is able to transfer transgenes into many different cell types. Using the pantropic pseudotyped defective retrovirus as a gene transfer vector, transgenes containing neoR or P-galactosidase have been introduced into zebrafish [23] and medaka [24]. Recently, the feasibility of using a pantropic pseudotyped retroviral vector for introducing genes into marine invertebrates has been tested in dwarf surf clams and the results have shown that transgenes can be readily transferred into clams at high efficiency (Lu et al., in review).
Characterization of transgenic fish Identification of transgenic fish The most time consuming step in producing transgenic fish is the identification of transgenic individuals. Traditionally, dot blot and Southern blot hybridization of genomic DNA were common methods used to determine the presence of transgenes in the presumptive transgenic individuals. These methods involve isolation of genomic DNA from tissues of presumptive transgenic individuals, digestion of DNA with restriction enzymes and Southern blot hybridization of the digested DNA products. Although this method is expensive, laborious and insensitive, it offers a definitive answer whether a transgene has been integrated into the host genome. Furthermore, it also reveals the pattern of transgene integration if appropriate restriction enzymes are employed in the Southern blot hybridization analysis. In order to handle a large number of animals efficiently and economically, a polymerase chain reaction (PCR) based assay has been adopted [15,16]. The strategy of the assay is outlined in Fig. 1. It involves isolation of genomic DNA from a very small piece of fin tissue, PCR amplification of the transgene sequence, and Southern blot analysis of the amplified products. Although this method does not differentiate whether the transgene is integrated in the host genome or remains as an extrachromosomal unit, it serves as a rapid and sensitive screening method for identifying individuals that contain the transgene at the time of analysis. In our laboratory, we use this method as a preliminary screen for transgenic individuals when screening thousands of the presumptive transgenic fish.
21 1
A
1
ANNEAL PRIMERS
RSV-LTR[ rtWlcDNA P2
1
PRIMER
EXTENSION
RS -rtGHlcDNA PCR PROWCTS
Fig. 1. Strategy for identifying the presence of transgenes in the presumptive transgenic fish by PCR and Southern blot hybridization. DNA samples were isolated from pectoral fin tissues of presumptive transgenic fish and subjected to PCR amplification. The amplified products were analyzed by electrophoresis on agarose gels and Southem blot hybridization. A: Strategy of PCR amplification; B: Southern blot analysis of PCR amplified products. Lanes 2-6 and 8-12, DNA samples from presumptive transgenic fish; lanes 1 and 7, transgene construct (RSVLTR-rtGHI cDNA ([16], with permission).
Expression of transgenes An important aspect of gene transfer studies is the detection of transgene expression. Depending on the levels of transgene products in the transgenic individuals, the methods commonly used for detecting transgene expression are: 1) RNA northern or dot blot hybridization; 2) RNase protection assay; 3) reverse transcription/polymerase chain reaction (RTPCR); 4)immunoblotting assay; and 5) other biochemical assays
212
A
4 RSV-LTR I
rtGH-cDNA
1
'
Transcription
Pa
AAAAA
1
I
GenomicDNA
mRNA
oligo-dT. dIiTF's. and ReverseTrposcriptnsc
ss-cDNA
rtGH-cDNA
Fig. 2. Strategy of detecting rtGH transgene expression by reverse transcription (RT)/PCR assay. A: Strategy of RTPCR. B: Detection of rtGH transgene expression in transgenic carp by RTPCR. Total RNA was isolated from liver, muscle, eyes, gut and testes of F, transgenic carp and controls following the acid guanidinium thiocyanate-phenol-chloroformmethod. Single-stranded cDNA was prepared by reverse transcription from each total RNA and used as a template for PCR amplification of rtGH using synthetic oligonucleotidesas amplification primers. The resulting products were analyzed by Southern blot analysis using radio-labelled rtGH cDNA as a hybridization probe (1161, with permission).
for determining the presence of the transgene protein products. Among these assays, RT/PCR is the most sensitive method and only requires a small amount of sample. The strategy of this assay is summarized in Fig. 2 [16]. Briefly, it involves the isolation of total RNA from a small piece of tissue, synthesis of single-stranded
213 cDNA by reverse transcription and PCR amplification of the transgene cDNA by employing a pair of oligonucleotide primers specific to the transgene product. The resulting products are resolved on agarose gels and analyzed by Southern blot hybridization using a radiolabeled transgene as a hybridization probe. Transgene expression can also be quantified by a quantitative RTPCR method [25]. Pattern of transgene integration Studies conducted in many fish species have shown that following injection of linear or circular transgene constructs into fish embryos, the transgenes are maintained as extrachromosomal units through many rounds of DNA replication in the early phase of the embryonic development. At later stages of embryonic development, some of the transgenes are randomly integrated into the host genome while others are degraded, resulting in the production of mosaic transgenic fish (for review, [ 121). In many fish species studied to date, multiple copies of transgenes were found to integrate in a head-to-head, head-to-tail or tail-to-tail form, except in transgenic common carp and channel catfish where single copies of transgenes were integrated at multiple sites on the host chromosomes [13]. Inheritance of transgenes Stable integration of the transgenes is an absolute requirement for continuous vertical transmission to subsequent generations and establishment of a transgenic fish line. To determine whether the transgene is transmitted to the subsequent generation, PI transgenic individuals are mated to nontransgenic individuals and the progeny are assayed for the presence of transgenes by the PCR assay method described earlier [15,16]. Although it has been shown that the transgene may persist into the F, generation of transgenic zebrafish as extrachromosomal DNA [26], detailed analysis of the rate of transmission of the transgenes to the F, and F, generations in many transgenic fish species indicates true and stable incorporation of the constructs into the host genome (for review, [10,12]). If the entire germline of the PI transgenic fish is transformed with at least one copy of the transgene per haploid genome, at least 50% of the F, transgenic progeny will be expected in a backcross involving a P, transgenic with a nontransgenic control. In many of such crosses, only about 20% of the progeny are transgenic [ 13,15,16,26-281. When the F, transgenic is backcrossed with a nontransgenic control, however, at least 50% of the F, progeny are transgenics. These results clearly suggest that the germlines of the PI transgenic fish are mosaic as a result of delayed transgene integration during embryonic development.
Application of transgenic fish in basic research Transgenic fish, like transgenic mice, can serve as excellent experimental models for a wide variety of basic scientific investigations. These studies include: 1) identifying
214 the regulatory elements of a gene; 2) examining the molecular genetics of early vertebrate development; 3) studying the functions of a gene product; 4) identifying the biological actions of hormones; 5) developing models for biomedical research; and 6) establishing models for environmental toxicant analysis. In higher vertebrates, growth is primarily modulated by the availability of growth hormone (GH) and insulin-like growth factors (IGFs) to their respective receptors. The secretion of GH from the pituitary gland and the binding of GH to its receptor, signals the production of IGF I mRNA and the corresponding polypeptide by the liver (endocrine production) and other tissues (autocrine/paracrine function). Although the influence of GH on IGF induction and the molecular mechanism that underlies the GH controlled IGF I gene expression have been under intensive investigation in higher vertebrates for many years, very little is known in lower vertebrates such as fish. Using rainbow trout as experimental animals, we are interested in studying the mechanism by which GH and IGF control growth in lower vertebrates.
Age-dependent, tissue-spec@ and growth-hormone-dependentexpression offish IGF genes As a step toward understanding the regulation of growth in fish by GH and IGFs, we initiated work to identify the presence of IGF'I and IGF 11 in rainbow trout by PCR and screening of a rainbow trout liver cDNA library. Two unique cDNA sequences have been identified. On the basis'of a 98.7% nucleotide sequence homology to coho salmon IGF I, one cDNA sequence was identified as rainbow trout IGF I. The second cDNA sequence shared 43.3% identity with trout IGF I at the predicted amino acid level and 53.6% identity with human IGF 11, and was identified as trout IGF I1 [29]. This was the first time that an IGF I1 was identified in a fish species. As a result of differential spiicing in the 5' untranslated region, signal peptide, E-domain, and 3' untranslated region, as -well as transcription initiated from more than one promoter, multiple size forms of IGF I and I1 mRNA have been detected in mammals [30,31]. To detect the presence of multiple size forms of IGF I and I1 mRNA in rainbow trout, an RTPCR method was adopted [32]. This assay employed two sets of primers each for IGF I and I1 so that small size differences of PCR products could be resolved on high concentration (e.g., 3%) agarose gels and the identity of each product could be confirmed by nucleotide sequence determination. The primer sets were designed to separately amplify the 5' region (predicted start codon to C-domain) or 3' region (C-domain to approximately 1 0 0 bp beyond the predicted stop codon) of both IGF I and 11. While only one size form of IGF I and II mRNA resulted from RTPCR with the 5' IGF I and both 5' and 3' IGF I1 primer sets, four size forms of IGF-I mRNA resulted from the 3' IGF I primer set. Results of nucleotide sequence determination of the four size forms of IGF I mRNA showed that the size differences were due to insertions or deletions in the E-domain. These four forms of IGF-I mRNA, in increasing nucleotide length, are designated as IGF IEa-1, -2, -3, -4. The predicted amino acid residues of the E-domains are 35,47, 62 and 74, respectively [32]. The entire nucleotide sequence for IGF IEa-2 and Ea-3
215 mRNA have been determined from their respective intact cDNA clones. Duguay et al. [33] recently detected three forms of IGF I mRNA for coho salmon by using an RTPCR assay and these three mRNA forms are equivalent to rainbow trout IGF IEa1, Ea-3 and Ea-4. By using the same approach, Wallis and Devlin [34] also detected three size forms of IGF I mRNA for chinook salmon. These three size forms correspond to rainbow trout IGF IEa-1, Ea-2, and Ea-4. The reasons for the absence of rainbow trout IGF IEa-2 and IEa-3 in the livers of coho salmon and chinook salmon, respectively, are unknown. It is conceivable that the missing forms were not resolved and therefore not recognized after agarose gel electrophoresis. Alternatively, the IGF I mRNA form absent in these two reports may not have been present or detectable in these fish, in which case it is surprising that the two salmonid species lack different analogues of rainbow trout IGF I. An RNase protection assay (RPA) was established to determine the mRNA levels of each of the four IGF I forms and IGF I1 in the liver, skeletal muscle, spleen, pyloric caeca (pancreatic tissue), heart, brain and gill of rapidly growing juvenile (7-8 months old) rainbow trout and sexually mature adults [32]. In this assay, probe templates were constructed by cloning the 3' region (from the C domain to approximately 100 bp into the 3' untranslated region) of each IGF I or IGF I1 into a Bluescript plasmid vector in order to generate a radiolabeled antisense cRNA probe and unlabelled sense cRNA as concentration standards by in vitro transcription with T 7 or T3 RNA polymerase. The protected fragments for the four IGF I mRNA forms and the IGF II mRNA were readily identifiable by resolving on a denatured polyacrylamide gel. RPA showed that at least one form of IGF I and IGF 11mRNA are expressed in all of the tissues examined in both developmental stages (Fig. 3). Liver is the site of greatest IGF mRNA abundance (p c 0.01), and the levels of total IGF I and I1 mRNA are one to two orders of magnitude higher than in other tissues examined. Furthermore, it is interesting to note that the levels of total IGF I and 11 mRNA are 2-fold higher in the adult liver than the juvenile liver (p < 0.01). In mammals, IGF I mRNA has been detected primarily in the postnatal liver, kidney, spleen, pancreas, lung, and testes of the mouse [35], the brain and several other regions of the central nervous system of the rat [31], and the placenta and whole premenopausal ovary of humans [36]. In chicken, IGF I mRNA has been detected in the eye, skeletal muscle and brain prior to hatching and the liver only after hatching [37]. IGF I1 mRNA has been detected in muscle, skin, lung, intestine, thymus, heart, kidney, brain and spinal cord of fetal/neonatal rats and in the brain and spinal cord of adult rats. However, it is interesting to note that except in the liver, levels of rtIGF I1 mRNA are much higher than those of the total rtIGF I mRNA in gill, kidney, heart, spleen, brain, muscle, pylorus, testes and ovaries. These results suggest that, in addition to IGF I, IGF I1 may play an important role in fish growth as well as in maintenance of osmotic balance. An in vivo study was conducted to determine the dependency of IGF mRNA accumulation upon GH treatment. In this study, yearling rainbow trout of about 150 g each maintained at 15°C were fasted for 5 days and each fish was injected with 10 pg/g fish body weight of bovine GH (bGH) or carrier solution as a control. Levels
216
Fig. 3. Levels of five forms of IGF mRNA in the tissues of juvenile and adult trout. The forms of IGF I mRNA are abbreviated as Ea-1, Ea-2, Ea-3 and Ea-4. Total IGF I, IGF 11, and total IGF levels are abbreviated as I, I1 and IGF, respectively ([80], with permission).
of IGF I and IGF I1 mRNA in different tissues were determined by RPA at different periods posthormone treatment. Levels of liver IGF I mRNA significantly increased 6 h postbovine GH treatment and remained significantly elevated at 12 h, while liver
217 IGF I1 mRNA levels were significantly elevated at 3 and 6 h posthormone treatment [38]. Both IGF I and IGF I1 mRNA levels responded with a 3- to 4-fold increase over mock injected controls. Although the levels of IGF I mRNA did not increase significantly in the pyloric caeca in response to bovine GH treatment, the levels of IGF I1 mRNA elevated at 12, 24, and 48 h by about 4-, 2- and 4-fold, respectively. To determine whether the response of IGF mRNA induction by GH is dose dependent, Shamblott et al. [38] conducted further in vitro studies in a rainbow trout primary hepatocyte culture maintained in a serum free medium supplemented with bGH. The results showed that both IGF I and IGF 11 mRNA levels responded to bovine GH treatment in a dose-dependent fashion with ED,, values of about 45 and 6 ng/ml, respectively. These results clearly showed that the synthesis of IGF I and IGF I1 mFWA in the liver of rainbow trout is under the modulation of GH. IGF I transgenic fish Although continuing study with the use of molecular biological approaches will shed light on the biological actions of IGFs, production of transgenic fish with elevated levels of these polypeptides by transgenic fish technology may generate alternative models for determining the biological effect of IGFs on fish growth and development. Toward this end, we have recently produced transgenic medaka by electroporating rtIGF I cDNA fused to a functional carp p-actin gene promoter (kindly donated by P. Hackett). Results of these studies have shown that more than 20% of the surviving embryos integrated the rtIGF I transgene in their respective genomes. Although the number of the PI transgenic individuals is still small, they are significantly larger than their nontransgenic controls (p c 0.01). Furthermore, it is interesting to note that both PI and F, IGF I transgenic fish hatched, on average, 2 days earlier than their nontransgenic controls. Unlike those studies conducted in mouse and rat, our results in medaka suggest that, in addition to regulating postembryonic somatic growth, IGF I may play an important role in embryonic development. As shown in Fig. 4A,B, several P, and F, transgenic female medaka exhibit unusually enlarged abdomens and these animals failed to spawn even after 6 months of age. Necropsy of these animals and histological examination have shown that the ovaries are filled with a gelatinous fluid and no mature eggs were observed. One IGF I PI transgenic female further developed an adenocarcinoma on the maxilla (Fig. 4C). This female was initially tumorless and spawned several batches of eggs. When the tumor appeared, it ceased spawning. Histological studies have shown that the lesion was comprised of epithelial cells forming small cysts. These cysts contained a pale eosinophilic material and some necrotic cells. The lesion was quite aggressive and replaced facial cartilage and bone, although it had not invaded the eye. Females with enlarged abdomens or bearing solid tumors are also observed among F, decedents from the above female (Fig. 4D).Although the above described observations are still preliminary and more detailed studies are required, it is clear that IGF transgenic fish can serve as models for studying the involvement of IGFs in: 1) normal growth and development; 2) reproduction; and 3) tumor development.
218
Application of transgenic fish in environmental toxicology The widespread use of polyhalogenated and polycyclic aromatic hydrocarbons (PCBs and PAHs) for various industrial purposes over the last 50 years, has resulted in the serious accumulation of toxic and carcinogenic compounds and their metabolites in aquatic ecosystems [39,40]. Although the toxicity of many chemicals is firmly established when present individually in high concentration, there is now a real and
Fig. 4. IGF I transgenic medaka with abnormal appearance in abdomen tumors. A: Female IGF I transgenic medaka showing an enlarged abdomen. B: Female IGF I transgenic medaka showing an enlarged ovary. C and D (next page): IGF I transgenic fish with tumors.
219
Fig. 4. C and D.
growing concern about the chronic, sublethal, synergistic effects of low level environmental contamination by xenobiotics and the subsequent links between environmental health and human health. Polyhalogenated and polycyclic aromatic hydrocarbons are hepatotoxic and have been directly or indirectly linked with reproductive and immunological dysfunction in a wide variety of animals such as birds, rodents, primates and fish [41].Although the development of highly sensitive analytical techniques now permits the detection of these compounds at infinitesimally low levels, little is known about their biological relevance at chronic sublethal levels, the early pathobiological effects, or the mechanisms of toxicity [42]. Development of an improved biomonitor Biomonitors are measurements that indicate sublethal levels of exposure to or effects
220 of toxic chemicals. These measurements can occur at any level of biological organization, from population and community levels to the molecular level. Generally speaking, observations of adverse effects in populations or communities occur when the pathobiological effects become severe and mortality ensues. It is unquestionably preferable to detect xenobiotic exposure and stress as early as possible, i.e., at the molecular level, in order to take any mitigating steps that could limit or prevent adverse effects. Since fish represent the largest and most diverse group of vertebrates, they provide an excellent model for assessing the impacts of environmental pollutants in aquatic ecosystems. The application of biomonitors in aquatic as well as mammalian toxicology has been widespread and the most commonly used biomonitor is the induction of CYPlA, the predominant hydrocarbon-inducible cytochrome P450 (for review; [43-451). Essentially all organisms have conseryed xenobiotic responsive genes that are inducible by a wide range of xenobiotic compounds. Cytochromes P450, a superfamily of monooxygenases, are one of the most important enzyme systems involved in the detoxification and activation of xenobiotics in eukaryotes [46]. CYPlA, the major PAH-inducible P450 form in mammals and fish, catalyzes the monooxygenase reactions ethoxyresomfin O-deethylation (EROD) and aryl hydrocarbon hydroxylation (AHH), both of which are strongly induced by PAHs, PCBs, and dioxin [47,48]. Cytochrome P450 genes are highly conserved across the evolution of vertebrate classes [49]. Among fish, the rainbow trout CYPIA cDNA was the first to be cloned and verified as an authentic homolog of the vertebrate CYPIA because of the high level of amino acid identity at functional domains necessary for P450 catalytic activity [50]. CYPlA cDNAs have subsequently been cloned from plaice, scup, toadfish and butterflyfish ([51,52]; Vrolijk and Chen, in review). The trout cDNA probe has been used to quantitate CYPlA mRNA levels after exposure to PAHs in a several fish species [53,54], demonstrating the utility of molecular probes in the rapid and sensitive biological detection of PAHs in aquatic environments. CYPlAl is also important in carcinogenesis by virtue of its central role in the metabolic activation of many PAHs into carcinogens [55,56]. For example, sequential P450 catalyzed biotransformation of benzo(a)pyrene (BP) to a highly electrophilic 7,8-dio1-9,10-epoxide is primarily responsible for initiation of BP-induced carcinogenesis. Correlation between hepatic neoplasms in fish and high levels of sediment PAHs has been widely established [57-591. However, the more common occurrence in the environment is low level, chronic PAH exposure. What effect low level exposure has on basic physiological functions such as reproduction, growth, or immune response, in addition to carcinogenic effects, is therefore a topic of growing concern. The detection of CYPlA at the protein level by enzymatic and immunological techniques has permitted significant advances to be made in our ability to detect exposure to PAHs, PCBs, and dioxins [60]. Improved sensitivity has further been achieved by application of CYPlA molecular probes [51,54]. However, a general criticism of using feral fish for biomonitoring studies, is the inability to determine
22 1
prior exposure and the effect of xenobiotic body burdens on enzyme activity and content as well as mRNA levels. One approach to dealing with this problem has been to cage fish of known exposure history in the habitat of concern [61]. Although this approach successfully addresses the problem of prior exposure, other problems still remain. Specifically, many fish must be collected and sacrificed if a biomonitoring scheme is employed on a large scale. In addition, inter-individual variation requires larger sample sizes for statistical comparison. An alternative to whole fish has been the development of cell culture systems for screening potential environmental toxicants [62,63]. This approach permits large scale screenings of xenobiotics as well as eliminates problems of individual variability. However, there are still several time consuming steps required for the isolation and quantification of protein or mRNA, a process that can result in degradation by proteolytic enzymes or nucleases, respectively. The alternative to isolation and quantification of a biomonitor’s gene expression is thus measuring gene expression in vivo. This can be accomplished by development of transgenic animals that utilize a biomonitor gene’s promoter domain with all of the appropriate regulatory sequences linked to a reporter gene that produces an easily quantifiable gene product [64,65]. For example, a CYPlAl transgenic human cell line has recently been developed for assessing the potential toxicity of organic compounds in environmental samples [66]. A transgenic C . elegans has also been produced in which a xenobiotic-inducible hspl6 promoter from C. elegans has been linked to the E. coli lacZ reporter gene [67]. These studies clearly demonstrate the viability and applicability of transgenic biomonitors and suggest that the application of transgenic fish technology to biomonitoring holds tremendous potential. The advantages are numerous, including the ability to monitor xenobiotic exposure in vivo over time in the same individuals in a homogeneous population. This system has the potential to be developed as a model system for research in environmental and human toxicology.
Transgenic fish as environmental biomonitors Since fish represent the largest and most diverse group of vertebrates, they provide an excellent model for assessing the impacts of environmental pollutants. In addition, their intimate association with aquatic pollutants makes them an excellent early warning system for environmental health problems that could potentially lead to human health concerns. Therefore, using fish as experimental animals, our goal is to develop a sensitive and reliable method for detecting biologically relevant levels of polycyclic aromatic hydrocarbon (PAH) exposure that is superior to presently utilized techniques. As a step toward this direction, we are establishing both a transformed fish cell line and producing transgenic fish stocks of Japanese medaka (Oryzaia latipes) that will be genetically engineered with transgenes containing the trout cytochrome P450 (CYPIAZ) promoter fused to the structural gene of the jellyfish green fluorescent protein (GFP) (CYP1A1-GFP; [68]). The entire sequence of the CYPlAl gene has been determined and characterized for rainbow trout [69]. The consensus sequence of the trout CYPlAl promoter
222 domain has also been identified by comparison with the mammalian CYPZAZ [70]. Regulatory elements present in the promoter region of mammalian CYPZAZ consist of 1) xenobiotic responsive elements (XREs; [7 11); 2) CYPZAZ xenobiotic consensus domains [72]; 3) a CAT-box [73]; and 4) the CAT-transcription factorhuclear factor I binding site [74]. These regulatory elements were also identified in the 5' flanking region of the rainbow trout CYPZAZ gene. Results of various studies showed that not only is the enzymatic activity of CYPZAI highly conserved, but the regulatory elements have also been conserved during vertebrate evolution. The prototype of the CYPlAl-GFP transgene construct is depicted in Fig. 5. It contains the promoter domain of CYPZAZ, the structural gene of GFP, a SV40 terminal intron, and a polyadenylation signal. By following the strategy outlined in
Fig. 5. Strategy of producing transgenic fish carrying P450 promoter-reportertransgene construct. P450, CYPlA gene; GFP, green fluorescent protein.
223 Fig. 5, transgenic medaka carrying CYPlAl-GFP transgene can be produced. These transgenic fish can be used to detect the presence of environmental xenobiotics in aquatic ecosystems. Transgenic fish as a model for characterizing the physiological effects of xenobiotics
Our second goal in application of transgenic fish technology in toxicology is to characterize the sublethal effects of PAHs on hepatic function and reproductive physiology in a model fish system. Reproduction in fish is adversely effected by hydrocarbon exposure [41,75]. At the molecular level, hydrocarbon exposure can interfere with the synthesis of vitellogenin (VG), the egg yolk precursor protein [42]. Vitellogenin is synthesized in the liver, secreted into the vascular system, and then deposited in the developing oocytes as lipovitellin and phosvitin [76]. In reproductively active females of oviparous vertebrates, production of vitellogenin is under the modulation of estrogen. In the absence of estradiol, however, VG is not detected in males or females. The synthesis of VG can be readily induced in the livers of juvenile male rainbow trout by administration of estradiol [77]. Juvenile rainbow trout fed food contaminated with sublethal levels of the hydrocarbon, Aroclor 1245, have reduced levels of estradiol-induced VG synthesis [78]. These results suggest a potential mechanistic link between hydrocarbon exposure and adverse effects on reproduction. We have isolated and characterized four VG genes from rainbow trout (Lin and Chen, unpublished results). Through nucleotide sequence comparison of these four genes and sequence alignment with VG genes of chicken [79], the consensus estrogen responsive elements (EREs) have been identified in the 5’ flanking region of each clone. One of these VG promoters can be used to construct a VG-LuxAB transgene construct. In conjunction with the CYPlAl -GFP transgene described above, a double transgenic cell line or medaka can be produced following the strategy described in Fig. 6. The resulting double transgenics can be used to determine the effect of experimental xenobics on reproduction at sublethal levels. Anticipated benefits
The development of transgenic fish holds unique potential for use in toxicology studies and presents numerous advantages over presently utilized techniques. Specifically, an in vivo assay system is the most biologically relevant way to test the effects of xenobiotics. The ability to measure light production in vivo will permit individual time course evaluations which have previously not been possible because fish will not have to be sacrificed. Development of a defined genetic stock of fish with known exposure history for screening chemical pollutants further eliminates the confounding variables, e.g., previous exposure associated with studies using feral fish. In addition, there will be a definite time as well as possible monetary savings because conventional enzyme, immuno-, and molecular assays will not have to be done. Medaka are the model species chosen for these studies because: 1) they are a model
224
Fig. 6. Strategy of producing transgenic fish carrying P450 promoter-reporter and vitellogenin promoterreporter gene double transgene constructs. P450, CYPlA gene; GFP,green fluorescent protein; VG, vitellogenin; LUX, bacterial luciftrase gene.
species in transgenic studies; 2) they are easily maintained and have a short (3 month) generation time; and 3) medaka are translucent, easily permitting the detection of light. In contrast to zebrafish, medaka can also withstand a wide range of salinities from freshwater to full strength seawater. This will potentially permit the application of these fish as biomonitors for water samples from any aquatic or marine environment.
225
Application of transgenic fish in biotechnology The initial drive for transgenic fish research came from attempts to increase production of economically important fish for human consumption. The worldwide harvest of fishery products traditionally depends upon natural populations of finfish, shellfish and crustaceans in fresh and marine water. In recent years, however, the total annual worldwide harvest of fish products has approached or even surpassed the maximal potential level of about 150 million metric tons (as calculated by the US Department of Commerce and the US National Oceanic and Atmospheric Administration). In order to cope with the worldwide demand of fish products and the escalating increase in price, many countries have turned to aquaculture for increasing production of fish products. In 1985, the world production of finfish, shellfish and macroalgae by aquaculture reached 10.6 million metric tons, or approximately 12.3% of the worldwide catch generated by international fishery efforts. Although aquaculture clearly has the potential for increasing worldwide fish production, innovative strategies are needed to improve efficiency. What can transgenic technology offer? Success in aquaculture depends on six factors: 1) complete control of the reproductive cycle of the fish species in culture; 2) excellent genetic background of the broodstock; 3) efficient prevention and detection of disease infection; 4) thorough understanding of the optimal physiological, environmental and nutritional conditions for growth and development; 5) sufficient supply of excellent quality water; and 6) application of innovative management techniques. By improving these factors, the aquaculture industry has developed to a remarkable extent during the last decade. To sustain this growth, however, newly developed technologies in molecular biology and transgenesis will have to be increasingly applied by the aquaculture industry. These technologies can be employed to enhance growth rates, control reproductive cycles, improve feed compositions, produce new vaccines, and develop disease resistant and hardier genetic stocks. In the last several years, we have been searching for strategies to increase fish production by manipulating fish growth hormone and growth factor genes. The feasibility of this approach is demonstrated below. Biosynthetic growth hormone and growth enhancement
In recent years, growth hormone (GH) cDNAs and genomic DNAs have been isolated and characterized for several fish species (for review; [80]). Expression of rainbow trout or striped bass GH cDNA in E. coli cells results in production of a large quantity of recombinant GH polypeptide [8 1,821. Since the GH polypeptide is highly hydrophobic and contains four cysteine residues, the newly synthesized recombinant GH polypeptide forms insoluble inclusion bodies in E. coli cells, rendering the hormone inactive. In an attempt to regain the biological activity of the recombinant hormone, Cheng et al. [82] developed a procedure for renaturing the protein. It involves dissolving the insoluble recombinant hormone in a buffer containing 8M urea and renaturing the polypeptide by slowly removing the urea from the protein
226 solution. The biological activity of the renatured protein was then assessed by an in vitro sulfation assay [83]. In a series of in vivo studies, Agellon et al. [81] showed that application of the recombinant trout GH to yearling rainbow trout resulted in a significant growth enhancement. After treatment of yearling rainbow trout with the recombinant GH for 4 weeks at a dose of 1 pg/g body weighvweek, the weight gain among the individuals of the hormone-treated group was 2 times greater than that of the controls (Fig. 7). Significant length gain was also evident in hormone-treated animals. When the same recombinant hormone was administered to rainbow trout fry (Table 2) or small juveniles by immersing the fish in a GH-containing solution, the same growthpromoting effect was also observed ([81]; Leong and Chen, unpublished results). These results are in agreement with those reported by Sekine et al. [84], Gill et al. [85] and many others [ 8 6 8 8 ] . However, it is important to mention that the growth enhancement effect of the biosynthetic hormone was markedly reduced when more than 2 pg/g body weight of the hormone was applied to the test animals [81]. Recently Paynter and Chen [89] have observed that administration of recombinant trout GH to spats of juvenile oysters (Crassostrea virginica) by the “dipping method” referred to above, also resulted in significant increases in shell height, shell weight, wet weight, and dry weight (Table 3). Furthermore, they also showed that oysters treated with recombinant trout GH, native bovine GH or bovine insulin consumed more oxygen per unit time than controls. The results summarized above clearly
b
220 t
I
0
1
2
3
4
5
8
7
8
Weeks after initial treatment Fig. 7. Effect of recombinant trout GH on growth of yearling rainbow trout. Groups of yearling rainbow trout received intraperitoneal injection of recombinant GH or control extract for 5 weeks. Wet weights of GH-treated and control fish are shown (mean f SE). Open symbols, GH-treated fish 0 , 0 . 2 pg/g body weight; 0 , 1.0 pg/g body weight; A, 2 pg/g body weight. Closed symbols, control fish: 0, mock-treated fish; W, untreated fish. The arrow indicates the time of the last hormone treatment ([81], with permission).
227 Table 2. Effect of GH treatment on the growth of rainbow trout fry ([81], with permission). Treatment
Weight (g)
Saline control GH (50 GH (500
wu
pa)
Initial
Final
% Gain
1.33 f 0.6b 1.29 f 0.7b 1.35 f 0.7b
3.94 f 1 . r 5.51 f 1.6' 5.30 f 1.3'
196 327 293
Values presented as mean f SD. Groups of rainbow trout fry (n = 15) were subjected to osmotic shock in the presence or absence of GH. Weight was measured prior to and 5 weeks posttreatment. Differences between mean weights of GH-treated and control groups were evaluated using Student's t test (a = 0.01). "Significantly different from the GH-treated groups (p < 0.01): brio significant difference between these groups; 'no significant difference between these two treatments.
Table 3. Effect of exogenously applied recombinant rainbow trout growth hormone on oyster growth ([89], with permission). Treatment
Initial ht (mm)
Final ht (mm)
Total wt (mg)
Shell wt (mg)
Dry wt (mg)
Control M 10.' M 10.' M
8.14 (0.25) 8.04 (0.27) 8.72 (0.18) 8.65 (0.32)
11.68 (0.27) 11.74 (0.23) 12.79 (0.27)ab 13.00 (0.36)ab
206 (1 1) 199 (9) 244 (20) 252 (13)b
136 (8) 131 (6) 171 (ll)b 189 (13yb
6.10 (0.66) 6.87 (0.66) 9.42 (0.41)sb 9.41 (0.74)Pb
"Significantly larger than the control group (t test: p < 0.05); bsignificantly larger than M treatment group (t test: p < 0.05). Initial ht reprents mean size at the beginning of the experiment and final ht, total wt, shell wt, and dry wt are mean values determined after the 5-week treatment cycle was concluded. Height (ht) was measured in mm from the umbo to the ventral shell margin; weight (wt) was measured in mg. Standard errors of the mean (SEM) are in parentheses.
suggest that exogenous application of recombinant fish growth hormone can enhance the somatic growth of finfish and shellfish.
GH transgenic fish Although exogenous application of biosynthetic GH results in a significant growth enhancement in fish, it may not be cost effective because of the following reasons: 1) producing purified biosynthetic GH is costly; 2) treating individual fish with the hormone is labor intensive; 3) the optimal hormone dosage for each fish species is difficult to identify; and 4) GH uptake into fish from an exogenous source is inefficient. If new strains of fish producing elevated but optimal levels of GH can be produced, it would bypass all of the problems associated with exogenous GH treatment. Moreover, once these fish strains have been generated, they would be far more cost effective than their ordinary counterparts because these fish would have their own means of producing and delivering the hormone and they could transmit their enhanced growth characteristics to their offspring.
228 Three aspects of fish growth characteristics that could be improved for aquaculture are: 1) initial growth rate so that they reach maturation earlier; 2) enhanced somatic growth rate as adults to provide larger body size for market; and 3) fish with improved feed conversion efficiencies. Among these three, enhanced somatic growth rates via manipulation of the GH gene show considerable promise. Zhu et al. [loo], reported the first successful transfer of a human GH gene fused to a mouse metallothionein (MT) gene promoter into goldfish and loach. Unfortunately, Zhu and his colleagues failed to present compelling evidence for integration and expression of the foreign genes in their transgenic fish studies. Recently, many laboratories throughout the world have successfully confirmed Zhu’s work by demonstrating that human or fish GH and many other genes can be readily transferred into embryos of a number of fish species and integrated into the genome of the host fish. While a few groups have demonstrated expression of foreign genes in transgenic fish, only Zhang et al. [ 131, Du et al. [ 141 and Lu et al. [ 151 have documented that a foreign GH gene could be: 1) transferred to the target fish species; 2) integrated into the fish genome; and 3) genetically transmitted to the subsequent generations. Furthermore, the expression of the foreign GH gene may result in enhancement of growth rates of both P, and F, generations of transgenic fish [13-151. In gene transfer studies conducted in common carp and channel catfish [ 13,16,90,91,941, about lo6molecules of a linearized recombinant plasmid containing the long terminal repeat (LTR) sequence of avian Rous sarcoma virus (RSV) and the rainbow trout GH cDNA were injected into the cytoplasm of one-cell, two-cell and four-cell embryos. Genomic DNA samples extracted from the pectoral fins of presumptive transgenic fish were analyzed for the presence of RSVLTR-rtGH1-cDNA by PCR amplification and Southern blot hybridization of the amplified DNA products using radiolabeled LTR of RSV and/or trout GH1 cDNA as hybridization probes. In the case of transgenic carp studies [13,16], about 35% of the injected embryos survived at hatching, about 10% of which the had stably integrated the RSVLTRrtGH1-cDNA sequence. A similar percentage of transgenic fish was also obtained when the RSVLTR-csGH-cDNA construct was injected into catfish embryos [91,92]. Southern blot analysis of genomic DNA samples of several transgenic carp revealed that a single copy of the RSVLTR-rtGH1-cDNA sequence was integrated at multiple chromosomal sites [ 131. The patterns of inheritance of RSVLTR-rtGH1 cDNA in transgenic common carp were studied by fertilizing eggs collected from nontransgenic females or P, transgenic females with sperm samples collected from several sexually mature P, male transgenic fish. DNA samples extracted from the resulting F, progeny were assayed for the presence of RSVLTR-rtGH1-cDNA sequence by PCR amplification and dot blot hybridization [16]. The percentage of the transgenic progeny resulting from nine matings were: 0, 32,26, 100 (4 progeny only), 25, 17, 31, 30 and 23%, respectively. If each of the transgenic parents in these nine matings carries at least one copy of the transgene in the gonad cell, about 5 E 7 5 % transgenic progeny would have been expected in each pairing. Out of these nine matings, two siblots, both control x P,, gave transgenic progeny numbers as large or larger than expected (p < 0.05) and the
229 remaining had lower than expected numbers of transgenic progeny. These results indicate that although most of these PI transgenic fish had RSVLTR-rtGH1 cDNA in their germline, they might be mosaics. Similar patterns of mosaicism in the germline of PI transgenic fish have been observed in many fish species studied to date [ 13,15,26,91,93,951. If the transgene carries a functional promoter, some of the transgenic individuals are expected to express the transgene activity. According to Zhang et al. [13] and Chen et al. [16], many of the PI and F, transgenic common carp produced rtGH and the levels of rtGH produced by the transgenic individuals varied about 10-fold. Chen et al. [ 161 recently confirmed these results by detecting the presence of rtGH mRNA in the F, transgenic carp using an assay involving reverse transcription (RT)/PCR amplification. They found that different levels of rtGH mRNA were detected in liver, eyes, gonads, intestine and muscle of the F, transgenic individuals. Since the site of transgene integration differs among individuals in any population of PI transgenic fish, they should be considered as totally different transgenic individuals and thus inappropriate for direct comparison of the growth performance among these animals. Instead, the growth performance studies should be conducted in F, transgenic and nontransgenic siblings derived from the same family. Recently Chen et al. [16] conducted studies to evaluate the growth performance of F, Table 4. Mean, standard deviation, coefficient of variation, and percent difference in body weight transgenic common carp, Cyprinus carpio, and their nontransgenic full-siblings. Family Mating 1
2
3
4
5
6
Geno- N type
Mean body weight (SD)
Coefficient of % Difference Range in body variation weight (g)
PI x control
T
31
120.6 (17.4)
14.4
NT
65
99.3 (14.7)
14.8
P, x control
T
11
206.0 (45.2)
21.9
NT
15
147.0 (48)
32.6
7
5.8 (3.4)
58.6
NT
21
7.9 (3.1)
39.2
T
28
66.1 (36.9)
55.8
NT
65
41.7 (27.8)
66.6
T
17
14.7 (6.8)
46.3
NT
82
12.1 (8.4)
69.4
T
97
114.2 (81.6)
71.5
215
133.6 (83.6)
62.5
T
15
72.2 (58.0)
80.3
NT
48
73.3 (47.6)
64.5
PI x PI
PI x PI
PI x PI
PI x PI
T
NT 7
of
PI x PI
20.8
95-173 65- 129
40.1
115-283 67-228
-26.6
1.8-11.3 3.3-17.9
58.5
18.5-338 8.3-141
21.5
6.5-30.4 3.9-56.1
-14.5
18.3-565.1 20.9-416.2
-1.5
7.1-214.4 8.7-203.3
Note: T = transgenic; NT = nontransgenic; N = number of fish; SD = standard deviation.
230
Fig.8. Body weight changes in F,.transgenic and nontransgenic common carp carrying RSV-LTR-rtGH transgene. Each family of fish is derived from a cross of F, transgenic and nontransgenic fish. All of these eight F, transgenics are derived from the same P, transgenic individual. Percent weight change in each F, individual is determined by comparing to the nontransgenic siblings.
transgenic carp in seven families. In these experiments, transgenic and nontransgenic full-siblings were spawned, hatched, and reared communally under the same environment. Results of these studies showed that growth response by families of F, transgenic individuals carrying these rtGHl cDNA varied widely. When compared
23 1 to nontransgenic full-siblings, the results of four out of seven growth trials showed 20, 40, 59, and 22% increases in growth, respectively (Table 4). In three of four families where F, transgenics grew faster than their nontransgenic full-siblings, the maximum and minimum body weights of the transgenics were larger than those of the nontransgenics. In the fourth family, the minimum, but not the maximum, body weight of the transgenics was larger than that of the nontransgenics. The same extent of growth enhancement was also observed in F, offspring derived from crossing the fast growing F, transgenics with nontransgenic controls (Fig. 8). Since the response of the transgenic fish to the insertion of the RSVLTR-rtGH1 cDNA appears to be variable as a result of random integration of the transgene, the fastest growing genotype will likely be developed by utilizing a combination of family selection and mass selection of transgenic individuals following the insertion of the foreign gene. More dramatic growth enhancement in transgenic fish was obtained by introducing Chinook salmon GH cDNA driven by the promoter of ocean pout antifreeze protein gene into Atlantic salmon embryos [14]. Some of these transgenic animals grew several times faster than their controls. In the studies of transgenic medaka carrying a chicken p-actin gene promoter/human GH gene construct, the F, transgenic individuals also grew significantly faster than the nontransgenic siblings [ 151. Manipulation of the GH gene is just one of many examples of improving the genetic traits of fish for aquaculture. Other important traits such as increased tolerance to lower oxygen concentration, increased resistance to bacterial, fungal, viral or parasitic infection, improved food conversion efficiency and increased tolerance to low or high temperature may also be altered by transgenic fish technology provided that the genes responsible for each of these traits are determined. Other biotechnological applications
Another important application of transgenic fish technology will be the generation of novel animals for producing pharmaceuticals of high economical value. Although no real example is available now, transgenic fish, like transgenic cow or sheep, may be used as bioreactors for large scale production of proteins such as human hemoglobin [95], human tissue plasminogen [96], human antihemophilic Factor IX [97], and human a-1-antitrypsin [98].
Acknowledgements This work was in part supported by grants from NSF (DCB-91-05719, IBN-93-17132) to T.T.C. and USDA (93-37205-9073) and BARD (US-2305-93RC) to T.T.C. and R.A.D.
References 1.
Gasser CS, Fraley RT. Genetically engineering plants of crop improvement. Science 1989;244:
232 2. 3. 4. 5.
6. 7. 8. 9. 10. 11. 12.
13.
14.
15.
16.
17. 18. 19. 20. 21.
22.
23. 24. 25.
1293- 1299. Stinchcomb DT, Shaw JE,Cam SH, Hirsh D. Extrachromosal DNA transformationof Caenorhabditis elegans. Molec Cell Biol 1985;5:3484-3496. Rublin GM, Spradling AC. Genetic transforamtion of Drosophila with transposable element vectors. Science 1982;1218:348-353. MaMahan AP, Novak JJ, Britten RJ, Davidson EH. Inducible expression of a cloned heat shock fusion gene in sea urchin embryos. Proc Natl Acad Sci USA 1984;81:749*7494. MaMahan AP, Flytzanis CN, Hough-Evans BR, Wakamatsu Y, Okasa TS. Production of cloned DNA in sea urchin egg cytoplasm: replication and persistence during embryogenesis. Devel Biol 1986;108:420-430. Etkin LD, Pearman B. Distribution, expression and germline transmission of exogenous DNA sequences following microinjection into Xenopus laevis eggs. Development 1987;99:15-23. Gordon JW.Transgenic animals. Int Rev Cytol 1989;155:171-229. Jaenisch R. Transgenic animals. Science 1990;240:1468-1477. Purse1 VG, Pinkert CA, Miller KA, Bolt DA, Campbell RG, Palmiter RD, Brinster RL, Hammer RE. Genetic engineering of livestock. Science 1989;244:1281-1 288. Chen ‘IT, Powers DA. Transgenic fish. Trends Biotechnol 19908:209-215. Fletcher GL, Davis PL. Transgenic fish for aquaculture. In: Setlow JK (ed) Genetic Engineering. New York: Plenum Press, 1991;13:331-370. Hackett PB. The molecular biology of transgenic fish. In: Hochachka P, Mommsen T (eds) Biochemiostry and Molecular Biology of Fish. Amsterdam: Elsevier Science Publishers B.V., 1993;2:207-240. Zhang P, Hayat M, Joyce C, Gonzales-VillasenorLI, Lin C-M, Dunham R, Chen ‘IT, Powers DA. Gene transfer, expression and inheritance of pRSV-Rainbow Trout-GH-cDNA in the carp, Cyprinus curpio (Linnaeus). Molec Reprod Devel 1990;25:313. Du SJ, Gong GL, Fletcher GL, Shears MA, King MJ, Idler DR, Hew CL. Growth enhancement in transgenic Altantic salmon by the use of an “all fish” chimeric growth hormone gene construct. Biotechnology 1992;10:176-181. Lu JK, Chrisman CL, Andrisani OM, Dixon JE, Chen ‘IT. Integration expression and germline transmission of foreign growth hormone genes in medaka, Oryzias latipes. Molec Mar Biol Biotechnol 1992;1:36&375. Chen TT, Powers DA, Lin CM, Kight K, Hayat M, Chatakondi N, Ramboux AC, Duncan PL, Dunham RA. Expression and inheritance of RSVLTR-rtGH1 cDNA in common carp, Cyprinus carpio. Molec Mar Biol Biotechnol 1993;2:88-95. Moav B, Liu Z, Groll Y, Hackett PR. Selection of promoters for gene transfer into fish. Molec Mar Biol Biotechnol 1992;1:338-345. Cotten M, Jennings P. Ribozyme-mediateddestructionof RNA in vivo. EMBO J 1989;8:3861-3866. Buono RJ, Linser PJ. Transient expression of RSVCAT in transgenic zebrafish made by electroporation. Molec Mar Biol Biotechnol 1992;1:271-275. Symonds JE, Walker SP, Sin FYT. Development of mass gene transfer method in chinook salmon: optimization of gene transfer by electroporated sperm. Molec Mar Biol Biotechnol 1994;3:104-111. Tseng FS, Lio IC, Tsai HJ. Introducing the exogenous growth hormone cDNA into lach (Misgurnus anguillicaudatus) eggs via electroporated sperms as carrier. 3rd International Marine Biotechnology Conference, Tromso, Norway, 1994;71. Bums JC, Friedmann T, Driever W, Burrascano M, Yee JK. VSV-G pseudotyped retroviral vector: concentration to very high titer and efficient gene transfer into mammalian and nonmammalian cells. Proc Natl Acad Sci USA 1993;90:8033-8037. Lin S, Gaiano N, Culp P, Bums JC, Friedmann T, Yee J-K, Hopkins N. Integration and germline transmission of a pseudotyped retroviral vector in zebrafish. Science 1994;265:666-668. Lu JK, Bums JC, Chen ‘IT. Retrovirusmediated transfer and expression of transgenes in medaka. 3rd International Marine Biotechnology Conference, Tromso, Norway, 1994;72. Ballagi-Pordany A, Ballagi-Pordany A, Funa K. Quantative determination of mRNA phenotypes by
233 26. 27. 28.
29. 30.
31. 32. 33. 34. 35. 36.
37. 38.
39. 40. 41. 42. 43.
44. 45. 46. 47.
48.
polymerase chain reaction. Anal Biochem 1991;196:88-94. Stuart GW, McMurry JV, Westerfield M. Replication, integration, and stable germline transmission of foreign sequence injected into early zebrafish embryos. Development 1988;109:403-412. Stuart GW, Vielkind JV, McMurray JV,Westerfield M. Stable lines of transgenic zebrafish exhibit reproduction patterns of transgene expression. Development 1990;109:293-296. Shears MA, Fletcher GL, Hew CL, Gauthier S, Davies PL. Transfer, expression, and stable inheritance of antifreeze protein genes in Atlantic salmon (Sulmo salur). Molec Mar Biol Biotechnol 1991;1:58-63. Shamblott M, Chen TT. Identification of a second insulin-like growth factor in a fish species. Proc Natl Acad Sci USA 1992;89:8913-8917. Bell GI, Gerhardt DS, Fong NM, Sanchez-Pescador R, Rall LB. Isolation of human insulin-like growth factor genes: insulin-like growth factor I1 and insulin genes are contiguous. Proc Natl Acad Sci USA 1985;82:6450-6455. Rotwein P, Burgess SK, Milbrandt JD, Krause JE. Differential expression of insulin-like growth factor genes in rat central nervous system. Proc Natl Acad Sci USA 1988;85:265-269. Shamblott MJ, Chen 'lT. Age-related and tissue-specific levels of five forms of insulin-like growth factor mRNA in a teleost. Molec Mar Biol Biotechnol 1993;2:351-361. Duguay SJ, Park LK, Samadpour M, Dickhoff WW. Nucleotide sequence and tissue distribution of three insulin-like growth factor I prohormones in salmon. Molec Endocrinol 1992;6:1202-1210. Wallis AE, Devlin RH. Duplicate insulin-like growth factor I gene in salmon display alternative splicing pathways. Molec Endocrinol 1993;7:409-422. Mathews LS, Norsted G, Palmiter RD. Regulation of insulin-like growth factor I gene expression by growth hormone. Proc Natl Acad Sci USA 1986;83:9343-9347. Hernandez ER, Hurwitz A, Vera A, Pellicer A, Adashi EY, Leroith D, Roberts CT Jr. Expression of the genes encoding the insulin-like growth factors and their receptors in the human ovary. J Clin Endocrinol Metab 1992;74419-425. Kikuchi K, Buonomo FC, Kajimoto Y, Rotwein P. Expression of insulin-like growth factor I during chicken development. Endocrinol 1990 128:1323-1328. Shamblott MJ, Cheng C, Bolt D, Chen TT.Insulin-like growth factor (IGF) mRNA is accumulated in the liver and pyloric caeca of a teleost in a growth hormone-dependent manner. Proc Natl Acad Sci USA 1995;92:694>6946. Malone TC, Bell WH. Environmental research, policy, and regulation: the Chesapeake Bay experience. Mar Pollut Bull 1991;23:497-501. Heinzow BGJ, McLean A. Critical evaluation of current concepts in exposure assessment. Clin Chem 1994;40:1368-1375. Safe S. Polychlorinated biphenyls (PCBs) and polybrominated biphenols (PBBs): biochemistry, toxicology, and mechanism of action. CRC Crit Rev Toxicol 1984;13:319-395. Chen TT. Investigation of effects of environmental xenobiotics to fish at sublethal levels by molecular biological approaches. Mar Environ Res 1988;24:333-337. Kleinow KM, Melancon MJ, Lech JJ. Biotransformation and induction: implications for toxicity, bioaccumulation and monitoring of environmental xenobiotics in fish. Environ Health Perspec 1987;71:105-119. Payne JF, Fancey LL, Rahimtula AD, Porter EL. Review and perspective on the use of mixedfunction oxygenase enzymes in biological monitoring. Comp Biochem Physiol 1987;86C:223-245. Jiminez BD, Stegeman JJ. Detoxification enzymes as indicators of environmental stress on fish. Am Fish SOCSymp 19908:67-79. Gonzalez FJ, Nebert DW. Evolution of the P450 gene superfamily. Trends Genet 1990;6:182-186. Stegeman JJ, Lech JJ. Cytochrome P-450monooxygenase systems in aquatic species: carcinogenic metabolism and biomarkers for carcinogen and pollutant exposure. Environ Health Perspec 1991;90: 101- 109. Stegeman JJ, Hahn ME. Biochemistry and molecular biology of monooxygenases: current perspectives on forms, functions, and regulation of cytochrome P450 in aquatic species. In: Malins
234
49. 50. 51.
52.
53. 54.
55.
56. 57.
58.
59.
60. 61.
62.
63.
64. 65. 66.
67. 68. 69.
DC, Ostrander GK (eds) Aquatic Toxicology; Molecular, Biochemical, and Cellular Perspectives. Boca Raton: Lewis Publishers, 1994;87-206. Nebert D, Gonzalez F. P450 genes: structure, evolution, and regulation. Ann Rev Biochem 1987;56:945-993. Heilmann LJ, Sheen Y, Bigelow SW, Nebert DW. Trout P450IA1: cDNA and deduced protein sequence, expression in liver, and evolutionary significance. DNA 1988;7:37’+387. Leaver MJ, Pimt L, George SG. Cytochrome P450 1Al cDNA from plaice (Pleuronectesplutessa) and induction of P450 1Al mRNA in various tissues by 3-methylcholanthrene and isosafrole. Molec Mar Biol Biotech 1994;2:338-345. Morrison HG, Oleksick MF, Cornell NW, Sogin ML, stegeman JJ. Identification of cytochrome P450 1A (CYPIA) genes from two teleost fish, toadfish (0. tau) and scup (S. chrysops), and phylogenetic analysis of CYPIA genes. Biochem J 1995;308:97-104. Haasch ML, Wejksnora PJ, Stegeman JJ, Lech JJ. Cloned rainbow trout liver P,450 complementary DNA as a potential environmental monitor. Toxicol Appl Pharmacol 1989;98:362-368. Kloepper-Sams PJ, Stegeman JJ. The temporal relationships between P450E protein content, catalytic activity, and mRNA levels in the teleost Fundulus hereroclitus following treatment with P-naphthoflavone. Arch Biochem Biophys 1989;268:525-535. Ionnides C. Induction of cytochrome P450 I and its influences in chemical carcinogenesis. Biochem Soc Trans 1990;18:32-34. Kawajiri K, Nakachi K, Imai K, Watanabe J, Hayashi S-I. The CYPlAl gene and cancer susceptibility. Crit Rev Oncol Hematol 1993;14:77-87. Myers MS, Landah, JT, Krahn MM, McCain BB. Relationships between hepatic neoplasms and related lesions and exposure to toxic chemicals in marine fish from the US west coast. Environ Health Perspec 1991;90:7-15. Collier TK, Singh SV, Awasthi YC, Varanasi U. Hepatic xenobiotic metabolizing enzymes in two species of benthic fish showing different prevalences of contaminant-associated liver neoplasms. Toxicol Appl Pharmacol 1992;113:31+324. Van Veld PA, Vogelbein WK, Smolowitz R, Woodin BR, Stegeman JJ. Cytochrome P450IA1 in hepatic lesions of a teleost fish (Fundulus hereroclitus) collected from a polycyclic aromatic hydrocarbon-contaminated site. Carcinogenesis 1992;13505-507. Goks~jyrA, Forlin L. The cytochrome P-450 system in fish, aquatic toxicology and environmental monitoring. Aquat Toxicol 1992;22:287-3 12. Lindstrom-Seppa P, Oikari A. Biotransformation and other toxicological and physiological responses in rainbow trout (Sulmo guirdneri, Richardson)caged in a lake receiving effluents of pulp and paper industry. Aquat Toxicol 1990 16:187-204. Hahn ME, Lamb TM, Schultz ME, Smolowitz RM, Stegeman JJ. Cytochrome P450IA induction and inhibition by 3,36,4,4O)-tetrachlorobiphenylin an Ah receptor-containing fish hepatoma cell line (PLHC-I). Aquat Toxicol 1993;26:185-208. Ryan JA, Hightower LE. Evaluation of heavy-metal ion toxicity in fish cells using a combined stress protein and cytotoxicity assay. Environ Toxicol Chem 1944,13:1231-1240. Meighen EA. Bacterial bioluminescence: organization, regulation, and application of the lux genes. FASEB J 1993;7:101&1022. Bronstein I, Fortin J, Stanley PE, Stewart G, Kricka. Chemiluminescent and bioluminescent reporter gene assays. Anal Biochem 1994;219:16+181. Anderson JW,Rossi SS, Tukey RH, Vu T, Quattrochi LC. A biomarker, P450 RGS, for assessing the potential toxicity of organic compounds in environmental samples. Environ Toxicol Chem 1995;(In press). Stringham EG, Candido EPM. Transgenic hspl6-lacZ strains of the soil nematode C. elegans as biological monitors of environmental stress. Environ Toxicol Chem 1994;13:1211-1220. Chalfie M, Tu Y, Euskirchen G, Ward, WW, Prasher DC. Green fluorescent protein as a marker for gene expression. Science 1994;263:802-805. Berndtson A, Chen TT.Two unique CYPZ genes are expressed in response to 3-methylcholanthrene
235 70. 71.
72.
73.
74. 75. 76. 77.
78.
79.
80.
81. 82.
83.
84.
85. 86.
87.
88.
in rainbow trout. Arch Biochem Biophys 199q310: 187-195. Jaiswal AK, Gonzalez FJ, Nebert DW. Human P,450 gene sequence and correlation of mRNA with genetic differences in benzo(a)pyrene metabolism. Nucl Acid Res 1985;13:4503-4520. Strom DK, Postlind H, Tukey RH. Characterization of the rabbit C Y f l A I and C Y P l A 2 genes: developmentaland dioxin-inducibleexpression of rabbit liver P4501A1 and P450IA2. Arch Biochem Biophys 1992;294:707-716. Ikeya KA, Jaiswal AK, Owens RA, Jones JE, Nebert DW, Kimura S. Human C Y f I A 2 : sequence, gene structure, comparison with the mouse and rat orthologous gene, and differences in liver 1A2 mRNA expression. Molec Endocrinol 1989;3:1399-1408. Gonzalez FJ, Kimura S , Nebert DW. Comparison of the flanking regions and introns of the mouse 2,3,7,8-tetrachlorodibenzo-p-dioxin-inducible cytochrome P,-450 and P,-450. J Biol Chem 1985;260:504(t5049. Jones KW, Whitlock J Jr. Functional analysis of the transcriptional promoter for the C Y P l A l gene. Molec Cell Biol 1990;10:5098-5105. Hansen PD, Von Westernhagen H, Rosenthal H. Chlorinated hydrocarbons and hatching success in Baltic herring spring spawners. Mar Environ Res 1985;15:59-76. Tata JR,Smith DP. Vitellogenesis: a versatile model for hormonal regulation of gene expression. Recent Prog Horm Res 1979;35:47-56. Chen TT, Reid PC, Van Beneden R, Sonstegard RA. Effect of Aroclor 1254 and Mirex on estradiol induced vitellogenin production in juvenile rainbow trout (Salmo gairdneri). Can J Fish Aqua Sci 1986;43:169-173. Chen TT, Howard DA, Agellon LB, Lin CM, Davies SL. Estrogen controlled gene expression: induction of two estrogen-responsivegenes in the liver of rainbow trout (Salmo gairdneri). Physiol ZOOI1988;62:25-37. van het Schip FD, Samallo J, Broos J, Ophius J, Mojet M, Gruber M, Ab G. Nucleotide sequence of a chicken vitellogenin gene and derived amino acid sequence of the encoded yolk precursor protein. J Molec Biol 1987;196:245-260. Chen TT, Marsh A, Shamblott MJ, Chan KM, Tang YL, Cheng CM, Yang BY. Structure and evolution of fish growth hormones and insulin-like growth factor genes. In: Hew CL, Sherwood N (eds) Fish Physiology. New York: Academic Press, 1994;179-209. Agellon LB, Emery CJ, Jones JM, Davies SL, Dingle AD, Chen TT.Growth hormone enhancement by genetically engineered rainbow trout growth hormone. Can J Fish Aqua Sci 1988;45:1 4 6 1 51. Cheng CM, Lin CM, Shamblott M, Gonzalez-Villasenor LI, Powers DA, Woods C, Chen TT. Production of a biologically active recombinant teleostean growth hormone in E . coli cells. Molec Cell Endocrinol 1995;108:75-85. Cheng CM, Chen TT. Synergism of growth hormone (GH) and insulin-like growth factor-I (IGF-I) in stimulation of sulphate uptake by teleostean branchial cartilage in vitro. J Endocrinol 1995;(In press). Sekine S, Miizukzmi T, Nishi T, Kuwana Y, Saito A, Sat0 M, Itoh H, Kawauchi H. Cloning and expression of cDNA for salmon growth hormone in E . coli. Proc Natl Acad Sci USA 1985;82: 4 3 0 6 4 3 10. Gill JA, Stumper JP, Donaldson EM, Dye HM. Recombinant chicken and bovine growth hormone in cultured juvenile Pacific salmon. Biotechnology 1985;3:430&43 10. Sat0 N, Murata K, Watanabe K, Hayami T, Kuriya Y, Sakaguchi M, Kimura S, Nonak M, Kimura A. Growth-promoting activity of tuna growth hormone and expression of tuna growth hormone cDNA in Escherichia coli. Biotechnol Appl Biochem 1988;10:385-392. Moriyama S, Takahashi A, Hirano T, Kawauchi H. Salmon growth hormone is transported into the circulation of rainbow trout (Oncorhynchus mykiss) after intestinal administration. J Comp Physiol B 1990160:251-260. Schulte PM, Down NE, Donaldson EM, Souza LM. Experimental administration of recombinant bovine growth hormone to juvenile rainbow trout (Salmo gairdneri) by injection or immersion. Aquaculture 1989;76:145-152.
236 89. Paynter K, Chen 'IT. Biological activity of biosynthetic rainbow trout growth hormone in the eastern oyster, (Crussostreu virginicu). Biol Bull 1991;181:45!3-462. 90. Chen TT, Lin C-M, Zhu Z, Gonzalez-Villasenor LI, Dunham RA, Powers DA. Gene transfer, expression and inheritance of rainbow trout growth hormone genes in carp and loach. In: Church R (ed) Transgenic Models in Medicine and Agriculture. New York: Wiley Liss Inc., 199Q127-139. 91. Dunham RA, Ramboux AC, Duncan PL, Hayat M, Chen TT, Lin C-M, Kight K, GonzalezVillasenor LI, Powers DA. Transfer, expression and inheritance of salmonid growth hormone genes in channel catfish, Ictuluruspuncturus, and effects on performance traits. Molec Mar Biol Biotechnol 1992;1:380-389. 92. Powers DA, Gonzalez-Villasenor LI, Zhang P, Chen TT,Dunham RA. Studies on transgenic fish gene transfer, expression and inheritance. In: First ML, Haseltine FP (eds) Transgenic Animals (Biotechnology series no. 16). Boston, MA: Butterworth-Heinmann, 1991;307-324. 93. Ozato K, Kondoh H, Inohabra H, Iwamatus Y, Okada TS. Production of transgenic fish and introduction and expression of chicken &crystallin gene in medaka embryos. Cell Differ 1986;99:237-244. 94. Dunham RA, Eash J, Askins J, Townes TM. Transfer of the metallothionein-human growth hormone fusion gene into channel catfish. Trans Am Fish SOC1987;116:87-91. 95. Swanson ME, Martin MJ, O'Donnell JK, Hoover K, Lago W, Huntress V, Parsons CT, Pinkert CA, Pilder S, Logan JS. Production of functional human hemoglobin in transgenic swine. Biotechnology 1992;10:557-559. 96. Pittius CW, Hennighausen L, Lee E, Westphal H, Nicols E, Vitale J, Gordon K. A milk protein gene promoter directs the expression of human tissue plasminogen activator cDNA to the mammary gland in transgenic mice. Proc Natl Acad Sci USA 1988;85:5874-5878. 97. Clark AJ, Bessos H, Bidhop JO, Hams S, Lathe R, McClenaghan M, Prowse C, Simons JP, Whitelaw CBA, Wilmut X. Expression of human antihemophilic factor IX in the milk of transgenic sheep. Biotechnology 1989;7:487-492. 98. Wright G, Carver A, Cottom D, Reeves D, Scott A, Simons P, Wilmut I, Gamer I, Colman A. High level expression of active human a-antitrypsin in the milk of transgenic sheep. Biotechnology 1991;9:83&834. 99. Powers DA, Hereford L, Cole T, Creech K, Chen TT,Lin CM, Kight K, Dunham RA. Electroporation: a method for transferring genes into gametes of zebrafish (Bruchydunio rerio) channel catfish (lctulurus punctutus), and common carp (Cyrinus curpio). Molec Mar Biol Biotechnol 1992;1:301308. 100. Zhu Z, Li G, He L, Chen SZ. Novel gene transfer into the goldfish (Curussius uurutus L 1758). Angew Ichthyol 1985;1:31-34.
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R. El-Gewely, editor.
237
Chitin biotechnology applications Shigehiro Hirano Department of Agricultural Biochemistry and Biotechnology, Tottori University, Tottori, Japan 680
Abstract. This review article describes the current status of the production and consumption of chitin and chitosan, and their current practical applications in biotechnology with some attempted uses. The applications include: 1) cationic agents for polluted waste-water treatment, 2) agricultural materials, 3) food and feed additives, 4) hypocholesterolemic agents, 5) biomedical and pharmaceutical materials, 6) wound-healing materials, 7) blood anticoagulant, antithrombogenicand hemostatic materials, 8) cosmetic ingredients, 9) textile, paper, film and sponge sheet materials, 10) chromatographic and immobilizing media, and 11) analytical reagents.
Key words: absorbable materials, affinity chromatographic media, agricultural materials, alkaline chitin, antibacterial agents, anticoagulantmaterials, antithrombogenicmaterials, antitumor agents, biological selfdefense function, biomedical materials, chelate complexes, chitin digestibility, chitin xanthate, chitin and chitosan films, chitin, chitinase, chitosan digestibility, chitosan, chitosan-coated papers, chitosanase, cholesterol, cholestyramine resin, CM-chitin, cosmetic ingredients, feed additives, food additives, gel permeation chromatographic media, HE-chitin, hemostatic materials, HP-chitin, hypocholesterolemic function, immobilizing media, immunoadjuvant activity, lysozyme, N-acetylchitosan, N-deaceylase, Nhexanoylchitosan, N-methylenechitosan, N-octanoylchitosan, pharmaceutical materials, polyelectrolyte complexes, PR-proteins, sludge dewatering agents, soil microbial flora, sponge sheets, textiles, uranium ion recovery, waste-water treatment, wound dressing, wound-healing materials. Abbreviations: BAEC, bovine arterial endodermic cells; CH, total cholesterol; CM, carboxymethyl; dp, degree of polymerization; ds, degree of substitution; ES, enzyme substrate; FFA, free fatty acids; HDL, high density lipoprotein; HE, hydroxyethyl; HP, hydroxypropyl; iv, intravenous; LMW, low molecular weight; LPL, lipoproteinlipase; PR-proteins, pathogenesis-related proteins; TG, triacylglycerol; UDP, uridine diphosphate; VSMC, vascular smooth muscle cells.
Introduction Chitin is a ( 1 ->4)-linked 2-acetamido-2-deoxy-~-D-glucan, and chitosan is Ndeacetylated derivatives of chitin (Fig. 1). Chitin and chitosan are the main structural components of the cuticles of crustaceans, insects and mollusks, and the cell walls of microorganisms [ 1,2]. UDP-N-acetyl-D-glucosamine is polymerized into chitin by chitin synthase (EC 2.4.1.16). Chitin-N-deacetylase (EC 3.5.1.41) catalyzes the Ndeacetylation reaction of chitin into chitosan. Chitinase (EC 3.2.1.14) and lysozyme (EC 3.2.1.17) catalyze the hydrolysis of chitin, and chitosanase (EC 3.2.1.132) catalyses that of chitosan to afford the corresponding oligosaccharides (Fig. 2). These enzymes are widely distributed in the tissues of plants, animals, insects and microorganisms in the soil-, hydro- and biospheres of the earth.
238 r
r
1
CHzOH
1
l H N 1,
l H IM J * c-0 I
CEG Chih
chitosan
Fig. I. Chemical structure for a repeating unit of chitin and chitosan.
The manufacture of chitin and chitosan Chitin and chitosan are commercially manufactured by a chemical method. Crab or shrimp shells are deprotenized by treatment with an aqueous 3-5% NaOH solution at room temperature overnight or at 80-90°C for a few hours. The resulting product is demineralized by treatment with an aqueous 3-5% HC1 solution at room temperature to afford a white or slightly pink sample of chitin. The N-deacetylation of chitin is performed by treatment with an aqueous 4&45% NaOH solution at 90-120°C for 4-5 h, and the insoluble precipitate was washed thoroughly with water to afford a crude sample of chitosan. The crude sample is dissolved in aqueous 2% acetic acid, and an insoluble material is removed. The resulting clear supernatant solution is neutralized with an aqueous NaOH solution to afford a purified sample of chitosan as a white precipitate. Chitin deacaylax
(EC 3.5.1.41)
Chitin I I ) Chitasan Chi(EC 3.2.1.14) ChitOsanaSc LysaymC
(EC 3.2.1.17)
chitin oligourcharid*i
Chitman oligosaccharidcs
1
N--113-D-glucospmimdasc (Ec 3.2.1.30)
1 2
N-Aoctyl-D-glucooaminc
Fig. 2. Enzymes hydrolizing chitin and chitosan.
D-GlWsamine
239 The chemical method consumes high energy and wastes a lot of concentrated alkaline solution, resulting in a pollution of the environment. The method will be replaced in the future with an enzymatic method. Crab or shrimp shells are treated by proteases under acidic conditions for the removal of both proteins and CaCO, at the same time to afford chitin, and the successive treatment with chitin-N-deacetylase for the removal of the N-acetyl group gives chitosan.
Molecular characteristics of chitin and chitosan Chitin and chitosan are: 1) the main component of crab and shrimp shells, which are abandoned form the processing companies of marine products; 2) naturally occurring rare aminopolysaccharides;3) biologically reproducible on the earth; 4)biodegradable on the earth; 5 ) biocompatible with the organs, tissues and cells of animals and plants; 6) almost no antigenic polysaccharides in animal tissues and organs; 7) almost no toxic in oral and implant administrations in animals; 8) able to be processed into several casting products including flakes, f i e powders, beads, membranes, sponges, cottons, fibers, and gels; 9) functional physically in high viscosity, moisturizing, metal chelatlng, polyelectrolyte-forming, affiity binding, etc.; 10) functional biologically in the organs, tissues, and cells of both animals and plants, and in soiland hydrospheres; 11) defiite in their chemical structures; and 12) modifiable chemically and enzymatically. Table 1. The estimated consumption of chitin, chitosan and their derivatives in Japanese markets in 1994. Uses
Consumption' (tonshear) ~
Cationic flocculating agents Living waste-water treatment Food manufacturing waste-water treatment Sugar manufacturing Food additives Food processing Functional health foods Agricultrural materials (e.g., plant seed coating, fertilizers) Feed additives for pets, fshes, and animals, etc. Textiles and fabrics Cosmetic ingredients for hair and skin cares d-Glucosamine and oligosaccharides Biomedical materials (e.g., adsorbable suture, wound dressings) Paint and dyeing Thickeners Membranes Chromatographic media and reagents (e.g., colloid titration, enzyme substrates, etc.) 'Estimated as chitosan.
60 50 40 13 20 10 10 1 1
Table 2. Changes in soil microbial flora after chitin is fertilized into soils of a farm."
Weeks after fertilizing
Actinomycetes (x104)
Mold fungi (x104)
Fusarium oxysporum (~10')
Added
None
Added
None
Added
24 8500 200000
24 22 17 12
24 1518
24 13 32 15
9 1
11700
7441 5340
None
0
0
"3%chitin by volume was added into farming soils, and the number of soil microorganisms per m3 in soils was counted.
The current status of the production and consumption of chitin and chitosan The main industrial source of chitin and chitosan are shells of crabs, shrimps, and
Fig. 3. Some commercial products for the improvement of farming soil bacterial flora. A: Granules of chitin and chitosan; B: crab shell flakes; and C: an aqueous chitosan solution in aqueous lactic acid.
24 1 Table 3. The enhancement of chitinase activity in young seedlings of the radish seed coated with chitosan or its derivatives. Chitinase activity in dry seedlings"
Chitosan for coating
Chitosan Oligosaccharidesb Carboxymethy1 HE Depoly merized' A B C d-Glucosamine Uncoated
mU/g tissue
mU/g protein
640 f 20 740 f 16 690 f 25 590 f 29
1.8 f 0.3 2.2 f 0.5 2.0 f 0.2 1.6 f 0.5
660 f 29 790 f 31 600 f 30 580 f 21 590 f 15
2.1 f 0.4 2.3 f 0.3 1.7 f 0.3 1.8 f 0.4 1.7 f 0.3
'Analyzed on the 4th day after the germination. One unit (U) of chitinase activity is defined as the release of reducing sugar value corresponding to 1 pmol of N-acetyl-d-glucosamine for 1 min at 37°C. bdp 2-7. T h e seeds were coated with 0.01 (A), 0.1 (B), and 1.0 (C), respectively, of depolymerized chitosan (average mol. wt. 3,000) per 5 ml of distilled water.
krills which are wasted from the processing companies of marine products. In Japan chitin is manufactured from crab and shrimp shells by six companies. A total of 11 companies, including the above six companies, manufacture chitosan and their derivatives, and supply to Japanese markets. In 1994 the estimated amount of 800 tons (as calculated as chitosan) of chitin and chitosan was manufactured by these companies, and all of them were consumed. Table 1 shows the estimated amount of consumption in various Japanese markets. The total production amount of chitin and chitosan was increased slightly every year, and their consumption tends to have value-added uses. Chitin and chitosan are also manufactured commercially by two companies in Canada, an institute in Poland, and one company each in Norway, India, Thailand, China, and Korea. In the world, the total estimated amount of over 1,OOO tons (calculated as chitosan) of chitin and chitosan was produced and consumed in 1994. Table 4 . An increase in potato yields by the fertilization of chitosan." Fertilizing methods
Control A B C
Total plants
54 54 54
54
Yield (kg/acre) Total weight (kg)
Total number of potatoes
Total weight/ Plant (g)
10.6 15.1 15.1 15.0
24 1 302 285 274
196 280 280 277
"A, chitosan (1 kdacre) was fertilized into soils; B, potato was soaked in 0.2% chitosan solution in aqueous lactic acid for 30 min just before plantation; C, the sliced section of potato was coated with fine chitosan powders.
242 Table 5. The digestibility of orally administrated and chitosan in several animals. Compound supplemented to the basal diet (g/head)
Feeding period (days)
Digestibility Rabbits
Hens
Broilers
Chitin 2 2
5 25
4 10 Chitosan
12
2 2 5 6
5 15 12 32 19
I
28-3 1 35 89 92 39 19 98 19 95
Some current applications of chitin and chitosan in biotechnology Several articles [3--61 have dealt with the practical applications of chitin and chitosan in wide fields. Some of them have been commercialized. Cationic agents for polluted waste-water treatment and dewatering Cationic chitosan forms polyelectrolyte complexes with polyanionic polymers and the chelate complexes with metal ions to afford precipitates [7-91. These reactions have been used for the clarification of polluted waste water. In 1975, chitosan acetate salt was first introduced by a Japanese company as a natural cationic agent for flocculating and sludge dewatering. The system is still used for the treatment of living waste water, the recycling of waste water (e.g., in swimming pools), the recovery of proteins and minerals from industrial waste water, the isolation of bioactive compounds from urines, and the removal of endotoxins from aqueous solutions [1&12]. Chitosan is also usable as an adsorbent for the removal of certain harmful radioisotopes from polluted water and for the recovery of uranium from sea water and fresh water. Agricultural materials Mainly four fertilizing methods are used in agriculture: 1. Chitin and chitosan are fertilized as powders, flacks, or solutions into farming soils or into liquid culture media. 2. A chitosan solution is sprinkled over plant leaves. 3. Plant seeds are soaked in an aqueous solution of chitosan. 4. The surface of plant seeds are coated with a thin membrane of chitin or chitosan, or with their fine powders [13,14].
243 Crab and shrimp shells are composed of chitin, CaCO,, and proteins, and the shells have been used as a traditional fertilizer in agricultural farming for a great number of years. Chitin and chitosan are biodegraded within 2 months in farming soils in summer [15], resulting in an improvement of soil microbial flora. The number of useful microorganisms (e.g., Actinomycetes) increases, and that of the harmful ones (e.g., Fusuriurn) decreases in chitin- and chitosan-fertilized fields [161 (Table 2). Figure 3 shows several commercial products for agricultural farming uses. When the surface of plant seeds is coated with a thin membrane of chitosan and its derivatives, seed chitinase activity is enhanced during their germination stage (Table 3). The chitinase induction of plant seeds enhances the biological self-defense function of seeds by preventing their microbial infections [17], resulting in an increase in plant production (Table 4) [14].
Fig. 4. Some commercial foods and feeds containing chitin or chitosan as an additive. A: Cookies; B: noodles; C (next p a g e ) : soy sauce; and D (next p a g e ) cat food in cans.
244
Fig. 4. Continued.
Food and feed additives
Mushrooms, baker yeasts, and eatable soft shrimps are our daily food materials, and they contain 3-25% of chitin. Chitin is almost nontoxic, and LD,, for chitosan is 16 g/kg in mice. Chitin and its oligosaccharides are a growth factor for bifidus bacteria in animal intestines [19], resulting in the improvement of intestinal bacterial flora in animals. Orally administered chitosan is digested by chitinase and chitosanase, which are secreted from some intestinal bacteria in animal intestines, although animals do Table 6. Effects of a 2% chitosan-supplemented diet feeding on the serum and liver cholesterol levels of rabbits fed a 0.9% cholesterol-enriched diet." Diet Feeding period (days) A
B
0 39 O 39
Serum (ng/dl)
Liver (mg/g)
Total CH
HDL-CH TG
FFA
79 f 4 650f210 7 6 f 12 3 0 0 f 130
37 53 32 58
0.06 0.12 0.06 0.10
140f 8 320f62 120f7 210f40
Total CH TG
Liver weight (9)
14f2
12f2
9 7 f 17
8 f 2
8 f 2
9f21
"A, a 0.9% cholesterol-enriched diet; B, a 2% chitosan-supplemented and 0.9% cholesterol-enriched diet.
245
Fig.5. A view of an adsorbable suture for the clinical use of human beings.
not secret these enzymes themselves [20]. In hens and broilers (Table 5 ) , orally administered chitin and chitosan are almost 100% digested. In rabbits, orally administered chitin is 30% digested, and the digestibility is almost unchanged even after daily feeding for 25 days. On the other hand, orally administered chitosan is digested at 35% in rabbits after daily feeding for 5 days, and the digestibility is increased to 80% after daily feeding for 15 days. This is probably due to an increase Table 7. The enhancement of serum lysozyme activity in the rabbits, which chitin- and chitosan oligosaccharides were iv injected.
Compound" injected
Blood lysozyme activity (U/mlserum) Day after the last injection
Saline (control) Chitosan-oligisaccharides Chitin-oligosaccharides
1st
3rd
5th
6th
4.4 f 1.2 9.2 f 2.2 4.4 f 1.5
4.4 f 2.0 7.1 f 2.2 3.7 f 1.6
ndb 6.9 f 2.4 nd
4.3 f 1.2 4.7 f 2.0 nd
Three to six rabbits weighing 3.5-4.2 kg each were used, and'the mean of three experiments is shown. bnd: not determined.
246
160
-
140
-
120
-
h
100
c
3-4
4-6 Chltlnollgosaccharides (0.1 mglml) 1
80
C 1 2 4 65-11 Chltosanollgosaccharldes (0.1 m g h l )
Fig. 6 . The stimulation of cell proliferation in the culture of rat vascular smooth muscle cells treated with chitin oligosaccharides or chitosan saccharides.
in intestinal bacterial flora secreting chitinase and chitosanase. No data are available on the digestion of chitin and chitosan in human intestines, and on the absorption of chitin- and chitosan-oligosaccharides through the intestines into the blood. Chitosan is used as a food additive for the improvement of food qualities in some commercial foods. Chitosan inhibits the growth of harmful bacteria and molds in foods even at low levels of NaCl, and salted fresh pickles (0.025%chitosan) and soy sauces have been commercialized [2I]. Several other biofunctional foods containing chitin or/and chitosan are commercialized in Japan (Fig. 4).
2
0 1 2 3 chitosanoligosaccharides( m g h l )
N A N :Control
B
C
A :Chitinoligosaccharides B :Chitosanoligosaccharides C :CM chitin
Fig. 7. The enhancementof extracellular lysozyme activity in the culture media of vascular smooth muscle cells (VSMC) and bovine arterial andodermic cells (BAEC) in response to chitinoligosaccharides,chitosan oligosaccharides and CM-chitin.
247
Fig. 8. Some wound dressings for human beings and animals. A: A nonwoven wound dressing of chitin for human tissue wounds. B: A composite dressing of chitosan and atero-collagen. C: A chitosan cotton dressing for animal tissue wounds.
248 Table 8. The blood anticoagualnt activity of sulfated chitin and chitosan. Derivative (molecular wight)
Anticoagualant activity"(units/mg)
0-Disulfated chitin (26.000) N,O-Disulfated chitosan ( 12,000) Heparin (11,ooo)
190-200 110-160 174
"Analyzed by the activated partial thromboplastin time ( A m ) .
Hypocholesterolemic agents
Orally administrated chitosan exhibits a hypocholesterolemic function in animal intestines, although chitosan is digestible in the intestines [22-291. In the rabbits fed a 0.9%cholesterol-enriched diet for 39 days, their serum cholesterol level increases from 79 to 650 mgjdl. In the rabbits fed 2% chitosan supplement to the above diet, the serum cholesterol level is depressed by up to 300 mddl without any significant decrease in useful HDL-cholesterol levels (Table 6) [20]. However, no hypocholesterolemic action is observed with the iv injection of chitosan oligosaccharides in rabbits, indicating that the hypocholesterolemic function is only in animal intestines [31]. Chitosan has a lower toxicity than nondigestible cholestyramine resin, which is used as a clinical hypocholesterolemic agent [26]. These data strongly indicate that chitosan is usable as a natural hypocholesterolemic agent. Biomedical and pharmaceutical materials
Chitin and chitosan are biQcompatible with organs, tissues and cells, and are usable in tissue implantatlons and in oral and iv administrations [32). Several attempts have been made for use of chitin and chitosan as a novel biomedical and pharmaceutical material [33,34]. Chitin and its derivatives are digestible in animal and plant tissues, resulting in an enhancement of the induction of biological defense proteins including lysozyme and chitinase. The rate of enzymatic digestibility of chitin is also controlled by the structure of an N-acyl group and by its ds. These unique properties are usable as biomedical materials for the clinical field, and as a controlled digestible material for the drug delivery system. Figure 5 shows a piece of an absorbable chitin suture, which is dissolved away in the tissues after the clinical operation. On the other hand, N-octanoyl and N-hexanoyl derivatives of chitosan are compatible with the blood, and are resistant to the lysozyme hydrolysis, indicating that these derivatives are usable for blood dialysis membranes and for artificial blood vessels [35]. Rabbit serum lysozyme activity is 4.4 U/ml. The activity is increased up to 9.2 U/ml on the 5th day, when a mixture of chitosan oligosaccharides (dp 2-9) is iv injected daily at a dose of 7.1-8.6 m@g. The serum lysozyme is secreted from the tissues and organs
249
Fig. 9. Some commercial products of chitinous cosmetics for skin and hair cares.
into the blood, and not from the blood cells [31]. The enhanced activity is kept for at least 5 days, and the injected oligosaccharides remain in the blood for up to 13 days after the last injection. However, the iv injection of chitin oligosaccharides does not enhance the lysozyme activity (Table 7). The enhancement of blood lysozyme activity results in the stimulation of the biological self-defense function of animals. Both antitumor [37] and antibacterial functions are also enhanced by the iv injection of both chitin- and chitosan-oligosaccharides [38]. The lipoprotein lipase activity in the blood is enhanced by the iv injection of N,O-sulfated chitosan [39], and the immunoadjvant activity is enhanced by the implantation of partially N-deacetylated chitin into animal tissues [4(t42]. A drug-impregnated chitosan film possesses an efficiency equivalent to the commercial tablet forms [43], and chitosan is used as an
250
Fig. 10. A: A synthetic fabric woven with a chitosan-coated layer. B: A windbreaker.
implantable vehicle for sustained release of anticancer drugs [a]. Chitosan is also formulated into an oral dosage form with drugs, resulting in an enhancement of drug adsorption into the blood [45].
25 1
Fig. Zl. Chitin and chitosan cottons, and some textile fabrics. A: A chitin cotton (100%). B: Chitincellulose composite cottons (1090, w/w). C: Fabrics not woven with cellulose and chitin (top) (99:1, w/w), and textile fabrics woven with a chitin-cellulose composite (bottom).
Wound-healingmaterials The tissue wounds of both plants and animals are covered with a sheet of the membrane or sponge of chitin and chitosan, or with their cottons or fine powders. The wounds are treated with the solutions or pastes of chitin and chitosan. As a result, cell proliferation in the wound tissues is stimulated, and extracellular chitinase or lysozyme is enhanced, resulting in an acceleration of the wound healing with the prevention of bacterial infections [46]. The cell proliferation is stimulated in a dpdependent manner of chitin, chitosan, and their derivatives in animal cell cultures (Fig. 6) [47], and the extracellular chitinase or lysozyme activity is also enhanced in a dose-dependent manner (Fig. 7) [47]. Several wound dressings (artificial skins) have been manufactured from chitin or chitosan, and are commercialized for the clinical use of both human and animals (Fig. 8) [48].
252
Fig. 12. A business card printed by using anionic ink on paper coated with chitosan. The result is a clearer print.
Blood anticoagulant, antithrombogenic and hemostatic materials N-octanoyl and N-hexanoyl derivatives of chitosan have an antithrombogenic function [35], and chitosan has a hemostatic function [49]. Sulfated derivatives of chitin and chitosan have a blood anticoagulant activity as analyzed by the activated partial thromboplastin time. N,O-sulfated chitosan has a lower anticoagulant activity than heparin, but O-disulfated chitin is slightly higher, although the latter compound has no N-sulfate group in the molecule (Table 8). The N-sulfate group in heparin is essential for the activity [50]. O-sulfated chitin has a low toxicity, LD,, 1.25-3.25 &g, in comparison with the toxicity of heparin, LD,, 1.59-2.00 &g. These sulfated derivatives of chitin and chitosan are usable as a novel heparinoid [51].
Cosmetic ingredients The organic acid salts of LMW chitosan are soluble in aqueous ethanol, and are used as an ingredient for hair-seking lotions. Anionic CM-chitin and cationic HP-chitosan are soluble in water and stable in a wide pH range, and they are used as a cosmetic ingredient for skin care. Chitosan, CM-chitin and HP- chitosan have a moisturizing function on skins, a protecting function of mechanical hair damages, and an antielectrostatic function on hairs. Their moisturizing property is compatible to those of an aqueous 20% propylenglycol solution and an aqueous hyaluronic acid solution [52]. These chitosan derivatives also protect against microbial infections on skins and activate skin cells, resulting in the prevention of the skin aging. Figure 9 shows some commercial products of chitinous cosmetics for skin and hair cares.
253
Fig. 23. A commercial product of chitosan beads.
Textile, paper, film and sponge sheet materials Several Japanese companies have manufactured chitosan- or chitin-coated synthetic fibers, and synthetic fabrics woven with a chitosan layer. These products have controlled moisturizing, antibacterial, and antifungal functions (Fig. 10). Chitin xanthate, alkaline chitin, N-acylchitosans gels, and chitosan are used as a manufacturing material for fibers, cottons, films, and sponges (Fig. 11) [53,54]. The physical strength of cellulose paper is increased by chitosan-coating [55], and anionic printing with anionic inks on cationic chitosan-coated paper results in a fine clear printing (Fig. 12). A chitosan solution in aqueous acetic acid is spread as a thin layer on a glass plate, the layer is then air-dried to give a transparent film of chitosan. Similarly, a thin layer of N-acylchitosan gels is air-dried or lyophilized to afford a series of transparent films or opaque sponges [56]. Some novel blend films are also prepared from chitin or chitosan with albumin [57], cellulose [58-611, silk fibroin [62], and several synthetic polymers. Chromatographic and immobilizing media N-Methylene- and N-acyl-chitosan gels are usable as a novel medium for gel
254
Fraction number
Fig. 14. Affinity chromatography of egg-white lysozyme on some N-fatty acyl derivatives of chitosan as an adsorbent. Column: 1.6 x 2.7 cm. Adsorbents: N-acetyl (C2), N-propionyl (C3), N-butyryl (C4), N pentanoyl (C5) derivatives of chitosan, and natural chitin.
permeation chromatography [63,64]. These gels are also usable as an adsorbent medium for the immobilization of enzymes, cells and drugs [65]. Various sizes of the beads of chitin and chitosan are commercialized as an novel adsorbent for metal ions, uranium ions [66], and urine kallikrein, urokinase and erythropoetin (Fig. 13) [67]. Chitinase and lysozyme form an ES complex on a series of N-lower fatty acyl derivatives of chitosan. The N-butyryl and N-hexanoyl derivatives are resistant to these enzymatic hydrolyses, and they form the ES complexes, indicating the suitability of these derivatives as a novel affinity chromatography adsorbent for the isolation and purification of these enzymes (Fig. 14) [68]. Analytical reagents
Chitosan, HE-chitosan, and methylchitosan react stoichiometrically with anionic polymeric colloids in aqueous solutions to afford a neutralized precipitate. This reaction is known as the colloid titration and has been used in analytical chemistry [69]. Colloidal chitin, regenerated chitin (N-acetylchitosan) and its gel, CM-chitin, and HE-chitin are used as enzymatic substrates, which are digestible more than natural chitin in the enzymatic reactions by chitinase and lysozyme.
255
Conclusion Chitin and chitosan are neither medicines, foods, feeds, fertilizers, insecticides, nor fungicides. These polysaccharides are biosynthesized and biodegraded on the earth in an estimated annual amount of 100 billion tons without their excess accumulation. This is called “the natural circulation of chitin on the earth”, and a balanced circulation plays a fundamental role for the conservation of the natural environment and the ecosystem. These ecologically and environmentally activepolysaccharides should be used, in the present and future, from low to high valued applications in wide fields without disordering the balanced circulation of chitin and chitosan on the earth.
Acknowledgements The present work was sponsored by the New Energy and Industrial Technology Development Organization (NEDO), and the Research Institute of Innovative Technology for the Earth (RITE), Kyoto, Japan.
References 1. Muzzarelli RAA. Chitin. Oxford: Pergamon Press, 1937;18+265. 2. Jpn SOCChitidChitosan (ed). Chitin and Chitosan (in Japanese). Tokyo: Gihodo Publishing Co., 1988. 3. Knorr D. Use of chitinous polymers in foods - a challenge for food research and development. Food Techno1 1984;38:85-89. 4. Jpn S o c ChitidChitosan (ed). Applications of Chitin and Chitosan (in Japanese). Tokyo: Gihodo Publishing Co., 1990. 5 . Nihon Kogyokai (ed). Applications of Chitin and Chitosan (in Japanese). Tokyo: Nihon Kogyokai, 1987. 6. Jpn SOCChitidChitosan (ed). A Handbook of Chitin and Chitosan (in Japanese). Tokyo: Gihodo Publishing Co., 1995. 7. Shinoda K, Nakajima A. Complex formation of heparin, sulfated cellulose, hyaluronic acid or chondroitin sulfate. Bull Instit Chem Res Kyoto Univ 1975;53:392-408. 8. Kikuchi Y. Polyelectrolyte complex of heparin with chitosan. Makromol Chem 1975;175:220+ 2211. 9. Hirano S, Mizutani C, Yamaguchi R, Miura 0. Formation of the polyelectrolyte complexes of some acidic glycosaminoglycans with partially N-acylated chitosans. Biopolymers 1987;17:805-810. 10. Hashimoto M. Waste-water treatment materials. In: Jpn Soc ChitidChitosan (ed) A Handbook of Chitin and Chitosan (in Japanese). Tokyo: Gihodo Publishing Co., 1995;485-504. 11. Hashimoto M. Applications of chitosan as an agent for the polluted waste-water treatment and the removal of nucleic acids and endotoxins. In: Nihon Kogyokai (ed) Applications of Chitin and Chitosan (in Japanese). Tokyo: Nihon Kogyokai, 1987;288-311. 12. Sat0 M. An application of chitosan as a flocculating agent. In: Jpn Soc ChitidChitosan (ed) Applications of Chitin and Chitosan (in Japanese). Tokyo: Gihodo Publishing Co., 199Q211-235. 13. Hirano S , Hayashi M, Murae K, Tsuchida H, Nishida T. Chitin and its derivatives as activators of plant cells in tissues and seeds. In: Gebelein CG, Carraher CE Jr, Foster VR (eds) Applied Bioactive
256 Polymeric Materials. New York: Plenum Press, 1988;45-59. 14. Tsugita T. Agricultural materials. In: Jpn SOCChitinKhitosan (ed) A Handbook of Chitin and Chitosan (in Japanese). Tokyo: Gihodo Publishing Co., 1995;44&458. 15. Hirano S, Koishibara Y, Kitaura S, Taneko T, Tsuchida H, Murae K, Yamamoto T. Chitin Biodegradation in Sand Dunes. Biochem Syst Ecol 1991;19:37+384. 16. Hirano S. Chitin and chitosan as an agricultural material (in Japanese). Nogyo Gijutsu Taikei 1992;7:156/18-24. 17. Hirano S, Yamamoto T, Hayashi M, Nishida T, Inui H. Chitinase activity in seeds coated with chitosan derivatives. Agric Biol Chem 199054:2719-2720. 18. Hadwiger LA, Fristensky B, Riggleman RC. Chitin, a natural regulator in plant-fungal pathogen interactions, increases crop yields. In: Zikakis JP (ed) Chitin, Chitosan and Related Enzymes. Orlando: Academic Press Inc., 1984;291-302. 19. Austin PR, Brine CJ, Castle JE, Zikakis JP. Chitin: new facets of research. Science 1981;212:749750. 20. Hirano S, Itakura C, Seino H, Akiyama Y, Nonaka I, Kanbara N, Kawakami T. Chitosan as an ingredient for domestic animal feeds. J Agric Food Chem 1990;38:1214-1217. 21. Hirano S. Some commercial chitinous products (in Japanese). MOL 1989;4:27-31. 22. Sugano M, Fujikawa T, Hiratsuji Y, Hasegawa Y. Hypocholesterolemic effects of chitosan in cholesterol-fed rats. Nutr Rep Int 1978;18:531-537. 23. Kobayashi T, Otsuka S, Yugai Y. Effect of chitosan on serum and liver cholesterol levels in cholesterol-fed rats. Nutr Rep Int 1979;19:327-334. 24. Nagyvary JJ, Falk JD, Hill ML, Schmidt ML, Wilkins A, Brodburg EL. The hypolipidemic activity of chitosan and other polysaccharides in rats. Nutr Rep Int 1979;20:677-684. 25. Sugano M, Fujikawa T, Hiratsuji Y, Nakashima Y, Fukuda N, Hasegawa Y. A novel use of chitosan as a hypocholesterolemic agent in rats. Am J Clin Nutr 1980;33:787-793. 26. Vahouny GV, Satchithanadam S, Cassidy MM, Lishtfort FB, Furda I. Comparative effects of chitosan and cholestyramine on lymphatic absorption of lipids in the rat. Am J Clin Nutr 1983;38:278-284. 27. Furda I. Aminopolysaccharides- their potential as dietary fiber. ACS Symp Ser 1983;214105-122. 28. Sugano M, Watanabe S, Kishi A, Izume M, Ohtakara A. Hypocholesterolemic action of chitosan with different viscosity. Lipids 1988;23:187-191. 29. Fukada Y, Kimura K, Ayaki Y. Effect of chitosan feeding on intestinal bile acid metabolism in rats. Lipids 1991;26:396399. 30. Maezaki Y, Tsuji K, Nakagawa Y, Kawai Y, Akimoto M, Tsugita T, Takekawa W, Terada A, Hara H, Mitsuoka T. Hypocholesterolemic effect of chitosan in adult males. Biosci Biotechnol Biochem 1993;57:1439-1444. 31. Hirano S, Akiyama Y. Absence of a hypocholesterolemic action of chitosan in high-serumcholesterol rabbits. J Sci Food Agric 1995;69:91-94. 32. Hirano S, Seino H, Akiyama Y, Nonaka I. Chitosan: a biocompatible material for oral and iv administrations. In: Gebelein CG, DUM RL (eds) Progress in Biomedical Polymers. New York: Plenum Press, 1990283-290. 33. Hirano S, Hirochi K, Hayashi K, Mikami T, Tachibana H. Cosmetic and pharmaceutical uses of chitin and chitosan. In: Gebelein CG (ed) Cosmetic and Pharmaceutical Applications of Polymers. New York: Plenum Press, 1991;95-104. 34. Hirano S. Chitin and chitosan. In: Sen0 M, Otsubo 0 (eds) A Handbook of Recent Biomedical Materials and their Applications. Tokyo: R&D Planning, 1986;235-245. 35. Hirano S, Noishiki Y. The blood compatibility of chitosan and N-acylchitosans. J Biomed Mater Res 1985;19:413-417. 36. Hirano S, Iwata M, Nakayama K, Tanaka H, Toda T, Inui H. Enhancement of serum isozyme activity by injecting a mixture of chitosan oligosaccharides intravenously in rabbits. Agric Biol Chem 1991;55:2623-2625. 37. Suzuki K, Mikami T, Okawa Y, Tokoro A, Suzuki S, Suzuki M. Antitumor effect of N-heaxaacetyl-
257 38.
39.
40. 41. 42. 43.
44.
45. 46.
47.
48. 49. 50. 51.
52. 53. 54. 55.
56. 57. 58. 59. 60.
chitohexaose and chitohexaose Carbohydr Res 1986;151:403-408. Tokoro A, Kobayashi M, Tatewaki N, Suzuki K, Okawa Y, Mikami T, Suzuki S, Suzuki M. Protective effect of N-acetylchitohexaose on Listeria monocyrogenes infection in mice. Microbiol Immunol 1989;33:357-3677. Hirano S, Kinugawa J, Nishioka A. Sulfated derivatives of chitosan, and their characterization with respect to biological activity. In: Muzzarelli RCC, Jeuniaux C, Gooday GW (eds) Chitin in Nature and Technology. New York: Plenum Publishing Co., 1986;461-468. Nishimura K, Nishimura S, Nishi N, Saki I, Tokura S, Azuma I. Immunological activity of chitin and its derivatives. Vaccine 1984;2:93-99. Nishimura K, Nishimura S, Nishi N, Numata F, Tone Y, Tokura S, Azuma I. Adjuvant activity of chitin derivatives in mice and guinea pigs. Vaccine 1985;3:37%384. Nishimura K, Nishimura S, Sen0 H, Nishi N, Tokura S, Azuma I. Macrophage activation with multiporous beads prepared from partially deacetylated chitin. Vaccine 1987;5:136- 140. Miyasaki S, Yamaguchi H, Takada M, Hou WM, Takeichi Y, Yasubuchi H. Pharmaceutical application of biomedical polymers. 29. Preliminary study on a film dosage form prepared from chitosan of oral drug delivery. Acta Pharm Nord 1990;2:401-406. Song Y, Onishi H, Nagai T. Conjugate of mitomycin C 20 with N-succinylchitosan: in vitro drug release properties, toxicity and antitumor activity. Int J Pharmacol 1993;98:121-130. Miyasaki S, Yamaguchi H, Yokouchi C, Takada M, Hou WM. Sustained release of indomethacin from chitosan granules in beagle dogs. J Pharm Pharmacol 1988;40:642-643. Hirano S, Inui H, Kosaki H, Uno Y, Toda T. Chitin and chitosan: ecologically bioactive polymers. In: Gebelein CG, Carraher C (eds) Biotechnology and Bioactive Polymers. New York: Plenum Press, 1994;43-54. Inui H, Tsujikubo M, Hirano S. Low molecular weight chitosan stimulates mitogenic response to platelet-derived growth factor in vascular smooth muscle cells. Biosci Biotechnol Biochem 1995;59:2111-2114. Kifune K. Biomedical materials. In: Jpn Soc ChitidChitosan (ed) A Handbook of Chitin and Chitosan (in Japanese). Tokyo: Gihodo Publishing Co., 1995;323-354. Malette WG, Euigleym HJ Jr, Gaines RD, Johnson ND, Rainer WG. A hemostatic activity of chitosan. Ann Thorac Surg 1983;35:55-58. Hirano S, Tanaka Y, Hasegawa M, Tobetto T, Nishioka A. Effect of sulfated derivatives of chitosan on some blood coagulant factors. Carbohydr Res 1985;137:205-215. Hirano S, Noishiki Y, Kinugawa J, Higashijima H, Hayashi T. Chitin and chitosan for use as a novel biomedical material. In: Gebelein CG (ed) Advances in Biomedical Polymers, New York: Plenum Publishing Co., 1987;285-287. Chikamatsu Y. Cosmetic materials (in Japanese). In: Jpn SOCChitin/Chitosan (ed) A Handbook of Chitin and Chitosan. Tokyo: Gihodo Publishing Co., 1995;355-382. Makagawa Y. Textiles. In: Jpn SOCChitin/Chitosan (ed) A Handbook of Chitin and Chitosan (in Japanese). Tokyo: Gihodo Publishing Co., 1995;474-483. Rathke TD, Hudson SM. Review of chitin and chitosan as fiber and film formers. JMS Rev Macromol Chem Phys 1994;C34:375-437. Kobayashi Y, Nishiyama M, Matsuo R, Tokura SM, Nishi N. Application of chitin and its derivatives to paper technology. In: Jpn S o c Chitin/Chitosan (ed) Chitin and Chitosan. Tokyo: Gihodo Publishing Co., 1982;239-243. Hirano S. A facile method for the preparation of novel membranes from N-acyl and M-arylidenechitosan gels. Agric Biol Chem 1978;42:193%1940. Chandy T, Sharma CP. Chitosan as a biomaterial. J Colloid Interface Sci 1989;130:331-340. Hosokawa J, Nishiyama M, Yoshihara K, Kubo T. Biodegradable film derived from chitosan and homogenize cellulose. Ind Eng Chem Res 1990;29:800-805. Hosokawa J, Nishiyama M, Yoshihara K, Kubo T. Reaction between chitosan and cellulose on biodegradable composite film formation. Ind Eng Chem Res 1991;30:788-792. Nishiyama M. Papers and pulps. In: Jpn SOCChitin/Chitosan (ed) A Handbook of Chitin and
25 8 61. 62. 63. 64. 65. 66. 67. 68. 69.
Chitosan (in Japanese). Tokyo: Gihodo Publishing Co., 1995;45+473. Hasegawa M, Isogai A, Onabe F, Usuda M, Atalla RH. Characterization of cellulose-chitosan blend films. J Appl Polym Sci 1992;45:1873-1879. Liang CX, Hirabayashi K. Studies on the mechanical properties of silk fibroin membranes. Part 3. The mechanical properties of fibroin-chitosan membranes. Sen4 Gakkaishi 1991;47:334-338. Hirano S, Matsuda M, Tanaka T. N-methylenechltosan gels and some of their properties as a medium for gel chromatography. Carbohydr Res 1979;71:344-348. Hirano S, Nishiguchi Y. N-Carboxyacyl-N-acylchitosangels as novel media for gel chromatography. Carbohydr Polym 1985;5:13-15. Yamaguchi R, Arai Y, Hirano S, Ito T. Incorporation of glucoamylase into some polyelectrolyte complexes. Agric Biol Chem 1978;42:1297-1299. Hirano S, Kondo Y, Nakazawa Y. Uranylchitosan complexes. Carbohydr Res 1982;100:431-434. Seo H. Porous bead materials. In: Jpn SOCChitin/Chitosan (ed) A Handbook of Chitin and Chitosan (in Japanese). Tokyo: Gihodo Publishing Co., 1995;505-537. Hirano S, Kaneko H, Kitagawa M. N-lower fatty acyl derivatives of chitosan as adsorbents for lysozyme and chitinase. Agric Biol Chem 1991;55:1683-1684. Senju R. Colloid Titration. Tokyo: Nankodo, 1969.
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R. El-Gewely, editor.
259
ADPglucose pyrophosphorylase: basic science and applications in biotechnology Jack Preiss Department of Biochemistry, Michigan State University, East Lansing, Michigan, USA
Abstract. The enzymatic reactions of bacterial glycogen and plant starch synthesis are similar and some of the properties of the biosynthetic enzymes are compared. Regulation occurs at the synthesis of ADPglucose and in almost all cases, ADPglucose pyrophosphorylase, is allosterically activated about 10to over 40-fold by glycolytic intermediates and inhibited by AMP, ADP or Pi. The activator specificity of the ADPglucose pyrophosphorylase varies with respect to the source of enzyme and can be correlated to the major assimilation pathway occurring in the organism. For example, ADPglucose pyrophosphorylases from plants and other oxygenic photosynthetic organisms are activated by 3-phosphoglycerate. Organisms using glycolysis for carbon assimilation have ADPglucose pyrophosphorylases with fructose1,6-bis-phosphateas the major activator. Chemical modification and site-directed mutagenesis studies that have determined the activator binding sites for some enzymes are described. The structural genes of Escherichiu coli ADPglucose pyrophosphorylase allosteric mutants which no longer require activator for activity have been isolated. Transformation of plant systems with an allosteric bacterial mutant gene (but not with the wild-type gene) increases their starch content. Transformed potato tubers can have 25-60% more starch than the normal tuber indicating the importance of allosteric regulation of ADPglucose synthesis. The increase of a normal plant product by transformation of the plant with a gene encoding the rate-limiting enzyme in starch synthesis is an important biotechnological advance and suggests the possibilities of changing starch composition (extent of branching and chain sizes) via transformation with the starch synthase and branching enzyme genes.
Key words: ADPglucose pyrophosphorylase, bacterial glycogen synthesis, branching enzyme, control coefficient analysis of starch synthesis, starch synthesis, starch synthase.
Introduction: the involvement of ADPglucose in synthesis of bacterial and plant a-1,4-glucan The biosynthesis of bacterial glycogen and of starch in algae and higher plants share common reactions and they are listed below. 1. ATP + a-Glc-l-P ADPGlc + PPi 2. ADPGlc + a-1,4-glu~an+ ADP + a-1,4-glucosyl-a- 1 , 4 - g l u c ~ 3. Elongated a-1,4-glucan chain + Branched a-1,6- a-1,4-glucan First, the sugar nucleotide, ADPGlc, is synthesized from ATP and glucose-l-P via a reaction (reaction 1) catalyzed by ADPglucose pyrophosphorylase (ADPGlc PPase; glucose-l-P adenyltransferase, E.C. 2.7.7.27). This reaction was first described by
Address for correspondence: Jack Preiss, Department of Biochemistry, Michigan State University, East Lansing, MI 48824, USA. Tel.: +1-517-353-3137. Fax: +1-517-353-9334. E-mail:
[email protected]
260 Espada in 1962 in soybean [ 13 and was subsequently found in many plant tissues and in bacterial extracts. The ADPGlc PPase of many systems was found to be highly regulated [2-91, glycolytic intermediates are potent allosteric activators and either AMP, Pi or ADP can be inhibitors. Thus, the regulatory kinetic properties of the enzyme from many systems were characterized in detail. The major activator and inhibitor for the ADPGlc PPase isolated from different systems can vary and the correlation of the nature of the activator with the metabolism of that system will be discussed in a later section. The activator can have a multifold effect on enzyme activity, either increasing catalytic efficiency via increasing V,,, and/or increasing the apparent affinity of the substrates for the enzyme. The activator can also reverse the effects of the allosteric inhibitor. It has been postulated that modulation of ADPGlc synthesis is due to variation of the ratio of activator concentration to inhibitor concentration [2-91. Indeed, a class of bacterial or algal mutants having altered glycogen or starch levels compared to the parent strain have been isolated which have an ADPGlc PPase modified with respect to its allosteric kinetic properties (for reviews see [2,3,5-91). Those mutants having an ADPGlc PPase with a higher affinity for the activator or lower affinity for the inhibitor synthesize their a-1,4-glucan at a higher rate than the normal parent strain. Conversely, those mutants having an ADPGlc PPase with a lower affinity for activator synthesized a-l,4-glucan at a lower rate. The subunit structures of the bacterial and plant ADPGlc PPases have been studied to some extent [3,5-91. For all the bacterial enzymes studied to date, including some of the cyanobacteria, the enzyme has one subunit with a mass of about 50 kDa and the native enzyme is a homotetramer. In contrast, the plant enzyme consists of two related but different subunits having masses in the 5CF60 kDa range. The smaller subunit (5CF55 kDa) is highly conserved with an amino acid sequence identity varying from 85-95% when different plant sources and tissues are compared [ 101. In contrast, the larger subunit of the ADPGlc PPase is less conserved, 5&60% [ 101. The identity between the large and small subunits is also about 5CF60% and the amino acid sequence identity of the E. coli enzyme with the plant enzymes is about 30% [lo]. Thus, it has been postulated that the higher plant enzyme genes evolved from the bacterial ADPGlc PPase gene via gene duplication and then diversion [lo]. Recent preliminary experiments [ 111 suggest that the plant ADPGlc PPase small subunit functions as the catalytic subunit while the large subunit is involved in the allosteric regulation of the activity. In the next reaction in the pathway (reaction 2) the glucosyl moiety of the sugar nucleotide is transferred to a maltodextrin, glycogen or starch to give rise to a new a-1,4-glucosyl linkage. There are some differences seen at this step with respect to the bacterial and plant systems. In the bacterial systems (e.g., E. coli) only one glycogen synthase (or glycogen synthase gene) has been found [12]. However, in plants there may be up to four different starch synthases (E.C. 2.4.1.21) found in a cell that can be differentiated either immunologically or by isolation of different cDNAs or genomic genes encoding them [2,5,6,8,9]. The functions of these starch synthases in the synthesis of the starch granule components, amylose and amylopec-
26 1 tin, are not well understood. Some starch synthases are bound to the starch granule (and can only be solubilized by a-amylase digestion) while others, noted as soluble starch synthases (SSS), are found in the soluble portion of the extract [8,9]. It is known that those mutant plants (waxy mutants) deficient in the granule-bound starch synthase isozyme (GBSS), do not synthesize amylose. The mechanism for amylose synthesis is not clear but these waxy mutants certainly suggest that the various granule-bound and soluble starch synthases play different roles in the synthesis of amylose and amylopectin. The starch synthase isozymes in maize endosperm have different molecular masses. The GBSS isozyme I has a molecular mass of 60 kDa, that of GBSSII 95 kDa, the SSS I a molecular mass of 72 kDa, and SSS 11,95 kDa [8,9]. Mu et al. [13], have reported the molecular mass of maize endosperm SSS I as 76 kDa which is similar to the value reported previously for SSS I (reviewed in [8,9]). These molecular mass values for the starch synthases are all higher than that of the E. coli glycogen synthase with a molecular weight of 52 kDa [12]. There is a report that in pea embryo some of the soluble starch synthases may also be bound to the starch granule [14] and that in maize endosperm some of SSS I adheres to the starch granule [ 131. The conclusions in the pea embryo study [ 141 are based on positive immunoblots obtained after electrophoresis of the soluble starch synthase with antibody prepared against the GBSS and also on similarity of the amino acid sequence of three peptides obtained from protease SV8 digests of the soluble starch synthase. This clearly shows there is a close relationship between SSS I1 and GBSSII but does not indicate that they are identical proteins. It is also not clear how much of the SSS I1 activity is present as granule-bound activity and how much is soluble activity. It is also not surprising that some of the SSS I1 is present as starch-granule bound as it does have affinity for its substrate, starch. The evidence for the maize study rests on the observation of a positive immunoblot with antibody prepared against a 76 kDa protein obtained from the starch granule with the SSS I on electrophoresis; the antibody also neutralized the SSS I activity. Since SSS I has affinity for the granule one would expect to have some of the SSS I protein bound to the granule and the question is how much is bound and whether the binding is similar to binding of the GBSS to the starch granule. There is no question that in maize, the GBSSI is immunologically distinct from the SSS I (reviewed in [8,9]). Obviously, more detailed studies remain to be done to establish the relationships between the granule-bound starch and the soluble starch starch synthases. It is clear they will have common sequences but it still remains to be established whether some of the soluble starch synthases and granule-bound starch synthases are the same proteins. They obviously must have different functions in synthesis of the starch components, amylose and amylopectin. This is certainly suggested by the recent studies of starch synthesis in Chlumydomonus [15-171 where it has been shown that SSII may be involved in synthesis of the intermediate size chains of amylopectin [ 161 and that granule-bound starch synthase is not only involved in amylose synthesis but also in amylopectin synthesis [ 171. Certainly, the various granule-bound and soluble
262 starch synthases must be separated from each other and their properties examined with respect to their chain-lengthening properties; to what chains they prefer to transfer glucosyl residues to (the A, B1, B2, B3 or B4 chains of amylopectin [18]) as well as the optimal length of glucosyl residues they can efficiently synthesize. A number of similarities have been seen with the bacterial glycogen synthase and the plant starch synthases with respect to amino acid sequences and they are shown in Fig. 1. There are three regions of high conservation and at least one of them has
Region 1
E. coli glycogen synthase
1MQVLHVCSEMFPLLKTGGLADVIGALP
Potato tuber wx protein
4MNLIFVGTEVGPWSKTQaLGDVLRGLP
Cassava Wx protein
4MNLIFVGAEVGPWSKTGGLGDVLGGLP
Maize Wx protein
5MNVVFVGAEMAPWSKTGGLGDVLGGLP
Barley Wx protein
6MNLVFVGAENAPWSKTGGLGDVLGGLP
Wheat Wx protein
7MNLVFVGAEMAPWSKTGGLGDVLGGLP
Rice Wx protein
6MNWFVGAEMAPWSKTGGLGDVLGGLP
Rice soluble starch synthase
E. coli glycogen synthase
I
I
I
I
I
I
I
I
I
I I
I
I
I
I
I I I
I I I
I
I I I
IIIIIII
I1
IIIII I I
I1
I1
IIIIIII
IIIIIII
II
I11111IIII
I I
IIIIIII
II
IIIIII
II
20RSWFVTGEASPYAKSGGLGDVCGSLP Region 2
Region 3
372VPSRFEPCGLTQL
397RTGGLADTV
Potato tuber wx protein
IIIIIIIIII II
Ill1 I l l
397VPSRFEPCGLIQL
422STGGLVDTV
Cassava Wx protein
IIIIIIIIII II
IIII Ill
398VPSRFEPCGLIQL
423STGGLVDTV
Maize Wx protein
I IIIIIIII II
IIII Ill
398VTSRFEPCGLIQL
423STGGLVDTV
Barley Wx protein
I IIIIIIII II
Ill1 Ill
396VTSRFEPCGLIQL
421STGGLVDTV
Wheat Wx protein
I IIIIIIII II
IIII Ill
410VTSRFEPCGLIQL
435STGGLVDTU
Rice Wx protein
397VPSRFEPCGLIQL
Rice soluble starch synthase
IIIIIIIII II
IIII Ill
372MPSRFEPCGLNQL
397GTGGLRDTV
1 IIIIIIII II
IIII Ill 422STGGLVDTV
Fig. 1. Conserved regions of amino acid sequences of the E. coli glycogen synthase, some granule-bound
starch synthases (also known as waxy (Wx) proteins) and rice seed soluble starch synthase. The numbers preceding the sequence indicate the residue number from the putative N-terminus in the sequence. The sequence in bold, KTGGL,has been shown for the E. coli glycogen synthase to be involved in binding of the sugar nucleotide substrate (19,20). References to the other sequences may be obtained from [9].
263 been shown to be involved in binding the substrate, ADPglucose [ 19,201.This region, region I, is at the N-terminal. The possible functions for regions 2 and 3 are not known. However, in region 2 only one or two of the 13 amino acids in the sequences known for granule-bound starch synthases, and of the only complete known sequence for one soluble starch synthase [21],are different from the E. coli glycogen synthase sequence. In region 111 all the GBSSs are identical with respect to the amino acid sequence while the E. coli sequence differs in only two of the nine amino acids, Arg for Ser and Ala for Val. The soluble starch synthase has Gly for that Ser and an Arg residue instead of Val. In addition, Lys residue 277 of the E. coli glycogen synthase, which has also been shown to be involved in catalysis [22],is also conserved in the granule-bound and soluble starch synthases. Many questions remain with respect to protein-structure-function relationships among the three types of a-1,4 glucan synthases with respect to the primer binding site and amino acid residues involved in catalysis. In rice seed, there is no question that the soluble starch synthases are different from the granule-bound starch synthases in that there is only 2!&37% identities with the rice GBSSs [21]. To date, the only a-1,4glucan synthase reported to be overexpressed with high activity is the E. coli enzyme [12].This system should be further exploited with respect to the methodologies of chemical modification, site-directed mutagenesis and attempts to determine its three-dimensional structure. The third reaction of the pathway is catalyzed by branching enzyme ( B E E.C. 2.4.1.18),and responsible for the synthesis of the a-1,6linkages found in amylopectin and in glycogen. In E. coli or other bacteria only one branching enzyme and one gene is present while in plants genetic and biochemical evidence indicate there can be two to four forms of branching enzyme. In maize endosperm [23-251 and in rice seed [26-28],two different cDNAs have been isolated encoding two branching enzyme isozymes. Both in vitro [29,30]and in vivo [31,32]studies suggest that in the case of maize endosperm the size of chain length transferred by the two branching enzyme isozymes are different. Maize BE I transfers long chains with sizes of about 3Ck100 glucose units long, while BE IIa and IIb transfer short chains about 6 1 4 glucose units long [29].BE I has 30 times more activity on amylose than amylopectin while BE IIa and IIb have 2 times more activity on amylopectin than on amylose [30].Thus, BE I may function in the synthesis of the interior structure of amylopectin synthesizing the B chains while BE IIa and BE IIb may be involved in synthesizing the exterior chains, mainly the A chains. The genes for the branching enzymes from various bacteria as well as cDNA clones of genes representing the BE isozymes from numerous plants have been isolated and references to these studies may be found in the recent review of Preiss and Sivak [9].Their deduced amino acid sequences show high identity. Outside of the studies discussed above with respect to size of the chains transferred by the maize endosperm BE to form the a-1,6linkages, no studies have been conducted with respect to identifying amino acid residues involved in catalysis. However, sequence comparisons done by a number of groups, especially by Svensson and her colleagues [33],indicate that the various starch and glycogen branching enzymes contain
264 concensus sequences to the four regions that are postulated to be the catalytic regions of the a-amylase family of enzymes. This family includes pullulanase, isoamylase, glucosyl transferase and cyclodextrin glucanotransferase (Fig. 2). The four regions are in the central portion of the amino acid sequences of these enzymes. The conservation of the putative catalytic sites of the a-amylase family in the starch and glycogen branching enzymes should be no surprise as BE catalyzes two consecutive reactions for synthesis of a-1,6 glucosyl linkage; cleavage of the a-1,4 glucosyl linkage and then transfer to a C-6 hydroxyl group of a glucose residue in the growing polysaccharide. These reactions most probably are very similar to those catalyzed by the other a-amylase family enzymes, namely cleavage and transfer to another glucose residue or to H,O. Of interest is that some of the eight highly conserved amino acid residues of the a-amylase family may also be functional in branching enzyme catalysis. Preliminary unpublished results (T. Kuriki and J. Preiss) using site-directed mutagenesis suggest
B. subtilis a-1-amylase B. sphaericus cyclodextrinase P . amyloderamosa isoamylase K . pneumonia pullulanase Maize endosperm BE I Maize endosperm BE I1 Potato tuber BE Rice seed BE 1 Rice seed BE 3 E. coli glycogen BE
B . subtilis a-amylase B. sphaericus cyclodextrinase P. amyloderamosa isoamylase K . pneumonia pullulanase Maize endosperm BE I Maize endosperm BE I1 Potato tuber BE Rice seed BE 1 Rice seed BE 3 E. coli glycogen BE
Region 1
Region 2
97 238 291 602 277 315 355 271 337 335
171 323 370 67 3 347 382 424 341 404 400
DAVINH DAVFNH DVWNH DVVYNH DVVHSH DWJHSH DVVHSH DVVHSH DVVHSH DWVPGH
GFRFDAAKH GWRLDVANE GFRFDLASV GFRFDLMGY GFRFDGVTS GFRFDGVTS GFRFDGITS GFRFDGVTS GFRFDGVTS ALRVDAVAS
Region 3
Region 4
204 350 412 702 402 437 453 396 459 453
261 414 499 826 470 501 545 461 524 517
FQYGEILQ IIVGEVWH RILREFTV YFFGEGWD TWAEDVS VTIGEDVS VTMAEEST TIVAEDVS ITIGEDVS VTMAEEST
LVTWVESHD SFNLLGSHD SINFIDVHD VVNYVSKBD CIAYAESHD CVTYAESHD CVTYAESHD CVTYAESHD CVTYAESHD NVFLPLNHD
Fig.2. Primary structures of various branching enzymes compared with the a-amylase family 4 conserved regions. The sequences have been derived from references cited in the text and in Svensson [33]. Four types of enzymes from the amylase family are compared with the branching enzymes. However, Svensson [33] shows comparisons of over 40 enzymes ranging from amylases, glucosidases, various a-1,6debranching enzymes as well as four cases of branching enzymes. The invariant amino acid residues are in bold letters.
265 that the conserved Asp residues of regions I1 and IV and the Glu residue of region I11 are important for BE activity. Their exact functions are unknown and further experiments such as chemical modification and analysis of the BE three-dimensional structure are required for determination of their role as catalytic residues and the detailed mechanism of reaction. Other studies with phenylglyoxal, a reagent for modification of arginine residues, have shown that the maize endosperm BEs can be inactivated (H. Cao and J. Preiss, unpublished results). The phenylglyoxal inactivation can be prevented by addition of amylose to the modification reaction mixture. Thus, arginine residues in BE may be involved in the binding of the substrate or in catalysis. Similarly, diethyl pyrocarbonate, a reagent specific for histidine residues also inactivates maize BE activity and this inactivation can also be prevented by the presence of amylose (K. Funane and J. Preiss, unpublished experiments). It should also be noted that the N- and C-terminii of the various BEs are quite dissimilar in amino acid sequence. These regions may be important with respect to substrate specificity as well as size of chain transferred and to extent of branching.
ADPGlc PPase: variation of its allosteric activator specificity with respect to its source of isolation Fructose-1,6-bis-P, fructose-6-P, pyruvate or 3-P-glycerate are the major activators for many of the ADPGlc PPases and it is possible to correlate the activator with the major carbon assimilation pathway in the organism or plant [34]. The cyanobacteria, green algae and higher plants assimilate CO, during photosynthesis to form 3-Pglycerate (3PGA) and the major major activator for the ADPGlc PPase of these oxygenic photosynthetic organisms is 3PGA [2-6,8,9]. Bacteria, such as E. coli and other enterics assimilate glucose via glycolysis and have as a major activator for their ADPGlc PPase, fructose-1,6-bis-P [3,6,7]. Regulation of the glycolytic pathway appears to be at the site of fructose-1,6-bis-P synthesis, i.e., the phosphofructokinase step. For organisms where the predominant pathway is not glycolysis, but rather the Entner-Doudoroff pathway, the activators for their ADPGlc PPase are fructose-6-P and pyruvate [3,6,34]. In this pathway fructose 1,6-bis-P is not a metabolite as glucose-6-P is converted first to 6-P-gluconate and then to 2-keto,3-deoxy,6-Pgluconate. These organisms contain no phosphofructokinase but have ample phosphoglucoisomerase. A bacterium such as Rhodospirillum rubrum cannot metabolize glucose but grows anaerobically on pyruvate or lactate or on CO,. Under these conditions high amounts of glycogen accumulate and under anaerobic photosynthesis pyruvate has been shown to be a product of CO, fixation. Of interest is that pyruvate is the sole activator of the R. rubrum. ADPGlc PPase [35]. Thus, in many cases there seems to be a correlation between the nature of the activator seen for the organism’s ADPGlc PPase and its major assimilation pathway. Consistent with this is the demonstration that Rhodobacter spheroides, a highly adaptable organism, has an ADPGlc PPase that is effectively activated either by fructose-l,6-bis-P, fructose-6-P or by pyruvate [36]. This organism can metabolize glucose by glycolysis
266 or, under other physiological conditions, by the Entner-Doudoroff pathway and it can also assimilate CO, during anaerobic photosynthesis. Because of its adaptability in carbon assimilation, utilizing under different conditions different pathways it has evolved a “flexible” ADPGlc PPase with respect to its activation specificity. It would certainly be of interest to determine the nature of the R. spheroides ADPGlc PPase allosteric binding sites and compare them with those elucidated for the E. coli, Anabaena and spinach leaf ADPGlc PPases.
Identification of allosteric binding sites of ADPGlc PPase The activator binding site of the E. coli, Anabaena and spinach leaf ADPGlc PPases have been determined using pyridoxal-phosphate (PLP) as a chemical modifying agent. PLP was found to be an activator for a number of ADPGlc PPases [3,37-421. As shown in Fig. 3, the physiological activators, fructose-1,6-bis-P, and 3-P-glycerate (3PGA) have two anionic residues, either two phosphates or a phosphate and carboxylate residue. The anionic residues of the activators most probably are important for binding to basic amino acid residues of the ADPGlc PPase at the allosteric site, i.e., basic amino acid residues are responsible for the binding of the activators. With PLP, the aldehyde residue in addition to the phosphate residue may be important for binding as it can react with the epsilon amino acid residue of a lysyl group of the enzyme. Pyridoxate-5-P, the carboxylate analog of PLP also is an effective activator of the E. coli ADPGlc PPase. Three-dimensional models show that the distance between the two phosphates in the furanose or open chain forms of fructose-1,6-bis phosphate, or between the carboxylate and phosphate residues of 3PGA are approximately equal to the distance from the aldehyde to the phosphate group in pyridoxal-P. In the case of fructose-1,6-bis-P it is most probably the open chain form that is the active structure. This is based on the fact that 1,6-hexanediol-bis-P is just as active as fructose-1,6-bis-P in stimulating the enzyme activity (-3O-fold), but only 5 pM is required for half-maximal stimulation as compared to the 68 pM required of FBP. Only about 5% of the FBP is in open chain form and this may be the reason that a CHzOPO,-B 0
-3
-
b H,OPO,-H
l c=o I HO-CH I H C-OH
H+
cooI I
H C V H CH20P03-B
I
HC-OH I
CH,OPO,-B
Pyridoxal-P
Fructose-1,6-bis-P
3-P-Glycerate
Fig. 3. Structures of fructose- 1,6-bis-phosphate, pyridoxal-5-phosphate and 3-phosphoglycerate.
267 17-fold higher concentration of FBP is required for half-maximal stimulation. If reductive phosphopyridoxylation occurs at the activator binding site then it would be expected that the modified enzyme would have either no requirement or a lesser requirement of activator for maximal activity. Moreover, the modification should be prevented by either inclusion of activator or inhibitor during the modification process. Both of these effects have been seen for the E. coli,Anabaena and spinach leaf ADPGlc PPases. These studies showed that K39 in the E. coli ADPGlc PPase [39,40], K440 in the small subunit of the spinach leaf enzyme [41] and K419 of the Anabaena enzyme [42] were involved in the binding of their respective activators. Site-directed mutageneses of these sites in the Anabaena [42] and in the E. coli enzymes [43] showed that the positive charge of the lysyl residue being the most effective amino acid with respect to the apparent affinity of the activator. For example, site-directed mutagenesis of the Anabaena ADPGlc PPase Lys419 residue to mutants Arg419 or to Ala419 [42] resulted in ADPGlc PPases with a 25- and 150-fold less apparent affinity for 3PGA, respectively. The effect was specific as the Ki for Pi, the inhibitor, or the Km for the substrates, ATP or glucose1-P are not affected to any great extent. However, the mutant enzymes are still activated by 3PGA and PLP providing the concentrations were high enough. This suggested that an additional binding site for the activator was also present. Reductive phosphopyridoxylation of the K419R mutant produced an enzyme less dependent on the presence of activator and the chemically modified lysine was K382 [42]. Thus, two activator binding site regions for the cyanobacterial ADPGlc PPase have been elucidated and these results are in agreement with the findings of the activator sites found for the spinach leaf ADPGlc PPase [41,44]. As seen in Fig. 4, the spinach leaf, Anahaena and potato tuber amino acid sequences are nearly identical and these sequences are highly conserved in almost all of the higher plant ADPGlc PPases [9,10]. Whereas both activator sites are on the same peptide for the Anahaena
Spinach 51 kDa (small) Potato 50 kDa (small) Spinach 54 kDa(large) Anahaena Synechocystis Escherichia coli Salmonella typhimurium
Activator site 1
Activator site 2
SGIVTVIKDALIPSGTVI SGIVTVIKDALIPSGIII SGITVIFKQATIKDGW SGIVWLECNAVITDGTI I NGIVWIKNVTIADGTVI RLKDLTNKRAKPAVHFGG RLKDLANKRAKPAVHFGG
IKRAIIDKNAR IKRAI IDKNAR IKDAIIDECNAR QRRAIIDKNAR IRRAIIDKNAR -
Fig. 4. Comparison of plant and bacterial ADPglucose pyrophosphorylase activator binding sites. The
sequences are listed in one letter code and the plant sequences were taken from Smith-White and Preiss, (10). The Lys residues that are in bold indicate that studies have shown they are covalently modified by pyridoxal-P and the chemical modification of the Lys residue is prevented by 3PGA and Pi, or that sitedirected mutagenesis experiments have identified their involvement in the binding of the activator. Athough the sequences of sites 1 and 2 are almost identical in the large and small subunits of the plant ADPGlc PPases, only site 1 of the small subunit and site 2 of the large subunit are involved in the binding of the activator. The bacterial enzymes have only one subunit and are homotetrameric. To date, only one activator binding site has been identified for the E. coli enzyme.
268 enzyme, in the spinach leaf enzyme one activator site is present in the small subunit (equivalent to Anabaena Lys419) while the other is present in the large subunit (equivalent to Anabaena Lys382). It should be noted that the activator binding site of the E. coli ADPGlc PPase (and presumably, the activator site of Salmonella typhimurium) is near the N-terminal portion of the amino acid sequence [6,45] while for the higher plant and oxygenic photosynthetic ADPGlc PPases the activator binding site is close to the C-terminal region of the sequence. Thus even though there are relationships between the enzymes with respect to the substrates binding sites and their relative positions in the subunits, it appears that the relative positions of the activator binding sites have changed.
ADPGlc PPase allosteric mutants Chemical mutagenesis with nitrosoguanidine has led to a class of mutants that store glycogen either at a faster or a slower rate than the wild-type strain and where the mutation affects the allosteric properties of the ADPGlc pyrophosphorylase. Table 1 shows that mutants SG5, CLll36 and 618 have about 2-4 times the glycogen synthetic rate than the parent wild-type strain. Correlated with the increase in glycogen synthetic rate is a higher apparent affinity for the activator and a decreased affinity for the inhibitor. In contrast, mutant SG14 synthesizes glycogen at about 30% the rate observed for the WT organism and has about a 12-fold lower apparent affinity for the activator. Not indicated in Table 1 is that the mutants have substantial activity in the absence of activator, in contrast to that observed for the WT enzyme. The WT enzyme has only 3% of its maximal activity in the absence of activator. However, the SG5 mutant enzyme has 12% of its maximal activity and the CL1136 and 618 mutant enzymes have about 65% of their maximal activities in absence of activator. The specific activities of these mutant enzymes when purified are the same as the wildtype enzyme (-100 pmol ADPGlc/min/mg). Thus, the mutant enzymes, particularly Table I. Regulatory kinetic constants of the ADPGlc PPases of allosteric mutants of E. coli. Also listed are the relative rates of glycogen synthesis of the organisms in stationary phase with the wild-type strain taken as 100%. The amino acid substitution due to mutation is also indicated for the mutant ADPGlc PPase. Strainlmutation
,
A, Fru- 1,6-bisP (PM)
I,, AMP
(PM)
Relative glycogen accumulation rate (%)
E. coli wild-type
68
75
100
SG14, Ala44Thr
820
500
28
5
680
370
SG5, Pro295Ser
22
170
175
618, Gly336Asp
15
860
350
CLI 136, Arg67Cys
269 those of CL1136 and 618, do not require activator for substantial activity as compared to the WT enzyme. The ADPGlc PPase of a number of allosteric mutants have been cloned [7,45-501 and the amino acid changes due to the mutations have been elucidated. In almost all cases these mutations occur at sites different from the allosteric and substrate binding sites. Thus the regions where the mutations have occurred may be involved in maintaining the conformation of the inactive WT enzyme. Figure 5 shows the sequences surrounding the affected amino acids in the allosteric mutants characterized to date and the effects of the amino acid change on the allosteric kinetics. Allosteric mutants of S. typhimurium ADPGlc PPase [51] and of the green alga, Chlamydomonas reinhardtii ADPGlc PPase [52] have been isolated and reported. The Salmonella mutants were overproducers of glycogen and one mutant contained an ADPGlc PPase not requiring activator for maximal activity and was less sensitive to AMP inhibition than the WT strain. The other S. typhimurium mutant had an ADPGlc PPase with the same affinity for the activator but had 4.5-fold less affinity for the inhibitor than the parent strain [51].
Evidence that the allosteric effects seen for the algal and higher plant ADPGlc PPases are functional in the in vivo regulation of starch synthesis There are some experiments [53,54] that employed the Kacser-Bums control analysis method [55,56] to determine if certain enzymes in the pathway towards starch synthesis were rate-limiting. The control analysis method is where the activity of an enzyme is varied, either by using mutants deficient in that enzyme or by varying the physiological conditions, and the effect of these changes on the rate of a metabolic process (e.g.. starch synthesis) is measured. If the enzyme activity is rate-limiting or regulatory of the metabolic process, then a large effect on that process should be seen. Conversely, if there is no or little effect, then the enzyme level of activity is not considered to be rate-limiting for the metabolic process being measured. In a Kacser-Bums analysis experiment, it was shown that in Arabidopsis thaliana the ADPGlc PPase is a major site of regulation for starch synthesis [53] and that
,
WT Mutants
WT Mutants
27 30 35 40 45 61 65 70 GGRGTRLKDLTNKRAKPAV----CINSGIRRYKAE-----GGRGTRLKDLTNKRAKPTV----CINSGICRYKAE-----SG14 CL1136 290 295 330 335 340 LASWPELDMY---LNSLVSGGCVI LASWSELDMY---LNSLVSMCVI SG5 618
Fig.5. Amino acid sequences of the E. coli ADPGlc PPase showing changes occumng in the different allosteric mutant enzymes. The top line shows the sequence of the wild-type enzyme and the bottom line shows the amino acid substitutions occitmng in the mutant strains.
270 regulation of the enzyme by 3PGA is an important determinant of the rate of starch synthesis in vivo [54]. A. thaliana mutant strains containing only 7% of the normal activity of ADPGlc PPase and a hybrid strain between the mutant and normal strain having 50%, had 90% and 39% reduction in the starch synthetic rate respectively, as compared to the wild-type [53]. In experiments using a mutant of Clarkia xantiana deficient in leaf cytosolic phosphoglucoseisomerase (having only 18% of the activity seen in the wild-type) sucrose synthetic rates were lower and the rate of starch synthesis increased [54]. The chloroplastic concentration of 3PGA increased about 2fold, suggesting that the increase of starch synthetic rate measured in the mutant deficient in cytosolic phosphoglucoseisomerase is due to activation of the ADPGlc PPase by the increased 3PGA concentration and the 3PGAPi ratio. Table 2 lists the various flux control coefficients obtained for various enzymes in the pathway towards starch synthesis. The enzymes that could be studied were those where variation of activity was achieved via mutants of ADPglucose pyrophosphorylase [53,57], chloroplastic phosphoglucomutase [58], branching enzyme [59,60] and chloroplastic phosphoglucoseisomerase [61].The values were obtained at high light where photosynthesis and starch synthesis were at a maximum for the normal plant and at low light where the photosynthetic and starch synthetic rates were lower. In low light very little effects are seen with variation of enzyme activity and the variation of starch synthetic rate. Most probably this is due to the lower rate of CO, fixation and the preferential use of this carbon for other processes such as sucrose biosynthesis. Some small but a significant control coefficient is, however, seen with ADPGlc PPase. In high light where higher CO, fixation occurs and there is greater starch synthesis a very small flux control coefficient effect is seen with BE and this will be discussed later on. The largest and most significant control coefficient is seen with ADPGlc PPase thus indicating the enzyme activity is most rate-limiting and exerts major control. It is important to note that the ADPGlc PPase flux control coefficient may also be underestimated because the of the allosteric properties of the enzyme. In flux control analysis the maximal enzyme activity is measured. In the case of an allosteric enzyme the potential maximal enzyme activity may not be as important as the allosteric effector concentrations that affect the enzyme activity. Thus one may not see with an allosteric enzyme a meaningful flux control coefficient that is just based Table 2. Estimated control coefficients of some enzymes involved in starch synthesis.
Enzyme
Flux control coefficient
Low light
High light
Chloroplast P-glucoseisomerase (Clarkia xantiana)
0.0
0.35
Chloroplast P-glucomutase (A. thaliana)
0.01
0.21
ADPGlc PPase (A. thaliana)
0.28
0.64
Branching enzyme (pea seed)
0.02
0.13
27 1 on potential maximum activity. With ADPGlc PPase we can have an activation by the 3PGA with various ADPGlc PPases anywhere from 10- to 100-fold. Moreover, there is also inhibition by the allosteric inhibitor, Pi. Thus, a flux coefficient control value based on only the potential maximal activities of the A. thaliana mutants and normal ADPGlc PPases can underestimate the regulatory potential of the ADPGlc PPase step. A significant finding was made by Ball et al. [52] who isolated a starch-deficient mutant of C. reinhardtii in which the defect was shown to be in the ADPGlc PPase, which could not be effectively activated by 3PGA. The enzyme showed poor activation with the activator, 3-PGA [52]. Recently, another putative ADPGlc PPase allosteric mutant, isolated from a mutant maize endosperm, which had 15% more dry weight than the normal endosperm, has been described [62]. The mutant allosteric ADPGlc PPase was less sensitive to Pi inhibition than the normal enzyme. Thus, the Chlamydomonas starch-deficient mutant and higher dry weight maize endosperm mutant studies strongly suggest that the in vitro regulatory effects observed with the photosynthetic and nonphotosynthetic plant ADPGlc PPases are highly functional in vivo and that ADPGlc synthesis is ratelimiting for starch synthesis. In short, data continue to accumulate showing the importance of the plant ADPGlc PPase in the regulation of starch synthesis, and that 3PGA and Pi are important allosteric effectors in vivo, in photosynthetic as well as in nonphotosynthetic plant tissue.
Are other starch biosynthetic enzymes rate-limiting? Can starch synthase and branching enzyme also be rate-limiting under certain situations? The wrinkled pea has a reduced starch content; about 6 6 7 5 % of that seen in the round seed, and whereas the amylose content is about 33% in the round form it is 60-70% in the wrinkled pea seed. Edwards et al. [63] measured the activities of several enzymes involved in starch metabolism in wrinkled pea at four different developmental stages. In this variety it was found that branching enzyme activity was, at its highest, only 14% of that seen for the round seed. The other starch biosynthetic enzymes and phosphorylase had similar activities in the wrinkled and round seeds. These results were confirmed by Smith [59] who also showed that the r (rugosus) lesion (as found in the wrinkled pea of genotype rr) was associated with the absence of one isoform of branching enzyme. Edwards et al. [63] proposed that the reduction in starch content observed in the mutant seeds is caused indirectly by the reduction in BE activity through an effect on the starch synthase. The authors suggested that, in the absence of branching enzyme activity, the starch synthase forms an a-1-Aglucosyl elongated chain which is a poor glucosyl-acceptor (primer) for the starch synthase substrate, ADPGlc, therefore decreasing the rate of a-1->Cglucan synthesis. Indeed, in a study of rabbit muscle glycogen synthase [64]it was found that continuous elongation of the outer chains of glycogen caused it to become an
272 ineffective primer, thus decreasing the apparent activity of the glycogen synthase. The observation that ADPglucose in the wrinkled pea accumulated to higher concentrations than in the round or normal pea, was considered evidence that activity of the starch synthase was restricted in vivo. Under optimal in vitro conditions, in which a suitable primer like amylopectin or glycogen is added, starch synthase activity in the wrinkled pea was equivalent to that found in the wild-type. Smith [59] showed that in mutant rr leaves, in high light intensity, there was a 40% decrease in the rate of starch synthesis. Control coefficient analysis reported later [60] showed that in low light intensity there was essentially no effect on the rate of starch synthesis while in high light intensity the flux control coefficient value was 0.13, a small value (meaning very little control) and only one-fifth the value seen for ADPGlc PPase [52]. Thus a 86% reduction of branching enzyme activity had a small effect on regulation of starch synthesis. It has been suggested that when plants are subjected to high temperatures, starch synthase activity may be rate-limiting [65-691. At temperatures above 30°C both maize [69] and wheat endosperm [65-681 had a reduction of starch deposition compared to lower temperatures. In wheat the starch biosynthetic enzyme affected was soluble starch synthase [65-681. Using flux control coefficient analysis, Keeling et al. [65] showed a control coefficient close to one between rate of starch synthesis and the level of starch synthase activity in the wheat endosperm extracts. It was also shown that in vitro, the endosperm starch synthase activity was sensitive to heat treatment in the range of 3CF40"C if the treatment was for longer than 15 min. A similar study was done with maize endosperm and there was a reduction of starch synthetic rate and a decrease in starch synthase activity in the heat-stressed maize endosperm [69]. It was also noted, however, that in the heat-stressed maize, the endosperm ADPGlc PPase activity was also reduced and even to a greater extent than the soluble starch synthase [69]. Thus, in wheat and maize there may be a relationship, under some environmental conditions, between reduction of starch synthase activity and decreased starch synthesis. However, as the maize data suggests [69], other factors besides starch synthase activity, not yet studied, may be the primary reason for the reduction of starch synthesis in the heat-stressed plants. In the case of maize endosperm another enzyme involved in starch synthesis, ADPGlc PPase, is also affected in the heat-stressed plant. It is also quite possible that other critical steps leading to starch biosynthesis are also affected in both plants, such as carbon flow (sucrose transport?) from source to sink tissues. Those processes were not studied in the heat-stressed plants. Thus, I believe it is premature on the basis of the published experiments to designate starch synthase as a major control point and other data should be obtained. Flux control coefficients for an enzyme can only be determined for any process if only that enzyme's activity is affected. In the case of the heat-stressed plants it has not yet been shown that only the starch synthase activity is affected. A crucial question is whether starch synthetic rate can be increased by overexpressing soluble starch synthase activity in the amyloplast? As will be discussed later, starch accumulation can be increased by expressing a bacterial ADPGlc PPase allosteric mutant in plants [70].
273
Transformation of plants with an E. coli allosteric mutant glg C gene increases starch content As previously discussed, there is a preponderance of evidence indicating that the ratelimiting and regulatory enzyme of starch synthesis in algae or bacterial glycogen synthesis is ADPGlc PPase [2--9,521. With respect to higher plants, control analysis experiments have shown that ADPGlc PPase is important in regulation of leaf starch synthesis [53]. Also, reduced ADPGlc PPase activity in the mutants led to a reduction in the rate of starch synthesis in Arubidopsis leaves [57] as well as in potato tubers [71]. Therefore, it was of interest to see if starch content in a plant could be augmented by increased expression of activity of one of the enzymes involved in starch biosynthesis. Overexpression of a plant ADPGlc PPase activity, however, would require expression of two distinct genes to reconstitute its ADPGlc PPase activity. Moreover, it is possible the plant would compensate for the overexpression by altering the ratio of the effector metabolites, 3PGA and Pi, so that starch synthesis would not be elevated. Thus, a different strategy was chosen: an E. coli ADPGlc PPase, glg C gene of allosteric mutant 618, referred to as G l g C16 [46], which encodes for an enzyme independent of the presence of an activator for activity, was used for the transformation. Expression of the bacterial mutant gene would have two advantages. First, only one gene has to be expressed for ADPGlc PPase activity and, second, the mutant enzyme would be less sensitive to inhibition by its allosteric inhibitor, S’AMP, insensitive to the plant enzyme’s inhibitor, Pi, and independent of activator for good activity (Table 1 and [46]). Thus, a collaboration with the Monsanto group was initiated to transfect plant systems with G l g C16 to see if the starch content of plants could be increased [70]. Starch synthesis occurs in the plastid and therefore a nucleotide sequence encoding transit peptide of the Arubidopsis ribulose 1,5-bisphosphate carboxylase chloroplast transit peptide was fused to the translation initiation site of the glg C16 gene (Fig. 6). The chimeric gene was then cloned behind either a cauliflower mosaic virus (CaMV) enhanced 35s promoter or a tuber-specific patatin promoter or, in the case of tomato plants, the Arubidopsis plant promoter from the rbcS gene ([70] Fig. 6). A polyadenylation signal from the nopaline synthase gene (Nos) was fused on at the 3‘ end of the chimeric gene. The chimeric gene containing promoter was placed in a cloning vector with a 35sneomycin phosphotransferase gene as a selectable marker [70] and used for transformation of tobacco calli, tomato cotyledons and potato plants [70]. The starch levels were increased over the controls lacking the glgC gene product by about 1.7 to 8.7-fold in tobacco calli where the glgC gene product activity was detected [70]. The CaMV-chimeric gene was electroporated into tobacco protoplasts and, as shown in Table 3, extracts of the transformed protoplasts gave rise to ADPGlc synthesis resistant to Pi inhibition and activated by fructose 1,6-bis-P. ADPGlc synthesis in the control protoplast extract was totally inhibited by Pi as expected since the tobacco and almost all plant ADPGlc PPases are most sensitive to inhibition by Pi. Examination and comparison via the light microscope of transgenic tobacco with control calli, show a very large increase in the number of starch granules [70].
274 Cleavage site
Promoter
Arabidopsis Ribulose bis-P carboxylase small subunit transit peptide
Additional cleavage site
23 amino acids of ADPGlc N-terminus of Ribulose-bis-P PPase carboxylase small subunit g/g C16 transit peptide gene
Nos Terminator
Fig. 6. Construction of the synthetic promoter-plastid transit peptide-glg C 16 ADPGlc PPase gene. The
chimeric gene contains the Arabidopsis thaliana chloroplast transit peptide portion of the ribulose bis-P carboxylase gene modified to contain a duplicated cleavage site to eliminate the 23 amino acids of the N-terminal of the small subunit (70) so it would not interfere with the catalytic or regulatory activity of the g/g C16 gene product. The Nos terminator is the nopaline synthase 3' poly A signal. The promoter can be. either a constitutive promoter such as e35S or a tissue specific one as patatin which in potato is tuber specific.
Similarly, tomato shoots excised from the transformed calli containing the transit peptide-Glg C16 gene stained black with I, while the controls were essentially negative [70]. Similar results have been obtained for Russet-Burbank potato tubers where the chimeric gene with transit peptide under control of a tuber-specific patatin promoter, increased starch in the tuber 2 5 6 0 % over controls not containing the bacterial enzyme ([70]; Table 4).If the bacterial ADPGlc PPase Glg C16 gene was expressed in the tuber lacking the transit peptide gene portion, no increase in starch content was noted (Table 4). Most probably the ADPGlc PPase which was expressed was not present in the amyloplast and was not able to supply ADPGlc to the starch synthases which are localized in the amyloplast. If the wild-type E. coli glg C gene was used for transformation, no increase in starch was noted [70]. This suggested that the allosteric properties of the ADPGlc PPase were important for regulation and that only alteration of the allosteric properties was necessary to increase starch levels. Some relationship between the expression levels of the ADPGlc PPase of G l g C 16 Table 3. Activation and inhibition of ADPGlc PPase activity in tobacco protoplasts transformed with glg C16". Protoplast extracts from and conditions
ADPGlc formed (nmol)
Nontransformed cells, + 10 mM inorganic phosphate Transformed + 2.5 mM fructose 1.6-bis-P Transformed + 2.5 mM fructose 1,6-bis-P + 10 mM inorganic phosphate Transformed + 20 mM + 3-P-glycerate Transformed + 10 mM inorganic phosphate
0.0 20.2 18.0 18.4 6.4
"Data from [70].
275 Tubk 4. Starch content in potato tubers transformed with the glg C16 and glg C genes’. Transformation with:
Average starch content, % wet wt.
A. Control; untransformed Chlorplast transit peptide-glg C16 glg C16, no transit peptide
12.3 f 1.15 16.0 f 2.00 12.4 f 0.24
B. Control; untransformed Chlorplast transit peptide-glg C
13.2 f 0.12 13.1 f 0.07
“Datafrom [70].
as measured by Western blotting of the potato extracts and the increase in starch content was demonstrated. This was noted particularly in tubers at the lower range of starch content [70]. Lower levels of the expressed ADPGlc PPase resulted in increases of 21-63% increase in starch, intermediate levels of the expressed ADPGlc PPase gave increases of 33--118% in starch and the high expressed levels of the transit peptide-Gfg C16 resulted in increases of 33-167%. It is of interest to note that when the wild-type E. cofi ADPGlc PPase gene was expressed in the tuber no increase in starch was noted ([70]; Table 4). Thus, an important factor in increasing starch synthesis is to transform the tuber with an ADPGlc PPase with allosteric properties minimized to permit higher rates of ADPGlc synthesis under physiological conditions. These results indicate that the bacterial enzyme can be expressed in plant tissues and stimulate greater starch production. These results also strongly suggest that the ADPGlc PPase is a rate-limiting enzyme for starch synthesis even in nonphotosynthetic plant tissues. The data obtained by the Monsanto group also indicate that transfection of a plant with a bacterial ADPglucose PPase increases the starch content of an important crop product. Further studies are currently being carried out to study the relationship between increased enzyme activity (due to the transformation of the bacterial gene) and rate of starch synthesis by feeding labeled glucose or sucrose to potato slices. The possibility of other genes involved in starch metabolism being indirectly affected by the Gfg C 16 transformation is also being investigated. Preliminary results indicate that correlation exists between increased allosteric mutant enzyme expression and increased rate of starch synthesis. Therefore it is conceivable that similar methods can be used to change in addition ‘to quantity, starch quality via expression/transformation of the isoforms of starch synthase and branching enzymes in plants. These “new starches” may have greater usefulness in food and industrial processes. The production of modified “specialty” starches via molecular biology techniques is promising, and perhaps more beneficial and more economical than the chemical production of modified starch. Since there has been an increased demand for starch in the past decade [72] for both specialized industrial and food uses, it appears that the study of basic questions on the structure-function relationships of the allosteric regulation of an enzyme involved in sugar nucleotide synthesis now may have a great impact on both
276 agriculture and industry. It is of interest that this research, which started about 30 years ago to study the routes and mechanism of regulation of bacterial glycogen and starch synthesis at the molecular level, has lead to opportunities to improve the quality of uses of starch in industrial and food processes. This was never the original purpose of the studies but is an example of how basic science which tries to answer basic questions, may lead to methods where nature can be manipulated for beneficial purposes.
Acknowledgements The author is indebted to the support that he has obtained from various governmental agencies (USPHS AI 05520 and A1 22835; NSF 78-16127, 82-05705,85-10088 and 86-10319; DOE FG02-93ER20121 and USDA 9301525 and 9501085) in the past and present for the research studies cited in this review.
References 1. Espada J. Enzymic synthesis of adenosine diphosphate glucose from glucose 1-phosphate and adenosine mphosphate. J Biol Chem 1962;237:3577-3581. 2. Preiss J. Regulation of the biosynthesis and degradation of starch. Ann Rev Plant Physiol 1982;54:431-454. 3. Preiss J. Bacterial glycogen synthesis and its regulation. AM Rev Microbiol 1984;38:419-458. 4. Preiss J. Biosynthesis of starch and its regulation. In: Preiss J (ed) The Biochemistry of Plants, 14. New York: Academic Press, 1988;181-254. 5 . Preiss J. Starch biosynthesis and its regulation. In: Miflin J (ed) Surveys of Plant Molecular and Cell Biology. Oxford: Oxford University Press, 1991;7:59-114. 6. Preiss J, Romeo T. Physiology, biochemistry and genetics of bacterial glycogen synthesis. Adv Microbiol Physiol 1989;30:83-238. 7. Preiss J, Romeo T. Molecular biology and regulatory aspects of glycogen biosynthesis in bacteria. Prog Nucl Acid Res Molec Biol 1994;47:299-329. 8. Sivak MN, Preiss J. Starch synthesis in seeds. In: Kigel J, Galili G (eds) Seed Development and Germination. New York: Marcel Dekker, Inc., 1994;139-168. 9. Preiss J, Sivak MN. Starch synthesis in sinks and sources. In: Zamski E, Schaffer AA (eds) Photoassimilate Distribution in Plants and Crops: Source-Sink Relationships. New York: Marcel Dekker, Inc., 1996;(In press). 10. Smith-White BJ, Preiss J. Comparison of proteins of ADP-glucose pyrophosphorylase from diverse sources. J Molec Evol 1992;34:449-464. 11. Ballicora MA, Laughlin MJ, Fu Y, Okita TW,Barry GF, Preiss J. ADPglucose from potato tuber. Significance of the N-terminal of the small subunit for catalytic properties and heat stability. Plant Physiol 1995;109:245-251. 12. Kumar A, Larsen CE, Preiss J. Biosynthesis of bacterial glycogen. Primary sbucture of Escherichia coli ADPglucose: a-1,4-glucan, 4-glucosyltransferasededuced from the nucleotide sequence of the glg A gene. J Biol Chem 1986;261:14634-14639. 13. Mu C, Ham C, KO Y-T, Singletary GW, Keeling PL, Wasserman BP. Plant J 1994;6:151-159. 14. Denyer K, Sidebottom C, Hylton CM, Smith AM. Soluble isoforrns of starch synthase and starch branching enzyme also occur within starch granules in developing pea embryos. Plant J 1993;4:191196.
277 15. Delrue B, Fontaine T, Routier F, Decq A, Wieruszeski J-M, Van Den Koomhuyse N, Maddelein
16.
17.
18. 19.
20.
M-L, Foumet B, Ball S. Waxy Chlamydomonas reinhurdrii: monocellular algal mutants defective in amylose synthesis and granule-bound starch synthase activity accumulate a structurally modified amylopectin. J Bacteriol 1992;174:3612-3620. Fontaine T, D’Hulst C, Maddelein M-L, Routier F, Pepin TM, Decq A, Wieruszeski J-M, Delrue B, Van Den Koomhuyse N, Bossu J-P, Foumet B, Ball S. Toward an understanding of the biogenesis of the starch granule. Evidence that Chlamydomonas soluble starch synthase I1 controls the synthesis of intermediate size glucans of amylopectin. J Biol Chem 1993;268:16223-16230. Maddelein M-L, Libessart N, Bellanger F, Delrue B, D’Hulst C, Van Den Koomhuyse N, Fontaine T, Wieruszeski J-M, TM, Decq A, Ball S. Toward an understanding of the biogenesis of the starch granule. Determination of granule-bound and soluble starch synthase functions in amylopectin synthesis. J Biol Chem 1994;269:2515*255157. Hizukuri S. Polymodal distribution of the chain lengths of amylopectins, and its significance. Carbohydr Res 1986;147:342-347. Furukawa K, Tagaya M, Inouye M, Preiss J, Fukui T. Identification of Lysine 15 at the active site in Escherichia coli glycogen synthase. Conservation of a Lys-X-Gly-Gly sequence in the bacterial and mammalian enzymes. J Biol Chem 1990265:208&2090. Furukawa K, Tagaya M, Tanizawa K, Fukui T. Role of the conserved Lys-X-Gly-Gly sequence at the ADP-glucose-binding site in Escherichia coli glycogen synthase. J Biol Chem 1993;268:23837-
23842. 21. Baba T, Nishihara M, Mizuno K, Kawasaki T, Shimada H, Kobayashi E, Ohnishi S, Tanaka K, Arai
22. 23. 24. 25. 26. 27. 28.
29. 30. 31. 32.
33. 34.
Y. Identification,cDNA cloning and gene expression of soluble starch synthase in rice (Oryza sutivu L.) immature seeds. Plant Physiol 1993;103:565-573. Furukawa K, Tagaya M, Tanizawa K, Fukui T. Identification of Lys2” at the active site of Escherichia coli glycogen synthase. J Biol Chem 1994;269:868-871. Guan HP, Baba T, Preiss J. Expression of branching enzyme I of maize endosperm in Escherichia coli. Plant Physiol 1994;104:1449-1453. Guan HP, Baba T, Preiss J. Expression of branching enzyme II of maize endosperm in Escherichia coli. Cell Molec Biol 1994;40981-987. Fisher DK, Boyer CD, Hannah LC. Starch branching enzyme II from maize endosperm. Plant Physiol 1993;102:1053-1054. Mizuno K, Kimura K, Arai Y, Kawasaki T, Shimada H, Baba T. Starch branching enzymes from immature rice seeds. J Biochem 1992;112:643-651. Nakamura Y, Yamanouchi H. Nucleotide sequence of a cDNA encoding starch branching enzyme or Q-enzyme I from rice endosperm. Plant Physiol 1992;99:1265-1266. Mizuno K, Kawasaki T, Shimada H, Satoh H, Kobayashi E, Okamura S, Arai Y, Baba T. Alteration of the structural properties of starch components by the lack of an isoform of starch branching enzyme in rice seeds. J Biol Chem 1993;268:19084-19091. Takeda Y, Guan HP, Preiss J. Branching of amylose by the branching isoenzymes of maize endosperm. Carbohydr Res 1993;240:253-263. Guan H-P, Preiss J. Differentiation of the properties of the branching isozymes from maize endosperm. Plant Physiol 1993;102;1269-1273. Boyer CD, Preiss J. Evidence for independent genetic control of the multiple forms of maize endosperm branching enzymes and starch synthases. Plant Physiol 1981;67:1141-1145. Guan HP, Kuriki T, Sivak MN, Preiss J. Maize branching enzyme catalyzes the synthesis of glycogen-like polysaccharide in glgB deficient Escherichia coli. Proc Natl Acad Sci USA 1995;92;964-967. Svensson B. Protein engineering in the a-amylase family: catalytic mechanism, substrate specificity and stability. Plant Molec Biol 1994;25:141-157. Preiss J. The regulation of the biosynthesis of a-1,4glucans in bacteria and plants. In: Stadtmen ER, Horecker BL, (eds) Current Topics of Cellular Regulation, vol 1. New York: Academic Press, 1969;125- 161.
278 35.
36.
37. 38.
39.
40.
41. 42.
43.
44. 45.
46.
47.
48.
49.
50. 51. 52.
53.
Furlong CE, Preiss J. Biosynthesis of bacterial glycogen synthesis VII. Purification and properties of adenosine diphosphoglucose pyrophosphorylase of Rhodospirillum rubrum. J Biol Chem 1969;244:253!&2548. Yung S-G, Preiss J. Biosynthesis of bacterial glycogen: purification and structural and immunological properties of Rhodopseudomonas sphaeroides ADPglucose synthetase. J Bacteriol 1982;151:742-749. Gentner N, Greenberg E, Preiss J. TPNH and pyridoxal-5’-phosphate: Activators of ADP-glucose pyrophosphorylase of Escherichia coli B. Biochem Biophys Res Commun 1969;36:373-380. Haugen T, Ishaque A, Preiss J. ADP-glucose pyrophosphorylase: evidence for a lysine residue at the activator site of the Escherichia coli B enzyme. Biochem Biophys Res Commun 1976;69:346353. Parsons TF, Preiss J. Biosynthesis of bacterial glycogen. Incorporation of pyridoxal phosphate into the allosteric activator site and an ADP-glucose-protected pyridoxal phosphate binding site of Escherichia coli B ADP-glucose synthase. J Biol Chem 1978;253;6197-6202. Parsons TF, Preiss J. Biosynthesis of bacterial glycogen. Isolation and characterization of the pyridoxal-P allosteric activator site and the ADP-glucose protected pyridoxal-P binding site of Escherichia coli B ADP-glucose synthase. J Biol Chem 1978;253:7638-7645. Morel1 M, Bloom M, Preiss J. Affinity labeling of the allosteric activator site(s) of spinach leaf ADPglucose pyrophosphorylase. J Biol Chem 1988;263:633-637. Chamg Y-Y, Iglesias AA, Preiss J. Structure-function relationships of cyanobacterial ADP-glucose pyrophosphorylase: site-directed mutagenesis and chemical modification of the activator-binding sites of ADP-glucose pyrophosphorylase from Anubaenu PCC 7120. J Biol Chem 1994;269:24107241 13. Gardiol A, Preiss J. Escherichia coli E-39 ADPglucose synthetase has different activation kinetics from the wild-type allosteric enzyme. Arch Biochem Biophys 1990;280:175-180. Ball KL, Preiss J. Allosteric sites of the large subunit of the spinach leaf adenosine diphosphate glucose pyrophosphorylase. J Biol Chem 1994;269:247&247 11. Leung P, Preiss J. Biosynthesis of bacterial glycogen. Primary structure of Salmonella typhimurium ADPglucose synthetase as deduced from the nucleotide sequence of the glg C gene. J Bacteriol 1987;169:4355-4360. Leung P, Lee YM, Greenberg E, Esch K, Boylan S, Preiss J. Cloning and expression of the Escherichia coli glg C gene from a mutant containing an ADPglucose pyrophosphorylase with altered allosteric properties. J Bacteriol 1986;167:82-88. Kumar A, Ghosh P, Lee YM, Hill MA, Preiss J. Biosynthesis of bacterial glycogen: determination of the amino acid changes that alter the regulatory properties of a mutant Escherichia coli ADPglucose synthetase. J Biol Chem 1989;264:10464-10471. Ghosh P, Meyer C, Remy E, Peterson D, Preiss J. Biosynthesis of bacterial glycogen: cloning, expression and nucleotide sequence of glg C gene from an allosteric mutant of Escherichia coli B. Arch Biochem Biophys 1992;296:122-128. Meyer CR, Ghosh P, Remy E, Preiss J. Cloning, expression and nucleotide sequence of a mutant glg C gene from Escherichia coli B; substitution of a proline to serine at position 295 results in altered allosteric properties of ADPGlc synthetase. J Bacteriol 1992;17434509-45 12. Meyer CR, Ghosh P, Nadler S, Preiss J. Cloning, expression and sequence of an allosteric mutant ADPglucose pyrophosphorylase from E. coli B. Arch Biochem Biophys 1993;302:64-7 1. Steiner KE, Preiss J. Biosynthesis of bacterial glycogen genetic and allosteric regulation of glycogen biosynthesis in Salmonella typhimurium LT.2. J Bacteriol 1977;129:246-253. Ball S, Marianne T, Dirick L, Fresnoy M, Delrue B, Decq A. A Chlamydomonus reinhardtii lowstarch mutant is defective for 3-phosphoglycerate activation and orthophosphate inhibition of ADPglucose pyrophosphorylase. Planta 1991;185:17-26. Neuhaus HE, Stitt M. Control analysis of photosynthate partitioning: impact of reduced activity of ADPglucose pyrophosphorylase activity or plastid phosphoglucomutase on the fluxes to starch and sucrose in Arabidopsis. Planta 1990;182A4-454.
279 54. Neuhaus HE, Kruckeberg AL, Feil R, Stitt M. Reduced activity mutants of phosphoglucoseisomerase
55.
56. 57.
58.
59. 60. 61.
62. 63.
64. 65. 66. 67. 68. 69. 70. 71.
72.
in the cytosol and chloroplast of Clurkiu xunrhiuna II. Study of the mechanisms which regulate photosynthate partitioning. Planta 1989;178:1l(t122. Kacser H. Control of metabolism. In: Davies DD (ed) The Biochemistry of Plants, vol 11. New York Academic Press, Inc., 1987;3+67. Kacser H, Bums JA. Control of flux. Symp SOCExptl Biol 1973;27:65-107. Lin TP, Caspar T, Somerville C, Preiss J. A starch deficient mutant of Arubidopsis thuliunu with low ADPglucose activity lacks one of the two sununits of the enzyme. Plant Physiol 1988;88:11751181. Caspar C, Huber SC, Somerville C. Alterations in growth, photosynthesis and respiration in a starch mutant of Arubidopsis thuliunu (L.) Heynh deficient in chloroplast phosphoglucomutase activity. Plant Physiol 1986;79:1-7. Smith AM. Major differences in isoforms of starch branching enzyme between developing embryos of round and wrinkled-seeded peas (Pisum sutivum L.). Planta 1988;175:27&279. Smith AM, Neuhaus HE, Stitt M. The impact of decreased activity of starch branching enzyme on photosynthetic starch synthesis in leaves of wrinkled-seeded peas. Planta 1990;181:3 10-3 15. Kruckeberg AL, Neuhaus HE, Feil R, Gottlieb LD, Stitt M. Decreased-activity mutants of phosphoglucose isomerase in the cytosol and chloroplast of Clurkiu xuntiunu. Impact on mass-action ratios and fluxes to sucrose and starch and estimation of flux control coefficients and elasticity coefficients. Biochem J 1989;261:457-467. Giroux MJ, Shaw J, Barry G, Cobb BG, Greene T, Okita T, Hannah LC. A single mutation that increases maize seed weight. Proc Natl Acad Sci USA 1995;(In press). Edwards J, Green JH, Ap Rees T. Activity of branching enzyme as a cardinal feature of the Ra locus in Pisum surivum. Phytochem I988;27: 1615- 1620. Carter J, Smith EE. Actions of glycogen synthase and phosphorylase of rabbit-skeletal muscle on modified glycogens. Carbohyd Res 1978;61:395-406. Keeling PL, Bacon PJ, Holt DC. Elevated temperature reduces starch deposition in wheat endosperm by reducing the activity of soluble starch synthase. Planta 1993;191:342-348. Keeling PL, Banisadr R, Barone L, Wasserman BP, Singletary GW. Aust J Plant Physiol 1994;21:807-827. Hawker JS, Jenner CF. High temperature affects the activity of enzymes in the committed pathway of starch synthesis in developing wheat endosperm. Aust J Plant Physiol 1993;20197-209. Jenner CF. Starch synthesis in the kernel of wheat under high temperature conditions. Aust J Plant Physiol 1994;21:791-806. Singletary GW, Banisadr R, Keeling PL. Heat stress during grain filling in maize: Effects on carbohydrate storage and metabolism. Aust J Plant Physiol 199+21:82+841. Stark DM, Timmerman KP, Barry GF, Preiss J Kishore GM. Role of ADPglucose pyrophosphorylase in regulating starch levels in plant tissues. Science 1992;258:287-292. Miiller-Rober BT, Sonnewald U, Willmitzer L. Inhibition of ADPglucose pyrophosphorylase in transgenic potatoes leads to sugar-storing tubers and influences tuber formation and expression of tuber storage protein genes. EMBO J 1992;11:122+1238. Katz FR. In: Goldberg I, Williams R (eds) Biotechnology and Food Ingredients. New York: Van Nostrand Reinhold, 1991;315-326.
This Page Intentionally Left Blank
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R. El-Gewely, editor.
28 1
The chemical degradation of starch: old reactions and new frontiers Rawle I. Hollingsworth Departments of Chemistry and Biochemistry, Michigan State University, East Lansing, Michigan, USA
Abstract. The chemistry of starch degradation is reviewed from the standpoint of the general utility of the reactions and the proposed mechanisms. The limitations and potential of the various chemical transformations are discussed. The reactions discussed include hydrolysis, oxidations, base-catalyzed degradations, halogenations and radiolysis. The potential scope for starch-derived materials in industry is examined. The potential use in the manufacture of fine chemicals is explored.
Key words: amylose, brominolysis,chiral synthons,chlorinolysis,degradation,depolymerisation, dextrins, halogenation, hydrolysis, lactones, mechanisms, oxidative cleavage, radicals, radiolysis, (S)-3,4dihydroxybutyric acid, starch.
Introduction The chemical degradation and modification of starch has been a much studied area primarily with a focus on structural alterations leading to new physicochemical properties. One of the largest economic forces driving such studies has traditionally been the paper-making industry where starches both native and modified are used for coatings and sizings. The chemistry of starch that is best known is the acid catalyzed hydrolysis leading to depolymerization to form dextrins. Less practiced are reactions that lead to scission of the individual glycosyl rings to form fragments with less than six carbon atoms. Such processes include oxidation with periodate or lead tetraacetate. Another well-studied process is the radiation-induced, solvent-mediated scission of both inter-ring and intra-ring bonds. More recently, the base-catalyzed, oxidative degradation of starch and starch hydrolysates has been commercialized to afford entry into high-priced chiral chemicals including hydroxy aqids, hydroxy amides, lactone, and various chiral nitrogen heterocycles. These aspects of starch chemistry will be discussed in this review article. No attempt will be made to compehensively review the enormous literature (dating back well over 150 years) on starch degradation. Instead I will try to point out new avenues and directions that starch chemistry has taken or can take. I will also discuss current views on mechanistic aspects of established starch chemistry.
Address for correspondence: Rawle I. Hollingsworth, Departments of Chemistry and Biochemistry, Michigan State University, East Lansing, MI 48824, USA. Tel.: +1-517-353-0613.Fax: +1-5 17-353-9334. E-mail:
[email protected]
282
Degradative reactions Acid hydrolysis This is the most practiced method of starch degradation. It has been described in the literature as far back as the early 19th century [1,2]. The acid catalyzed hydrolysis of starch leads to the formation of lower molecular weight polymers or oligomers called dextrins. The mechanism of acid hydrolysis involves protonation of the anomeric oxygen and loss of the glycosyl residue to form an oxonium or oxycarbenium intermediate species. This intermediate then reacts with water to form the terminal hemiacetal group (Fig. 1A). The intermediate can also react with some other nucleophile such as an alcohol. In this case, the product is a new glycoside. If the hydroxyl group that reacts with the carbenium ion is the 6-hydroxyl group of the same glucosyl residue, a 1,6-anhydro linkage is formed and the chain terminates but with a nonreducing residue (Fig. 1B). If, however, the reacting hydroxyl group comes from another glycosyl component, the process (called transglucosidation) results in a different branching pattern or even in an increase of the chain length (Fig. 1C). In a related process, if the oxycarbenium ion is formed by loss of the anomeric hydroxyl group from the reducing terminus of a starch chain, combination with an alcoholic function of another starch chain leads to an increase in molecular weight. This is the mechanism of the phenomenon called reversion (Fig. 1D). Transglucosidation and reversion both act to counteract depolymerization during acid-catalyzed hydrolysis. There are several commercial uses for starch that has been hydrolyzed to varying extents. These include paper coating, textile sizing, adhesives, thickeners, binders, fillers in the pharmaceutical industry and gels and sweeteners in the food industry. Starch hydrolysates are also used in the fermentation industry as a carbon source. Commercial aspects of the practice of starch degradation by acid are discussed in a review by Fleche [3]. A good discussion of the chemical and physicochemical properties of starch has been written by Greenwood [4]. A classic pair of volumes on the very early chemistry of starch has also been written [ 5 ] .
Oxidative cleavage There are several well-characterized oxidative degradation methods that have been applied to starch and other polysaccharides. One of the best known of such methods is the periodate oxidation. 'In this reaction (Fig. 2), vicinal glycols (i.e., the 2 and 3 hydroxyl groups of starch) are cleaved to yield a dialdehyde and iodate via a cyclic intermediate. This reaction has tremendous analytical significance since it can be used to ascertain what proportion of glucosyl residues might be blocked at 0 2 or 0 3 in a modification reaction. Substitution of either or both by hydroxyl group renders the ring insensitive to periodate oxidation. Once the glucosyl ring is cleaved by periodate, the glycosidic linkage involving that residue can be selectively cleaved by very mild acid hydrolysis after reduction of the aldehyde groups to alcohols or oxidation to carboxylic acids. Treatment of the oxidized product with base leads to further
283
\ON
+
Roll
li
A
h
011
C
Fig. 1. Possible outcomes of the attempted acid-catalyzed hydrolysis of starch. A: Simple hydrolysis via an oxycarbenium intermediate. B: 1,6-anhydro ring formation leading to chain termination (competes at low water content). C: Transglucosidation. D: Reversion.
degradation via p-elimination. The formation of the dialdehyde function also affords a way of modifying starch by reactions such as reductive amination with amines. One of the drawbacks of the periodate reaction from a commercial standpoint is that the reagent is extremely expensive and ways of regenerating it have to be perfected. Lead tetra-acetate oxidation is very similar in many respects to the periodate oxidative cleavage method. The products from the reaction of starch are largely the
284
Fig. 2. Mechanism of periodate cleavage of glucosyl rings to form dialdehydes.
same. This reaction is also thought to proceed by a cyclic intermediate species (Fig. 3). Lead tetra-acetate oxidation is subject to more stringent conformational constraints. Trans-diols are oxidized much more slowly than are cis-diols. The C2 and C3 hydroxyl groups of glucose in starch have a trans-relationship. Oxidation of starch with halides or hypohalites is also a very commercially significant process. Such reactions are usually effected by the action of bromine in aqueous solutions at different pH values. The optimum pH value is on the basic side where the hypohalite concentration is highest. In the case of sodium hypochlorite treatment, the primary result is the scission between the 2 and 3 carbons of the glucosyl residues of starch to form a diacid [6,7]. Similar products have been shown to predominate in hypobromite oxidations [8-111. There are, however, some important differences between the actions of the two reagents. Whereas hypochlorite oxidation takes place optimally at a pH value of just over 7, the optimal pH for
Fig. 3. Mechanism of lead tetra-acetate cleavage to form dialdehydes.
285 hypobromite oxidation is 9 [ 111. Besides the formation of D-erythrOniC acid and glycolic acid by C2/C3 scission, other competing processes include simple oxidation of C2 or C3 to keto groups, oxidation of C6 to form glucuronic acid and oxidation of C1 to form gluconic acid. Hypobromite is a more powerful oxidizing agent than is hypochlorite and reaction of the former at its pH optimum of 9 is 4 times faster than the latter at its pH optimum of just over 7. In one study, it was determined that amylose is more readily degraded by bromine at neutral pH than is amylopectin [ 101. There are important differences between the products of oxidation by aqueous bromine at different pH values. At pH 7, the concentration of hypobromite is essentially 0 and the oxidant is primarily bromine [ 111. These conditions lead to the formation of 2-uloses. The regiochemistry of bromine oxidation is controlled by stereochemical factors. Attack at the 3-position is hampered by the presence of the axial glycosidic substituent [12,131. The mechanism of oxidation of carbohydrates by aqueous solutions of halogens is a matter of some dispute. A very wide variety of mechanisms have been proposed. These range from ionic to radical to mixed mode processes. One of the earlier mechanisms proposed for the oxidation of carbohydrates by aqueous bromine involved the loss of a skeletal hydrogen atom as a hydride ion followed by subsequent loss of a proton from the alcohol group [14]. This study was based primarily on kinetic isotope effects and linear free energy relationships using 2propanol as a model compound. The study was limited to oxidations in the low pH regime. A similar mechanism involving a cyclic transition state has been proposed for the oxidation of cyclohexanol by bromine [15]. Another study, also using linear free energy relationships, could not really distinguish between a mechanism in which there was a rate-limiting hydride transfer or protonation but concluded that a hypobromite ester was not involved [16]. Such an ester has been proposed [17]. A radical mechanism is quite reasonable for hypochlorite oxidations at high pH. One unusual oxidation that has not been used much but might have some potential for the degradation of starch is chlorinolysis in acetic acid [18]. This reaction has been applied to glycosides to produce 1-chloro-1-deoxy compounds and a hypochlorous ester (Fig. 4). The reaction when applied to amylopectin yielded fragments containing the chlorodeoxy glycoside and a 4-ulose. The photochemical oxidation of starch with chlorine has also been demonstrated and is known to yield uloses [19]. This is certainly by a radical mechanism and should involve oxidation at C 1. In a recent development, it was demonstrated that 4-linked aldohexoses can be converted in high yield to optically pure (S)-3,4-dihydroxybutyric acid or its lactone
bn
Fig. 4. Chlorinolysis of glycoside linkages to form glycosyl chlorides and hypochlorite esters.
286 in high yield [20-22]. This chemistry is based on the isomerization of the aldehydo function to a ketose and subsequent loss of the 4-alkoxy substituent by base catalyzed j3-elimination to form an a-diketone. This is then cleaved by an oxidant, such as hydrogen peroxide, to form the dihydroxy butyric acid and glycolic acid (Fig. 5 ) . This chemistry has been applied to the degradation of starch and starch oligosaccharides (dextrins) on a commercial stage to produce the chiral acid and lactone in high yield. It constitutes the entry of starch as a primary raw material in the very high priced, competitive (but always growing) fine chemical market. This lactone and the trio1 obtained from its reduction is the central component in a number of literature
/
J
OH
J
vfi
I
w14
Fig. 5. Mechanism of the base catalyzed, oxidative degradation of an amylose chain to yield optically pure (S)-3,4-dihydroxybutyricacid.
287 syntheses some of which have high commercial significance [23-281. Figure 6 shows some of the chemical intermediates that are now routinely made from this starting material. The glycolic acid that is formed as a by product already has some use in the preparation of specialty polymers. It is usually made by the halogenation of acetic acid followed by base treatment.
Cleavage by radiolysis This constitutes one method of enormous potential in the degradation and structural modification of starch. Starch is rapidly degraded by y-radiation, such as that from 6oCosources to form a myriad of functionalities [2+33]. This degradation is evident by a marked decrease in viscosity and an increase in reducing sugar content. Irradiation of water leads to the formation of hydroxyl radicals, hydrogen radicals and solvated electrons. The cleavage of glycosides to form reducing sugars has been rationalized (Fig. 7) by a process involving electron capture by the glycoside to form a radical anion followed by loss of the glycosidic alkoxy function to form a radical at the glycosidic center. This radical is then annihilated by a hydroxyl radical to form the hemiacetal function of the reducing sugar [34]. Several processes take place concurrently in the radiation-induced decomposition of starch besides the cleavage
Fig. 6. Some products that have been obtained in high (>75%) yield from the base catalyzed, oxidative degradation of starch oligomers.
288 ,011
OR
/. Fig. 7. Mechanism of the radiation induced cleavage of glycosides.
of glycosidic linkages to yield-reducing sugars. In addition to this, there are several other processes that result in the oxidation of the carbohydrate ring to yield uloses. These eventually undergo ring scission to give pentoses, tetroses, acids and glyoxal in addition to several other 2, 3 and 4-carbon products. Hydroxyl radicals have been demonstrated to be the major species mediating the radiation induced decomposition of carbohydrates in aqueous solution [35].There is a pronounced effect of oxygen concentration on the effect of y-radiation on carbohydrates. The conversion rate of molecules per 100 electron volts of energy (G value) is markedly less in the presence of oxygen. This is not surprising since oxygen exists in a triplet ground state and will therefore quench radical species that are formed during the radiolysis. Degradation is highest at low water content [36,37]. An extensive discussion on the effects of radiation on carbohydrates is available [34].
The outlook for starch chemistry It is clear that the chemistry of starch degradation has not undergone any dramatic changes in the last few decades. This is surprising on the one hand because it is one of the most abundant substances in nature and can be obtained in a pure form. It is understandable on the other because it is extremely difficult to surmount the problems caused by the high density of functional groups and their redundancy. Because of its limited solubility in nonaqueous solvents, it is not possible to apply many of the newer organic reactions to starch since only a small proportion of these were devised with aqueous solvent systems in mind. Despite this, the prospect of being able to degrade starch or maltodextrins to high value chiral intermediates for use in the chemical industry is a tantalizing one that merits much attention. There is much commercial potential in this area for starch because of the high proportion of chiral centers bearing hydroxy groups and the potential functionalities that they can be
289 converted to. This is well illustrated by the wide diversity of intermediates (some of which have bulk prices >US $4,000 per kg) that have been successfully made from starch (Fig. 1). The challenge is to devise ways that are selective enough. One possibility is to use the relatively higher accessibility and reactivity of the primary hydroxyl groups. A possible class of reactions that may accomplish this is transition metal catalyzed oxidations. New ring scission reactions that lead to cleavage between C2 and C3 can also be developed. Electrochemical methods that regenerate the oxidants can also be investigated. Although it has been fairly well studied, the radical chemistry (either radiation or chemically induced) of starch and maltodextrins, especially in different solvents, can be further explored. The use of chemically reactive solvents for such reactions may lead to effective quenching of intermediate radical species to yield new derivatives. The chemical degradation of starch by ring oxidative ring scission will lead to hydroxy aldehydes and hydroxy acids as the most obvious products. These have tremendous potential in the area of polymers, a potentially high volume application. Hydroxy acids can be used for the fabrication of polyesters. They can be dehydrated to yield functionalized a$-unsaturated acids (or aldehydes in the case of hydroxy aldehydes) thus opening the road to a broader area of polymer chemistry via wellestablished radical polymerization methods. Unlike polyesters, such polymers will have carbon atoms along the entire carbon skeleton but will have hydroxy groups at periodic sites to enable grafting or application-specific modifications. The aldehyde function can be modified by reductive amination to yield cationic polymers. Despite the fact that it is fraught with many technical challenges, the outlook for industrial applications involving degradative starch chemistry is extremely bright. The complex nature of starch with an amylose component and an amylopectin component (branched at the 6-position) as well as various degrees of phosphorylation further complicates the issue. This should not, however, cloud optimism. There is a need to provide a more homogeneous starting material for chemical reactions. Hence in the future, the ability to debranch the starch polymer (enzymatically) will be a great advantage. Another approach would to be genetically engineer plants that have very low amylopectin content. There is also a need to change the current attitudes towards starch chemistry. In the past, the primary objective was to manipulate the physical properties of the polymer. Modifications were camed out and some properties such as solubility, viscosity or gelling ability monitored. There was not much of an attempt to really understand and manipulate functionality in the traditional chemical sense. Starch has traditionally been viewed as a chemically intractable substance and there has not been much reason for optimism with respect to its use as a chemical feedstock. As the steady march of chemical and biochemical methods advances, there is every reason to be optimistic.
Acknowledgements This work was supported by grant # DE-FG02-89ER14029 from the U.S. Department
290 of Energy and by the Michigan State University Research Excellence Fund through the Center for New Plant Products.
References 1. Kirchkoff GSC. Mem Acad Imp Sci Petersbing 1811;427. (Cited in: Fleche G. Chemical modification and degradation of starch. Food Sci Technol 1985;14:73-99.) 2. Bouillon-Lagrange M. Bull Pharm Paris 1811;3:395-398. (Cited in: Fleche G. Chemical modification and degradation of starch. Food Sci Technol 1985;14:73-99.) 3. Fleche G. Chemical modification and degradation of starch. Food Sci Technol 1985;1473-99. 4. Greenwood CT.Starch and Glycogen. In: Pigman W, Horton D (eds) The Carbohydrates,Chemistry and Biochemistry, 2nd edn. New York: Academic Press, 1970;(IIB):471-5 13. 5. Whistler RL, Paschal1 EF (eds) Starch: Chemistry and Technology, vols I and 11. New York: Academic Press, 1966. 6. Whistler RL, Schweiger L. Oxidation of amylopectin with hypochlorite at different hydrogen ion concentrations. J Am Chem Soc 1957;79:6460-6464. 7. Whistler RL,BeMiller JN. Alkaline degradation of polysaccharides. Adv Carbohydr Chem Biochem 1958;13:289-329. 8. Eliassafat J, Be1 Ayche J. The interaction of starch with bromine in acid solution. Carbohydr Res 1967;5:470-476. 9. Eliassafat J, Be1 Ayche J. The iodine affinity of some kinds of starch. Starke 1965;17:389-390. 10. Torneport LJ, Salomonsson BA, Theander 0. Chemical characterization of bromine oxidized potato starch. Starke 1990;42413-417. 11. Doane WM, Whistler RL. Oxidation of amylopectin with hypobromite at different hydrogen ion concentrations. Starke 1964;16:177-180. 12. Lam 0, Scholander E, Theander 0. Bromine oxidation of methyl a- and P-pyranosides of Dgalactose, D-glucose, and D-mannose. Carbohydr Res 1976;49:6+77. 13. Scholander E. Bromine oxidation of a,a- and P,P-trehalose. Carbohydr Res 1979;73:302-308. 14. Swain CG, Wiles RA, Bader FW. Use of substituent effects on isotope effects to distinguish between proton and hydride transfers, Part I. Mechanism of oxidation of alcohols by bromine in water. J Am Chem Soc 1961;83:1945-1950. 15. Barker IRL,Overend WG, Rees CW. The oxidation of cyclohexanol and related compounds with bromine. J Am Chem Soc 1964;3263-3267. 16. Venkatasubramanian N, Thiagarajan V. The mechanism of oxidation of alcohols by bromine. Tetrahedron Letts 1968;14:1711-1714. 17. Den0 NC, Potter NH. The mechanism of oxidation of alcohols by aqueous bromine. J Am Chem Soc 1967;89:3555-3556. 18. Whistler RL, Mittag TW, Ingle TR. Chlorinolysis of glycosidic bonds. J Am Chem Soc 1965;87:4218-4219. 19. Meiners AF, Moms FV.The light-catalyzed oxidation of starch with aqueous chlorine. J Org Chem 1964;29:449-452. 20. Hollingsworth RI. Process for the Preparation of 3,4-Dihydroxybutanoic Acid and Salts Thereof. U.S. Patent 5,292,939 (1994). 21. Hollingsworth RI. Process for the Preparation of 3,4-Dihydroxybutanoic Acid and Salts Thereof. U.S. Patent 5,319,110 (1994). 22. Hollingsworth RI. Process for the Preparation of 3,4-Dihydroxybutanoic Acid and Salts Thereof. U.S. Patent 5,374,773 (1994). 23. Corey El,Niwa H, Knolle J. Total synthesis of (S)-12-hydroxy-5,8,14-cis,-lO-trans-eicosatetraenoic acid (Samuelsson’s H E E ) . J Am Chem SOC1978;100:1942-1943. 24. Uchikawa 0, Okukado N, Sakata T, Arase K, Terada K. Syntheses of ( S ) - and (R)-3-hydroxy-4-
29 1
25. 26. 27. 28. 29. 30. 31. 32. 33. 34.
35. 36. 37.
butanolide (2S,4S)-, (2R,4S)-, (2S,4R)-, and (2R,4R)-2-hydroxy-4-hydroxymethyl-4-butanolide and their satiety and hunger-modulating activities. Bull Chem Soc Jpn 1988;61:2025-2029. Hayashi H, Nakanishi K, Brandon C, Marmur J. Structure and synthesis of dihydroxypentyluracil from bacteriophage SP-15 deoxyribonucleic acid. J Am Chem SOC1973;95:874+8757. Danklmaier J, Honig H. Synthese und struktur diastereomerenreiner 2,6-disubstituierter 3morpholinone. Liebigs Ann Chem 1988;1149-1153. Mori Y, Kuhara M, Takeuchi A, Suzuki M. Stereoselective reduction of P-alkoxy ketones: a synthesis of syn-l,3-diols. Tetrahedron Letts 1988;29:5419-5422. Shieh HM, Prestwich GD. Chiral, biomimetic total synthesis of (-)-aplysistatin. Tetrahedron Letts 1982;23:4643-4646. Hofreiter BT. Starch and amylose degradation byoC@ ' y-irradiation. J Polymer Sci 1974;12:27552766. Tyler BS, Munno FJ, Cadman TW. Effects of radiation on corn starch sols. Environ Sci Technol 1968;2:628-632. Tollier PMTh, Guilbot A. Studies on the action of y-irradiation on starch. Starke 1972;24:285-290. El Saadany RMA, El Saadany FM, Foda YH. Degradation of corn starch under the influence of gamma irradiation. Starke 1976;28:208-211. Scherz Von H. Irradiation of starch. Starke 1971;23:25+294. Philips GO. The effects of radiation of carbohydrates. In: Pigman W, Horton D (eds) The Carbohydrates,Chemistry and Biochemistry,2nd edn. New York: Academic Press. 1980;(IB):12171297. Philips GO, Griffith W, Davies JV. Radiation chemistry of carbohydrates, Part XVI. The contribution of OH radicals to the radiolysis of aqueous solutions. J Chem Soc (B) 1966;194-200. Ehenberg L, Jaarma M, Zimmer KG. The influence of water content on the action of ionizing radiation on starch. Acta Chem Scand 1957;11:950-956. O'Meara JP, Sheen TM Detection of free radicals in irradiated food constituents by electron paramagnetic resonance spectroscopy. Food Technol 1957;11:132-136.
This Page Intentionally Left Blank
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R.El-Gewely, editor.
293
Biotechnological applications of the disaccharide trehalose Carmen L.A. Paiva' and Anita D. Panek2 'Instituto Biornkdico, CCBS, Universidade do Rio de Janeiro, Rio de Janeiro; and 'Departamento de Bioquimica. Instituto de Quimica, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil
Abstract. Trehalose is a disaccharide present in a variety of anhydrobiotic organisms which have the ability to promptly resume their metabolism after addition of water. It has been successfully used as a nontoxic cryoprotectant of enzymes, membranes, vaccines, animal and plant cells and organs for surgical transplants. It has been predicted that trehalose can also be used as an ingredient for dried and processed food. Therefore, the recent biotechnological applications of trehalose have imposed the standardization of methods for its production, as well as for its specific quantification. Key words: anhydrobiotic organisms, bank cells, cryoprotectant, disaccharide, E. coli trehalase, environmental stress, food, freeze-drying,liposomes, membranes proteins, organs, S . cerevisiae trehalase, Saccharomyces cerevisiae, stabilization dehydration, thermotolerance, trehalase, trehalose, trehalose quantification, trehalose purification, vaccines, yeast.
Biochemistry and biotechnology were born out of studies with yeast cells. In fact biotechnology preceded the basic science. Early Egyptians already prepared their dough, a simple mixture of flour and water and kept it warm until bubbles began to form. They also delighted in drinking the fermented fruit juices accepting them as a gift from the Gods. Throughout the first half of the 19th century those few scientists who had grasped the truth about yeast had difficulties in imposing their views: fermentation was considered to be a purely chemical reaction in which life was not involved. Only when Pasteur in 1850 turned his attention to fermentation and brewing was it shown that yeast cells were capable of transforming sugar in alcohol. Biochemistry was born soon after, when Edward Buchner in 1896, ground yeast cells with sand and pressed them to obtain an extract for pharmaceutical purposes. In order to preserve it he added sugar. To his surprise, even though the yeast integrity had been destroyed, the fluid was still able to ferment. The enzymes had remained active. The biochemical process involved was identified for both fermentation and raising of dough. Yeast are racing with bacteria for the lead position in biotechnology. They share with animal cells many of the molecules and processes that are subject of intensive study in molecular biology. Moreover, unlike bacterial cells, which must be broken to retrieve a produced heterologous protein, yeast cells are able to secrete the product beyond their cell wall. Furthermore, there is another advantage to the use of yeast cells in biotechnology, it is a psychological advantage. While people tend to associate bacteria with diseases, yeast has been associated, over millennia, with the good things of life, i.e., bread and wine.
294 Over the last 10 years scientists have come to the conclusion that yeast cells can provide mankind with another gift from nature: trehalose, a very unique sugar capable of protecting life against environmental stresses. The aim of this review is to guide the reader through the evidence that lead to this conclusion and to show how this knowledge is being applied to preserve biological materials. Due to the vast interest evoked by trehalose over the last years, this review does not intend to include all of the literature on the subject.
Trehalose Trehalose (a-D-glucopyranosyl a-D-glucopyranoside) was described in the early 19th century to be present in the ergot of rye [l]. Later it was found to be widely distributed in different organisms [2] yet it has been mostly studied in yeast cells where it may reach over 20% of the dry weight under certain physiological conditions [3]. Yeast cells accumulate two carbohydrates during their cell cycle - a fact which intrigued researchers for many years. According to Lillie and Pringle [4] the pattern of glycogen accumulation and utilization is compatible with it serving as a source of energy for Saccharomyces cerevisiae (S. cerevisiae) both during respiratory adaptation and during starvation. In contrast, the authors claim, trehalose seems to play a role only during starvation. This observation taken together with more detailed studies of other anhydrobiotic organisms [5,6] led us to envisage a more specific role for trehalose in yeast cells under different stress conditions. Water is usually thought to be required for the living state, however, numerous organisms are capable of surviving complete dehydration without dying. They commonly survive in this state which is known as anhydrobiosis, even when >99% of their body water is removed. The dry but viable tissues contain as little as 0.1 % water, a condition that would normally be thought not to be consistent with life. Many of these organisms, such as plant seeds, yeast cells, fungal spores are familiar in daily life, but many less familiar organisms also exhibit this phenomenon - microscopic animals, such as certain nematodes, rotifers, tardigrades and cysts of some crustacean embryos. The dry organisms may remain in anhydrobiosis for decades or, perhaps, even centuries under favorable conditions. When water again becomes available they rapidly swell and resume active life [6]. In the past decade, studies of these organisms have established some of the major mechanisms that permit them to survive dehydration. Trehalose emerged as a characteristic constituent of anhydrobiotic organisms [7]. Trehalose is among the most chemically unreactive and stable sugars in nature. Because the two glucose moieties are joined through the 1,l carbon atoms of the two glucopyranose rings, it is completely nonreducing. The glycoside oxygen bond joining the two hexose rings has a very low bond energy (<1 kcal/mol) which renders the disaccharide structure very stable. Trehalose does not dissociate into two reducing glucose molecules except under extreme hy-
295 drolysis conditions or in the presence of the specific hydrolytic enzyme trehalase. In comparison, another nonreducing disaccharide, sucrose, has a high energy bond of more than +27 kcal/mol. Moreover, although sucrose is stable as a pure substance, in the presence of chemically reactive amino groups of proteins, it is readily split to glucose and fructose - both reducing monosaccharides. Prolonged storage of dried proteins in sucrose leads to progressive chemical damage due to browning reactions [81. The absence of direct internal hydrogen bonds produces flexibility around the disaccharide bond. All four hydrogen bonds are formed with the two associated water molecules in trehalose dihydrate, the form in which trehalose is naturally found. The two glucose rings are thus able to conform to the irregular surface of proteins when hydrogen binding. Less flexible disaccharide, such as maltose and sucrose, is stabilized in a rigid conformation by direct internal hydrogen bonds. Using Fourier transform infrared spectroscopy (FTIR) to characterize the interaction between dried proteins and sugars, Carpenter and Crowe [9] demonstrated that hydrogen bonding between sugars and proteins is a requisite for labile proteins to be preserved during drying. Again trehalose emerged as the most efficient protectant. Recent IR and Raman spectroscopic studies of the interaction of trehalose and lysozyme indicate that the effect of the disaccharide is to concentrate what water remains close to the protein rather than binding to it [lo]. Freezing and drying have been commonly assumed to exert similar stresses on biological materials based on the fact that both involve removal of water from the system. However, it has been demonstrated that different mechanisms are involved in these stresses and therefore equally different mechanisms are required to obviate them [ l l ] . In order to protect a freeze-dried protein stabilizing additives must hydrogen-bond to the dried protein thus serving as a “water substitute” in the dry state. However, if only this interaction were needed for protein preservation during freeze-drying then glucose, a monosaccharide, should provide the same protection. This is not the case, as was shown when phosphofructokinase was freeze-dried [ 121. The enzyme which is totally inactivated if freeze-dried without stabilizers can be recovered with 60% of its activity when trehalose is used as cryoprotectant whereas the recovered activity is less than 5% if glucose is used. Obviously, there is more than one explanation for trehalose protection during freezing. Prestrelski et al. [13] using FTIR spectroscopy showed that besides hydrogen bonding a good cryoprotectant has to ensure the maintenance of the native structure of the protein in the dried state. Trehalose fulfills this role more efficiently than other carbohydrates. Let us now consider the consequences of dehydration on membranes. The primary destabilizing forces during dehydration are fusion and lipid phase transitions which lead to leakage upon rehydration [ 141. It was shown that trehalose is capable of preventing this damage by replacing water molecules during dehydration, by inhibiting fusion and by avoiding phase transition during the rehydration step. The phospholipids in the membrane which normally undergo phase transition at a certain temperature are maintained by trehalose in the liquid crystalline phase even though they are dry. Thus when the membranes are rehydrated the phase transition is avoided [15].
296 The most effective sugars in stabilizing dry liposomes and native membranes are disaccharides. Sucrose is less effective than trehalose. Trehalose and maltose are equally effective but maltose is a reducing sugar and might, therefore, enter into secondary reactions which could lead to damage such as browning reactions [16]. It appears that it is this unique collection of properties that ensured the parallel evolution of trehalose synthesis in so many diverse anhydrobiotic organisms. The same properties confer, on trehalose, its remarkable ability to protect valuable biomolecules as well as cells and organs. More recently an additional metabolic role was described for trehalose, rather for its immediate precursor, trehalose-6-phosphate. It has been common knowledge that glucose-6-phosphate regulates glucose uptake in mammalian cells by inhibiting hexokinase, whereas in yeast cells none of the three hexokinases respond to this control [ 171. In 1985 [ 181 we showed that the potential effects of catabolite repression on mitochondrial development were minimized when maltose was the carbon source for growth due to a bypass leading glucose-6-phosphate into trehalose instead of feeding glycolysis directly. A strong support for this regulatory role has now come from the demonstration that trehalose-6-phosphateinhibits sugar phosphorylation [ 191. The strongest effect was observed upon hexokinase 11 which is the kinase that also controls catabolite repression. Preliminary reports from our laboratory also point towards a regulatory role for trehalose-6-phosphate during progression of the G1 phase of the yeast division cycle (unpublished results). It has become quite evident that trehalose metabolism plays a very significant role in the control of various stages of the yeast cell cycle and physiology besides protecting the cell membrane and its proteins against stresses caused by low and high temperatures, dehydration, osmolarity and high ethanol concentration.
Trehalose synthesis Trehalose is generally formed from uridine 5'-(a-D-glUCOpyranOSyl pyrophosphate) and D-glucose-6-phosphate by the action of trehalose-6-phosphate synthase (TPS) followed by a specific phosphatase (TPP) which ensures the equilibrium of the reaction in favor of the formation of trehalose phosphate [20]. The synthase was purified from brewer's yeast and showed maximal activity at pH 6.6 in the presence of a 25 mM concentration of Mg2+ions. Trehalose synthase activity is generally measured by a coupled reaction with pyruvate kinase and colorimetric determination of pyruvate [2 11 or by coupling lactate dehydrogenase to measure the pyruvate formed. This latter procedure allows kinetic studies and is much faster and more reliable than the colorimetric determination [22]. Many years elapsed from the time of the identification of the synthesizing enzymes and the unravelling of what turned out to be a very complex system. The purification of a proteolytically modified form of the TPS/TPP complex was reported by Londesborough and Vuorio [23]. The authors describe a protein complex of about
297 800 kDa composed of three polypeptides of 57, 86 and 93 kDa. In addition, a protein dimer of approximately 110 kDa was isolated and was shown to activate TPS but to have no effect on TPP. Using a different approach - a strain lacking vacuolar proteases - Bell et al. [24] obtained a multimeric protein of 630 kDa and identified the TPS1 gene which encodes the smallest subunit (56 kDa) of the TPS complex. In 1989 Paschoalin et al. [25] identified an alternative enzyme for trehalose synthesis in our laboratory. The donor of the glucose moiety is ADPG instead of UDPG. By fast protein liquid chromatography we isolated a protein of 48 kDa clearly distinct from the 56 kDa protein which corresponds to the UDPG-dependent trehalose synthase activity [26]. Assessing the function of this ADPG-dependent enzyme is a difficult task. It is always present and active. It does not suffer the effects of catabolite inactivation and repression in contrast to what has been described for the classical enzyme [26,27]. Furthermore, neither a glucose signal nor a heat shock elicit the deactivation that has been described for the UDPG-dependent enzyme [28]. However, when a diploid was constructed from two haploid strains deficient in the UDPG-dependent trehalose-6phosphate synthase the new strain proved to be able to sporulate and, thereafter, spores responded to a glucose signal by accumulating trehalose [28]. It seems, therefore, that the ADPG-dependent trehalose-6-phosphate synthase is a more primitive enzyme than the UDPG-dependent protein, yet it has a major ecological significance for the yeast cell since it guarantees viability after a period of starvation.
Trehalose catabolism An enzyme able to hydrolyze trehalose was first found in Aspergillus niger [29]. Since then, trehalase has been detected in many organisms of both the plant and animal kingdoms [2]. Activity is generally determined measuring glucose formation by a modification of the glucose oxidase-peroxidase method [30]. In many invertebrates, including insects, trehalose is the major hemolymph sugar and serves as an indispensable substrate for energy production and biosynthesis of macromolecules [31]. The pivotal role of trehalose is achieved by the action of trehalase which is localized in almost all tissues, in different soluble and particulate forms throughout the life cycle. In Japan some efforts have been made to synthesize trehalase inhibitors aiming at the use of such chemicals as fungal insecticides [32]. Trehalose mobilization in fungi, in general, was long considered to be a relatively simple process and no studies of the regulation of trehalase appeared in the literature for many years. In 1974 the first evidence for the involvement of cAMP and protein kinase in the activation of “cryptic” trehalase was described [33]. These studies were followed by the new approach of using mutant strains of S. cerevisiae harboring lesions in the cAMP protein kinase cascade which definitely established the regulatory mechanism, demonstrating that the phosphorylated form is the active form of the enzyme [30,34]. In a series of papers these results were corroborated by
298 demonstrating that both addition of glucose to starved cells and a 2-min heat shock at 52°C caused an increase in CAMP levels which correlated with a transient activation of trehalase [35,36]. Cryptic neutral trehalase from S. cerevisiue was purified about 3,000-fold. Active trehalase, obtained through phosphorylation of the cryptic enzyme by CAMPdependent protein kinase, was isolated by chromatography on DEAE-cellulose. A major phosphorylated protein with an apparent value of 86 kDa was detected after SDS-polyacrylamide-gel electrophoresis. Multiple forms of cryptic trehalase were characterized by centrifugation of a glycerol density gradient, with values of 320,160 and 80 kDa, respectively. However, upon activation of each of these forms with protein kinase, a single form of 160 kDa was obtained [37]. These results were corroborated by another laboratory [38]. Recently the gene for this neutral trehalase was cloned and sequenced and it was shown to harbor the putative CAMP-dependent phosphorylation consensus sequence, RRGS, from amino acid positions 22 to 25 [39]. Another trehalase activity was found to be present in the vacuoles of the yeast cell [40,41]. This protein is not regulated by phosphorylation and, in contrast to the cytosolic neutral trehalase, it has a pH optimum of 4.5. The physiological function of this vacuolar trehalase has yet to be established, moreover, trehalose has not yet been localized inside the vacuole.
Protection against stress As mentioned above, trehalose accumulation occurs in a variety of plants and animals. This correlation suggested but did not prove that the sugar was responsible for the absence of desiccation damage in anhydrobiotic organisms. In 1988 we reported studies with mutant strains of S. cerevisiue unable to synthesize trehalose due to a defect in the trehalose-synthesizingpathway. These strains failed to survive storage in a frozen, dried state (Fig. 1) [42]. This data proved unequivocally, for the first time, that trehalose is indeed necessary for cells to survive anhydrobiosis. Yeast cells are capable of protecting their membranes and proteins during various environmental stresses. With what mechanisms has nature endowed these cells to allow them to survive? High temperatures, dehydration, high osmolarity and the stress caused by ethanol are conditions that have to be faced and fought by the yeast cells both in their natural habitats and during biotechnological processes. Cells accumulate trehalose when submitted to a temperature stress during a shift from 28 to 40"C, for example; a condition which occurs during fermentation and which, furthermore, enables them to survive a subsequent more severe, generally lethal heat stress of 50°C/8 min [43,44]. Some of the heat shock proteins which are also synthesized during induction of thermotolerance seem to be crucial in recovery from the heat-stressed state whereas trehalose, most probably, provides a means to protect proteins, mainly enzymes, against denaturation during the rise in temperature
299
so S288C
30
=.--
10
5
50
h
n o
Q 6R2
$ 30 10
7
14
21
days Fig. 1. Cells were dried at 37°C for 6 h in the absence (0)and presence of 5°C trehalose (W). They were then stored at 5°C and at the indicated times resuspended in 0.05 M phosphate buffer, pH 6.0, in the absence or presence of 5% trehalose. The percentage of viable cells was determined by colony formation on YPD plates. (Extracted from [42].)
[45]. On the other hand, the enzymes involved in trehalose synthesis, during heat shock, should be considered as heat shock proteins (hsp). They are present in small amounts in cells growing on glucose and their activities are strongly enhanced during a heat shock [46]. Isoform PI1 of hexokinase which we have shown to be implicated in the pathway of trehalose synthesis during heat shock has been confirmed as an hsp [47]. The same can be said for the 56 kDa subunit of trehalose synthase which shows an increase in the transcription of its gene during heat shock [24]. During fermentation, when ethanol concentration is high, yeast experience inhibition of cell growth, loss of viability loss of nutrient uptake, decreased proton fluxes and impaired fermentation performance. The plasma membrane has been considered as one of the major target sites of ethanol toxicity [48]. Trehalose inhibits ethanol-induced leakage from intact cells as well as from liposomes suggesting that the sugar enhances ethanol tolerance [49]. Moreover, the high glucose concentration (20%) in the fermentors for ethanol production, by itself, also leads to trehalose accumulation as shown when 18% sorbitol, which is not metabolized by the cells, was used as an experimental model [50]. Similar results have been published for the osmotic effects of NaCl on cells of Escherichia coli (E. coli) which respond to the stress by accumulating trehalose [Sl]. From these results it becomes quite clear that mankind has benefited from alcohol
300 fermentation for centuries mainly due to the fact that S. cerevisiue is able to respond to the environmental stress by accumulating trehalose, thereby, reducing damage caused to membranes and proteins by the increase in temperature, osmotic pressure and ethanol concentration which are all inherent to the process. The role trehalose plays in protecting yeast cells against dehydration is an important aspect since it has lead to many technological applications. Trehalose exerts its effects by the binding of its hydroxyl groups to the polar head groups of phospholipids at locations otherwise occupied by water [5]. Studies with unilamellar liposomes showed that it was necessary to have trehalose on both sides of the bilayer for maximal efficiency in retaining entrapped solutes during dehydration [8]. If trehalose protects anhydrobiotic organisms by accumulating in the cytoplasm, how does it get to the outer side of the plasma membrane? Again the answer to this question came from yeast cells. Most diploid strains isolated from nature are able to transport trehalose although this carbohydrate is not normally found as a carbon source. Intriguing as it may seem, this specific transporter appeared to be the ideal candidate for demonstrating that accumulated trehalose, inside the yeast cell, has to reach the outer membrane in order to guarantee viability after dehydration. The trehalose transporter showed low activity in the presence of glucose but was induced after glucose depletion coincidentally with accumulation of internal trehalose [52]. Thus, appearance of transport activity can be related to a condition of stress either by the presence of ethanol formed from glucose or by the total absence of a good carbon source. Yeast cells are well known for their ability to survive complete dehydration and this capacity is strongly linked to the presence of trehalose inside the cell (Fig. 2). Production of dried yeast for bread making is based on this capacity. However, when mutant strains lacking the trehalose transporter are dried at 30°C and then rehydrated they are not able to survive in spite of the fact that they accumulate trehalose inside the cell (Fig. 3) [53]. Addition of external trehalose restores their ability to survive dehydration [53]. It is our belief that the trehalose carrier plays a crucial role in transporting accumulated trehalose to the outer side of the membrane when yeast cells are submitted to the stress of dehydration (Fig. 4). During their normal life cycle cells of S. cerevisiue accumulate trehalose during sporulation (when the stress is starvation) in order to protect their membranes during germination when some sugar becomes available again [28]. Mutant strains which lack the trehalose transporter fail to germinate normally after sporulation, a capacity which can be restored by the addition of external trehalose. Again trehalose and its specific carrier are shown to be essential for survival of the species [54]. These results, taken together, suggest that introduction of trehalose into the cytoplasm of cells which do not normally synthesize this molecule might confer resistance to damage from dehydration and high temperatures. Therefore, two lines of research are being currently developed for isolating the trehalose canier protein: identification of the protein in membrane preparations after induction of transport
301 100
Control Strains
n
b
Y
m
- 80
80
r( d
n
a3
cn
0
Y
60
T
S
T
S
- 60 3 *$
T
40
i!
T
20
-40
v,
w
- 20
Fig. 2. Trehalose concentrations and dehydration survival of strains. Cells were harvested in stationary phase, washed twice with cold, sterile 50 mM phosphate buffer (pH 6.0). and 100 mg samples were dried to constant weight at 3OOC. Rehydration was carried out with the initial volume of the same buffer wanned at 40°C. Proper dilutions were plated before and after dehydration to assess survival rates. Trehalose was determined as described by Eleutherio et al. [53]. (Extracted from [53].)
activity [55] and cloning of the gene responsible for coding of the transporter using mutants which are deficient in this activity [54]. As one might imagine all these results have recently led to new methods for stabilization of biological materials opening new horizons for industrial and medical applications which we shall now discuss.
Biotechnological applications In recent years the literature has been enriched by references to trehalose as a potent protectant of proteins, enzymes and membranes as well as of cells and organs for surgical transplants. Some of the applications, however, have been protected by patents. Because the three-dimensional topology of highly hydrated proteins, like antibodies and enzymes, confers affinity and specificity to their sites, any distortion of their topology may result in loss of function. In order to minimize damage freezedrying has been used for concentrating such compounds. It is now recognized that removing the residual water in the presence of trehalose can produce stable freezedried products with very long shelf lives at room temperatures. It is important to mention that the only constraint on the simplicity of the procedure is that the
302
Mutant Strains
T 30 h
vl
W .-(
(d
M
\
;20
20
0
$ 7
4
v)
(d
4
K 10
10
M
E 0
0 Klg-102
N o Trehalose Synthesis
T82
308Tl 30812
308T5
No Trehalose
Transporter
Fig. 3 . Trehalose concentrations and dehydration survival of mutant strains the same buffer warmed at 40°C. Proper dilutions were plated before and after dehydration to assess survival rates. Trehalose was determined as described by Eleutherio et al. [53].(Extracted from [53].)
trehalose solution must contain all components required for the active molecular configuration [56]. One of the many examples was reported by Prestrelski et al. in 1993 [13]. They demonstrated that 10 mM trehalose added to 1% PEG can preserve the structure of the labile enzymes lactate dehydrogenase and phosphofructokinase, when submitted to freezing and drying. Under these conditions essentially total enzymatic activity was recovered upon rehydration. The authors claim that for labile proteins, preservation of the native structure during lyophilization is a requisite for recovery of activity following rehydration. An important group of labile enzymes, such as those which are employed in molecular biology for gene cloning, has also been studied concerning stabilization by trehalose. A remarkable result of protection of the notoriously unstable DNA restriction endonucleases EcoR I, Bgl 11, Pst I and Hind 11, was described by Colaqo et al., in 1992 [57]. A preparation of each enzyme tested could withstand prolonged exposure to a temperature as high as +70"C, when dried in the presence of 15% trehalose, but not with other sugars. The restriction enzymes could still act accurately on DNA after being stored for 35 days at +70"C. They also showed no loss of enzymatic activity after storage at 37°C for 9 months. According to Colaqo et al. [57] not only endonucleases but also other DNA-
303 +T
+T f
80
‘;ii 60
*E 7
M 20
0
D273-10B 1412-4D
Control Strains
lag- 102
No Trehnlose Synthesis
T82
308Tl 808T2 308TCS
No Trehalose
Transporter
Fig. 4. Effects of outside addition of trehalose on dehydration survival. Conditions were the same as described in the legend of Fig. 2. Trehalose (10%)was added to the buffer before dehydration as well as to the dehydration buffer (+) and survival compared to controls without exogenous trehalose (-). (Extracted from [53].)
modifying enzymes, that are extremely useful tools for recombinant DNA technology, have been tested for their stability in the presence of trehalose. It was found that T4 DNA ligase and T7 DNA polymerase, both dried with 15% trehalose, could be stored for 1 week at room temperature showing no loss of activity. The authors anticipate that these results may have an enormous impact on the development of science, since new types of automatic processes for genome mapping and sequencing can be designed, as well as a wide range of important biomolecules can be stabilized by trehalose. The applications of stabilizing labile molecules can reach areas ranging from basic science, through health care and agriculture, to bioelectronics. The results reported above are specially important for the development of science in the Third World countries, where enzymes for molecular biology are not often produced and need to be shipped and stored at very low temperatures. These handicaps usually make their utilization very costly and difficult. Trehalose has been shown to protect various kinds of biomolecules under conditions such as desiccation, freezing, and high temperatures. Anti-blood-group antibodies air-dried at room temperature or at 37°C in the presence of trehalose were preserved even after several years of storage [56,58]. It has been suggested that the mechanism of preserving molecules by trehalose may be universally applicable, since a variety of different antibodies and proteins,
304 e.g., enzymes and blood coagulation factors, could be protected. Liposomes when subjected to dehydration or freeze-thawing can also be protected by trehalose and the cryoprotectant, glycerol. These molecules keep vesicle integrity during dehydration and freezing, respectively [59,60]. Harrigan et al. [61] examined the susceptibility to leakage induced by both dehydration and freezing. They came to the conclusion that the vesicle size has a great influence on leakage induced by both dehydration an freezing and that the smallest system (7CF100 nm diameter) is the most stable. In the presence of trehalose fusion is avoided in addition to leakage. Not only antibodies but also antigens, such as attenuated viruses for vaccine production, would benefit from the protection given by trehalose. Roser and ColaGo [62] have predicted that health care in developing countries would be much improved if the majority of vaccines could be kept stable at room temperature. Encouraging results with drying in the presence of trehalose were obtained for the oral polio vaccine, since a single dose formulation was stable at 45°C as compared to the liquid vaccine chilled at 4°C [62]. We are at present testing the influence of trehalose in the stabilization of the yellow fever vaccine. This is a joint project with Biomanguinhos, in Rio de Janeiro, Brazil. The stability of antiophidic serum is also being tested in the presence of trehalose by another associated group at the Instituto Butantii, Sao Paulo. Considering that our country has continental dimensions, bringing vaccines and antisera to the very interior is a major task, specially if they need to be frozen. Stabilizing these life-saving molecules by trehalose is our present target and dream. Not only viruses are expected to be stabilized by trehalose. Baker's yeast, which can be stored dry on the kitchen shelf for years can begin to ferment within minutes after addition of water. This phenomenon is caused by high trehalose accumulation inside the yeast cell. Coutinho et al. [42], in our laboratory, investigated the viability of yeast cells submitted to different drying techniques. It was found that yeast strains dried in layers, at 37°C for 6 h, did not loose viability, however, they died thereafter at 5°C unless trehalose was used for re-suspending the cells before drying. A remarkable result was obtained when cells wefe frozen at -120OC in the presence of 10% trehalose. Trehalose was also tested as the sole cryoprotectant during freezing of carrot and tobacco cells. No viable cells were found following thawing. However, if a pretreatment of the carrot cells with 5-10% trehalose for 24 h was applied, prior to freezing in the presence of 20 or 40% trehalose, 50% viability was obtained. The observation that trehalose is effective as cryoprotectant as long as applied as a pretreatment of cells, indicates either that the cells need to be adapted prior to freezing or that the disaccharide must also be taken' up and located inside the plasmalemma to be effective. Our point of view is that the latter is true. Furthermore, high external trehalose concentration may be necessary for optimal osmolarity [63]. In vitro manipulation of animal cells and the preservation of these cells by
305 different agents also play important roles in the development of investigations in areas such as medicine, biology and biotechnology. The preservation of genetically engineered or selected cells for different applications such as the production of monoclonal antibodies, viral vaccines and hormones imposed on investigators the search of nontoxic cryoprotectants. Although, DMSO (dimethylsulfoxide) is known as a good protectant concerning cell viability it induces modifications in cell behavior, therefore being extremely toxic. Our laboratory, in another joint project with the Cell Bank of Rio de Janeiro, has been investigating the influence of trehalose on the conservation of hybridomas as well as on the preservation of human cells for autologous transplantation. Preliminary results indicate a possibility of reducing the normally used DMSO concentration (unpublished results). Noteworthy is the fact that trehalose was found to be a cryoprotectant not only of cells but also of organs for surgical transplant. Various kinds of solutions can preserve organs such as kidneys for transplantation [@I. However, the usual preservative solutions have been unable to protect lungs in a clinically useful condition, because of the unique structure and requirements of the lung [65]. Hirata et al. [66], in 1994, reported a very encouraging result using trehalose for preserving canine lungs. They found that trehalose was effective in avoiding ischemia, for up to 12 h, when canine lungs were perfused with a buffer containing 35 mgil trehalose. Bando et al. [67] in 1994, also reported the effectiveness of two types of solution, ET-K and IT-K, which contain 4.1% trehalose, hydroxyethylurea and gluconate. By scanning electron microscopy they were able to suggest that trehalose may produce a stable environment around the endothelial cell membrane and act as a thermoprotectant against lung injury. They demonstrated that ET-K provides a reliable 20-h lung preservation of canine lungs, however, further studies are needed to examine the efficacy of ET-K for longer periods of time. They expect that ET-K may become of clinical use in the future. The importance of increasing the duration time of donor organs is obvious, specially for long-distance transportation. Proteins, enzymes, virhes, membranes, cells and organs have been preserved by trehalose. Therefore, these results point towards the application of these,findings not only in medical and biological research but also for industrial use. If the biotechnological applications of trehalose, described here, have been predicted to revolutionize health care, it can also be predicted that they will bring enormous benefit to the food industry. However, the two main problems concerning the utilization of trehalose-drying methods are that trehalose still has a high cost and that the air-dried foods are too harsh. The usual alternative for drying is to use low temperatures, but this requires vacuum equipment which is expensive to buy and to operate. To overcome this problem, captive atmosphere partial pressure (CAPP) drying was developed in which very dry air is circulated to remove water vapor [62]. Trehalose drying was tested for preserving different kinds of food stuff. Roser and
306 Colaqo [62] added trehalose to liquefied foods, when it can come into close contact with food molecules. They also suggest it could be used in dried-fruit purees and in egg powder which recovered their original color and taste, after prolonged storage, as soon as water was added. Trehalose is “a sweet target for agrobiotechnology” [68]; trehalose is a “sweeter way to fresher food” [62], trehalose, “the sweet protectant of life”, “trehalose drying: a novel replacement for freeze-drying” [56] are some of the titles found in the literature. Mysterious because not known by common people trehalose is, in fact, the natural defense of the desert resurrection plant. Since Mother Nature has chosen this sugar to protect life, it is our duty to find a low-cost procedure of producing, extracting and purifying this extremely useful disaccharide.
Trehalose production As the biotechnological applicability of this disaccharide seems vast, Kidd and Devorak [68] have anticipated that the first products containing trehalose might appear on the market around 1999. Therefore, producing trehalose or engineering trehalose accumulation in plants, for food production, seems a good investment. Calgene (Davis, CA) and Quadrant Holding (Cambridge, UK) have formed the joint venture Osmotic Foods (Davis, CA) which will develop trehalose-based foods and food ingredients. Also, MOGEN International (Leiden, The Netherlands) and D.J. van der Have (Kapelle, The Netherlands) a seed company subsidiary of Suiker Unie, the largest sugar cooperative in The Netherlands, are collaborating in the production of trehalose in plants. Osmotica Foods has engineered trehalose accumulation into tomato and is researching its introduction into other fruits. Furthermore, MOGEN has been trying trehalose accumulation in sugar beet and potato. This company began trehalose research in 1992 and has filed for patent protection on its technology [68]. We have also made efforts to produce, extract and purify trehalose from yeast using an inexpensive procedure. To date the cost of 1 kg of commercialized trehalose may reach US $700. The production of this protectant by a developing country, such as Brazil, will provide means of avoiding import of this expensive disaccharide. Furthermore, the inherent importance of developing different kinds of technology for trehalose production has encouraged us to invest in this area. We chose yeast cells for trehalose production based on the following facts: yeast cells accumulate intracelldar trehalose in response to biological stresses [69] and they can be easily cultivated in different media. Our laboratory has adopted an economic method for producing and extracting this disaccharide from a S. cerevisiue strain, which overproduces trehalose in response to an environmental stress. The substrate used was cane juice, a cheap and abundant raw material in Brazil. At the appropriate growth conditions the cells accumulate 20% of their dry weight as trehalose, which can be extracted and purified for further use. This process of trehalose production was filed for patent protection in Brazil in 1993 [70]. Immobilization on solid supports also leads to trehalose accumulation as described
307 by Doran and Bailey [71]. It is important to mention that placing cells on a surface, or in a matrix, may alter their metabolic state substantially, and further growth under those conditions induces responses which may cause fundamental changes in their behavior. Evidence of altered metabolism in immobilized microorganisms suggests that immobilization can affect cell function and growth [72]. It is known that immobilized yeast progeny produces 17-fold higher amounts of trehalose than the nonimmobilized ones. Therefore, it is of technological importance to develop systems that can give rise to progeny yeast cells, which can be spontaneously liberated from the solid support and collected for trehalose extraction and purification. Although a great number of scientific papers have been published about immobilization of cells on solid supports and its applicability to industrial purposes, very little has been reported concerning the investigation of their physiology under the stress conditions caused by immobilization. Parascondola et al. [72] have shown that immobilized S. cerevisiue cells grown on glucose or maltose accumulate storage carbohydrates, such as trehalose and glycogen more conspicuously than the structural carbohydrates manan and glucan. Furthermore, trehalose accumulates in higher yield than glycogen. They suggest that cell viability under nutritional starvation, caused by diffusional limitations imposed by the support, may be maintained by the accumulated trehalose. To find the best conditions in which free or immobilized yeast cells produce higher amounts of trehalose is the aim of projects dealing with metabolic engineering, which may introduce genetic and/or environmental modifications in the cells for achieving higher yields of this sugar [73].
Trehalose quantification by trehalase The present biotechnological applications of trehalose have imposed the standardization of a specific and sensitive enzymatic method for its identification and quantification. The nonenzymatic methods for quantifying the disaccharide are nonspecific and detect total carbohydrates, not only trehalose. Therefore, the anthrone and the phenol sulfuric methods can be used for detecting trehalose, as long as other sugars and sugar phosphates are previously destroyed by acid and alkaline hydrolysis. Nowadays, many investigators have been determining trehalose by HPLC analysis or by a specific enzymatic method which employs the enzyme trehalase [74]. The enzymatic method for quantifying trehalose resides in the fact that the product of trehalose hydrolysis by trehalase, glucose, can be detected by the glucose oxidaseperoxidase method [75]. Since not only S. cerevisiue, but also other yeast, fungi, bacteria, insects and some plants produce the disaccharide trehalose, the enzyme trehalase is expected to be found in all of these living organisms. It is worth mentioning that in Euglena grucilis a trehalose phosphorylase is involved in trehalose degradation [76]. In S. cerevisiae a vacuolar and a cytosolic trehalase have been described, as previously reported in this paper, and in Neurosporu crussa the enzyme was found to be intimately associated with the inner layer of the
308 ascopore wall [77]. Furthermore, the purification and properties of the soluble midgut trehalase from gypsy moth, Lymanrriu dispar, were described by Valaitis and Bowers, in 1993 [78]. In 1987, Boos et al. [79] reported the mapping and cloning of the E. coli trehalase structural gene. They identified the enzyme as a periplasmic protein induced under high osmolarity growth conditions. Therefore, the enzyme could be easily liberated by osmotic shock, which made purification easier. Using polyacrylamide gel electrophoresis, they found a protein of 58,000 molecular weight among the periplasmic proteins of the pTRE5 carrying strain. They also used minicells containing the treA' plasmid for producing trehalase. This system also provided a 58,000 dalton trehalase proving that the plasmid carried the structural gene for the periplasmic trehalase and not just a gene involved in the regulation of the enzyme. On the other hand, App and Holzer [38] have purified a cytosolic neutral trehalase from a strain of yeast (ABYS1) harvested in stationary phase. The purified, electrophoretically homogeneous preparation of phosphorylated neutral trehalase exhibited a molecular weight of 160,000, under nondenaturing gel electrophoresis, and 80,000, in sodium dodecyl sulfate-gel electrophoresis. Yeast trehalase was also purified in our laboratory by Araujo et al. [80]. This enzyme was thought to be useful for quantifying trehalose, however, its activity can be affected by regulatory mechanisms such as allosteric effectors, covalent modification by phosphorylation and different expression throughout the growth period. Thus, the control of all these influences made the process less attractive for industrial use. Therefore, we took the advantage of the fact that E. coli has a trehalase, which is I
I
I
1
J
0.20 c
. 0.05
-
0 0
11.0
34.0
51.0
68.0
85.0
m m Gel Fig. 5. Localization of trehalose activity after gel electrophoresis. The gel was scanned, 2-mm slices were extracted and trehalase activity was located in the position indicated by the arrow. (Extracted from [fill.)
309 the major protein of the periplasmic space and can be easily released by osmotic shock. Tourinho dos Santos et al. [Sl], in our laboratory, partially purified this enzyme (Fig. 5) from a transformed Mph2 strain, which carries a multicopy plasmid, pTREl1, that harbors the TreA' gene and overexpresses it. This gene is inserted into a portion containing the tetracycline resistance gene in the plasmid which also carries the ampicillin resistance gene. (E. coli transformation with the pTREl1 plasmid was performed by Elvira Carvajal, in our laboratory; the strain and plasmid were a kind gift from Peter Postma, University of Amsterdam, The Netherlands.) In 1994, Tourinho dos Santos et al. [81] obtained a trehalase preparation that, although purified only 4-fold, proved to be suitable for trehalase determination in biological samples. It was assayed for substrate specificity at pH 6.0 in the presence of 50 mM of each of the following sugars: lactose, maltose sucrose, galactose, adenosine diphosphate glucose and uridine diphosphate glucose, exhibiting total specificity. Amounts as low as 6 nmol trehalose could be measured reproducibly by the enzymatic method, while the anthrone reagent did not permit detection below 29 nmol. The apparent Michaelis constant was 0.78 mM trehalose, at pH 6.0, with a corresponding V,, of 13.9 nmol/min, as calculated with a Lineweaver-Burk plot. This Km value is 6-fold lower than that described for trehalase extracted from Frankia (4.7 mM) and 205- to 640-fold lower than that for the enzyme extracted from Ectothiorhodopsiu halochloris (0.16-0.5 M) (for a review, see Herzog et al. [82]). Having purified trehalase, we also demonstrated that the preparations produced were suitable for immobilization on solid supports such as nylon-6, spherisorb 5NH, [81] (Fig. 6), glass beads and chitin [83,84]. Among the supports tried, glass beads and chitin, the latter treated with
100
s 60
0
5
10
1s Time(days)
20
25
30
Fig. 6. Stability of immobilized trehalaseactivity during 28 days. The trehalase-spherisorbpreparation was stored in 50 mM maleate buffer, pH 6.0, containing 7 mM P-mercptoethanol, 1 mM EDTA and 50% glycerol, at 5°C. (Extracted from [81].)
3 10 hexamethylenediamine, proved to be the most adequate for our purpose, since both enzyme-support complexes were more stable than the other materials tested (unpublished results). The immobilization was tried with the aim of producing reusable microreactors for analytical purpose, e.g., for routine specific quantification of trehalose molecules in biological samples. The results obtained with immobilized trehalase proved to be adequate for routine use in baker’s yeast production in Brazil. Having successfully immobilized trehalase the process was filed for patent protection [84]. It seems unbelievable that the first recipe for brewing malt should be 5,000 years old and that we are still toiling with the understanding of yeast physiology. Research in this area, in the last 10 years, has taken a large step forward confirming the yeast cell as a model for investigation in basic science and that it can provide us with a sweet protector of life besides bread and wine (for reviews see [85,86]).
Acknowledgements The authors express their gratitude to all the students and collaborators who have joined the group and helped to build up this review with their relevant results. The authors are also grateful to the Brazilian agencies, CNPq, CAPES and PADCT/FINEP for their financial support.
References 1.
2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
Wiggers HAL. In: Wolfrom ML (eds) Trkhalose dans le sbgie ergott. Annales, 1832;1:129. (As cited by: Birch GG. In: Advances in Carbohydrate Chemistry New York: Academic Press, 1963;18:201-225.) Elbein AD. The metabolism of ap-trehalose. In: Tipson S, Hotton D (eds) Advances in Carbohydrate Chemistry and Biochemistry 30. New York: Academic Press, 1974;227-256. Grba S, Oura E, Suomalainen H. On the formation of glycogen and trehalose in baker’s yeast. Eur J Appl Microbiol 1975;2:2+31. Lillie SH, Pringle JR. Reserve carbohydrate metabolism in Succhuromyces cerevisiue response to nutrient limitation. J Bacteriol 1980;143:1384-1394. Crowe JH, Crowe LM, Chapman D. Preservation of membranes in anhydrobiotic organisms: the role of trehalose. Science 1984;223:701-703. Leopold AC (ed). Membranes, Metabolism, and dry organisms. Ithaca: University Press, 1986. Westh P, Ramlov H. Trehalose accumulation in the tartigrade Adorybiotus coronifer during anhydrobiosis. J Exp Zoo1 1991;258:303-311. Crowe JH, Crowe LM. Stabilization of membranes in anhydrobiotic organisms. In: Leopold AC (ed) Membranes, Metabolism, and Dry Organisms. Ithaca: University Press, 1986;188-210. Carpenter JF, Crowe JH. An infrared spectroscopic study of the interactions of carbohydrates with dried proteins. Biochemistry 1980;8:39163922. Belton PS, Gil AM. IR and raman spectroscopic studies of the interaction of trehalose with hen egg white lysozyme. Biopolymers 1990;34:957-961. Crowe JH, Carpenter JF, Crowe LM, Anchordoguy TJ. Are freezing and dehydration similar stress vector? A compararison of modes of interaction of stabilizing solutes with biomolecules. Cryobiology 1994;27:21+231.
311 12. CarpenterJF, Crowe LM, Crowe JH. Stabilizationof phosphofructokinase with sugars during freezedrying: characterizationof enhanced protection in the presence of divalent cations. Biochim Biophys Acta 1987;923:10+115. 13. Prestrelski SJ, Arakawa T, Carpenter JF. Separation of freezing and drying-induced denaturation of lyophilized proteins using stress-specific stabilization. Structural sudies using infrared spectroscopy. Arch Biochem Biophys 1993;303:456-464. 14. Crowe JH, Crowe LM, Carpenter AS, Rudolph AS, Aurell-Wistrom C, Spargo BJ, Anchordoguy YJ. Interactions of sugars with membranes. Biochim Biophys Acta 1988;947:367-384. 15. Crowe LmM, Womersley C, Crowe JH, Reid D, Appel L, Rudolph A. Prevention of fusion and leakage in freeze-dried liposomes by carbohydrates. Biochim Biophys Acta 1986;861:131-140. 16. Crowe JH, Crowe LM, Jackson SA. Preservation of structural and functional activity in lyophilized sarcoplasmic reticulum. Arch Biochem Biophys 1983;220:477-484. 17. Sols A. The pasteur effect in the Allosteric era. In: Komberg A, Horecker BL, Comudella L, Or6 J (eds). Reflections in Biochemistry. Oxford: Pergamon Press, 1976;19+206. 18. Panek AD. Trehalose metabolism and its role in Saccharomyces cerevisiae. J Biotechnol 1985;3:121-130. 19. Blizquez MA, Lagunas R, Gancedo C, Gancedo JM. Trehalose-6-phosphate,a new regulator of yeast glycolysis that inhibits hexokinases. FEBS Lett 1993;329:51-54. 20. Cabib E, Leloir F. The biosynthesis of trehalose phosphate. J Biologic Chem 1958;231:25+275. 21. Operti MS, Oliveira DE, Freitas-Valle AD, Oestreicher EG, Mattoon JR, Panek AD. Relationships between trehalose metabolism and maltose utilization in Saccharomyces cerevisiae. Curr Genet 1982;5:69-76. 22. Panek AC, Araujo PS, Moura-Net0 V, Panek AD. Regulation of the trehalose-6-phosphate synthase complex in Saccharomyces cerevisiae. Curr Genet 1987;11:45+465. 23. Londesborough J, Vuorio 0. Trehalose-6-phosphate synthase/phosphatasecomplex from bakers’ yeast: purification of a proteolytically activated form. J Genet Microbiol 1991;137:323-330. 24. Bell W, Klaassen P, Ohnacker M, Boller T, Herweijer M, Schoppink P, van der Zee P, Wiemken A. Characterization of the 56-kDa subunit of yeast trehalose-6-phosphate synthase and cloning of its gene reveal its identity with the product of CIFl, a regulator of carbon catabolite inactivation. Eur J Biochem 1992;209:951-959. 25. Paschoalin VMF, Silva JT, Panek AD. Identification of an ADPG-dependent trehalose synthase in Saccharomyces. Curr Genet 1989;16:8 1-87. 26. Yoneshigue FW.PurificaGao parcial e regulaGao da atividade trealose-6-fosfato sintase ADPGdependente em Saccharomyces. Master’s thesis, UFRJ. Rio de Janeiro, 1995. 27. FranGois J, Neves MJ, Hers HG. The control of trehalose biosynthesis in Saccharomyces cerevisiae: evidence for a catabolite inactivation and repression of trehalose-6-phosphate synthase and trehalose6-phosphate phosphatase. Yeast 1991;7:575-587. 28. Ferreira JC, Panek AD. Trehalose metabolism during sporulation in Saccharomyces cerevisiae. Biochem Molec Biol Int 1993;31:1081-1090. 29. Bourquelot ME. Sur un ferment soluble nouveau dddoublant le tr6halose en glucose. Comptts Rendus de 1’Academie des Sciences 1994;116:826828. 30. Ortiz CH, Maia JCC, Tenan MN, Braz-Padrao GR, Mattoon JR, Panek AD. Regulation of yeast trehalase by a monocyclic, cyclic AMP-dependent phosphorylation-dephosphorylation cascade system. J Bacteriol 1983;153:64&651. 31. Wyatt GR. The biochemistry of sugars and polysaccharides in insects. Adv Insect Physiol 1967;4:287-360. 32. Ogawa S, Uchida C. Synthesis of aminocyclitol moieties of trehalase inhibitors. Trehalostatin and trehazolin. Confirmation of the correct structure of the inhibitor. J Chem SOCPerkin Trans 1992;I:193+1942. 33. Van der Plaat JB, van Solingen P. Cyclic 3‘5’-adenosine monophosphate stimulates trehalose degradation in baker’s yeast. Biochem Biophys Res Commun 1974;56:58&586. 34. Uno I, Matsumoto K, Adachi K, Ishigawa T. Genetic and biochemical evidence that trehalase is a
312 substrate of CAMP-dependent protein kinase in yeast. J Biologic Chem 1983;285:10867-10872. 35. Thevelein JM, Jones KA. Reversibility characteristics of glucose induced trehalase activation associated with the breaking of dormancy in yeast ascospores. Eur J Biochem 1983;136:583-587. 36. Thevelein JM. Activation of trehalase by heat shock in yeast ascospores: correlation with total cellular cyclic-AMP content. Curr Microbiol 1984;lO:15F164. 37. Dellamora Ortiz GM, Ortiz CHD, Maia JCC, Panek AD. Partial purification and characterization of the interconvertible forms of trehalase from Saccharomyces cerevisiae. Arch Biochem Biophys 1986;251:205-214. 38. App H, Holzer H. Purification and characterization of neutral trehalase from the yeast ABYSl mutant. J Biol Chem 1989;264:17583-17588. 39. Kopp M, Muller H, Holzer H. Molecular analysis of the neutral trehalase gene from Saccharomyces cerevisiae. J Biol Chem 1993;268:4766-4774. 40. Keller F, Schellenberg M, Wiemken A. Localization of trehalase in vacuoles and of trehalose in the cytosol of yeast Saccharomyces cerevisiae. Arch Microbiol 1982;131:298-301. 41. Londesborough J, Varimo K. Characterization of two trehalases in baker’s yeast. Biochem J 1984;219:511-5 18. 42. Coutinho CC, Bemardes E, Felix D, Panek AD. Trehalose as cryoprotectantfor preservation of yeast strains. J Biotechnol 1988;7:23-32. 43. Hottiger T, Schmutz P, Wiemken A. Heat-induced accumulation and futile cycling of trehalose in Saccharomyces cerevisiae. J Bacteriol 1987;1695518-5522. 44. Eleutherio ECA, Araujo PS, Panek AD. Protective role of trehalose during heat stress in Saccharomyces cerevisiae. Criobiology 1993;30:591-596. 45. Hottiger T, De Virgilio C, Bell W, Boller T, Wiemken A. The 70-kilodalton heat shock proteins of the SSA subfamily negatively modulate heat-shock-induced accumulation of trehalose and promote recovery from heat stress in the yeast Saccharomyces cerevisiae. Eur J Biochem 1992;210:125-132. 46. Ribeiro MJS, Silva JT, Panek AD. Trehalose metabolism in Saccharomyces cerevisiae during heat shock. Biochim Biophys Acta 1994;1200:13!&147. 47. BataillC N, RCgnacq M, Boucherie H. Induction of a heat-shock-type response in Saccharomyces cerevisiae following glucose limitation. Yeast 1991;7:367-378. 48. Thomas DS, Rose AH. Inhibitory effect of ethanol on growth and solute accumulation by Saccharomyces cerevisiue as affected by plasma-membrane lipid composition. Arch Microbiol 1979;122:4F55. 49. Mansure JCM, Panek AD, Crowe LM, Crowe JH. Trehalose inhibits ethanol effects on intact yeast cells and liposomes. Biochim Biophys Acta 1994;1191:309-316. 50. Panek AD, Mansure JJC, Eleutherio ECA. The Role of trehalose in yeast cells under stress. Criobiology 1993;30:238-239. 51. Dinnbier U, Limpinsel E, Schmid R, Bakker EP. Transient accumulation of potassium glutamate and its replacement by trehalose during adaptation of growing cells of Escherichia coli K-12 to elevated sodium chloride concentrations. Arch Microbiol 1988;150:348-357. 52. Crowe JH, Panek AD, Crowe LM, Panek AC, Araujo PS. Trehalose transport in yeast cells. Biochem Int 1991;24:72 1-730. 53. Eleutherio EA, Araujo PS,.Panek AD. Role of the trehalose carrier in dehydration resistance of Saccharomyces cerevisiae. Biochem Biophys Acta 1993;1 156:263-266. 54. Cuber R. Physiological and metabolic role of the trehalose transporter in Saccharomyces: Master’s thesis, UFRJ. Rio de Janeiro, 1994. 55. Araujo PS, Panek AC, Crowe JH, Crowe LM, Panek AD. Trehalose transporting membrane vesicles from yeast. Biochem Int 1991;24:731-737. 56. Roser B. Trehalose drying: a novel replacement for freeze-drying. Biopharm 1 9 9 1 5 4 7 4 3 . 57. Colago C, Sen S,Thangavelu M, Pinder S, Roger B. BioRechnology 1992;10:1007-1011. 58. Blakerley D, Tolliday B, Colago CAIS, Roser B. Dry instant blood typing plate for bedside use. Lancet 1990;336:854-855. 59. Crowe J, Crove L, Carpenter JF, Aurell Wenstron C. Stabilization of dry phospholipid bilayers and
313 60. 61. 62. 63.
64. 65. 66. 67.
68. 69.
70. 71.
72.
73. 74. 75. 76. 77. 78. 79.
80. 81.
82.
83.
proteins by sugars. Biochem J 1987;242:1-10. Rudolph AS, Crowe JH. Membrane stabilization during freezing: the role of two natural cryoprotectants trehalose and proline. Cryobiology 1985;22:367-377. Hanigan PR, Maddeu TD, Cullis PR. Protection of liposomes during dehydration or freezing. Chem Phys Lipids 1990;52:139-149. Roser B, ColaGo C. A sweeter way to fresher food. New Scientist 1993;15:25-28. Bhandal IS, Hauptmann RM, Widholm JM. Trehalose as cryoprotectant for freeze preservation of carrot and tobacco cells. Plant Physiol 1985;78:43+432. Collins GM, Bravo-Shugarman M, Terasaki PI. Kidney preservation for transportation. Initial perfusion and 30 hours ice storage. Lancet 1969;2:1219-1222. Harverich A, Aziz S, Scott WC, Jamieson SW, Shumway NE. Improved lung preservation using Euro-Collins solution for flush perfusion. J Thorac Cardiovasc Surg 1986;34:368-376. Hirata T, Fukuse LCJ, Muro K, Yokomise H, Yage K, Inui K, Hitomi S, Wada H. Effects of trehalose in canine lung preservation. Surgery 1994;115:102-107. Bando T, Kosaka S, Liu C, Hirai T, Hirata T, Yokomise H, Yagi K, Inui K, Hitomi S, Wada H. Effects of newly developed solutions containing trehalose on 20-hour canine lung preservation. J Thorac Cardiovasc Surg 1994;108:92-98. Kidd G, Devorak J. Trehalose is a sweet target for agbiotech. Bionechnology 1994;12:1328-1329. De Virgilio C, Hottiger T, Domingues J, Boller T, Wiemken A. The role of trehalose synthesis for the acquisition of thermotolerance in yeast. I. Genetic evidence that trehalose is a thermoprotectant. Eur J Biochem 1994;219:179-186. Meleiro CRM, Paschoalin VMF, Panek AD. Process0 para obten@o de trealose de Saccharomyces cerevisiae. Patent no. 9303490, INPI, Brazil, 1993. Doran PM, Bailey JE. Effects of immobilization on growth, fermentation properties and macromolecular composition of Saccharomyces cerevisiae attached to gelatin. Biotechnol Bioeng 1986;28:73-77. Parascandola P, Alteris E de, Scardi V. Invertase and acid phosphatase in free and gel immobilized cells of Saccharomyces cerevisiae grown under different cultural conditions. Enzyme Microbiol Techno1 1993;15:42-49. Galazzo JL, Bailey JE. Growing Saccharomyces cerevisiae in calcium alginate beads induces cell alterations which accelerate glucose conversion to ethanol. Biotechnol Bioeng 1990;36:417-426. Shulze U, Larsen ME, Villadsen J. Determination of intracellular trehalase and glycogen in Saccharomyces cerevisiae. Anal Biochem 1995;228:143- 149. Zimmermann FK, Eaton NR. Genetic of induction and catabolic repression of maltose synthesis in Saccharomyces cerevisiae. Molec Gen Genet 1974;134:26 1-262. Marechal LR, Belocopitow E. Metabolism of trehalose in Euglena gracilis. J Biologic Chem 1971;25:3223-3228. Thevelein JM. Regulation of the trehalose mobilization in fungi. Microbiol Rev 1984;48(1):42-59. Valaitis AP, Bowers DF. Purification and properties of the soluble midgut trehalase from gypsy moth Lymantria dispar. Insect Biochem Molec Biol 1993;23:599--606. Boos W, Ehmann U, Bremer E, Middendorp A, Postma P. Trehalase of Escherichia coli. Mapping and cloning of its structural gene and identification of the enzyme as a periplasmic protein induced under high osmolarity growth conditions. J Biologic Chem 1987;262:1312-1318. Araujo PS, Panek AC, Ferreira R, Panek AD. Determination of trehalose in biological sample by a simple and stable trehalase preparation. Anal Biochem 1987;176:432-436. Tourinho-dos-Santos CF, Bachinski N, Paschoalin VMF, Silva JJ, Panek AD. Periplasmic trehalase of Escherichia coli - characterizationand immobilization on spherisorb. Brazil J Med Biologic Res 1994;27:627-636. Herzog RM, Galinsh EA, Trupper HG. Degradation of the compatible solute trehalose in Ecfothiorhodopsia halochloris: isolation and characterization of trehalase. Arch Microbiol 1990 153: 6W606. Bachinski N. Purification and immobilization of trehalase from Escherichia coli. Master’s thesis,
314 UFRJ. Rio de Janeiro, 1994. 84. Bachinski N, Paiva CLA, Paschoalin VMF, Panek AD. Purification and immobilization of Escherichia coli trehalase for specific determination of the disaccharide trehalose. Patent no. 9303489, INPI, Brazil, 1993. 85. Panek AD. Yeast - 100 years of contribution to Biochemistry. Brazil J Med Biologic Res 1993;26:337-341. 86. Panek AD. Trehalose metabolism - new horizons in technological applications. Brazil J Med Biologic Res 1995;28:169-181.
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R. El-Gewely, editor.
315
Protein electrostatics Paulo J. Martel', Ant6nio Baptista and Steffen B. Petersen MR-Center, SINTEF UNIMED, Trondheim, Norway. 'Present address: Instituto de Tecnologia Quimica e Bioldgica, Apartado 127, P-2781 Oeiras, Portugal
Abstract. Electrostatic interactions are among the strongest and long-ranged inter- and intramolecular forces, and are thus a major determinant of the properties of proteins and other biomolecules, namely function and stability. This article gives an overview on the modeling of electrostatic interactions and pH effects in molecular systems, emphasizing the different physical idealizations assumed by the different methods. These are then illustrated with several applications, including pK, prediction and the general influence of electrostatics on diffusion, binding, catalysis, stability and extraction processes. Key words: electrostatic interactions, enzymatic activity, molecular dynamics, pH, Poisson-Boltzmann equation, protein folding, protein engineering, protein stability, simulation methods, solvation, structurefunction relationship.
1 Introduction
A comprehensive understanding of biological organisms and materials necessarily involves a clear understanding of the properties of its molecular constituents. Among these, proteins have a crucial role because of their involvement in virtually any functional aspect of biological systems. The understanding of protein properties at an atomic scale is essentially an understanding of the interplay of inter- and intramolecular forces and of their consequences. Among those forces, electrostatic interactions emerge as particularly relevant because of their strength and long-ranged nature. They can act as a determinant steering force in the guidance of substrates (or other ligands) towards enzyme active sites (or other binding sites) [l-31. In many cases the diffusion of the substrate can be the limiting reaction step in enzymatic catalysis, and an increased activity can only be obtained by improving diffusion, as illustrated by the successful charge engineering of superoxide dismutase [4]. In ligand-receptor binding or in generic protein-protein association, the diffusion rate completely determines the binding rate and can be strongly affected by electrostatics, as shown in the association between yeast iso- 1-ferrocytochrome c and bovine liver ferrocytochrome b, [5]. The final binding occurring at the end of the diffusional process does not only require complementarity in terms of shape but also, in many cases, of electrostatics. This may be particularly important for highly charged molecules such as DNA, as observed between E. coli trp repressor and its operator [6]. Besides their
Address for correspondence: Steffen B. Petersen, MR-Center, SINTEF UNIMED, N-7034 Trondheim, Norway.
316 diffusional effects on enzymatic reactions, electrostatic interactions can be fundamental in the catalytic mechanism itself. The stabilization of reaction intermediates, often involving transient charge distributions, may require that the necessary electrostatic environment be provided by the protein, as seems to be the case in serine proteases [7,8]. Electrostatic effects can be of major importance in processes where charge alteration occur, as, for example, in the phosphorylation of glycogen phosphorylase [9] or the electron transfer in the cytochrome subunit of the Rhodopseudomonas viridis reaction center [lo]. Also, electrostatic homology may exist despite extensive structural differences between proteins with identical function, as is found for bacterial and eukaryotic DNA polymerases [ l l ] . An aspect which can be hardly dissociated from electrostatics is the effect of pH upon proteins, because of the charge variability of titrable residues. The resulting changes in the interplay of electrostatic forces can result in a sometimes dramatic dependence of protein function (e.g., substrate binding [12]) and stability [13] on pH. Thus, the understanding of the molecular basis of pH effects on proteins can only be obtained through electrostatics, together with a correct treatment of protonation equilibrium. Given this ubiquitous role of electrostatic interactions, their proper treatment is an essential requisite in molecular modeling studies aimed at the understanding or prediction of the function of native or novel proteins. Actually, all the examples given above also illustrate the use of computer modeling in the study of electrostatic effects. Computer modeling methods are now a common tool in biomolecular studies [14-161, the most familiar ones being molecular dynamics (MD) and energy minimization. There are several ways of modeling electrostatic interactions, each of which is often aimed at particular properties and reflects different levels of approximation in the representation of the molecular system. The use of such methods should be based on their ability to address the particular aspects under study, and it is thus convenient to know their major strengths and limitations. In the first two sections of this article we give an overview of the treatment of electrostatic interactions and pH effects in the different existing methods. We have tried to emphasize the physical meaning of their assumptions and how they relate to the behavior of real protein systems, while avoiding unnecessary theoretical details. In the third section we give an overview of the different uses of electrostatic methods and illustrate them with several selected cases. Although we have tried to cover most types of applications, it is not our intention to present here an exhaustive list of the studies on protein electrostatics. Instead, we have tried to show the large scope of electrostatic modeling methods in the study of proteins and biological systems in general, and therefore many interesting studies have necessarily been left out. The more practically oriented reader may skip the first two modeling sections and go directly to the applications section, since reference to the former is made whenever necessary. We note, however, that the application of theoretically based methods is certainly most fruitful with a reasonable understanding of the general principles behind them; and those principles are not the incidental mathematical apparatus in itself, but rather the physical basic ideas it tries to convey, which we have tried to present here in a clear and simple, although rigorous, manner.
317 2 Modeling electrostatic interactions in proteins The approaches to model electrostatic interactions in chemical and biochemical systems, either from a purely theoretical or a computational viewpoint, can be divided in two broad types. The first type of models that were developed avoided a description at the atomic level and instead treated the solute and solvent as different dielectric regions, where charges were distributed in a discrete or continuous fashion [ 17-20]. In this way the treatment of atomic electrostatic interactions was reduced to a problem of classical continuum electrostatics (CE). With the advent of computers, new methods were introduced for calculations based on simulations at the atomic level, namely Monte Car10 [21] and molecular dynamics [22,23]. These atomic-level methods became a common practice in chemical physical studies and were later extended to a wide range of systems of chemical and biological interest [ 14,16,24,25]. The atomic detail of these methods lead to a neglect of CE-based methods, whose less detailed nature became regarded as an approximation. However, the development of fast numerical and computational methods made it possible to achieve a quantitative level in CE calculations and caused a revival in the use of CE methods [26,27]. As we intend to show in this section, both types of approach play a role in the understanding of electrostatic interactions in biomolecular systems. Moreover, they can be combined in several ways in order to produce new methodologies, as discussed in section 2.3. More extensive discussions on the modeling of electrostatic interactions in proteins can be found in several reviews [2,28-331, as well as general works on the application of simulation methods in biomolecular systems [ 14-16]. An overview over current issues can also be found on the Current Opinion in Structural Biology series [34-371. 2 .I Molecular mechanics
2.1 .I Force fields The most familiar simulation method in the study of biomolecular systems is molecular dynamics (MD) [14-161. This method can be seen as a particular case of a more general approach usually referred to as molecular mechanics (MM), to which, e.g., energy minimization and Brownian dynamics also belong. The principle behind MM is the application of classical mechanical concepts at the molecular level, such as potential energies, forces, point masses, etc. In MM the system is characterized by a potential energy or “energy function”, which depends on the nuclear atomic coordinates and describes all intra and intermolecular interactions. The form of the energy function is to some extent arbitrary, since it is simply a classical parameterization of the underlying quantum mechanical behavior of the system. A typical energy function for a system having N atoms with coordinates ( r , ,rZ,...,rN) is
318
c K$
-S) J +
[ 1 +cos(rn@
dihedral
The first two sums correspond to harmonic restoring forces for the deformation of bond lengths b and contiguous bond-bond angles 8; the third sum accounts for the eventual existence of multiple minima in dihedral angles @; the fourth sum refers to the van der Waals interactions between pairs of nonbonded atoms, here modeled with a Lennard-Jones 12-6 potential; the last sum accounts for the existence of electrostatic interactions among pairs of nonbonded atoms with partial charges 4;and qj, assuming the medium is characterized by a dielectric constant E. (The interactions corresponding to these last two sums are usually referred to as “nonbonded”.) Additional terms may be included to treat explicitly special types of interactions (e.g., planar constraints or hydrogen bonds) or to account for coupling between different types of interaction. The parameters in the energy function are obtained by fitting the results from simulations using Eqn. 1 to experimental data. The term “force field” usually refers to both the energy function and the corresponding parameters. An important consequence of the parameterization process is that the force field is only appropriate for the type of properties and systems used in the fitting. For example, a force field developed for hydrocarbons will be inappropriate for proteins, and one developed for bulk water may be unable to predict its correct properties near surfaces. Even when the “right” systems and properties are used, one should not forget that the force field is an approximate effective potential, which tries to express not only the average effect of the electronic degrees of freedom (which are not included in an explicit way), but also the average nature of the set of compounds used in the fitting procedure. In essence, all terms in Eqn. 1 are of electrostatic nature, because they all arise from the electrostatic interaction of electrons and nuclei at the quantum level. One can thus regard the choice of a particular energy function as a way of grouping the (electrostatic) terms which give rise to different qualitative effects. Of the resulting groups in Eqn. 1, only one retains an electrostatic appearance, the one with the pairwise Coulombic interactions. As stated above, the form of the energy function is, to some extent, arbitrary and one may ask if the inclusion of other electrostatic terms will not be desirable. In fact, the placement of fixed charges at the nuclei positions may seem unrealistic, since electron clouds can be distorted by the effect of other charges. This electronic polarizability can be included by using an inducible point dipole at the nucleus, which can reorient in response to electrostatic interactions (see below). Some MM simulations using inducible dipoles have been done for simple water systems (see, e.g., references in [34,37]), but presently no properly parameter-
319
ized force field with explicit electronic polarizability exists for biomolecular systems, and the Coulombic interaction is still the only source of electrostatic effects in pure MM simulations of biomolecules. However, a static model based on inducible point dipoles has been proposed by Warshel et al. [28], called the protein dipoles Langevin dipoles (PDLD) model. Each protein atom is attributed a point charge and a point dipole, and the surrounding water is represented as a grid of point dipoles. All atomic positions are fixed and the only dynamical feature of the model consists on the reorientation of the dipoles on the combined charge and dipole electric field. In principle, the correct electrostatic field can be obtained by a self-consistent calculation of dipole orientations and field, though a more simple scheme is usually used [28]. Although it uses an atomic-level description, this static approach cannot be classified as being of the MM type, being in some sense more similar to the CE models to be discussed below, as noted in section 2.3.
2.1.2 Sampling configurations Equation 1 can be used to perform energy minimizations, i.e., to find a configuration of the system which is a (local) minimum of potential energy. These minimizations and other energy-based procedures are an essential tool in the refinement of the molecular structures obtained by X-ray diffraction and nuclear magnetic resonance (NMR) structural studies. Unfortunately, the appealing molecular structures emerging from these studies led too often to the wrong idea that biological macromolecules are static entities, and that to find their “correct” structure was simply a matter of finding the energy minimum of Eqn. 1. Despite some resilience of this static view, there is today a large body of experimental and theoretical evidence showing that these molecules, particularly proteins, can display extensive conformational freedom at physiological temperatures [ 14,151. The fundamental reason why energy minimization does not produce satisfactory results is that proteins under physiological conditions are not isolated, but subjected to a solvent environment which acts as a heat bath and whose collisions lead to conformational fluctuations of the protein molecule. This can be stated in thermodynamic terms by saying that the system is not characterized by the energy of a single conformation, but rather by its free energy, where entropic effects are included. At the MM level these entropic effects arise from the fluctuations of the protein and solvent configurations. Therefore, Eqn. 1 alone is not sufficient for realistic studies and it has to be combined with methods that can yield a good sample of configurations. Simulation methods must include proper sampling, if results are to be compared with thermodynamic data, and without it the usefulness of the MM approach would be dramatically reduced. In a system at temperature T the configurations will be populated with probabilities
where kB is Boltzmann’s constant and
320
and the sum is over all possible configurations of the system. (For simplicity of notation, we assume discrete configurations and neglect the fact that the system is often also at constant-pressure conditions.) The most obvious sampling procedure is to try to mimic the true temporal evolution of the system, as in the MD method. In this simulation method the energy function of Eqn. 1 is taken as the potential energy of the system and Newton’s equations of motion d2r m -. = -Vr V(rl,r2,. .., r N ) ‘dt2
(3)
are simultaneously solved for the N particles using a numerical method. This procedure leads to a realistic temporal behavior and satisfies Eqn. 2. More simplified methods can be used which try to describe the temporal evolution of only part of the system (see below). Other sampling methods can be used (e.g., Monte Carlo) that produce a sample which is not time-ordered but simply reflects the global population of configurations in Eqn. 2. These latter methods can only be used to compute “static” properties like averages and the magnitude of fluctuations. True temporal properties, such as the characteristic times of fluctuations, can only be obtained from MD simulations. Even though temporal properties are not among the most commonly computed ones in biomolecular simulations, MD is the standard method in this type of systems, because it is the most efficient for conformational sampling of large flexible molecules such as proteins. All sampling methods are ultimately based on statistical mechanical arguments that ensure a proper sampling of configurations [24], at least when the simulation time becomes very large. In quantitative terms, the sampling of configurations allows us to compute thermodynamic macroscopic properties, which are usually the averages, fluctuations, etc. of the microscopic properties. For instance, if we sampled N , configurations of the system, the thermodynamic energy of the system is simply the arithmetic average of the microscopic (potential + kinetic) energy E(rl, r2,...,rN):
E
1 = -
N,
C sampleof
E(rl,r2,..:,rN).
(4)
lr,,r*,...,rNl
Average quantities may be of interest even if they do not correspond to any thermodynamic property, such as the average protein conformation. In many cases a more thorough analysis of the sample may prove very informative, as it can reveal multiple alternative conformations relevant to the protein function (a classical example is the relaxed/tense conformational equilibrium involved in hemoglobin cooperativity).
321
2 .I .3 Solvent efsects An accurate simulation of a protein should try to reproduce its physiological or experimental environment as closely as possible. The typical environment of a globular protein is an aqueous solution, where various counterion and other small solute species may be present. Simulations in vacuum lead to a destabilization of groups at the surface of the protein, which tend to turn into the protein interior seeking nonbonded attractive interactions. This usually results on a shrinkage of the molecule and on an abnormally high number of intramolecular hydrogen bonds. The destabilization of the structure and its consequent distortion is particularly drastic for polar and charged groups at the surface, because without the shielding provided by the solvent, their electrostatic interaction can result in excessive attractions and repulsions. Therefore, the inclusion of the solvent effect in MM simulations seems a necessity. The usual procedure is to place the protein in a simulation box which is then filled with water molecules. This box usually has periodic boundary conditions, with the box being part of a periodic lattice made up of similar boxes. (In practical terms this is equivalent to say that an atom leaving the box by one side will enter again from the opposite one.). The use of periodic boundary conditions solves the problem of surface effects resulting from an abrupt end of the solvent region. If one wants to study the protein crystal, this periodicity is, in fact, ideal for the simulation. Otherwise, it can lead to unwanted periodicity effects in the system, unless a fairly large box is used. On the other hand, the box size has to be kept small in order to make the computation times realistic. The problem is that, because of the pairwise interactions in the last terms in Eqn. 1, the computational needs of a simulation will increase with the square of the number of atoms. Usually, the box is made only big enough as to leave a minimum of 10-20 A between the protein molecules in neighboring boxes. This is often combined with the use of a cutoff radius around 7-1 2 A, beyond which nonbonded interactions are neglected (or computed only periodically). The use of a cut off radius drastically reduces the number of pair interactions and also ensures that neighboring protein images do not interact with each other. Unfortunately, though this truncation procedure is acceptable for van der Waals interactions, it is known to lead to miscalculations in the long-ranged electrostatic forces [38]. A common approach is to use a truncation scheme based on groups with a zero net charge, since the leading term in the electrostatic interaction of these groups should be a dipole-dipole interaction, which falls of as r-3, instead of the r-' dependence of the Coulombic interaction. However, a typical protein always bears a large proportion of charged residues whose net charge cannot be split in neutral groups of atoms, and whose interaction will be strongly affected by the use of a short cut off radius. Yet, the simulation of an average-sized protein molecule with explicit inclusion of water molecules is prohibitive without the use of a reasonable cut off radius, so that one often faces the decision of whether to do a simulation with incorrect electrostatic interactions or not to do it at all. For the case of very large systems, an alternative to the truncation of electrostatic interactions is the fast multipole method [39], which treats the electrostatic interactions in a nonpairwise manner.
322 Besides these practical problems, the use of water molecules in protein simulations requires a force field with parameters for both the protein and water. Normally, such force field is obtained by simply combining two different force fields, one parameterized for protein-like environments and the other for bulk water. However, the surface of the protein is likely the disturb the bulk water structure near the interface. In fact, even in the case of monatomic fluids, the force fields developed for the bulk liquid using only pairwise interactions, are known to be unable to predict surface properties [40]. This problem is probably more serious for water, because of its molecular asymmetry, so that there is a risk of modeling the protein-water interaction in an incorrect way. Also, protein force fields are often fitted against experimental data for amino acid crystals and are probably more well suited to the solid-like interior of the protein than to its solvent-exposed surface. Thus, the present parameterization of the force fields being used in protein simulations can certainly be improved, and it should always be tested and further improved by comparison with experiments, whenever possible. An alternative to the explicit inclusion of water molecules in an MD simulation is the use of implicit solvation methods. In these methods the protein is simulated in vacuum but the force field is provided with additional or modified terms intended to implicitly include the effect of the solvent. These methods usually focus on the shielding of electrostatic interactions between charged groups, which, as referred to above, is one of the crucial solvent effects. Since this electrostatic screening is felt more strongly for charges interacting at large distances, the pairwise interactions appear to be described by a dielectric constant that gets larger with increasing distance. This suggests the use of a distance-dependent dielectric “constant” in the Coulombic interactions. The most popular method is the use of a dielectric “constant” proportional to the distance between the interacting charges, though more elaborated functional dependencies were proposed [28,41]. Since surface groups are more exposed to the shielding effect, another way of simulating it is to scale the atomic (partial) charges as a function of the distance from the protein surface [42]. Another approach is simply to use E > 1 in Eqn. 1, which will equally shield all Coulombic interactions, regardless of the solvent exposure of the charges involved. All these methods can include the solvent screening effect in a qualitative manner, but they cannot correctly describe the global and local effects of the solvent upon the protein structure. Implicit solvation methods based on CE are discussed in section 2.3. The presence of the solvent also affects the protein dynamics through frictional forces and random collisions. If one is interested in true temporal properties besides configurational sampling (see section 2.1.2), these effects should be included in the simulation. They are obviously present in MD simulations with explicit water molecules, but more simplified ways exist to include them. One of these is the Brownian dynamics (BD) method, where force terms for friction and random collisions are added to the force caused by the explicit atoms in the simulation. Instead of Newton’s equation (Eqn. 3) the Langevin equation is used:
323 where 5; is the friction coefficient felt by atom i andfi(t) is a random force due to collisions. Instead of the potential energy V(r),a potential of mean force (PMF) v(r) is used to compute the force due to the explicit atoms in the system. The original potential energy is not appropriate because the solvent, besides acting directly through the frictional forces and collisions, also modifies the interaction between the explicit atoms (e.g., by shielding their electrostatic interaction). The PMF v(r)is an “effective potential energy” that includes those effects (see section 2.3). The implicit-solvation methods described in the previous paragraph are an example of ad hoc PMF’s; a more rigorous approach is described in section 2.3. In BD simulations of intermolecular processes, the PMF is usually taken as the electrostatic energy obtained from a CE method (see section 2.2), as in the diffusion simulations of section 4.5. Besides the solvent itself, the aqueous solution surrounding the protein usually contains counterions of various kinds. These ions can be explicitly included in an MD simulation in amounts that reflect the solution ionic strength [43,44]. However, it is not clear if the presently available simulation times are long enough to provide a good sampling of the ion configurations around the protein, and thus properly reflect the effect of ionic strength. Only solvation models based on CE can consider counterion effects in an implicit manner, as discussed in section 2.3.
2.1.4 Free energy calculations As stated above, real systems are usually not characterized by their energy but rather by their free energy, where entropic effects are included. Experimental macroscopic quantities are often closely related with these free energies. A typical example is the binding constant in a protein-ligand system, which is directly related to the free energy difference between the bound and the free species. In such cases we are not interested in the configurational sampling per se, but rather on the resulting free energy value. Actually, as the binding example illustrates, one is usually interested in computing the free energy change of a thermodynamic process and not to compute the free energy value of a thermodynamic state. Although the free energy can be expressed as an average value [24], similarly to the thermodynamic energy in Eqn. 4,its calculation cannot be done with an MD simulation because the sampling of the latter precisely avoids the configurations which most contribute to the average. Thus, MD by itself is of no use for free energy calculations and one has to resort to specialized methods. Beveridge and DiCapua have reviewed MM-based free energy methods [45],whose details will not be of concern to us here. The more common method with biomolecules is thermodynamic integration, where an artificial path is defined between the initial and final states and successive MD simulations are then run along the path. This means that one may need 5-10 MD simulations to obtain one free energy difference (the more steps we use, the higher the accuracy). In addition, we are often interested in differences of free energy differences. A typical example is the study of the effect of a mutation on a binding constant, represented in Fig. 1. To obtain the new binding constant,
AGL
=
AG,+AG,-AG,,
324
Fig. I. Thermodynamic cycle to compute the effect of a mutation on the binding constant of a protein P and a ligand L.
we need to perform two free energy calculations (for AG, and AGJ or even four, if we cross-check the results by computing the reverse processes, as is often convenient. In conclusion, MM-based free energy methods are extremely demanding in terms of computer time and, though several studies have been done with biomolecular systems [45], their use is still not as common as normal MD.
2.2 Continuum electrostatics
2.2 .I Continuum concepts at the microscopic level MM and CE models share the fact of each being rooted in one of the fundamental theories of classical physics - the first in classical mechanics and the second in classical electromagnetism or, more exactly, in classical electrostatics, the particular case where charges are static and no magnetic effects take place. The application of both approaches to phenomena at the molecular level is meaningful because many of such phenomena can in fact be understood with a classical or semiclassical treatment. From a classical viewpoint, both theories are largely independent in the sense that classical mechanics is not concerned with the nature of the interactions in a system, only with its mechanical consequences. Thus, the inclusion of electrostatic terms in a MM force field, as the Coulombic ones in Eqn. 1, does not require any special assumption and is largely a matter of choice (guided by physical intuition) and proper parameterization. As we then noted, besides the charge-charge Coulombic terms, others may be included to describe higher-order electrostatic interactions: chargedipole, dipole-dipole, etc. But even an elaborated force field with such higher terms will remain a point-based one, with point charges, point dipoles, point quadrupoles, etc., since the continuum concepts from classical electrostatics are not being considered at all. This is consistent with MM’s purpose of describing the molecular properties as a consequence of the movements of point mass atoms. The concepts from CE are not appropriate to describe such a system, because they derive from a macroscopic level of description, where charge densities and dielectric properties can be seen as continuously changing properties in space. The fact that these concepts are adequate at a macroscopic level means that they reflect the behavior of the microscopic system, i.e., they already reflect the averaging expressed by Eqn. 2. The charge density is simply the average charge that arises at a particular point as a result
325 of the microscopic charge distribution caused by atomic movements. The dielectric constant reflects the fluctuation of dipoles at a given region when exposed to an external electric field. In principle, there is nothing which prevents these averages to be taken at a molecular scale, so that one may speak of a molecular-scaled charge density and dielectric constant. As stated above, the basic idea behind CE models is precisely to apply CE concepts at the molecular level in order to average out some of the microscopic properties, namely the solvent configuration. The task is then to choose the dielectric constant and charge density in a physically meaningful manner, and in doing this we implicitly renounce the MM picture of moving point atoms. The dielectric properties of a material are a consequence of the fluctuation of dipoles at the molecular level. These dipoles are essentially of two types: permanent and induced. Permanent dipoles occur when the distribution of charge over neighboring atoms is not symmetric; typical examples are the peptide bond and the water molecule. The relative freedom of the latter in the bulk liquid allows for a high dipolar rotation, which leads to its high dielectric constant (40).On the contrary, the peptide bond is considerably constrained in the protein backbone, specially in the protein interior, and its contribution to the overall dielectric constant of the protein is believed to be very small. In fact, its contribution is probably of the same magnitude as the one due to the induced dipoles in the protein interior. Induced dipoles arise from electronic polarization, i.e., the distortion of electron clouds by an electric field. Electronic polarization typically leads to a value of 2-4 for the dielectric constant [3 13, which is the value usually assigned to the protein molecule [46]. The resulting dielectric regions are shown in Fig. 2, a cavity with a small dielectric constant % (the protein) immersed in a medium with a high dielectric constant E, (the solvent). The boundary between both regions is usually taken as the surface accessible to the solvent, defined as the one generated by the center of a probe sphere with the size of the water molecule rolled over the protein surface [47].
Fig. 2. Continuum electrostatic model of a protein in solution.
326 The charge density in the protein molecule can simply be assumed to consist of the formal charges of titrable sites and bound ions or, in a more detailed model, of the partial charges of all atoms. It is usually convenient to make a distinction between the charges belonging to titrable sites and the rest, the background charges (see section 3). To average out the protein conformational freedom we can assign the charge positions using the protein average conformation, obtained experimentally (e.g., by X-ray or NMR) or from an MD simulation. In contrast, the free solvent molecules contribute, on the average, with a zero charge to the solvent region. It may seem that, by stating this we are neglecting the preferential orientations of water molecules that may be induced near the protein surface because of the presence of charges in the protein. Actually, this effect is already included in the model through the existence of the protein-solvent dielectric interface, and need not be considered again (see below). What should be taken into account is the average behavior of solution counterions, whose distribution around the protein will be affected by the charges in the latter. Thus, the proper counterion charge density (i.e., the average charge taken over counterion configurations) around the protein should also be included in the model. The counterions cannot approach the protein more than allowed by their radii, which defines an ionic boundary as shown in Fig. 2. The counterion distribution is usually assumed to be determined simply by the electrostatic potential and the solution ionic strength I , as in the familiar Debye-Huckel theory of ionic solutions (see below). In this way we complete the specification of the charge density of the system. The final CE model of the protein system is shown in Fig. 2.
2.2.2 Electrostatic quantities and free energies After having characterized a given system by its dielectric constant ~ ( rand ) charge density p(r) at each point r in space, the electrostatic potential $(r)can be determined as the solution of Poisson's equation for an inhomogeneous medium [48]:
For a homogeneous medium of dielectric constant E with a set of point charges [ q l , q2,...,qM}placed at the positions [rl, r2,...,r M } the , solution to the Poisson equation is the familiar Coulombic potential:
When considering the presence of counterions, the local ion concentration is usually assumed to be determined only by the local electrostatic potential, as in the Debye-Huckel theory of electrolytes [49,50], and Eqn. 7 becomes the PoissonBoltzmann equation (PBE), whose linear form (LPBE) is
327
where the charge density p,(r) refers only to the protein charges, i.e., the counterion effect is totally contained in the second term of the equation. In the counterion region K is the inverse Debye length
and is zero otherwise. Alternatively, one may use the nonlinear form of the PBE (NLPBE), which for the case of a 1:l salt takes the form
v*[~(r)V$(r)] - &(r)K2(r)sinh[@(r)e/kBT]kBT/e+4scpp(r) = 0,
(1 1)
For small values of the potential, second and higher order terms of the sinh expansion can be dropped, yielding again Eqn. 9. One of the consequences of the NLPBE is the loss of linearity of CE, i.e., the superposition of the potential arising from independent charges. There is some controversy about the validity of the NLPBE [49,51], but this will not be of concern to us here. Unless otherwise stated we will consider here only the case where linearity holds (i.e., the Poisson or LPB equations). From the electrostatic potential most electrostatic properties can be computed. The electric field is the negative gradient of the electrostatic potential:
E(r) = -V$(r). E is also the force experienced by a unit charge at r (assuming that the charge does not perturb the field), so that
is the force acting on a charge q placed at r. The electrostatic energy can be seen as the reversible work required to bring all charges from zero to their actual values. For both the Poisson (where p = p,) or LPB equations the electrostatic energy of the system is given by the volume integral
328
which for a protein with a set of point charges ( q l , q2,...,qH} placed at the positions ( r l , r2,...,rM) simplifies to
Because of the linearity inherent of CE (see above), the potential at each point can be decomposed in the several contributions arising from each charge in the system, and we can write the energy as
where CP(r,,rj) is the potential produced at point ri by a unit charge placed at rj. In this way, the electrostatic energy can be seen as a sum of pairwise terms, an interpretation which will prove very useful. It is often convenient to group separately the terms with i = j and the ones with i # j . Each term with i = j reflects the seljenergy arising from the interaction of the charge with the polarization charge it induces at the dielectric boundary (which is implicitly included in the calculations by the use of two dielectric constants) and with the ionic charge density it induces in the solvent. This self-energy also includes the (infinite) Coulombic interaction of the charge with itself. The terms with i + j reflect the interactions among different protein charges, both directly (as pairwise Coulombic terms) and indirectly (with the polarization charge and ionic charge density they induce). If we group the contributions for all charges we can thus write:
When computing free energy differences, the infinite Coulombic self-energy term can often be canceled, as in the examples below. We now examine the physical meaning of the electrostatic energies computed within a molecular CE model. In fact this quantity is not a real energy, since some of the solvent and protein configurational freedom has been implicitly averaged through the choice of dielectrics and charges, thus making it resemble a free energy. Yet, we cannot simply write W = G. In the first place, Eqn. 15 would imply that the free energy of an uncharged protein is zero, which is obviously false. This is actually not a problem, since one is usually interested in computing free energy differences,
329 as noted in section 2.1.4. Thus, the relevant quantity is the electrostatic energy difference between the final and initial states, so that we may write, tentatively, AW = AG. However, this still is not a proper relation, because some aspects of the protein behavior cannot be included through electrostatic concepts. A particularly important aspect which is not included in the CE model is the apolar interaction with the solvent, which gives rise to the so-called hydrophobic effect. Thus, the electrostatic energy differences computed for a CE model should be viewed as an electrostatic free energy change, i.e., as the electrostatic contribution to the free energy change: AW . When the process under study is believed to have other important contributions (e.g., hydrophobic), it can usually be split into several steps whose free energy changes can be taken as being purely electrostatic, hydrophobic, etc., and which can be computed separately. A typical example is the free energy of solvation of a charged molecule, i.e., the free energy change when it is transferred from gas to liquid phase (e.g., water). This free energy change can be computed using a thermodynamic cycle involving the discharging and recharging of the molecule, as shown in Fig. 3. Steps (1) and (3) can be computed with a CE method as
Step (2) can be obtained with, e.g., a method based on surface areas [52]. Some aspects of the treatment of electrostatic interactions in CE methods may actually be more accurate than in MM-based free energy methods (section 2.1.4), because the latter relies on the use of painvise interactions which neglect polarization and manybody effects on water, effects that are implicitly included in CE methods through the use of the dielectric constant. In practical terms, a very attractive feature of CE methods is that they require 3-4 orders of magnitude less computer time than the
Fig. 3. Thermodynamic cycle to compute a solvation free energy.
330 MM free energy methods. However, when using CE methods one should always make sure which free energy terms are being considered and which ones neglected. Particular care must be taken when comparing CE calculations with experimental or MM results. One additional comment should be made about the relation between the free energy differences obtained from CE and MM methods. In a strict sense, in a free energy all energetic terms contribute in a nonadditive manner, and it is in general not possible to separate them [53]. Thermodynamically, this can be seen as a consequence of the fact that, although the total free energy is a state function whose value is path-independent, the individual contributions depend on the order and nature of the corresponding subpaths. Therefore, it is not possible to say in a unambiguous way what is the “electrostatic contribution” in a free energy difference. In fact, one may say that what CE methods do is to define what that contribution is, by writing AW - AGeleCtr. In terms of MM methods, AGelectrcould then be computed as the free energy difference between the real protein and a hypothetical uncharged one. However, it is important to clearly state what “uncharged” means: a protein without formal charges or a totally uncharged one without atomic partial charges. The two cases may give different AGeleCtr values, both for the CE and MM calculations. In the remainder of this article, and unless otherwise stated, the term “electrostatic free energy difference” will be understood as defined by some particular CE model (which includes the definition of the uncharged form) through the above equality.
2.2.3 A summary of CE methods The solution of either Eqn. 7, 9 or 11 can be obtained from analytical or numerical solutions, depending on the complexity of the problem. When the system has some symmetry it is usually possible to express the solution in an analytic form. But symmetry is not a common feature of real proteins, and its presence in a protein CE model is the result of more or less extensive approximations. Historically, the first and most simple CE model is due to Born [17] in his treatment of ionic solvation. In Born’s model, the spherical symmetry occurs naturally for both the low dielectric cavity and the charge distribution, yielding a very simple analytical solution. The solvation free energy is computed as the difference in electrostatic energies of the ion cavity in the gaseous low dielectric media, and the water higher one (i.e., step 2 in Fig. 3 is neglected). An important step was taken by Debye and Huckel in their theory of electrolyte solutions [ 181 by including ‘theeffect of the ion concentration trough the formulation of the LPBE (9). Again, the natural spherical symmetry of ions leads to an analytical solution for the electrostatic potential. In the Debye-Huckel theory the aim is to obtain the excess chemical potential due to the ionic charge, which can be computed as the work of charging the ions of each type from a hypothetical neutral state to their full charge. The first introduction of asymmetry in CE models was done by Kirkwood [20], who extended the Debye-Huckel model to the case of a set of point charges arbitrarily placed inside a spherical cavity. Although the electrostatic potential can be
33 1 expressed analytically with the help of some special mathematical functions (spherical Bessel functions and spherical harmonics), these have to be computed somehow, which means that in practice some numerical calculations are necessary. The first application of CE models to proteins was done by Linderstrom-Lang [54],who modeled a protein molecule as a sphere with its total charge smeared uniformly over the surface. This corresponds to assuming that charged groups (in particular titrable sites) are equally likely to lie at any position on the surface, a reasonable assumption considering that at the time the ideas on protein structure were mostly speculative. The model and the corresponding solution for the electrostatic potential are very similar to the ones in the original Debye-Huckel theory. The electrostatic energy of the system can then be used to predict protein titration curves, as explained in section 3. The asymmetric Kirkwood model is better suited to a well-defined protein native structure than the Linderstrom-Lang one, and can be applied directly to proteins once the charge positions on the protein are known. This was the model used by Tanford and Kirkwood in their theory of protein titration [55] (see section 3.3), and is shown in Fig. 4 (cf. Fig. 2). The use of a spherical shape poses some problems when applying the method to real proteins. The radius b of the sphere can be chosen from the mass/density ratio of the protein, its radius of gyration, etc. Usually, all charges are considered to be at the same depth from the surface of the sphere [56] (i.e., the same d,) and separated by the experimental (e.g., crystallographic) rij distances. An ad hoc procedure to account for deviations from spherical shape in the TanfordKirkwood (TK) model has been suggested by Gurd et al. [57], which uses the solvent accessibility of the residues. The electrostatic energy term corresponding to the interaction between two sites i and j , W,,(see section 3.3) is corrected with the relative (i.e., compared to fully accessible) solvent accessible areas (SA, and SA,) of the respective amino acid residues:
Fig. 4. The Tanford-Kirkwood model.
332
Wi,! = Wij[1 - (SA, + SA,)/2].
(19)
This approach is usually referred to as the Modified Tanford-Kirkwood Method (MTK). Another modified version of the original TK model can be obtained by placing some of the charges in the outer (solvent) region [58]. In general, although the spherical TK model is very consistent with its original physical assumptions, the inclusion of further atomic detail is always somewhat artificial and cumbersome. A more realistic representation of the protein molecule, corresponding to the type of model shown in Fig. 2, usually implies a loss of any geometrical symmetry, meaning that one has to resort to numerical methods to solve the Poisson or PB equations. The method of finite differences is the most common one in protein applications and was used to solve the Poisson [59], LPB [60,61] and NLPB [62] equations, generically referred to as FDPB (finite difference Poisson-Boltzmann) methods. Other numerical methods have been used, namely finite elements [63] and boundary elements [64-661. The choice of a particular CE method is usually the result of a compromise between the atomic detail of the model and the computation time. With the current computer power, a finite difference calculation for any fairly large protein should be feasible. However, for some applications (see section 3) a huge number of successive CE calculations is necessary and the use of the most detailed methods may be a problem. Also, the high detail of some of the methods does not necessarily imply a high accuracy, because a proper parameterization is necessary (like in MM methods). For example, the results differ when partial charge sets from different MM force fields are used, which suggests that a specific parameterization for each CE method may be the more correct procedure [52]. Another criterion for the choice of a particular CE method should be the electrostatic quantities one is interested in. For example, though the visualization of the electrostatic potential around the protein can give valuable insight on its function (see section 4.1), its calculation is only meaningful when some molecular detail is included. More simple methods like the TK one can be used to compute electrostatic energies, but their electrostatic potentials are not particularly useful, since the spherical approximation makes the method inappropriate to map atomic-level properties. In general, if computer time is a problem, a combination of different CE methods can be a good solution to compute different electrostatic quantities (see section 4.5).
2.3 Hybrid methods and overview The distinction between MM and CE methods is not as sharp as the foregoing discussion may suggest. This is clearly illustrated by force fields that try to include the solvent effect in an implicit way (see section 2.1.3). The use of high or distancedependent dielectric constants in these force fields intends to describe the solvent (or at least its shielding effects) in an average manner through the use of CE concepts,
333 while keeping a MM approach to describe the protein conformational freedom. In these force fields the energy has some of the characteristics we attributed to the electrostatic energy computed with CE methods (section 2.2.2), i.e., it is like a partial free energy that already includes the average effect of part of the system (the solvent) or, more exactly, it is a PMF. The use of CE-like electrostatic terms to describe the solvent shielding was recently brought to its “logical” conclusion: to obtain the electrostatic term of the force field energy from a complete CE calculation [67-751. In this way not only the effect of the solvent is included, but also the one of ionic strength. This still, is an area in development and the heavy computational needs will certainly preclude an “exact” application of the CE part (one CE calculation per protein conformation) for some time. Moreover, such hybrid force fields will require a proper parameterization: one cannot simply strip out the electrostatic terms from an existing force field and append the CE-computed electrostatic energies. An additional problem with this approach is that other (nonelectrostatic) solvent-protein interactions are not considered, so that, e.g., hydrophobic-like effects could not be reproduced in a simulation. Also, even if the “energy” or PMF of such hybrid force fields somehow manages to include all types of interactions in a correct way, there is no guaranty that the temporal behavior is maintained, i.e., they may generate a good sample of protein conformations but be unable to compute true temporal properties (see 2.1.2). In any case, this combination of CE and MM is probably the more promising route to bring force fields with implicit solvation into a quantitative stage. Another method which shares similarities with both MM and CE models is the PDLD method referred to in section 2.1.1. Despite the atomic level of the model, typical of MM, the protein static structure and the inclusion of dipolar orientation effects closely resembles CE models. In fact, as pointed out before [2], the use of point dipoles in a static structure is roughly equivalent to the use of a spatially dependent dielectric constant. As for the water dipoles, fixed in a grid, they should probably be seen as representing the average effect of the solvent, and not as individual solvent molecules, whose positions will most certainly be inaccurate. In overview, the choice of a model to describe electrostatic interactions in proteins should be based on the kind of properties one is interested in and on the computer power that is available. If conformational changes are believed to be important, an MM or hybrid method should be used. A pure MM approach is the only safe route to examine the temporal behavior and probably the one that better accounts for details of the protein-solvent interactions (e.g., hydrogen bonding). If the computation time is a problem one may resort to the use of implicit solvation, with MM-CE hybrid methods emerging as a particularly promising (though still somewhat heavy) choice. If conformational changes are unlikely to be of interest, the use of pure CE methods is probably the best solution. In fact, for a static protein, the CE treatment of electrostatic interactions may actually be better than that of the MM approach (see section 2.2.2). An additional advantage is that in CE methods the ionic strength effect can be easily included, whereas in MD simulations the sampling of ion positions is probably insufficient to yield a proper effect. The choice within CE methods was already discussed in section 2.2.3.
334 Finally, we stress the necessity to clearly identify what is being neglected or approximated in each method and how this is reflected on the particular case being studied. Incorrect modeling can lead to erroneous ideas or conclusions and is often worse than no modeling at all.
3 Modeling the effects of pH on proteins The direct result of a pH change is a modification in the equilibrium concentrations of the protonated and deprotonated forms of titrable sites, whose most pronounced consequence is a corresponding change in the balance of charges. Thus, as noted in the introduction, the consideration of pH effects on proteins is to a great extent a problem in electrostatics. Furthermore, all the methods in the previous section assume a given set of charges on the protein, whose choice will certainly depend on pH. Therefore, besides being necessary to understand pH effects, a discussion of protein electrostatics is somewhat incomplete without considering these same effects. At a first glance the theoretical task of explaining and predicting these pHdependent electrostatic changes may seem straightforward - given the pKa of the titrable residues (available in any biochemistry handbook) it would be a trivial matter to tell whether a given group is charged or not at a particular pH value. Unfortunately, the situation is far more complicated because the other charged sites and the local environment in the protein may shift the pKa of a given site from its typical value by several pH units (e.g., [76-781). In fact, as shown below, even the usual concept of pKa becomes, to some extent, inappropriate. In this section we present some general considerations on the problem and an overview of the existing methods to deal with it.
3.1 The protein titration 3.1.I Intrinsic pKa’s Let us start by considering a single titrable site. In this case the protonation equilibrium is fully described by the pK, of the site, through the familiar HendersonHasselbalch equation of acid-base equilibrium: pKa = pH+log-
f
,
1 -f
where f is the degree of protonation, i.e., the fraction of molecules that has the site protonated. From this equation is also obvious the usual meaning of the pKa as the pH value at which the site is half-protonated (the logarithm vanishes). The “typical” pKa values of titrable amino acids (i.e., the ones obtained with analogous model compounds) can be found in any biochemistry handbook, but, as noted above, this is not of much help when the amino acid is part of a protein molecule (or, in general,
335
AH
1
*en"( AH) P-AH
pKmod ___)
A+Ht
1
A P n v(A)
+ P-A+H+ flint
Fig. 5. Thermodynamic cycle to compute the effect of inserting a titrable amino acid (A) in a protein molecule (P).
part of a macromolecular complex). The reason is that the environment around the titrable site changes when the residue is "transferred" from the solution into the protein, so that its protonation equilibrium becomes perturbed. Thus, the pKa value measured for the model compound in solution (pK"'&) reflects an aqueous environment which is different from the one in the protein molecule [79]. In general, the charged form becomes less stable in the apolar interior of the protein, unless it is stabilized by nearby dipolar groups. The pKa shift occumng because of this insertion in the protein can be computed with the help of the thermodynamic cycle in Fig. 5. The pK, value of a titrable site in an otherwise neutral protein is usually referred to as the intrinsic pKa [55], p P , and is given by
where MG""' is the free energy change of removing the proton in the protein environment relative to solution: M G """ = AG "'(A) - AG ""'(AH).
(22)
Therefore, to obtain the pKiintof any protein titrable site i, we only need to devise a method of computing M G Y . .3.1.2 Multiple protonation If we had only one titrable site in the protein molecule, the protonation equilibrium would still be given by Eqn. 20, with pKa = pK'"'. However, when other titrable sites exist, the interaction between them has also to be considered. The important point to note is that such interaction is mainly electrostatic and will depend on the particular charges borne by the sites. Thus, the way in which the p P of a given site is affected by a nearby one, depends on whether the latter is charged or not. But, conversely, the protonation state of the second group will also depend on the protonation of the first. The way to avoid circularity is to consider all possible protonation states simultaneously.
336
A protein with s titrable sites has 2s possible protonation states, which can be conveniently represented with a vector n = (n,,n2,...,nJ, where ni= 1 or 0, depending on whether site i is protonated or not. To characterize the protonation equilibrium of a single titrable site, we had to specify the populations of each of its two forms at each pH value through Eqn. 20. Now we need to specify the populations of each of the 2" forms of the protein at each pH value. This task is very similar to the one of specifying the populations of alternative configurations, discussed in section 2.1.2. In this case we have that the probability (i.e., the fraction of the total population) of each protonation state is [55,77,8&82]
with
AG(n) is the free energy of the protonation reaction where the fully deprotonated protein, P(O), acquires a protonation state n:
+ nH' + P(n),
P(0)
(24)
ci
ni is the number of protons necessary for the reaction. While the and n = distribution of configurations in Eqn. 2 was only dependent on the temperature, we see that the distribution of protonation states depends on both the temperature and PHFrom Eqn. 23 all protonation dependent properties can be computed. For example, the protonation degree of a given site,i, (see Eqn. 20) is simply the mean protonation of the site:
-
J. = n; =
c
p(n)n,.
n
In spite of the apparent simplicity of this equation, the dependence off, on pH is quite involved (through Eqn. 23). We can still define an effective pK, to characterize the protonation equilibrium of a given site, (cf. Eqn. 20): eff
pK,
f.
= pH+logL. 1 -J.
This pKrn can in general be written as the pKF plus an additional interaction with the other sites:
337 pKfff = pK;int --AG 1 2.3kBT
inter
,
(27)
but the interaction term would be a pH-dependent quantity ultimately contained in Eqn. 23. Thus, since the pKieff becomes itself pH-dependent, it can be no longer equated with the pH corresponding to half protonation. Yet, this half-protonation meaning is the basis of experimental pK, determinations, and so it is convenient to define also an apparent pK,, pPpp,as the pH value at which half protonation occurs, i.e., when = 0.5. In conclusion, the two physical meanings of the usual pK, no longer coincide when multiple titrable sites exist. To speak of the protonation equilibrium we have to use the pK""; to speak of the half-protonation we have to use the pPpp.Furthermore, in some extreme cases there may exist several pH values at which half protonation occurs [81,83], rendering the concept of p P p puseless. It is also worth noting that the experimental values determined by NMR are usually obtained through fitting, using Eqn. 20, and therefore do not necessarily correspond to the exact titration midpoint. Even assuming that we can compute all the pKjintand AG(n) values (see below), the task of solving Eqn.s 23 or 25, with 2$terms, is a heavy one. Several approximate methods have been proposed: the reduced-site method [80], the Monte Carlo method [84,85], partial or full mean field approximations [56, 80-821 and the predominant state approximation [82]. The reduced-site method avoids to treat sites that are fully protonated or deprotonated at the considered pH value, instead considering them as background (i.e., nontitrable) charges. The Monte Carlo method has already been referred apropos of configurational sampling (section 2.1.2) and can be used to estimate mean values like in Eqn. 25 without summing over all the 2s states. The mean field approximation assumes that, although influencing each other, the sites titrate independently, so that each of them is only affected by the mean protonation of the others (see section 3.3). The predominant state approximation consists in assuming that only the most populated protonation state exists. In practice this method does not really avoid the use of some sampling procedure, since the predominant state has itself to be determined somehow.
3.2 Molecular mechanics The MM approach is suited to deal with only one protonation state n at the time. The usual way of including pH effects in MM simulations is to choose a reasonable n for the pH region of interest. The typical choice is charged Asp, Glu, Lys, Arg and Cand N-termini, and neutral His, Tyr and (free) Cys. If more acidic conditions are pretended a charged His is sometimes used. However, this corresponds to using pK, values of model compounds, while we have seen above that the values in proteins may differ from these by several pH units. Thus, the inclusion of pH in MM simulations is not only approximate (by using a single protonation state) but also somewhat arbitrary (by using model pK,'s). Gilson [82] has suggested the use of the
338 predominant charge set obtained from a CE-based calculation (see the next section), but this approach may prove problematic because those calculations are conformationdependent (see section 3.4). Another way of using MM to address the effect of pH in proteins, would be to compute the quantities M G Y and AG(n), presented in the previous section. These could in principle be computed using the free energy methods referred to in section 2.1.4. However, as we then noted, these methods are computationally very demanding and the 2s calculations required for the AG(n) values are obviously not feasible. In conclusion, MM in its current form is not of much help in dealing with the problem of pH effects. (See however section 3.4).
3.3 Continuum electrostatics Most of the considerations of the previous section are actually applicable to CE methods as well - they cannot directly include the effect of pH in a single calculation and the only alternative would be to use them to compute all the s M G Y ' s and 2s AG(n)'s. The main difference is that this last route is feasible with CE methods, as we now show. By refemng to Fig. 5 we see that MG""' arises mainly because of the presence or absence of one charge and thus is essentially of electrostatic nature. Therefore, we can compute it using a CE method (see section 2.2.2) as: M G '"'= [W(P-A)-W(A)] - [W(P-AH) - W(AH)].
(28)
In this way we can obtain all the s p P ' values through Eqn. 21. This explicit calculation of p P " s was done with finite difference methods using an atomic-level model (see section 2.2.3) [77,81,83,86]. In these studies CE calculations were done for all the 4 molecular species in Eqn. 28, using the appropriate dielectric regions and charges in each case. It is usual to consider 2 different terms in MG'"':
which correspond to the differences of, respectively, the cost of "solvating" the residue in the protein dielectric region (MGBom),and the interaction with the background nontitrable charges (MGback).The usual procedure with the TK method (see section 2.2.3) is to neglect MG'"' and consider p P ' = Kmd. This corresponds to assuming that the site's environment is not significantly changed upon its inclusion in the protein, which may be a reasonable assumption for solvent-exposed sites. Even for buried sites there often seems to exist a stabilization of the charged form by neighboring dipolar groups, making the shift from p P d very small (see the lysozyme and bovine pancreatic trypsin inhibitor (BPTI) examples in section 4.3). The problem of computing pK"' values is also not addressed by the Lindertrgm-Lang model (see
339 section 2.2.3). The calculation of AG(n) can also be obtained by considering a cycle analogous to the one in Fig. 5 that includes all the sites that become protonated in reaction (Eqn. 24), which again results in a difference of free energies that is essentially of electrostatic nature. Additionally to the self-energy terms considered for the pKL""s, we now have also charge-charge interaction terms (see section 2.2.2), and AG(n) can then be written as [55,80]
c
-AG(n) = 2.3kBT
nipKF -
i
-c 1 2
i
yi
ziziWij,
where zi and zi are the formal charges (in protonic units, e) of the sites in the considered protonation state n i.e.,
zi =
ni-1 if the site is anionic, ni if the site is cationic,
and W,, can be seen as measuring the electrostatic interaction of the positively charged sites i and j. For the simplest models where titration is represented by the placement or removal of a single charge, W , is simply the function a@,, q) of Eqn. 16. If we use a more detailed model where all atomic partial charges are considered, Eqn. 30 can still be used, though W , acquires a slightly different form [81,83]. The explicit calculation of AG(n) through the W,,'s has been done in several studies,
<
using some of the approximate methods referred together with the calculation of to in section 3.1.2 [77,81,83,86,87]. A simpler way to compute the 7 ' s is to use the mean field approximation mentioned in section 3.1.2, which does not actually use Eqn. 23. In the mean field approximation, only the mean individual protonations are necessary, as if the resulting mean global state were the (unique) actual protonation state. The interaction term in Eqn. 27 then simply reflects the electrostatic interaction between the mean-charged sites [55,80], yielding
--c
pKieff= pKiint
1 2.3kBTyi
yii,
In terms of a physical model this is the same as considering that the gradual titration actually corresponds to a continuous change of the charges. The value of pK,"" can be obtained by solving Eqn.s 32 and 26 until self-consistency, starting with pK," as a first guess. This self-consistent algorithm based on the mean field approximation is the usual procedure with the TK model and its modifications
340 [29,56,57,88]. It has also been used with more detailed CE models (e.g., finite differences), yielding results as good as the more exact methods (e.g., Monte Carlo), except when some of the sites interact strongly and titrate in the same pH region [80,8 11. In the original Linderstrom-Lang model the total protein charge is not assigned to individual charged sites, but rather smeared on the surface of the sphere representing the protein [54] (see section 2.2.3). Although the sites are assumed to titrate independently, as in the TK model, there are no pairwise interactions being switched on or off by titration, but only the continuous change of the total mean charge of the sphere, Z , yielding
where W is the electrostatic energy of the sphere and can be easily obtained from the Debye-Huckel theory of electrolytes:
(34)
with K being the ionic strength dependent Debye-Huckel parameter (Eqn. 10) and a and b having the same meaning as in Fig. 4. Given the crudeness of this model, it is not surprising that it was unable to describe the effects of ionic strength and group interactions, as pointed out by Linderstrom-Lang himself [54]; the best results were obtained by treating it as an empirical model with adjustable parameters [89]. In fact, the historical importance of the Linderstrom-Lang model does not lay on its exactness or predictive power, but rather on the establishment of a general theoretical approach to the relation between electrostatics and titration, which served as the basis for the later developments we have described in this section.
3.4 Overview From this resumed account of the ability of MM and CE methods to model pH effects on proteins, we see that the latter are, at present, much better suited for the task. We note that this success of CE methods is ultimately due to the linear nature of the electrostatic interactions, expressed in Eqn.s. 15 and 16. This linearity is introduced through Eqn. 30, which reduces the problem of computing 2s terms into a pairwise s2 problem. Thus, although free energies are in general not decomposable in individual terms (e.g., [53]), they can be if linear CE is used. If linearity is lost, as it happens if one uses the NLPBE (Eqn. 1 l), some additional approximations are necessary in order to make the calculations feasible [90]. The choice of a particular method to model pH effects should be based on two
34 1 major aspects. The first is the large number of CE calculations involved if one wants to perform a detailed calculation of all the p P and Wijvalues. The criteria presented in section 2.2.3 are still valid, but the large number of calculations may be the determinant factor in deciding for a specific CE method. If we know that there are no large pK, shifts (e.g., from NMR data), we may simply not do the calculation of the pKL""s and use values instead; this will mean saving a significant amount of computation time. The second aspect is the sampling of the protonation states. If there are no strongly interacting sites titrating in the same pH region, the use of the mean field approximation is obviously the most economic choice. Otherwise, the Monte Car10 method has much to recommend, since it is always valid, does not require any type of additional criteria (as some of the other methods do) and the computation time increases linearly with the number of titrable sites (compared with the 2s dependence of the exact calculation). A final note should be made about the protein conformation used to perform the calculations. It has been observed that the computed p P P values show a close dependence on the protein conformation. This dependence has been observed when using different crystal forms or refinements [77,86]or different conformations from MD simulations [13,83,91], and can go up to 5 pH units! This does not mean that all proteins will show such sensitivity, but suggests that one should check the calculations with alternative conformations, when these are available. These results should not be very surprising, since we have seen in section 2.3 that the dependence of CE calculations on protein conformation has actually been used to develop implicit solvation methods. The problem is that the aforementioned pH modeling methods neglect conformationalfreedom entirely. We have recently developed a method which simultaneously considers conformational freedom and titration, which should help to solve this problem (submitted for publication).
4 Applications of protein electrostatics In this section there is a clear emphasis on the use of CE methods. This simply reflects the fact that they address directly the modeling of electrostatic interactions, contrary to MM methods, and are therefore the natural route to study the importance of electrostatic effects in proteins. With MM methods it is not easy to obtain the average effect of the solvent on the electrostatic interactions. Even free energy methods (see section 2.1.4) are not of much help in this respect, since it is not easy to define what the "electrostatic contribution" to the free energy is, as discussed before (see section 2.2.2). Therefore, CE methods have had, until now, a leading role in the study of electrostatic effects in proteins. 4.1 Visualizing electrostatics As discussed in section 2, the presence of charges in protein molecules creates an electrostatic field which spans the protein volume as well as the volume surrounding
342 it. The study of the electrostatic properties of a protein and their connection with stability and function will then require methods of visualization of this field or its effects. The electrostatic field E is equal to the symmetric of the gradient of the electrostatic potential at each point, as expressed in Eqn. 12, and it is common to visualize the potential 4) around a protein rather than the electric field. This is partly due to the fact the 4) is a scalar function, making it much easier to represent than a vector quantity like E. The field can be depicted using surfaces of constant potential, also known as equipotentials. This is exemplified in Fig. 6, where the electrostatic potential around trypsin is shown by picturing three equipotential surfaces at -lkBT/e (red), 1 k,T/e (blue) and 0 k,T/e (white). It is interesting to note how this simplified representation is sufficient to show the dipolar character of the molecule. The significance of the f l kBT/e energy levels comes from the fact that the average thermal energy of the particles in a solvent at temperature T is -k,T. Since the electrostatic energy W of a particle experiencing an electrostatic potential 4) is given by
Fig. 6; Surfaces of constant electrostatic potential field around trypsin. The contour values are -1 k,T/e (red), lk,T/e (blue) and 0 k,T/e (white). Calculations and display done with GRASP [92].
Fig. 7. The electrostatic potential on the molecular surface of acetylcholinesterase.Negative regions are colored red, and positive regions are colored blue. Some field lines are shown in green. Calculations and display done with GRASP [92].
343
the regions inside the k,T/e surface contours are those where the electrostatic energy of a charged particle is above the thermal noise, and therefore ready to be electrostatically driven by the action of the protein field. Mapping the electrostatic potential on the molecular surface of the protein is another possibility - this can be particularly useful when one is looking for electrostatic “hot spots” on the protein surface, like an ion binding site or the active site for a charged substrate. In Fig. 7 the electrostatic potential on the surface of acetylcholinesterase is depicted, showing the negatively charged region where the active site resides (red region on right half), ready to interact with its positively charged substrate [93]. There are situations where we want to look directly at the electrostatic forces acting on given parts of our molecular system or simply want to picture the electrostatic field itself. In this case one can use electric field lines to depict the trajectories of hypothetic positive probe charges placed at given points in space and moving under the sole influence of the electrostatic field, so that the latter will always be tangent to the field lines. The field lines are normally oriented according to the direction of movement of positively charged probe (from + to -). In Fig. 7 a set of field lines converging to the active site is shown;’which correspond to possible trajectories of the positive substrate. Picturing the field vectors directly is another possibility, but unless we are looking at a small set of atoms or locations, the display can easily be cluttered. Since the protonation or deprotonation of the titrable sites in a protein causes a change in their charge, the electrostatic profile which can be obtained from the previous visualization methods is actually dependent on pH. This dependence can be introduced by using either the predominant or mean charges of the residues at the pH of interest. The advantage of using the mean charges is that partial protonation is naturally accounted for and the resulting electrostatic potential is the mean potential at that pH value (assuming that the superposition principle is valid; see section 2.2.2). The mean charges can be obtained by one of the methods discussed in section 3. If computation time is a problem, a convenient solution is to compute the mean charges with a less detailed method like the MTK method (see section 2.2.3) and make only one PB calculation with those charges, in order to obtain the potential to be visualized. We have used this approach to compute equipotential surfaces for hen egg-white lysozyme (HEW)at several pH values, some of which are shown in Fig. 8. A previous study comparing lysozymes from different organisms suggests that the strong electrostatic potential gradient existing in the active site has an important catalytic role [94]. In Fig. 8 we see that this gradient is formed at pH 4, but then it gradually disappears with increase of pH, as the active site (the upper left crevice) becomes uniformly negative. Therefore, this pH-dependence of the potential gradient may help to explain the pH-dependence of lysozyme activity, whose optimal pH = 5 [95].
344
Fig. 8. Electrostatic equipotential contours for H E W , as a function of pH (colors as in Fig. 6). From top to bottom, pH values of 3, 4 and 7. The mean charges were computed with TITRA [96], and the equipotentials computed with DELPHI [97,98] and displayed with INSIGHT I1 (Biosym Technologies, San Diego).
345
4.2 The computation of pKa shifts in proteins In section 3 we discussed the general problem of modeling pH effects on proteins, in particular the determination of pKa values. The models then discussed addressed the problem from an ab initio point of view, i.e., they tried to include the various factors affecting all the titrable groups. Here we will start by looking at a related but much simpler problem, the relative changes of pKa, the so-called pKa shifts. If we introduce or remove charge from a protein through site-directed mutagenesis, a shift in pKa of the ionic residues is expected to occur, due to the associated potential change at their location. If A@is the change in potential at the location of the group, the expected pK, shift is given by [56] ApKa =
-
eA@ 2.303 k,T ’
(36)
where e is the protonic charge and z the fractional (positive or negative) charge of the group. If we take the electrostatic potential of the system to be given by the Poisson or LPB equations, we can separate the contribution of the charge mutation from the rest of the charge set, and our A@ is simply the potential at the ionic site created by a charge of the intended sign placed at the mutated site [99]. The ability of predicting pKa shifts in proteins is of particular importance in the field of enzyme engineering, since enzymatic activity is often dependent of the protonation state of a few key residues [loo]. 4.2.1 Subtilisin The bacterial serine protease subtilisin uses His64 as a general base during catalysis, accepting a proton from Ser221 as it forms a bond with the substrate [loo]. In order for catalysis to occur, His64 must be in the deprotonated form, so the pK, of this residue can be determined by measuring the change in subtilisin’s activity with pH. Russell et al. [loll used this method to determine the pKa of His64 in both the wildtype enzyme and single/double charged mutants. In this way the effect of a charge mutation upon the pKa of His64 could be measured directly. The effective dielectric constant of particular interaction can be computed using the expression
& =
e2 r,,2.3 kJI ApKJ
(37)
where z, and z2 are the charges of the two sites and rI2is the separation distance. Using this expression, constants ranging from 4 E 5 0 up to 90 were computed. This indicates clearly that a simple Coulombic model would not be able to predict the shifts correctly. First, the effective constant varies depending on the particular mutation/ApKapair. Second, for some pairs the effective dielectric constant is higher
346 than both the solvent and protein dielectric. This may be surprising at first, but it is a well known property of classical electrostatic dielectric systems [30]. The dependence of the pKa shifts on the ionic strength was also studied as expected from an electrostatic effect, their magnitude decreased with increasing ionic strength. Given the shortcomings of the Coulombic approach, it was natural to try the protein-solvent dielectric model using the PBE. Gilson and Honig used their PBE solver package DELPHI [97,98], and they found reasonably good agreement between calculated and experimental shifts for the mutations Asp99 + Ser and Glu156 + Ser [102]. The ionic strength dependence could also be predicted with good accuracy. Overall, the error in the predictions was near 0.1, which can be considered good, more so if we take into account that the error of the experimental observations was around 0.05. Gilson et al. [97] found also that the magnitude of the shifts changed only slightly when the value of the protein dielectric was changed from 2 to 10. Stemberg et al. [lo31 did another set of calculations, this time using the Poisson equation, so no attempt was made to simulate the ionic strength dependence. Instead, only the experimental results at the lowest ionic strength were modeled. The error in the computed pKa shifts was of the same magnitude as the one in the DELPHI calculations. There was one mutation (Lys213 + Thr),where the calculated shift had a larger error, but this was for a residue in a flexible region of the protein, so uncertainty in the spatial location of the charge could have been the explanation. As with the previous case, the method was able to estimate the shifts even when the effective dielectric constant was in excess of both that of the solvent and protein. For the double mutants the experimental shifts were found not to be additive, and since the calculations were done assuming additivity, the agreement was poor. A comparison of the experimental results with the two sets of calculations is show in Table 1. Overall, these results showed that the Poisson-Boltzmann approach is capable of correctly estimating the electrostatic interaction between two sites in a protein. 4.2.2 Barnase Fersht et al. [lo41 measured the pKa of His18 in the active site of barnase by fluorescence titration, using the quenching effect of the protonated form of the active Table 1. pK, shifts of subtilisin as a function of ionic strength. Values calculated with the DELPHI program [97,98] (adapted from [102]).
+ Ser
Glu256 + Ser
Ionic strength
Asp99
Experimental
Calculated
Experimental
Calculated
0.500 0.100 0.025 0.010 0.005 0.001
0.09 0.26 0.36 0.42 0.31 -
0.10 0.18 0.25 0.29 0.38 0.34
-
0.19 0.27 0.34 0.37 0.39 0.42
0.25 0.41 0.42 0.39
347 His in the fluorescence of the neighbor Trp94. This is a sensitive’method that can detect pKa changes down to about 0.03 unit. Measurements were made, for both the wild-type enzyme and a number of charged mutants prepared by site-directed mutagenesis, yielding pKa shift values that could be used to estimate the effect of the long-range charge interactions on the pKa of Hisl8. Both single, double and triple mutants were prepared, the basic idea being to preserve the wild type structure as much as possible and the mutations to be far enough from the active site that the effect on His18 could be assumed to come only from the long-range electrostatic interaction. Theoretical calculations were done with the program DELPHI [97,98], and the agreement between measured and computed pKa’s is quite good. In most cases, the calculated and experimental values are the same within experimental error (see Table 2). The magnitude of all His-charge interactions measured was small, less than 0.33 kcal/mol in every case. A theoretical calculation of the summed effect of all barnase charges upon the stability of the charged form of His18 gave also a slight destabilization of 0.3 kcal/mol. This is much too small to explain the high pKa value (7.7) of His18 in native barnase. In the case of the double and triple mutants the agreement was good insofar as the measured pK,’s were additive, because the LPB model used in DELPHI, assumes independent contributions of the charges (as in the case of subtilisin).
4.3 Titration curves and the modeling of pKa’s Presently, the most common test on the models used to treat pH effects on proteins is the prediction of p r P pvalues, i.e., the pH values corresponding to the titration midpoint (see section 3.1.2). However, in the days before NMR and other modem high-resolution techniques, computation of isolated pK,’s values could not be done in a systematic way, and was dependent on the particular protein and chemical nature of the residue (or residues) under study. Therefore, hydrogen titration curves were the general way to characterize the ionization behavior of proteins and the natural data against which to test theoretical models - the original articles of the TK model were entitled “Theory of Protein Titration Curves”. Even today, the correct prediction of titration curves is an important test on the existing methods, since the p P P pvalues only characterize the group at a particular pH value; the characterization is only Table 2. Measured vs. calculated pK, shifts of His18 in barnase charged + neutral mutants. Computed values were produced with DELPHI (adapted from [104]). Mutant
Measured ApK,
Calculated ApK,
Asp8 + Ala Asp12 + Ala Asp22 + Met Lys27 + Ala Lys49 + Leu Lys66 + Ala ArgllO + Ala
-0.14 -0.23 -0.1 1 0.06 0.01 0.16 0.18
-0.08 -0.17 -0.12 0.04 0.07 0.13 0.17
348 complete by specifying its pK"" at all pH values (see section 3.1.2). In this section we present some cases that illustrate the level of agreement between theory and experiment. 4.3.1 Lysozyme The availability of lysozyme, together with its great stability towards pH changes, has made it into a model system particularly well suited for titration and electrostatic studies (as well as many other aspects of protein science). Those characteristics can make one fairly confident that lysozyme does not suffer significant conformational changes throughout the pH range, and also that the solution structure should not differ significantly from the crystal structure [56]. As referred in section 3, a simplistic approach to protein titration would be to consider the titrable sites to behave in a totally independent way and similarly to what is observed with isolated model compounds. Quantitatively this means that the global titration curve should be the sum of the titration curves of the independent groups, each given by Eqn. 20 with pK, = p P & . Figure 9 shows a comparison between such hypothetical curve and two experimental ones, for H E W . It is obvious that the titration behavior deviates significantly from the one obtained with the p P d values. The experimental data is shown at two different ionic strengths, and the data at the higher ionic strength is closer to the independent-groups curve, which is to be expected if the perturbation is due to electrostatic interactions between the titrating groups, since the ionic strength has a shielding effect on the electrostatic interactions 20
15
10
5
9 $0
0
N
0
O
-5 0
2
4
6 PH
0
10
12
Fig. 9. Comparison between the experimental titration curves of H E W at two ionic strengths with the curve for the hypothetical case of independent, noninteracting titrable groups (experimental data from [561).
349 between them. Roxby and Tanford [lo51 did titration studies of GuC1-denatured lysozyme and they found that the titration curve could be fitted to the curve in Fig. 9, using previously measured pK"'Od'sfor individual groups [79]. Besides showing that lysozyme is denatured in GuCl, it also indicated that the previously determined pK"'O" values were consistent enough to be used for electrostatic calculations in this protein. The results shown in Fig. 9 are typical of protein titrations, with the protein curve "flatter" than the one obtained with the independent groups. This is a consequence of the pK, shifts already mentioned, which cause the pKa's to be more spread over the whole pH range, instead of being clustered at a few points where sharp charge transitions will then occur. In 1972, Tanford and Roxby applied the TK model to the titration curve of HEWL, using the then newly determined high-resolution X-ray data. They used as intrinsic pKa's the Nozaki and Tanford model compound (pK"O") values, to find if the titration curve could be derived from the effect of electrostatic interactions alone. Rather than trying to compute the energies for the 2s protonation states (see Section 3.1.2), they used the self-consistent algorithm referred to in section 3.3. The various parameter in the TK model (see Fig. 4) were set as follows: the spherical proteins radius was set to 17.5 8, on the basis of hydrodynamic measurements, the ion exclusion radius a was set to 20 A, the water dielectric constant was assigned the experimental value at 298 K (78.5) and the dielectric constant of the protein interior was set to that of liquid acetamide (4.0). The value of the depth parameter di is a more problematic choice, because the calculated energies are very sensitive to its value, as it had been found out in previous calculations [87,106]. The 1.0 8, value used by Kirkwood in small molecule calculations was found to give unacceptably large energies, in good agreement with Orttung's calculations in hemoglobin [ 1071, who had found that values between 0.5 down to 0.0 8, gave the best results. A 0.4 8, value gave the best fit to the experimental titration curves, both at 0.1 and 1.0 M ionic strength. The titration curve fit, as seen in Fig. 10 is reasonably good, and the ionic strength dependency is very well accounted for. However, five pKa values had to be set in advance, because there was no way their value could be accounted for by electrostatic interactions alone. Glu35 is noteworthy, with its 6.3 experimentally measured value well above the model compound value, and without an abnormal environment to justify it. The a-amino group, too, had to be assigned a value 0.7 above its pK"O 'd value, but in this case a hydrogen bond to Thr40 was put forth as a likely explanation. Finally there are three tyrosyl groups whose spectrophotometrically measured pKa values could not be accounted for. The necessity of preassigning the above pKa values, together with the arbitrariness in the value of di, which works as an adjustable parameter, led the authors to conclude that the TK model was inappropriate to the description of electrostatic effects in proteins. The empirical value of 0.4 8, for the parameter d seemed in contradiction with the 1.0 8, value used with success in small molecule calculations, and very difficult to account for in physical terms. The authors pointed out as a possible explanation of the failure of the model, the predominance of charged groups
350 19 17 15 13
11 9
7 5
N
3
1 0
-1
Fig. 10. Comparisonbetween experimental and calculated titrations of HEW at two ionic strength values. The Tanford-Roxby method was used, and the computation was made using the TITRA program [96].
protruding into the solvent, where they are effectively surrounded by the high dielectric region rather than the low dielectric interior. The MTK method (see section 2.2.3) was able to circumvent, to the same extent, the problems with the surface description on the original TK model. This led to the successful modeling of the pH-dependent behavior of proteins with well exposed residues, where site-site interactions dominate the pH-dependent behavior and the p P ' values deviate little from the corresponding p P d values (see below). In the case of lysozyme, things are more difficult because, as we have seen, many sites are partly buried and display pK, alterations which cannot easily be explained in terms of sitesite interactions. It is clear that a more detailed model was required, both in terms of p P ' calculation (inclusion self-energy terms) and interaction of the charged sites with polar, nontitrable groups in the protein. The latter poses serious problems when used together with a sphere model, because the polar atoms are not necessarily exposed and they may be at any depth under the protein surface, making the choice of a value for di problematic. States and Karplus [58] tried to overcome the problems with the protein surface definition in the sphere model by defining the sphere radius in terms of the protein core density, and placing the residues according to the X-ray coordinates. The transition between the protein interior and the solvent was set at the point where the protein density drops to about half the core value. In this way residues can be inside or outside the dielectric sphere, depending on the degree of exposure to the solvent.
35 1 This approach has the advantage of dispensing with both the depth parameter and the solvent accessibility correction of Shire et al. [57]. Calculation of the electrostatic interaction terms is slightly more complicated than in the original TK model, but once it is done the unit charge interaction terms Wijcan be stored in a matrix once and for all. From there the procedure follows the algorithm of Tanford and Roxby [56]. In spite of the improved physical model, the results were similar to those of Tanford and Roxby, in that it was necessary to make ad hoc modifications of the pK"' values to obtain near quantitative agreement between experimental and calculated p P pvalues. As previously discussed, this is due to the fact that not all pKa shifts can be explained in terms of charge-charge interactions- self-energies and nonelectrostatic effects were not taken into account, although the model did include a background interaction term between the charged sites and polar atoms in the protein core. When the calculations were run on triclinic and trigonal crystal forms of lysozyme, some significant differences in the pPppvalues were observed, as well as a slight change in shape of the overall titration curve. This showed that an accuracy comparable to that of the Tanford and Roxby calculations could be obtained, without the need for the adjustable parameter d. Spassov et al. [ 1081used an MTK-like algorithm with peptide bond dipole charges adjustment. The effect of the peptide bond and an empirical correction term for PIT@' dipoles was modeled by placing -0.5 and +0.5 charges at the termini of each helix, in a manner similar to by Friend and Gurd [109]. Empirical pK"' values were obtained from the pP@' using the expression pKiint= pK,"@'fa(l-AA,)
(38)
where AA,is the relative atomic accessibility of the charge atom(s) in each site, and a an adjustable parameter whose best value was found to be 1. In contrast with the previous calculations on lysozyme, these workers did not try to reproduce the pKa values and match the titration curve using the TK potential, but instead they tried to fit the interaction potential to an expression of the form 3
(39) where ru is the distance between sites and the parameters a,, a, and u3were optimized trough minimization of the difference between the experimental and calculated titration curves. The final values of a,, a2and u3 gave an excellent fit to the titration curve, and the calculated pPpp values were within 0.5 units of the experimental values for all but two cases. When Eqn. 38 was not used and the pK"' were simply made equal to the PIT@' the agreement with experiment was worse. The large Glu35 upward pKa shift in native lysozyme was very well accounted for, coming from both the correction of Eqn. 38 and the effect of a nearby helix dipole charge. The pK"'
352 correction was also necessary to account for the abnormal p P p pvalues of Tyr53 and Hisl5. Interestingly enough, the fit to the experimental titration is very good even if Eqn. 38 is not used, although the calculated p P p pare much worse (in this case the final a,, a, and a3values are of course different from those in the calculation run with corrected pK"' values). This highlights the incompleteness of the information contained in the titration curve. To our knowledge, the more detailed pKa calculation on lysozyme to date is that of Bashford and Karplus [77]. These workers used an FDPB method to calculate the various energies involved in computing the difference between pK"'"" and p F f fvalues (these are pH-dependent and should not be confused with the previously discussed pPpp; see section 3 for definitions). As discussed in section 3, the differences between p F " and pK""'" can be calculated by means of a thermodynamic cycle where 3 distinct energy terms appear, each one contributing to the total difference: MGBo" (the solvation term),MGback(interaction with background charges in the protein) and AG'""' (interaction with other titrable sites):
In order to evaluate the MGBornand MGbakterms, model compounds were built by "cutting off' each residue from the protein molecule and running separate finite difference runs. Calculation of the titration curve was done using a Boltzmann sum over the protonation state vector (see section 3 and references [55] and [80]), instead of the simpler self-consistent method of Tanford and Roxby [56]. Calculations were run on the triclinic and tetragonal lysozyme crystal forms, in order to test the sensibility of the calculations to variations in the structure, and significant differences were found in several cases (see below). For the triclinic structure calculations 11 out of the 21 calculated p P p p are within 1 unit of the experimental values; for the tetragonal structure, 10 of the values are within the same error level. This is reasonable considering that no a priori adjustments were made. In general it was observed that the differences between p P " " and p P t were larger than those between the latter and the pFff, indicating that solvation and background terms tend to be more important than site-site interactions. In most cases the Born and background terms have opposite signs, the unfavorable desolvation effect being compensated by favorable interactions with background charges (in agreement with the microscopic calculations of Warshel et al. in BPTI, see below). The active site Glu35 high pPPP value is computed as 6.3, in excellent agreement with the experimental value of 6.1. This is due to a loss of solvation which is incompletely compensated by the background terms. Tyr53 is overestimated by more than 6 pK units, but the authors suggest that difference between theory and experiment can be explained by a conformationalchange. Comparison of the triclinic and tetragonal calculations shows that only 8 of the 21 residues are within 1 pK unit of each other, the largest deviation being 3.3 pK units. Examination of the crystal structures shows the deviations to be
353 due to conformational differences in exposed side chains, with side chain interactions present in one form but absent in the other. This highlights the necessity of incorporating flexibility in the model, or sampling a family of conformations rather than working with a single structure. 4.3.2 Myoglobin and hemoglobin The solvent accessibility modification introduced by Gurd et al. [57] brought considerable improvement to the TK model (see section 2.2.3), their first applications being myoglobin and hemoglobin. Sperm whale myoglobin was chosen by these workers because of the availability of the structural coordinates and the monomeric, spherical structure (desirable if the TK model is to be applied). There were experimental titration curves available for several forms of the protein, as well as individual pKa values for the a-amino terminal, a number of histidines (NMRexperiments) and the iron-bound water. The choice of the parameters in the model was based on the known dimensions of the myoglobin molecule, and on values used in previous calculations with the TK model, but different values for the depth parameter were tried, in the range 0.0-1.0 A. When using the nonmodified TK method, the best agreement with the experiments was obtained using a depth value of 0.0, in agreement with Orttung's earlier calculations [107], but even then some experimental trends could not be correctly described. Introduction of the static solvent accessibility [47] correction led to results in excellent agreement with experiments, both in terms of pH and ionic strength dependence. However several assumptions had to be made concerning individual p p ' values. Some sites were assumed to be masked, based on reactivity studies, and where simply excluded from the calculation, while hydrogen bonded residues had their pK"' values adjusted by k0.5 pKa units. The pK"' for the iron-bond water was set to the experimental value at zero ionic strength. The arginine at position 45 was assumed to be ion-paired with a chloride ion and therefore was also excluded from the calculation. The p P t values for "normal" groups were assigned on the basis of model compound studies [79]. With this initial choice of parameters the correlation between the experimental and calculated titrations curves is very good [57]. Comparison with experimentally determined pK, values, when available, showed excellent agreement [ 1lo]. The ionic strength dependence of the iron-bound water pKa was also studied, and again the calculated values were close to the experimental data [110]. Botelho et al. [ 1113 did a comparative NMR study of myoglobins from 16 different species, and were able to assign resonances for 11 His residues based on sequence substitutions and chemical modification. Titration of these residue yielded p r f fvalues that could be compared with computations made with the MTK method of Shire et al. [57]. The calculations where based' on the sperm whale myoglobin structure, with the appropriate amino acid replacements for each species [ 1121. Calculated and observed p r f f values were in good agreement, the majority of the deviations being in the range 0.2-4.4 pKa units. Matthew et al. [ 113,1141 successfully applied the MTK method to human oxy- and deoxyhemoglobin. The difference between the titration curves of the two forms could
354 be simulated, providing an electrostatic basis for the alkaline Bohr effect [ 1151. pKa values for the histidine residues and the a-amino terminal group where found to agree well with experiment. It is noteworthy that the set of PIC"' values used in this calculation was the same used by Botelho et al. [112]. Again some residues were assumed to be masked and excluded from the calculation, based on the chemical or structural evidence (e.g., masking during monomer association). Bashford et al. [86] have recently done experimental and theoretical determinations of p r P pvalues for the histidine residues in carbon monoxy sperm whale myoglobin. The theoretical treatment was similar to the one used previously with HEWL [77] (see above), the differences being the assignment of partial charges to all atoms and the consideration of His neutral tautomers in the multiple titration of the protein. The predicted His p P P values were generally lower than the experimental values, although some qualitative agreement was observed. In contrast with the HEW study (see above), no compensation of the MGBom and MGhCkterms was generally observed, for His and other titrating groups. As for the H E W study, the theoretical p r P pvalues showed significant dependence on the crystal structures used in the calculation, with some very large differences (up to 5 units) being obtained even between independently determined structures of the same crystal form. This again points to the necessity of incorporating conformational effects in the existing models. In addition, the choices of the sets of atomic radii and partial charges were found to affect, respectively, the computed MGBomand MGback, which points to the necessity of developing parameter sets specific to CE methods. 4.3.3 Bovine pancreatic trypsin inhibitor BPTI is a small (58 residues) protein, with exceptional stability towards pH and temperature changes.The small size allowed for both X-ray and NMR structure determination [ 116- 1181, and for the assignment of individual pKffvalues for titrable residues [119,120]. The stability towards pH changes allows for broad range titration of BPTI, and makes it, like lysozyme, a good model system for electrostatic calculations. However, interpretation of the experimental data is compounded by a pH-linked structural transition near the neutral range, with the formation of a salt bridge between the amino and carboxylic terminal groups, which is absent from the X-ray structure, determined under alkaline conditions. Marsh et al. [120] assigned the individual resonances of titrable groups in BPTI using 13C-NMR, and performed titration experiments to assign pKff values for individual residues. They compared the experimental values with theoretical estimates obtained through use of the MTK method [57], obtaining good agreement between the two sets of values. Since the computation was based upon the alkaline crystal coordinates, some of the discrepancies could be ascribed to the postulated salt bridge between residues Lysl and Ala58 in the solution form. The pK"' values used where the same used in the above described myoglobin calculations from Gurd's group [ 111,1121. As in the previous applications of the MTK method, the effect of hydrogen bonding to titrable groups was included as a f0.5 perturbation in the computed pK"' values. Tyr23 was considered "masked" and excluded from the calculations. In any
355 case its inclusion would affect the other calculated values only slightly (10.04 units) [120]. Karshikov et al. [121] used their own adapted version of the MTK method to obtain estimates of the p r f fvalues for BPTI that showed somewhat better agreement with the experimental data than the above described calculations of Marsh et al. Use of slightly different parameters, the inclusion of the peptide bond partial charges and a different set of crystal coordinates were on the basis of the discrepancies between the two sets of calculations. Yang et al. [81] applied an FDPB method with full treatment of partial charges and protonation states (see Section 3 for explanation of methods) to BPTI, and obtained values in excellent agreement with experiment, except for the N-terminus and Asp5O. Evidence from NMR experiments indicates, as previously discussed, the presence of a salt bridge between the N- and C-termini at neutral pH in solution, and when this is taken into account, the computed p r f f for the termini becomes very close to the experimental values. In the case of Asp50, the error may be due to a salt bridge with Arg53, which is likely to be absent from the solution form, as both NMR and X-ray studies indicate. In Table 3 we compare the experimentally determined p F f fvalues from Marsh et al. [120] for BPTI with the various sets of computed values described above. As previously discussed, all calculations were based on the crystal coordinates, and no corrections were introduced to compensate for the deviations from the solution structure. In parallel with these calculations, all based on a continuum approach to the modeling of the solvent-protein system, Russell and Warshel [122] used their Table 3. Comparison between measured p r f f values for BPTI with computations done with various methods. See text for references.
Residue
Observed"
Marsh et al.
Karshikov et al.
N-term Glu7 Glu49 Asp3 Asp50 Tyr 10 Tyr21 Tyr23 Tyr35 Lysl5 Lys26 Lys41 Lys46 C-term
7.94 3.89 4.00 3.57 3.18 9.46 9.94 11.00 10.60 10.43 10.10 10.60 9.87 3.05
7.94 4.07 4.49 3.85 3.26 9.54 10.16 10.96 9.20 10.41 10.45 11.08 10.20 3.06
7.56 4.23 4.43 3.93 2.89 9.94 10.05 -c 10.22 10.43 10.43 10.70 10.35 3.49
"Data from Marsh et al. [120]. bTyr residues not included in the calculations. 'This site considered as masked by the authors.
Yang et al.b 7.0 3.4 4.5 3.6 1.7 -
10.7 10.8 10.3 10.3 3.6
356 microscopic method PDLD [123] (see also sections 2.1.1 and 2.3) to calculate estimates for the pk""' values of 4 acidic residues in BPTI, namely Asp3, Glu7, Glu49 and Asp5O. The calculated values were compared with the estimated value for the dissociation of an acidic group in water. The differences were of the order of 2-3 pK, units, in disagreement with the much smaller observed experimental differences. However, the results were regarded as reasonable because the detailed microscopic description of the system should be complemented with a sampling of conformational states, if a free energy is to be calculated (in the continuum methods the configurational averaging is partly implicit in the methodology, see section 3). On the other hand, the calculated energies were obtained as the cancellation of very large terms, the self-energy terms being canceled by the terms describing interactions between the sites and the surrounding charges. This fact correlates well with the observation that plTd values and pK"' are generally close, and explains why the MTK approach works so well in spite of the neglect of the self-energy terms [122]. 4.4 Protein stability
In discussing the influence of electrostatic interactions on protein stability we will address two distinct problems. Firstly, how significant is the contribution of electrostatic interactions in the maintenance of the folded state of a protein, and secondly, how do electrostatic interactions modulate the effect of environmental factors like pH or ionic strength on protein stability. In the first model of protein ionization of Linderstrcdm-Lang [54], the total charge of the protein ion was smeared on the protein surface, giving rise to an electrostatic energy proportional to the square of the protein charge (cf.Eqn. 34). The contribution of the electrostatic forces was always destabilizing, except at the isoelectric point where W = 0. In this way the isoelectric pH was predicted to be the point of highest electrostatic stabilization (or least destabilization). Although this model had some predictive power in titration modeling, it overlooks the fact that charges are concentrated on specific locations on the protein, and specific charge-charge interactions could counteract the overall destabilizing effect of the "smeared" field. The possibility that ion pairing could be the major force behind protein folding was considered in the past [124], however, different lines of evidence indicate that this is not so. Firstly, the analysis of known protein structures indicates that charges, either single or paired, prefer to be at the surface and are rarely found in the protein core [125,126]. Also, ion pairs are relatively rare, about one per 30 residues [126]. The charge reversal experiments of Hollecker and Creighton [ 1271 gave very small effects on changing the charge of groups taking part in ionic interactions. Recently, Sauer et al. [ 1281 did a series of mutations on the Arc repressor protein, replacing a buried salt bridge with different combinations of hydrophobic residues, with a significant increase in stability. The continuum electrostatic analysis of the energetics of salt bridges by Hendsch and Tidor [129] indicates that the energetic gain of a salt bridge is close to zero, and can even be slightly destabilizing, due to the desolvation penalty of burying the charged residues under the protein surface.
357 In spite of these facts, charge-charge interactions on the protein surface can have stabilizing effect, as can be seen by a number of experimental determinations. In 1972, Alan Fersht made the fi st measurement of the stabilization effect of a salt bridge in chymotrypsin, obtaining a value of the 2.9 kcal/mol [130]. The Asp70His31 salt bridge in T4 lysozyme has been found to stabilize the protein by 3-5 kcal/mol [131]. Other authors have found similar values in a number of systems [132]. Friend and Gurd [109,133] used the MTK model to compute the total chargecharge electrostatic energy of myoglobin, finding a net negative stabilization in a broad range of pH, with its maximum centered at 6.5, in good agreement with experiment. The large net electrostatic stabilization illustrates the fact that charged residues tend to be surrounded by residues of the opposite sign, as Wada and Nakamura demonstrated in their statistical analysis [ 1341. The stability maximum at pH 6.5 is well below the isoelectric pH of myoglobin, 8.30. This would be impossible in the smeared-charge model, which, as we have seen, predicts maximum stability at the isoelectric point. The main problem with electrostatic energy calculations done with the MTK method, is that only the charge-charge interaction terms are evaluated, leaving aside the interaction terms between sites, the background terms, and the selfenergy terms associated with the desolvation process [ 1351. The effective contribution of the electrostatic interactions to stabilization relative to the denatured state, can only be determined when all the three terms are included [ 1361. Yang and Honig [13] presented a detailed treatment of the effect of ionizable groups in the pH dependence of protein stability. Determination of the averages charge in the folded and unfolded states is done using methods previously described (see section 4.3), including FDPB calculations for solvation and background terms and solution of the multistate titration problem [80]. The difference of charge between the two states as a function of pH can then be used to calculate the pH-dependent electrostatic component of the free energy of unfolding, using an expression by Tanford [137]. Calculations on a model system illustrate the fact that unfolding transitions in a pH range should not be treated as a normal titration with a group titrating at the midpoint of the transition, but rather the ends of the transition indicate the pK, values in the folded and unfolded state. Calculations on lysozyme were able to reproduce qualitatively pH-stability experimental curves, and show that the electrostatic interactions are destabilizing at all pH values. The effect of specific groups on experimental stability curves can also be modeled - as an example, the 'authors consider the His3 1-Asp70 salt-bridge in wild-type T4 lysozyme. Based on studies on the H3 1ND70N mutant, Anderson et al. [ 1311 estimated the contribution of the His31-Asp70 to the stability of the native protein as being 3-5 kcal/mol. The shape of the experimental dependence of the stability on the pH was reasonably reproduced, and the pH of maximum stability of the mutant form matched the experimental value. The model predicts a stabilizing contribution for the salt-bridge, again in accord with the experiment. Yang and Honig used the above described methods to investigate the mechanism of acid denaturation of apomyoglobin. This protein displays a molten globule-like
intermediate (I)in the pH range 3-5, but outside this range only the native (N) and unfolded (0 forms are observed. The calculations of Yang and Honig indicate that the acid induced N + I transition is driven by the three residues His24, His64 and Hisll3, which display abnormally low pK, values in the N state. These low values are likely to result from a combination of charge interactions and desolvation. The second stage of acid denaturation, the transition I + U, seems to be due to the protonation of carboxylic groups with very low pKa’s, in particular Glu6 and Asp122. So in both cases the transitions seem to be driven by specific sets of residues displaying abnormal pKa values, and not by the overall increase in positive charge in acidic conditions. This indicates that a smeared-charged model (see above) would be unable to correctly describe the energy profile. Although assignment of denaturation effects to a few residues with abnormal pKa is in agreement with the conclusions of Friend and Gurd [109], based on the MTK model, the effect of charge-charge interactions is stabilizing in the range where denaturation occurs. This shows the importance of including desolvation effects in the model, one of the main problems with the MTK treatment [123]. In the case of the I + U transition, the effect seems to be due mostly to the desolvation of the carboxylic residues, charge-charge interactions playing little or no role. To summarize, the work of Yang and Honig shows that the pH-dependence of protein stability can be due to the contribution of a few residues rather than to overall charge stabilization/destabilization;the residues contribute to stability by displaying different pKa values in the native and unfolded or intermediate forms, and the pKa differences result from a combination of desolvation and specific charge-charge, charge-dipole or hydrogen bonding effects. The relative contribution of electrostatic interactions to protein stability is believed to be very small [124,136]. This comes from number of experimental and theoretical analyses, including those presented above. The negative charge-charge interaction energy that most globular proteins display near their isoelectric points is misleading, due to the large desolvation penalties associated with the burial of the polar groups. In a detailed analysis of the free energies contributions to folding, Honig and Yang [136] obtained a large positive electrostatic effect associated with the formation of tertiary structure, which offsets the large negative term coming from the burial of the hydrophobic surfaces. Theoretical calculations [ 1361 indicate that helices and sheets are only marginally stable, again with destabilizing electrostatic interactions being compensated by hydrophobic contacts. The emerging picture is that hydrophobic interactions create aggregation, while electrostatic interactions create structure, by minimizing charge-charge energies and hydrogen bond potentials.
4.5 Simulation of molecular encounters Given the long range nature of electrostatic interactions, we expect them to play an initiating role in many types of molecular encounters and more generally in the phenomenon of molecular recognition. Early electrostatic studies aimed at the recognition of complementary features in the electrostatic fields of approaching
359 molecules, especially in situations where the relative orientation of the two elements, is critical [27]. In the first approaches to the problem, simple Coulombic electrostatic models were used, with dielectric constants ranging from 1 (vacuum) to 80 (water), the potential simply being given by the Coulombic expression of Eqn. 8. With the advent of fast numerical methods for the solution of the PBE, the Coulombic equation calculations have become practically obsolete. The main problem with Coulombic models is that they fail to incorporate the effects of the presence of two regions with very distinct dielectric properties, the protein and the solvent (see section 2). As a consequence, distortions of the field at the boundary between the two media (the molecular surface), are completely overlooked. As we will see below, this boundary effect can have dramatic consequences for binding and catalysis.
4.5.1 Superoxide dismutase The enzyme superoxide dismutase (SOD) catalyzes the dismutation of the superoxide anion into 0, and hydrogen peroxide, thus protecting the cell from oxidative damage induced by superoxide. With an effective rate constant of 2 x lo9 M - k ‘ SOD is a near “kinetically perfect” enzyme [ 1381, meaning that its catalytic constant approaches the theoretical limit for a diffusion controlled reaction. However, the crystallographic structures of both human and bovine SOD showed the catalytic copper and zinc ions to be at the bottom of a narrow channel, making it difficult for the superoxide ion to find its way by nonfacilitated diffusion (Fig. 11). Early experiments showed that increasing the ionic strength causes a decrease in the reaction rate, and that the modification of charged residues alters both the rate and ionic dependence profile. These observations suggested electrostatic control of the reaction rate, apparently in contradiction with SOD’S total charge of 4 e at pH 7.0, 24
A” Glu 131
*
10
A”
Fig. 11. Schematic view of the active site of Zn,Cu-superoxide dismutase, showing the copper ion at the bottom of a narrow channel.
360 which would cause a net repulsion of the negatively charged superoxide ion and an increased rate at higher ionic strength (contradicting the experiments). However, there are patches of positive charge in the active site region, both due to the copper and zinc ions and various positively charged residues. Early Coulombic calculations showed a strong positive potential emanating from the active site [139]. Klapper et al. [61] used a numerical algorithm to solve the PBE for SOD at different ionic strengths. They observed that the particular shape of the active site cavity creates what they called “focusing” of the electrostatic field - the force field lines, diffracted by the dielectric boundary reaches further in the solvent then they would if the calculation was run with a single dielectric Coulombic model. In the case of SOD, this causes the positive patches of potential near the active site to be more pervasive (Fig. 12), an effect that could help the facilitated diffusion of the superoxide ion to the catalytic copper ions. Thus, this work provided some qualitative indication as to how SOD can achieve its high catalytic rate through electrostatic steering of the substrate, which compensates for the low accessibility of the catalytic region. The focusing and ionic strength effects lower significantly the energy banier that the superoxide ion has to overcome in order to reach the active site. In order to obtain quantitative estimates of the catalytic rates of SOD and demonstrate how the electrostatic field could provide the required rate enhancement, Honig et al. [ 1401 and Allison et al. [ 1411 made simulations of the enzyme-substrate molecular encounter, using a combination of Brownian dynamics and PoissonBoltzmann electrostatics on a cubic grid spanning the protein and the surrounding solvent. First the electrostatic potential was calculated at every point of the grid, using the LPBE (Eqn. 9), then superoxide ions, modeled as spherical negative unit charges were released at a fixed distance from the protein, and allowed to drift according to a simplified variant of Eqn. 5 (see [142]). Several thousands of trajectories were produced by releasing superoxide ions at a distance R, from the protein, and letting them drift in under the combined action of the electrostatic field and the Brownian force. Collision of the ion with the exposed copper ion surface was counted as a “productive” hit. If the substrate ions went further than a predetermined distance R,, from the protein, a new trajectory was initiated (Fig. 13). By computing the fraction of superoxide trajectories giving rise to productive hits, it was possible to estimate the diffusion rates to the active site. The values obtained with this method were 2 to 4 times larger than the experimental measured rate constants. This was to be expected, since the simulation only estimates the number of hits between the substrate and the active site, counting all hits as “productive”, the same as assuming a reaction probability of 1. Orientational and steric effects are likely to reduce the reaction probability significantly, explaining the difference between computed and measured values. The simulation could describe very well the dependence of the reaction rate on the ionic strength and the effect of some mutations on the charge residues around the active site. More recently, Getzoff et al. [4]produced mutants of human SOD with enhanced rates when compared to wild types. The rational for the mutations was to make the potential more positive at the entrance of the active site, but without disrupting the
361
Fig. 12. Comparison of Coulombic and Poisson-Boltzmann calculations of the electrostatic potential of SOD: Coulombic E = 80 (top); PB calculation E, = 2, E, = 80 (bottom). The contour values are -1 kBT/e, -0.5 kBT/e (red), 1 kBT/e, 0.5 kBT/e (blue). Calculations and display done with GRASP [92].
hydrogen bond network formed by a number of residues in the region. This was achieved through the mutations Glu132 + Gln and Glu133 + Gln, making the potential more positive but preserving the hydrogen bonding pattern of the residues Glu132, Glu133 and Lys136. The mutated forms showed a 2- to 3-fold catalytic rate increase. The Glu133 + Lys mutation was expected to produce an even greater rate increase, but a decrease was observed instead. This shows the importance of preserving the hydrogen bonding pattern, which may be responsible in turn for the correct alignment of the near active site residues [4].
362
Fig. 13. Brownian simulation of a molecular encounter with an electrostatic force. See text for the
meaning of the different parameters.
All in all these results show beyond doubt that the electrostatic field of SOD can produce a rate enhancement for the reaction catalyzed by this enzyme, and how the simulation of the electrostatic driven protein-substrate molecular encounter could provide a rational for creating “superperfect” SOD mutants [ 1431.
4.5.2 Triose phosphate isomerase Triose phosphate isomerase (TIM) is a glycolytic enzyme that catalyzes the interconversion of ~-glyceraldehyde-3-phosphate(GAP) and dihydroxyacetone phosphate. TIM is a remarkably efficient enzyme, with a measured rate constant of 4.8 x lo8 M-’s-’. The reaction seems to be controlled by diffusion [144], the diffusional encounter between TIM and GAP being the rate limiting step. Since GAP is a small negatively charged substrate, the possibility of electrostatic steering was worth being investigated. Luty et al. [145], used the UHBD simulation package [67] to simulate the diffusional encounter of GAP and TIM. UHBD implements the computational procedures described for SOD in an integrated manner. As before, the PBE was solved for the electrostatic potential at every point of a grid containing the protein. The GAP substrate was modeled as a -2e point charge, and the crystal
363 structure coordinates of TIM used to defined the excluded grid points. Reaction was assumed to occur whenever the substrate diffused to within 6.0 8, of either Ser-210 or Gly-232 in the active site. The R , and R,, values where set to 80 and 300 8, respectively. Two sets of 2000 trajectories were run, one with the inter-particle electrostatic force switched off (pure Brownian dynamics), and the other with full electrostatics. The calculated rate constants were 0.306 x 10" M - k ' for no electrostatics and 1.48 x 10" M-'S-' with electrostatics. These results make clear that electrostatic steering facilitates diffusion to the active site, but the calculated rates are two orders of magnitude above the experimental value. As with SOD, steric and orientational factors could explain the discrepancy. Firstly, the spherical model used for GAP disregards the orientational requirements for reaction initiation, which could lower significantly the reaction probability (like in SOD). Secondly, the rigid protein model does not take into account the mobile peptide loops that close over the active site upon substrate binding. These loops have to be in the open conformation for GAP to reach the active site - they could therefore act as "gates" that would explain the discrepancy between theory and experiment. To test this hypothesis, Wade et a1.([146]) did Brownian dynamics simulations of the flexible loops in TIM, but the gating period was shown to be much shorter than the enzyme-substrate relaxation time, and therefore this effect cannot account for the observed reduced rate. So at this time the more likely explanation is that orientational requirements are responsible for the lower observed rate [146].
4.5.3 Electrostatic efsects in enzymatic catalysis Electrostatic interactions can affect enzymatic activity and specificity in different ways. Specific charge-charge interactions can modulate the pH dependence of catalytic activity, by affecting the pKa's of key residues in catalysis. We have already discussed the case of His64 in subtilisin, where the pKa value can be estimated directly from the pH-dependent activity profile. The overall electrostatic field of the molecule can increase the binding rates of substrates trough electrostatic steering, as was discussed for Zn,Cu-superoxide dismutase and triose-phosphate isomerase. Electrostatic interactions may also provide stabilization for the transition intermediate, or for a particular protonation state of a catalytic residue. In the case of the serine proteases, Warshel et al. [7] used their PDLD method to calculate the stabilization effect of the buried active site aspartate on the positively charged catalytic histidine. The calculations showed that the negative charge on the aspartate contributes with a stabilization of 4 kcal/mol to the complex between the active His and the oxyanion intermediate, close to the experimental value of 6 kcal/mol [147]. In lysozyme, the distribution of charged residues creates a large potential gradient across the active site cleft, which could lower the energy barrier for the reaction by 9 kcal/mol or more. Lysozymes unrelated in sequence show similar distribution of charges and electric fields [94]. The gradient seems to be due both to the asymmetric distribution of charges and the focusing effect of the active site cleft. In section 4.1 we have shown how the pH-dependence of this gradient in HEWL seems to be in accord with its activity pH profile.
-
364 In an FDPB comparative study of rat and cow trypsin, Soman et al. [8] found that the potentials at the active site are very similar, even though the net charges of the two enzymes differ by 12.5 units. The geometry of the dielectric interface effectively shields the potentials from distant charged residues, so that the determinant of the potential in the active is the Asp102 residue. In agreement with the previously described calculations from Warshel et al. [7] the negatively charged aspartate is found to stabilize the transition complex by about 4 kcal/mol. The close agreement between the FDPB and PDLD method provides strong evidence for the stabilizing role of the catalytic aspartate in the serine proteases. Hol [ 1481 and Knowles [ 1491 pointed out the possible relevance of the a-helix macrodipole in the catalytic mechanism of various enzymes. In lysozyme, the calculations of Spassov et al. [lo81 (see section 4.3.1) indicate that the large upward pK, shift in the catalytic Glu35 residue is partly due to interaction with the macrodipole of helix 25-35. 4.6 Prediction of extraction yields in two-phase systems The use of biphasic systems is an important technique in the separation of proteins and other biological materials [ 150,1511. The observed partition reflects the relative preference of the protein for the two environments, whose origin lies on the different interactions it establishes with the respective phase components. Given the high strength and long-ranged nature of electrostatic interactions, they are expected to play a role in the extraction process. The electrostatic interaction between protein and micelle was actually the first proposed mechanism to explain the pH-dependent partition in reversed micellar systems [152]. A correlation between charge asymmetry and extraction yield was observed in this type of system for several proteins by Wolbert et al. [153]. This correlation is, however, not a general rule, as shown in a study with rat cytochrome b, charge mutants [154]. In this study, a more detailed analysis of the electrostatic interactions was carried out in our laboratory using a combination of the MTK and FDPB methods (as in the lysozyme example at the end of section 4.1). The extraction yields were found to be clearly correlated with both the total charge and the percent of molecular surface having a positive electrostatic potential, suggesting that a stronger and more extensive electrostatic interaction between the protein and the anionic surfactant facilitates the encapsulating process and/or the stability of the encapsulated form in the micellar phase. This could account very well for the relative extraction yields shown by the different mutants, as well as their pH dependence. The dipolar moment did not show any correlation with extraction, probably because of the somewhat anomalous situation of the positive pole being in a region of negative surface potential [154]. Another successful prediction of extraction yields was achieved for the Fusarium solani pisi cutinase extraction in polyethylene glycol/phosphate aqueous two-phase system (to be submitted). In this case we have extended a method of Eiteman [155], which was itself an extension to include charge effects on a hydrophobicity-based method [ 1561. Eiteman’s extension was obtained by relating the concentration ratio
365 of the neutral forms in the two phases (the ones considered in the original method) with the pH of the medium, through simple chemical equilibrium considerations; constant pK, values were assumed for the titrable sites. This method showed some qualitative agreement with experimental dipeptide partition yields in a polyethylene glycol/sodium phosphate system [ 1551. However, such approach cannot account for the extensive electrostatic effects occurring in more complex molecules such as proteins, which can shift the pK, values by several pH units, in a pH-dependent manner (see sections 4.2 and 4.3). In fact, even small solutes may display pK, shifts due to, e.g., the dielectric properties of the medium. In the cutinase study we computed the two-phase concentration ratio of the neutral, isoelectric form of the protein as a function of pH, ionic strength and dielectric constant of the medium (which depends on PEG concentration), using the TITRA program [96]. The method was able to satisfactorily predict the partition coefficients in a pH range of 6.0-9.0, where it gave consistently better results than the original, purely hydrophobic method. The results were remarkably good, considering the number of factors included in the calculation and the absence of any adjustment of the p P values required by TITRA (see, e.g., section 4.3.2). The method failed for pH = 4.5, possibly because of denaturation or aggregation of cutinase.
5 Summary and future prospects This overview of protein electrostatics shows that a wide range of methods exists, making possible a choice according to the complexity of the problem to be solved and the available computer resources. As usual, simpler models imply less accuracy, but even qualitative results can often give valuable insight on the problem at hand. When conformational freedom is believed to be important, an MM method should necessarily be used, preferably an MD simulation. Otherwise, CE methods are an attractive choice because they are much faster than MD and address directly the effect of electrostatic interactions; in fact, “protein electrostatics” is nowadays often used to refer to the use of CE methods alone. Although CE methods have been applied mostly to proteins in solution, applications to protein-membrane systems have also been published (e.g., [83,157]), and they can in principle be extended to other types of molecular associations. Also, nonaqueous solutions can be considered, as long as the solvent dielectric constant is known. Besides molecular aspects such as binding and catalysis, macroscopic properties can also be inferred from the molecular-level calculations, as clearly illustrated by the extraction predictions referred in section 4.6. Some of the future improvements in the treatment of electrostatic interactions in MM methods are more or less “obvious”, and they are mainly dependent on the increase of computer power (see below). These are the inclusion of many-body effects in the force fields, electronic polarizability in particular. Eventually, the “dynamics” of electrons could also be included directly in the simulations [ 158,1591. Also, the parameterization of such force fields should ideally be system-independent (possibly obtained from theoretical methods), so that a new empirical parameteriza-
366 tion would not be necessary for each new type of system (see sections 2.1.1 and 2.1.3). While we wait for enough computer power to make these improvements feasible, the efforts are focused on solving the problem of truncating electrostatic interactions (section 2.1.3) or finding ways of avoiding the explicit inclusion of solvent entirely (sections 2.1.3 and 2.3). Current CE methods may seem fast enough if one is interested in a single calculation, but CE-based solvation models (section 2.3) and the modeling of pH effects (section 3) clearly show the need for faster calculations. A promising new approach for CE models has been presented by Davis [160] and it is possible that new numerical techniques can further improve the computation times of this and the other usual methods. However, some serious practical problems exist with CE methods which cannot be solved with new formulations or numerical techniques, but rather arise from the underlying models. One of them is the definition of the boundary between the two dielectric regions when loosely bounded water exists, since it is not clear what dielectric value should be attributed to the water molecule region [81]. Another problem of this kind is the rigidity of the protein in CE calculations, which, as noted before (section 3.4), can lead to significant uncertainty in, e.g., computed pK, values. This assumption of rigidity on some of the applications of CE methods is at odds with its use on implicit solvation methods (section 2.3) and such inconsistency points to the necessity of clarifying which atomic properties can be included through a CE model and which ones cannot (section 3.4). In addition to the theoretical and methodological progress, computer power will determine to a great extent the progress on the modeling of electrostatic interactions. Computer power has been steadily increasing at a rate of about one order of magnitude each 5-7 years and one may confidently expect it to continue so [32]. This means that, e.g., MD simulations of the order of tens of nanoseconds or, instead, simulations on the nanosecond scale for systems 10 times larger or using much more detailed energy functions will be a trivial task in a few years. For CE methods this means higher accuracy of the Poisson and PB calculations (e.g., more grid points in FD methods) and the averaging over the multiple protonation states will no longer be a problem for pH-dependent calculations (section 3). Meanwhile, a wide range of biological and biotechnological problems are waiting to be probed with the available plethora of methods; this will hopefully result on increased biological knowledge, new technological advances and, of course, the testing of the methods.
Acknowledgements P.M. thanks Junta Nacional de Investigapo Cientifica e Tecnol6gica (JNICT), Portugal, and Instituto de Tecnologia Quimica e Biolbgica, Portugal, for grants. A.B. thanks JNICT for grants.
References I.
Berg OG, von Hippel PH. Diffusion-controled macromolecular interactions. Ann Rev Biophys
367 2. 3. 4.
5.
6. 7.
8. 9. 10. 11. 12.
13. 14. 15. 16. 17. 18.
19. 20. 21. 22. 23. 24. 25. 26. 27. 28.
Biophys Chem 1985;14:131-160. Davis ME, McCammon JA. Electrostatics in biomolecular structure and dynamics. Chem Rev 199O;90:50+521. Northrup SH. Hydrodynamic motions of large molecules. Curr Opin Struct Biol 1994;4:26+274. Getzoff ED, Cabelli DE, Fischer CL, Parge HE, Viezzoli MS, Banci L. Hallewell RA. Faster superoxidedismutase mutants designed by enhancing electrostatic guidance. Nature 1992;358:347351. Northrup SH, Thomasson KA, Miller CM, Barker PD, Eltis LD, Guillemette JG, Inglis SC, Mauk AG. Effects of charged amino acid mutations on the bimolecular kinetics of reduction of yeast iso1-femcytochrome c by bovine ferrocytochrome b,. Biochemistry 1993;32:6613-6623. Guenot J, Fletterick RJ, Kollman PA. A negative electrostatic determinant mediates the association between the Escherichia coli trp repressor and its operator DNA). Protein Sci 1994;3:1276-1285. Warshel A, Naray-Szabo G, Sussman F, Hwang J-K. How do serine proteases really work? Biochemistry 1989;28:362+3637. Soman K, Yang A-S, Honig B, Fletterick R. Electrical potentials in trypsin isozymes. Biochemistry 1989;28:9918-9926. Johnson LN, Barford D. Electrostatic effects in the control of glycogen phosphorylase by phosphorylation. Protein Sci 1994;3:1726-1730. Gunner MR, Honig B. Electrostatic control of midpoint potentials in the cytochrome subunit of the Rhodopseudornonas viridis reaction center. Proc Natl Acad Sci USA 1991;88:9151-9155. Krishna TSR, Kong X-P, Gary S, Burgers P, Kuriyan J. Crystal structure of the eucaryotic DNA polymerase processivity factor PCNA. Cell 1994;79:1233-1243. MacKerell Jr AD, Sommer MS, Karplus M. pH dependence of binding reactions from free energy simulations and macroscopic continuum electrostatic calculations: Application to 2’GMFWGMP binding to ribonuclease T, and implications for catalysis. J Mol Biol 1995;247:774-807. Yang A-S, Honig B. On the pH dependence of protein stability. J Mol Biol 1993;231:459-474. McCammon JA, Harvey SC. Dynamics of Proteins and Nucleic Acids. Cambridge: Cambridge University Press, 1987. Brooks In CL, Karplus M, Pettitt BM. Proteins: A theoretical perspective of dynamics, structure and thermodynamics. New York: Wiley, 1988. van Gunsteren WF,Weiner PK (eds). Computer Simulationof Biomolecular Systems, 1989. Leiden: Escom. Born M. Volumen und Hydratationswarme der ionen. Z Phys 1920;1:4548. Debye VP, Hiickel E. Zur Theorie der Electrolyte. I. Gefrierpunktsemiedrigung und verwandte Erscheinungen. Phys Z 1923;24:185-206. Onsager L. Electric moments of molecular liquids. J Am Chem Soc 1936;58:1486-1493. Kirkwood JG. Theory of solutions of molecules containing widely separated charges with special application to zwitterions. J Chem Phys 1934;2(7):351-361. Metropolis N, Rosenbluth AW, Rosenbluth MN, Teller AH, Teller E. Equation of state calculations by fast computing machines. J Chem Phys 1953;21:1087-1092. Alder BJ, Wainwright TE. Studies in molecular dynamics. I. General method. J Chem Phys 1959; 31:45+466. Rahaman A. Correlations in the motion of atoms in liquid argon. Phys Rev 1964,136A:405-411. Allen MP, Tildesley DJ. Computer simulation of liquids. Oxford: Clarendon,l987. Ciccotti G, Frenkel D, McDonald IR (eds). Simulation of Liquids and Solids. Amsterdam: NorthHolland, 1987. Rashin AA, Bukatin MA. A view of thermodynamics of hydration emerging from continuum studies. Biophys Chem 1994;51:167-192. Honig B, Nicholls A. Classical electrostatics in biology and chemistry. Science 1995;268:11441149. Warshel A, Russell ST. Calculations of electrostatic interactions in biological systems and in solutions. Q Rev Biophys 1984;17:283-422.
368 29. Matthew JB. Electrostatic effects in proteins. Ann Rev Biophys Biophys Chem 1985;14:387-417. 30. Rogers NK. The modelling of electrostatic interactions in the function of globular proteins. Prog Biophys Molec Biol 1986;48:37-66. 31. Harvey SC. Treatment of electrostatic effects in macromolecular modeling. Proteins 1989;5:78-92. 32. van Gunsteren WF, Berendsen HJC.Computer simulations of molecular dynamics: Methodology, applications, and perspectives in chemistry. Angew Chem Int Ed Eng 1990;29:992-1023. 33. Sharp KA, Honig B. Electrostatic interactions in macromolecules: Theory and applications.AM Rev Biophys Biophys Chem 1990;19:301-332. 34. Bashford D. ElectrostaticEffects in Biological Molecules. Curr Opin Struct Biol1991;1(2):175-184. 35. Moult J. Electrostatics. Curr Opin Struct Biol 1992;2:223-229. 36. Sharp KA. Electrostatic interactions in macromolecules. Curr Opin Struct Biol 1994;4:234-239. 37. Gilson MK. Theory of electrostatic interactions in macromolecules. Curr Opin Struct Biol 1995;5: 216-223. 38. Brooks III CL, Pettitt BM, Karplus M. Structural and energetic effects of truncating long ranged interactions in ionic and polar fluids. J Chem Phys 1985;83:5897-5908. 39. Greengard L, Rokhlin V. On the evaluation of electrostatic interactions in molecular modeling. Chemica Scripta 1989;29A:139-144. 40. Beme BJ, Wallqvist A. On modeling hydrophobic interactions. Chemica Scripta 1989;29A:85-91. 41. Solmajer T, Mehler EL. Electrostatic screening in molecular dynamics simulations. Protein Engng 1991;4(8):911-9 17. 42. Northrup SH, Pear MR, Morgan JD, McCammon JA, Karplus M. Molecular dynamics of ferrocytochrome c. Magnitude and anisotropy of atomic displacements. J Mol Biol1981;153:1087--1109. 43. Levitt M. Molecular dynamics of macromolecules in water. Chemica Scripta 1989;29A:197-203. 44. Mark AE, Berendsen HJC, van Gunsteren WF. Conformational flexibility of aqueous monomeric and dimeric insulin: A molecular dynamics study. Biochemistry 1991;30;1086610872. 45. Beveridge DL, DiCapua FM. Free energy via molecular simulation: Application to chemical and biomolecular systems. Ann Rev Biophys Biophys Chem 1989;18:431-492. 46. Gilson MK, Honig BH. The dielectric constant of a folded protein. Biopolymers 1986;25:20972119. 47. Lee B, Richards FM. The interpretation of protein structures: Estimation of static accessibility.J Mol Biol 1971;55:37+400. 48. Jackson JD. Classical electrodynamics. New York Wiley, 1975. 49. Bockris JO'M, Reddy AKN. Modem Electrochemistry. London: Macdonald, 1970. 50. McQuanie DA. Statistical Mechanics. New York: Harper, 1976. 51. Sharp KA, Honig B. Calculating total electrostatic energies with the nonlinear Poisson-Boltzmann equation. J Phys Chem 199Q947648-7692. 52. Sitkoff D, Sharp KA, Honig B. Accurate calculation of hydration free energies using macroscopic solvent methods. J Phys Chem 1994;98:1978-1988. 53. Mark AE,van Gunsteren WF. Decomposition of the free energy of a system in terms of specific interactions. J Mol Biol 1994240: 167-176. 54. Linderstrom-Lang K. On the ionization of proteins. C R Trav Lab Carlsberg 1924;15:1-29. 55. Tanford C, Kirkwood JG. Theory of protein titration curves. I. General equations for impenetrable spheres. J Am Chem SOC1957;79(20):5333-5339. 56. Tanford C, Roxby R. The interpretation of protein titration curves. Application to lysozyme. Biochemistry 1972;11(1 1):2192-2 198. 57. Shire SJ, Hanania GIH, Gurd FRN. Electrostatic effects in myoglobin. Hydrogen ion equilibria in sperm whale femmyoglobin. Biochemistry 1974;13(14):2967-2974. 58. States DJ, Karplus M. A model for electrostatic effects in proteins. J Mol Biol 1987;197:122-130. 59. 'Warwicker J, Watson HC. Calculation of the electric potential in the active site cleft due to alphahelix dipoles. J Mol Biol 1982;157:671-679. 60. Warwicker J. Continuum dieletric modelling of the protein-solvent system, and calculation of the long-range electrostatic field of the enzyme phosphoglycerate mutase. J Theor Biol 1986;121:19+
369 61.
62. 63.
64. 65. 66. 67. 68. 69. 70. 71. 72.
73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85.
210. Klapper I, Hagstrom R, Fine R, Sharp K, Honig B. Focusing of electrostatic fields in the active site of Cu-Zn superoxide Dismutase: effects of ionic strength and amino acid modification. Proteins 1986;1:47-56. Jayaram B, Sharp K, Honig B. The electrostatic potential of DNA. Biopolymers 1989;28:975-993. Orttung WH. Direct solution of the Poisson equation for biomolecules of arbitray shape, polarizability density and charge distribution. Ann NY Acad Sci 1977;303:22-37. Zauhar RJ, Morgan RS. A new method for computing the macromolecularelectric potential. J Mol Biol 1985;186:815-820. Zauhar RJ, Morgan RS. The rigorous computation of the molecular electric potential. J Comput Chem 1988;9(2):171-187. Zhou H-X. Boundary element solution of macromolecularelectrostatics:Interaction energy between two proteins. Biophys J 1993;65:955-963. Davis ME, McCammon JA. Calculating electrostatic forces from grid-calculated potentials. J Comput Chem 1990;11:401-409. Still WC, Tempczyk A, Hawley RC, Hendnckson T. Semianalytic treatment of solvation for molecular mechanics and dynamics. J Am Chem SOC1990112:6127-6129. Zauhar RJ. The incorporation of hydration forces determined by continuum electrostatics into molecular mechanics simulations. J Comput Chem 1991;12575-583. Sharp K. Incorporatingsolvent and ion screening into molecular dynamics using the finite-difference Poisson-Boltzmann method. J Comput Chem 1991;12(4):454-468. Gilson MK, Honig B. The inclusion of electrostatic hydration energies in molecular mechanics calculations. J Comput Aided Mol Des 199155-20. Niedenneier C, Schulten K. Molecular dynamics simulations in heterogeneous dielectrica and Debye-Huckel media: Application to the protein bovine pancreatic trypsin inhibitor. Mol Simulation 1992;8:361-387. Gilson MK, Davis ME, Luty BA, McCammon JA. Computation of electrostatic forces on solvated molecules using the Poisson-Boltzmann equation. J Phys Chem 1993;97:3591-3600. Smith KC, Honig B. Evaluation of the conformational free energies of loops in proteins. Proteins 1994;18:119-132. Abagyan R, Totrov M. Biased probability Monte Carlo conformational searches and electrostatic calculations for peptides and proteins. J Mol Biol 1994;235:983-1002. Mehler EL, Eichele G. Electrostatic effects in water-accessible regions of proteins. Biochemistry 198+23:3887-3991. Bashford D, Karplus M. pK,’s of Ionizable Groups in Proteins: Atomic Detail from a Continuum Electrostatic Model. Biochemistry 1990;29:1021F10225. Lederer F. Extreme pK, displacements at the active sites of FMN-depedent alpha-hydroxy acidoxidizing enzymes. Protein Sci 1992;1:54&548. Nozaki Y, Tanford C. Acid-base titrations in concentrated guanidine hydrocloride. Dissociation constants of the guanidinium ion and of some amino acids. J Am Chem SOC 1967;$9(4):736-742. Bashford D, Karplus M. Multiple-site titration curves of proteins: An analysis of exact and approximate methods for their calculation. J Phys Chem 1991;95:9956-9561. Yang A-S, Gunner MR, Sampogna R, Sharp K, Honig B. On the calculation of pKa’s in proteins. Proteins 1993;15:252-265. Gilson MK. Multiple-site titration and molecular modeling: Two rapid methods for computing energies and forces for ionizable groups in proteins. Proteins 1993;15266-282. Bashford D, Gerwert K. Electrostatic calculations of the pK, values of ionizable groups in bacteriorhodopsin. J Mol Biol 1992;224:473-486. Antosiewicz J, Porschke D. The nature of protein dipole moments, experimental and calculated permanent dipole of a-chymohypsin. Biochemistry 198%28:10072- 1W78. Beroza P, Fredkin DR, Okamura MY, Feher G. Protonation of interacting residues in a protein by a Monte Carlo method: application to lysozyme and the photosynthetic reaction center of
370 Rhodobacter sphaeroides. Proc Natl Acad Sci USA 1991;88:5804-5808. 86. Bashford D, Case DA, Dalvit C, Tennant L, Wright PE. Electrostatic calculations of side-chain pK, values in myoglobin and comparison with NMR data for histidines. Biochemistry 1993;32:80458056. 87. Tanford C. Theory of protein titration curves. 11. Calculations for simple models at low ionic strength. J Am Chem SOC1957;79(20):5340-5347. 88. Matthew JB, Gurd FRN, Garcia-Moreno EB, Flanagan MA, March KL, Shire SJ. pH-dependent processes in proteins. CRC Crit Rev Biochem 1985;188(2):91-197. 89. Nozaki Y, Tanford C. Proteins as random coils. II. Hydrogen ion titration curve of ribonuclease in 6 M guanidine hydrochloride. J Am Chem SOC1967;89(4):742-749. 90. Vorobjev YN, Scheraga HA, Hitz B, Honig B. Theoretical modeling of electrostatic effects of titrable side-chain groups on protein conformation in a polar ionic solution. 1. Potential of mean force between charged lysine residues and titration of poly(L-lysine) in 95% methanol solution. J Phys Chem 1994;98:1094(F10948. 91. Wendoloski JJ, Matthew JB.Molecular dynamics effects on protein electrostatics. Proteins 1989;5: 3 13-321. 92. Nicholls A, Sharp KA, Honig B. Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins 1991;11:281-296. 93. Gilson MK, Straatsma TP, McCammon JA, Ripoli DR, Faerman CH, Axelsen PH, Silman I, Sussman JL. Open “back door” in a molecular dynamics simulation of acetylcholinesterase. Science 1994;263:12761278. 94. Dao-Pin S, Liao D-I, Remington SJ. Electrostatic fields in the active sites of lysozymes. Proc Natl Acad Sci USA 1989;86:5361-5365. 95. Creighton TE.Proteins, 2nd edn. New York: Freeman, 1993. 96. Anthonsen HW, Baptista A, Drables F, Martel P, Petersen SB. The blind watchmaker and rational protein engineering. Biotechnol 1994;36:185-220. 97. Gilson MK, Sharp KA, Honig B. Calculating the electrostatic potential of proteins in solution: Method and error assessment. J Comput Chem 1987;9:327-335. 98. Nicholls A, Honig B. A rapide finite difference algorithm, utilizing successive over-relaxation to solve the Poisson-Boltzmann equation. J Comput Chem 1991;12(4):435-445. 99. Gilson MK, Honig B. Energetics of charge-charge interactions in proteins. Proteins 1988;3:32-52. 100. Fersht AR. Enzyme Structure and Mechanism, 2nd edn. New York Freeman, 1987. 101. Russell AJ,Thomas PG, Fersht AR. Electrostatic effects on modification of charged groups in the active site cleft of subtilisin by protein engineering. J Mol Biol 1987;193:803-813. 102. Gilson MK, Honig BH. Calculation of electrostatic potentials in an enzyme active site. Nature 1987;330:84-86. 103. Sternberg MJE, Hayes FR, Russel AJ, Thomas PG,Fersht AR. Prediction of electrostatic effects of engineering of protein charges. Nature 1987;330:86-88. 104. Loewenthal R, Sancho J, Reinikainen T, Fersht AR. Long-range surface charge-Charge interactions in proteins. Comparison of experimental results with calculations from a theoretical model. J Mol Biol 1993;232:574-583. 105. Roxby R, Tanford C. Hydrogen ion titration curve of lysozyme in 6 M guanidine hydrochloride. Biochemistry 1971;10(18):3348-3352. 106. Tanford C. The location of electrostatic charges in Kirkwood’s model of organic ions. J Am Chem SOC1957;79:5348-5352. 107. Orttung WH. Proton binding and dipole moment of hemoglobin. Refined calculations. Biochemistry 19709(12):2394-2402. 108. Spassov VZ, Karshikov A, Atanasov BP. Electrostatic interactions in proteins. A theoretical analysis of lysozyme ionization. Biochem Biophys Acta 1989;999:1-6. 109. Friend SH, Gurd FRN. Electrostatic stabilization of myoglobin. pH dependence of summed electrostatic contributions. Biochemistry 1979;18(21):46 12-46 19. 110. Shire SJ, Hanania GI, Gurd FRN. Electrostatic effects in myoglobin. pH and ionic strength
37 1
1 11.
112.
113. 114. 115. 116. 117. 118.
119. 120.
121.
122. 123. 124. 125. 126. 127. 128. 129. 130. 131.
132. 133.
variations of ionization equilibria for individual groups in sperm whale ferrimyoglobin. Biochemistry 1974;13(14):2974-2979. Botelho LH, Gurd FRN. Proton nuclear magnetic resonance study of histidine ionizations in myoglobin of various species. Specific assignement of individual resonances. Biochemistry 1978;17(24):5188-5 196. Botelho LH, Friend SH, Matthew JB, Lehman LD, Hanania GM, Gurd FRN. Proton nuclear magnetic resonance study of histidine ionizations in myoglobm of various species. Comparison of observed and computed pK, values. Biochemistry 1978;17(24):5197-5205. Matthew JB, Hanania GM, Gurd FRN. Electrostaticeffects in hemoglobin: Hydrogen ion equilibria in human deoxy- and oxyhemoglobin A. Biochemistry 1979;18(10):1919-1928. Matthew JB, Hanania GI, Gurd FRN. Electrostatic effects in hemoglobin: Bohr effect and ionic strength dependence of individual groups. Biochemistry 1979;18(10):1928-1936. Ho C, Russu IM. How much do we know about the bohr effect of hemoglobin?. Biochemistry 1987; 26:629+6305. Wlodawer A, Walter J, Huber H, Sjolin L. Structure of bovine pancreatic trypsin inhibitor. Results of joint neutron and X-ray refinement of crystal form II. J Mol Biol 1984;180:301-329. Wlodawer A, Deisenhofer J, Huber H. Comparison of two highly refined structures of bovine pancreatic trypsin inhibitor. J Mol Biol 1987;193:145-156. Bemdt KD, Guntert P, Orbons LM, Wiithrich K.Determination of a highquality nuclear magnetic resonance solution structure of the bovine pancreatic trypsin inhibitor and comparison with three crystal structures. J Mol Biol 1992;227:757-775. Wiithrich K, Wagner G. Nuclear magnetic ressonance of labile proteins in the bovine trypsin pancreatic inhibitor. J Mol Biol 1979;1301-18. Marsh KL, Maskalik D, England RD, Friend SH, Gurd FRN. Analysis of electrostatic interactions and their relationship to conformation and stability of bovine pancreatic trypsin inhibitor. Biochemistry 1982;21:5241-525 1. Karshikov AD, Engh R, Bode W, Atanasov BP. Electrostatic interactions in proteins: Calculations of the electrostatic term of free energy and the electrostatic potential field. Eur J Biophys 1989;17:287-297. Russell ST, Warshel A. Calculations of electrostatic energies in proteins. The energetics of ionized groups in bovine pancreatic trypsin inhibitor. J Mol Biol 1985;185:38+4O4 1. Warshel A, Russell ST, Churg AK. Macrospic models for studies of electrostatic interactions in proteins: Limitations and applicability. Proc Natl Acad Sci USA 1984;81:47854789. Dill KA. Dominant forces in protein fo1ding.Biochemistry 1990,29(31):7133-7155. Rashin AA, Honig B. On the environment of ionizable groups in globular proteins. J Mol Biol 1984; 173515-521. Barlow DJ, Thornton JM. Ion-pairs in Proteins. J Mol Biol 1983;168:867-885. Hollecker M, Creighton TE. Effect on protein stability of reversing the charge of amino groups. Biochem Biophys Acta 1982;701:395-404. Waldburguer CD, Schildbach JF, Sauer RT. Are buried salt bridges important for protein stability and conformational specificity? Nature Struct Biol 1995;2(2):122-128. Hendsch ZS, Tidor B. Do salt bridges stabilize proteins? A continuum electrostatic analysis. Protein Sci 1994;3:211-226. Fersht AR. Conformational equilibria in a-and y -chymotrypsin. The energetics and importance of the salt bridge. J Mol Biol 1972;64:497-509. Anderson DE, Becktel WJ, Dahlquist FW.pH-induced denaturation of proteins: A single salt bridge contributes %5 kcal/mol to the free energy of folding of T4 lysozyme Biochemistry 1990;29:24032408. Erwin CR, Barnett BL, Oliver JD, Sullivan JF. Effects of engineered salt bridges on the stability of subtilisin BPN'. Protein Engng 199O;4(1):87-97. Friend SH, Gurd FRN. Electrostatic stabilization in myoglobin. Interactive free energies between individual sites. Biochemistry 1979;18(21):462&4630.
372 134. Wada A, Nakamura H. Nature of the charged distributions in proteins. Nature 1981;293:757-758. 135. Gilson MK, Honig B. Calculation of the total electrostatic energy of a macromolecular system: Solvation energies, binding energies, and conformational analysis. Proteins 1988;4:7- 18 1. 136. Honig B, Yang A-S. Free energy balance in protein folding. Adv Protein Chem 1995;46:27-58. 137. Tanford C. Protein denaturation Part C. Adv Protein Chem 1970;25:1-95. 138. Albery JW,Knowles JR. Evolution of enzyme function and the development of catalytic efficiency. Biochemistry 1976;155631-5640. 139. Getzoff ED, Tainer JA, Weiner PK, Kollmann PA, Richardson JS, Richardson DC. Electrostatic recognition between superoxide and copper, zinc superoxide dismutase. Nature 1983;306:287-290. 140. Sharp K, Fine R, Honig B. Computer simulations of the diffusion of a substrate to an active site of an enzyme. Science 1987;236:146G1463. 141. Allison SA, Bacquet RJ, McCammon JA. Simulation of the diffusion controlled reaction between superoxide and superoxide dismutase. 11. D letailed models. Biopolymers 1988;27:251-269. 142. Ermak DL, McCammon JA. Brownian dynamics with hydrodynamic interactions. J Chem Phys 1978;69:1352-1360. 143. McCammon JA. Superperfect enzymes. Curr Biol 1992;2(11):585-586. 144. Davis ME, Madura JA, Sines J, Luty BA, Allison SA, McCammon JA. Diffusion-controlled enzymatic reactions. Meth Enzymol 1991;202:473-497. 145. Luty BA, Wade RC, Madura JD, Davis ME, Briggs JM, McCammon JA. Brownian dynamics simulations of diffusional encounters between triose phosphate isomerase and glyceraldehyde phosphate: Electrostatic steering of glyceraldehyde phosphate. J Phys Chem 1993;97:233-237. 146. Wade RC, Davis ME, Luty BA, Madura JD, McCammon JA. Gating of the active site of triose phosphate isomerase: Brownian dynamics simulation of flexible peptide loops in the enzyme. Biophys J 1993;64:+15. 147. Carter P, Wells JA. Dissecting the catalytic triad of a serine protease. Nature 1988;332(6164):564568. 148. Hol WJ. The role of the alpha-helix dipole in protein function and structure. Prog Biophys Mol Biol 1985;45:14+195. 149. Knowles JR.Enzyme catalysis: not different, just better. Nature .1991;350:121-124. 150. Kula MR. Liquid-liquid extraction of biopolymers. In: Cooney CL, Humphrey AE (eds) Comprehensive Biotechnology, vol 2. London: Pergamon Press, 1979;451 4 7 0 . 15 1. Cabral JMS, Aires-Barros MR. Reversed micelles in liquid-liquid extraction. In: Kennedy JF,Cabral JMS (eds) Recovery Processes for Biological Materials. London: Wiley, 1993;247-27 1. 152. Gocklen KE, Hatton TA. Protein extraction using reverse micelles. Biotechnol Progr 1985;1:6+74. 153. Wolbert RBG, Hilhorst R, Voskuilen G, Nachtegaal H, Dekker M, van’t Riet K, Bijsterbosch BH. Protein transfer from an aqueous phase into reversed micelles: The effect of protein size and charge distribution. Eur J Biochem 1989;184:627-633. 154. Pires MJ, Martel P, Baptista A, Petersen SB, Willson RC, Cabral JMS. Improving protein extraction yield in reversed micellar systems through surface charge engineering. Biotechnol Bioeng 1994;44: 773-780. 155. Eiteman MA. Predicting partition coefficients of multi-charged solutes in aqueous two-phase systems. J Chromatogr A 1994;668:21-30. 156. Eiteman MA, Gainer JL. A model for the prediction of partition coefficients in aqueous two-phase systems. Bioseparation 1991;2:31-41. 157. Karshikov A, Spassov V, Cowan SW, Landenstein R, Schirmer T. Electrostatic properties of two porin channels from Escherichia coli. J Mol Biol 1994;240:372-384. 158. LeSar R, Gordon RG. Density functional theory for the solid alkali cyanides. J Chem Phys 1982; 77:3682-3692. 159. Car R, Paninelo M. Unified approach for molecular dynamics and density functional theory. Phys Rev Lett 1985;55:2471-2474. 160. Davis ME. The inducible multipole solvation method: A new model for solvation effects on solute electrostatics. J Chem Phys 1994;7:5149-5 159.
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R. El-Gewely, editor.
373
Biochemical and molecular approaches for production of pravastatin, a potent cholesterol-lowering drug Nobufusa Serizawa Biomedical Research Laboratories, Sankyo Co., Ltd., Shinagawa-ku, Tokyo, Japan
Abstract. The intensive search for inhibitors of cholesterol biosynthesis by screening culture broths has spanned more than 20 years here at Sankyo. Resulting from our efforts, ML-236B was discovered in Japan as the f i st potent and specific inhibitor of HMG-CoA reductase. This compound contributed to the Nobel Prize-winning work of Goldstein and Brown in which they elucidated the mechanism of the LDL receptor pathway. After the discovery of ML-236B, many attempts were performed to find other HMGCoA reductase inhibitors, and some potent inhibitors including pravastatin have already been launched. HMG-CoA reductase inhibitors are in worldwide clinical use and play a pivotal role in the therapy of hyperlipidemic patients. Pravastatin is produced by a two-step fermentation, firstly ML-236B is produced by Penicillium citrinum followed by the hydroxylation of ML-236B by S. carbophilus to form pravastatin. Recent advances in the molecular characterization of the Cyt P-45Osca-2 and their responsiveness to ML-236B and PB in bacterial cultures should help elucidate the underlying cellular and molecular mechanisms of ML-236BNa and PB induction. In an effort to increase the productivity of this fermentation process, new technologies have been developed, and the mechanism of hydroxylation has been extensively investigated. Key words: actinomycete, cholesterol-loweringdrug, coronary heart disease (CHD), cytochrome P-450 gene, cytochrome P-450,heme binding site, HMG-CoA reductase inhibitor,ML-236B, negative regulation, P-450repressor, Penicillum citrinum, phenobarbital,pravastatin, Streptomyces carbophilus, Streptomyces lividans, substrate induction, transcriptional activation. Abbreviations: ML-236BNa, ML-236B sodium salt; pravastatin, pravastatin sodium; HMG-CoA reductase, 3-hydroxy-3-methylglutaryl-coenzymeA reductase; PB, phenobarbital; S . carbophilus, Streptomyces carbophilus; S. lividans, Streptomyces lividans; Cyt P-450,cytochrome P-450 gene; Cyt P450,cytochrome P-450;a.a., amino acid; ORF, open reading frame.
Introduction Coronary heart disease (CHD) is one of the major causes of death in both Western countries and Japan. Among the types of CHD known, ischemic heart disease (IHD) has been singled out as the major cause of death. In recent years the number of patients suffering from IHD in Japan has been on the rise. This trend is attributable, in part, to a growing elderly population and Westernized food intake. It is well known that the three major risk factors for IHD are hypercholesterolemia, hypertension and smoking. Hypercholesterolemia has long been considered to be the most
Address for correspondence: Nobufusa Serizawa, Biomedical Research Laboratories, Sankyo Co., Ltd., No. 2-58, Hiromachi-1-chome, Shinagawa-ku, Tokyo 140,Japan.
374 important of these factors [ 11. In order to reduce the risk associated with high serum cholesterol levels, the development of several hypolipidemic drugs and therapies have been explored among various research communities. Cholesterol is present endogenously by both absorption from diet and biosynthesis and is excreted mainly as bile acids into feces [2]. In order to reduce body cholesterol, three major strategies can be considered: 1) inhibition of cholesterol absorption by a compound such as p-sitosterol, 2) inhibition of bile acids reabsorption by a compound such as cholestyramine, and 3) inhibition of cholesterol biosynthesis. Since more than 70% of the total input of body cholesterol in humans is derived from de novo synthesis, it is expected that plasma cholesterol levels could be reduced by inhibition of cholesterol biosynthesis. The biosynthesis of cholesterol is traversed over an extensive biochemical process exceeding more than 20 steps starting from acetyl-coenzyme A (CoA). The ratelimiting enzyme of this pathway is 3-hydroxy-3-methylglutaryl(HMG)-CoA reductase (mevalonate: NADP+ oxidoreductase (CoA-acylating), EC 1.1.1.34) which catalyzes the reduction of HMG-CoA to mevalonate. Generally the later steps of this biosynthesis are the most suitable stage for targetting inhibition because the production of other substances is minimally disturbed. Hydrophobic substances such as cholesterol have been shown to be an exception to this idea due to the accumlation of hydrophobic intermediates. In 1971, the Sankyo Co., Ltd. launched a major research initiative to screen inhibitors for the cholesterol synthesis from the culture broth of microorganisms using a cell-free enzyme system from rat liver. After much rigorous screening of microorganims, ML-236B (mevastatin, Fig. 1) was discovered in the culture broth of Penicillium citrinum in 1975 [3]. It is noteworthy that compactin, a compound identical to ML-236B, was independently isolated by Beecham Pharmaceuticals as a weak antifungal antibiotic [4] 2 years after Sankyo’s patent filing. As shown in Fig. 1, a portion of the ML-236B structure resembles that of HMG (3-hydroxy-3-methylglutarate),the part of HMG-CoA that serves as the substrate of HMG-CoA reductase. Accordingly, ML-236B and the related compounds shown in Fig. 1 inhibit the enzyme in a competitive manner with respect to HMG-CoA. It has been well known for some time that the liver and intestine are the major
HMG-CoA
Pravastatin
Fig. 1. Structures of pravastatin, ML-236B. and HMG-CoA.
ML-236B
375 organs involved in de novo cholesterogenesis. Our drug discovery program has been focused on finding a compound which has enhanced target-organ directed characteristics, since target-organ directed inhibition would be expected to minimally disturb cholesterol metabolism in other organs, including hormone-producing organs. After screening microbial products as well as chemically and biologically modified derivatives of ML-236B, pravastatin was finally chosen as the candidate for development. Indeed pravastatin displays a stronger and more tissue-selective inhibition of cholesterol synthesis than ML-236B [5,6]. Pravastatin contains a hydroxyl group at the 6p position of its decaline structure (Fig. 1). In 1979 this drug was first found to be a minor urinary metabolite of ML236B in dogs. For the industrial hydroxylation of ML-236B, chemical syntheses had been initially attempted, but economic considerationjudged this method as infeasible. For this reason, microbial hydroxylation was chosen for the production of pravastatin. After screening for microorganisms capable of converting ML-236B to pravastatin, S. carbophilus was selected for the second step of the fermentation process [7,8]. Reference to the details of the biological activity of pravastatin can be found in other articles [9]. In 1981, pravastatin was chosen for development as a hypolipidemic drug, and clinical trials were started in 1984. Pravastatin was launched in 1989 as “Mevalotin” in Japan. The drug was licensed to the Bristol-Myers Squibb Company and has been developed worldwide and as a result of such effortsis currently vailable on the market of 45 countries including Japan. Pravastatin is produced by a two-step fermentation process (Fig. 2); the first step is production of ML-236B and the second one is hydroxylation of ML-236B. This review surveys the subject of pravastatin production with particular focus being given to the molecular mechanism of microbial hydroxylation of ML-236B.
Biochemical approach Microbial hydroxylation The hydroxylation reaction by which pravastatin is formed could not be performed easily through conventional synthetic methods, thus we set out to isolate microorganisms capable of hydroxylating ML-236B at the 6p position to form pravastatin. The first successful experiment reported for a microbial-mediated transformation of an exogenous compound was in an oxgenation reaction for steroid production by
ML-2368
Fig. 2. Two-step fermentation based production of pravastatin.
376 Rhizopus nigricans [lo]. This section focuses on the production of pravastatin by microbial hydroxylation. Discovery of microorganisms capable of hydroxylating ML-236B to pravastatin [5,11,121 Various microorganisms were tested for their capability to hydroxylate ML-236B at the 6P-position. Resulting from an extensive screening program, Mucor hiemalis SANK 36372 was found to be one of the most effective microorganisms for that hydroxylation. Mucor hiemulis SANK 36372 converted ML-236B mainly to pravastatin. None of these fungi, however, tolerated high concentrations of ML-236B in the culture broth, probably because of the antifungal activity of the substrate. ML-236B does not exhibit antibiotic activity in many strains of actinomycetes and bacteria. Identified cultures and isolates of these strains from soil samples were therefore tested for their ability to hydroxylate ML-236B. We collected soil samples from several areas of Australia. Isolates from soil samples collected in Australia had potent abilities to hydroxylate ML-236B at the 6P-position. The strains were identified as A. autotrophica. A. autotrophica were found to tolerate ML-236B and to have strong transformation activity, and all produced dihydroxylated by-product. The supply of pravastatin was sufficient for pharmacological tests and development studies.
S . carbophilus as a potent converter of ML-236B [7,8,11,13] In order to produce pravastatin on an industrial scale, we screened for microoganisms that had strong hydroxylating activity and produced few by-products. A. autotrophica grew well on D-trehalose media. After further screening for fresh isolates of actinomycetes that grew on D-trehalose, the strain SANK 62585, isolated from Australian soil, was discovered as a potent converter that produced only limited byproducts. The morphological and physiological characteristics suggested that strain SANK 62585 belonged to the genus Streptomyces. Taxonomically the train S A N K 62585 should be classified as a new species of the genus Streptomyces for which the name S. carbophilus has been given. Hydroxylation mechanism of ML-236B to pravastatin by a novel actinomycete Cyt P450 [7,8] A most noteworthy finding was the discovery of S. carbophilus as a potent converter. The discovery permitted the production of pravastatin on an industrial sale. Nevertheless an understanding of the hydroxylation mechanism for conversion of ML-236B to pravastatin is requisite for industrial-scale production. As mentioned previously, pravastatin was first found as a minor urinary metabolite of ML-236B in dogs, suggesting the participation of hepatic Cyt P-450s. When pravastatin was later obtained from microbes, a microbial counterpart to Cyt P-450 was anticipated to play a role in this hydroxylation. In the last two decades, microbial Cyt P-450s have been extensively studied. In
377 prokaryotes, the most thorough studies have been carried out on the camphor hydroxylation system in Pseudomonas putida [ 14-1 81. Other well-studied bacterial Cyt P-450s are the fatty acid hydroxylation system in Bacillus megaterium ATCC 14581 [19-221. Recently, Cyt P-450 (eryF) that catalyze the hydroxylation of 6deoxyerythronolide B, an intermediate of erythromycin A biosynthesis in Saccharopolyspora erythraea, have been characterized [23]. In actinomycetes, Cyt P-450dependent monooxygenase systems were found to participate in the detoxification of xenobiotics, and the induction of Cyt P-450s was also observed [24,25]. The mechanism by which ML-236B is hydroxylated to pravastatin by a novel actinomycete Cyt P-450 is discussed here. Characterization of Cyt P-450sca from S. carbophilus [7] The soluble Cyt P-450 induced by ML-236BNa was detected in the cell-free extract prepared from S. carbophilus. Table 1 shows hydroxylating activities in a supernatant, that was incubated with an NADH generation system. Increased rates were seen at high concentrations of ML-236Na, suggesting that the hydroxylation activity was induced by the substrate. A characteristic absorption maximum at 45 1 nm corresponding to Cyt P-450 appeared only when ML-236BNa was added to the medium, demonstrating the induction of Cyt P-450 by ML-236BNa. Since this actinomycete originates from S. carbophilus, we named it Cyt P-450sca. After purification, Cyts P-450sca-1 and P-450sca-2 were purified 441- and 491fold, respectively. Electrophoresis of Cyts P-450sca- 1 and P-450sca-2 revealed a single protein band of 46 f 1 kDa, indicating that Cyts P-450sca-1 and P-450sca-2 were identical. However, the amino acid compositions of Cyts P-450sca-1 and P450sca-2 differed. In Western blot analysis of Cyts P-450sca-1 and P-450sca-2, both proteins reacted similarly to the anti-(Cyt P-450sca-2) antibody. The reduced-CO vs. reduced difference spectra of both purified Cyt P-450sca-1 and P-450sca-2 showed absorbance maxima at 448 nm. The addition of ML-236BNa to a Cyt P-450sca-1 or Cyt P-450sca-2 solution caused an immediate shift to the high-spin form with the Soret peak appearing at 386 nm. Amino acid analysis showed that the composition of Cyts P-450sca-1 and P450sca-2 are closely related to mitochondria1 type Cyt P-450 and to other bacterial Cyt P-450s (Fig. 3). Cyts P-45Osca- 1 and P-450sca-2 contained hydrophobic residues in relatively high percentages (46 and 47%, respectively), as is generally observed in other Cyt P-450s.
Table I. Induction of hydroxylation activity. Inducer
Hydroxylation of ML-236BNa to pravastatin (nmol product/mg proteidmin)
None ML-236BNa
0 0.6
378 S. carbophilus Cyl P 4 5 0 . o l
Bovina adrenal cortex mitochondria CyI P-450 l l p
S. carbophilus Cyt P-450w.-2
S. erythraea P 4 5 0
Rat llver mlcrosoma Cyt P-450LM-3
Rat llver mlcrosome Qt P-450A Arg
Asx Thr
Phe Tv LSU
LeU Ise
1.W
CYs
Ile Met Val
Ala
Gb
CYs
Met Val Ala
Qb
Fig. 3. Histogram of the amino acid composition of bacterial and mammalian Cyt P-450proteins.
A two-component type Cyt P-450 in prokaryotes that catalyzes hydroxylation of ML236BNa to pravastatin [8] This section describes resolution of the Cyt P-450 monooxygenase system of S. carbophilus, purification of the NADH-Cyt P-450 reductase, and reconstitution of the hydroxylaton activity by endogenous components. Incubation of the cell-free extracts from S. carbophilus with pyrimidine nucleotides and ML-236BNa indicated that the Cyt P-450sca monooxygenase system required NADH as an electron donor. In general, actinomycetes Cyt P-450 systems require NADH. This is in contrast to mammalian Cyt P-450 systems in which NADH is not required [8]. NADH-Cyt P-450 reductase was separated from Cyt P-450sca by ion-exchange column chromatography. Finally, NADH-Cyt P-450 reductase was purified as a single polypeptide chain with a molecular weight of 5 1 kDa. The absorption spectrum of the purified protein showed typical flavoprotein absorption maxima at 370 and 446 nm, and a shoulder at 475 nm. The purified NADH-Cyt P-450 reductase protein contained FAD and FMN molecules. The FMN molecule was easily dissociated from the reductase (Kd = 5 x M). NADH-dependent reductase catalyzed the hydroxylation of ML-236BNa to pravastatin in the presense of a purified Cyt P-450sca. Elimination of Cyt P-450sca or reductase from this assay system abolished all hydroxylating activity. These two proteins were essential for the reconstitution of hydroxylation. A production mechanism for the synthesis of pravastatin in S. carbophilus is proposed (as shown
379
ML-236BNa Fig. 4. A production mechanism for the synthesis of pravastatin in S. curbophilus (Fp: flavoprotein).
in Fig. 4) based on these findings. Cyt P-450 monooxygenase systems can be classified as either two- or threecomponent type [26]. Three-component type systems are composed of Cyt P-450 reductase, Cyt P-450 and iron-sulfar protein, whereas two-component type systems are composed of Cyt P-450 reductase and Cyt P-450. The results of many previous works have shown that almost all prokaryotes have three-component (mitochondrial) type Cyt P-450 systems [27] except Cyt P-450BM-3 from B. megaterium. Cyt P450BM-3, from B. megaterium, exists as a single polypeptide that is capable of catalyzing the entire monooxygenase reaction of a substrate with the additon of only NADPH and 0, [22]. However, a two-component type Cyt P-450 monooxygenase system had not been reported for prokaryotes. The Cyt P-450sca monooxygenase system is composed of P-450 and flavoprotein, and the addition of ferredoxin did not cause any stimulation of hydroxylation activity. NADH-Cyt P-450 reductase from S. carbophilus contained both FAD and FMN. Therefore, the Cyt P-450sca monooxygenase system was classified as a two-component-type system. All twocomponent-type Cyt P-450 systems are membrane-bound. The Cyt P-450sca monooxygenase system may exist in soluble fraction. Therefore, the novel character of a two-component type in prokaryote was suggested for the Cyt P-450sca system. Our finding represents the first evidence of a two-component-type Cyt P-450 system in prokaryotes. Considered together, the intermediate character of the Cyt P-450sca monooxygen-
380 ase system between those of prokaryotes and eukaryotes may reflect the position of actinomycetes in its evolutionary pathway. In order to gain additional insights into P-450sca and the P-45Osca, the use of recombinant techniques will be required in future studies.
Molecular approach Cloning and expression of Cyt P-450sca-2 from S . carbophilus [28] The most extensive genetically molecular characterized bacterial Cyt P-450 is Cyt P450cam and Cyt P-450BM-3 [14-18,291. Additionally, some Cyt P-450s from actinomycetes have been purified and characterized, and their genes have been cloned. For example, Cyts P-45OSU 1 and P-45OSU2 from Streptomyces griseolus which metabolize herbicides, have been well-studied by Omer and O’Keefe et al. [30,311. Also, the Cyt P-450 (eryF) that catalyses the hydroxylation of 6-deoxyerythronolide B, an intermediate of erythromycin A biosynthesis in S. erythraea was also cloned [32]. The section on Biochemical approach showed the purification and characterization of Cyt P-450sca from S. carbophilus. Cyt P-450sca is induced 30-fold by its substrate ML-236BNa [7], and by PB [33]. Although PB is one of the major inducers for mammalian Cyt P-450s, its induction mechanism has not yet been fully elucidated, except in the case of barbiturate-inducible-CytP-450BM-3 from Bacillus megaterium [34]. As was described in the section on Biochemical approach, the Cyt P-450sca induction may serve as a model for mammalian PB-inducible Cyt P-450s [8]. In order to understand the induction mechanism, it is a prerequisite to determine the nucleotide sequence of the Cyt P-450sca. In this section we describe the cloning of Cyt P-45Osca-2 and its expression in S. lividans, frequently used as a host in the Streptomyces gene manipulation.
Cloning and sequence analysis of S . carbophilus Cyt P-450sca-2 Using a oligonucleotide corresponding to the N-terminal aa sequence of Cyt P450sca-2 as a probe for colony hybridization, the entire Cyt P-45Osca-2 was cloned. ORF of 1,233 bp encoding a 410-aa protein is preceeded by a potential ribosome binding site. The G + C content in ORF was 69% with the frequency of use of G and C at the third position of each codon being 91%. Cyt P-450sca-2 had the Streptomyces ribosome-binding site as five base pairs (-GAGGG-) upstream of the initiation codon. We also compared the HR2 region of Cyt P-450sca-2, which contains the heme-binding cysteine, with that of several prokaryotic and eukaryotic Cyt P-450s (Fig. 5). Cyt P-450sca-2 also contained the putative oxygen-binding site proposed in Cyt P-450cam of P. putida [16]. The residues forming the proposed oxygen-binding site in a variety of Cyt P-450s (-G/A-G-S-D/E-T-) are highly conserved as -A-G-H-ET- in Cyt P-450sca-2. Cyt P-450sca-2 shared its greatest homology (79%) with Cyt P-45OSU1 derived from S. griseolus [35]. However, the Cyt P-450 monooxygenase
38 1 Heme binding site
Actinomycetes
Bacterium
Rat
Identity(%)
L G Q N L A R L E L E V I
100
L G Q N L A R L E L E V I
95
L G Q P L A R V E L Q IA
71
M G R P L A K L E G E V A
57
-
62
L G Q H L A R R E I I V T
38
[:
F S T G K R I C L G E G I A R N E L F L F
43
F G L G L R R C I G E I P A L W E V F L F
33
C21
F G C G A R V C L C E S L A R L E L F V V
Bovine
U
62
Fig. 5. Comparison of the heme-binding region of bacterial and mammalian Cyt P-450 proteins.
system of S. griseolus is reported as a three component system. Thus, we can assume that the interactions of these two Cyt P-450s with their respective electrontransferring proteins are different.
iwr
0
5
I
I
I
10
15
20
OJ
25
Time (h) Fig. 6. Expression of the Cyt P-450sca-2 in S. lividans TK21/pSCA205. Relative production of pravastatin by S. carbophilus SANK 62585 (n),S. lividans TK21/pSCA205 (B) and S. lividans TK21 (0)is shown at various times after adding ML-236BNa to the medium.
382 Expression of Cyt P-450sca-2 in S. lividans TK21lpSCA205 Plasmid pSCA205 containing the ORF of the Cyt P-450sca-2 and its 1.0-kb 5’noncoding region, was constructed from pIJ702 [36] and S. lividans TK21 was transformed using this plasmid [37]. S. lividans TK21/pSCA205 converted ML236BNa to pravastatin in 8 h after the addition of ML-236BNa (Fig. 6). These results demonstrate that the 5’-upstream region of the Cyt P-45Osca-2 take on the function of a strong promotor in S. lividans. Substrate induction of Cyt P-450sca-2 [33] ML-236BNa is known to induce the production of Cyt P-450sca-2, but the mechanism of induction remains unclear. Therefore in attempts to clarify this situation, we assayed transcripts of Cyt P-45Osca-2 both in the presence and absence of ML-236BNa by northern hybridization. In the absence of ML-236BNa no transcripts of Cyt P-450sca-2 mRNA were detected, however, in its presence three different lengths of Cyt P-450 mRNA were found to be transcribed as having the values 1.8,2.8 and 4.6 kb. The amounts of these transcripts gradually increased after addition of ML-236BNa, reaching a maximum at 6 h. We consider that the 1.8-kb transcript is the major mRNA of Cyt P-450sca-2, because the amount of this transcript was the highest and its length is sufficient to incorporate the gene. The multiple transcripts may be due to difference in the lengths of the 3’ region, since a single transcriptional starting point was detected in the primer extension analysis. In 1992, Pate1 et al. [38] described the presence of an ORF downstream of Cyt P450SU in S. griseolus. Therefore the possibility arises whereby a gene in connection with the Cyt P-450sca monooxygenase system exists in the downstream region of Cyt P-45Osca-2. The above findings indicated that substrate induction is possibly regulated at the transcriptional level. Transcription was strongly induced in the presence of ML-236BNa in S. carbophilus, indicating Cyt P-450sca-2 induction was regulated at the transcriptional level. Molecular mechanism of Cyt P-45Osca expression [33] Several possible mechanisms whereby PB and “PB-like” inducers such as ML-236B activate transcription of the Cyt P-450 can be broadly categorised into the following two classes of mechanisms: 1) a receptor-dependent induction mechanism; and 2) a Cyt P-450-dependent induction mechanism [39], albeit the mechanism of induction remains unclear. The most actively studied barbiturate-inducibleCyt P-450 is Cyt P-450BM-3 from bacterial B. megaterium [40]. In B. megaterium, PB and other lipophilic barbiturates induces the expression of two Cyt P-450s, P-450BM-3 (gene CYPIO2), a naturally occurring and catalytically self-sufficient Cyt P-450-NADPH (P-450 reductase fusion protein of unusually high activity) and P-450BM-1 (gene CYPIO6) [41]. Thus the responsiveness of P-450s to PB is not confined to eukaryotes, and may have certain features that are broadly conserved in nature. A barbiturate-responsive regulatory region of the bacterial CYP102 has been localized through transformation studies of
383 a DNA segment spanning the region between 0.8 and 1.1 kb upstream from the transcription start site [42]. Although the functional importance of this DNA segment has yet to be demonstrated, it seems likely that it contributes to PB-dependent P-450 regulation; negative regulation in the case of the B. megaterium genes and positive regulation in the case of the rat genes. Interestingly, in vitro treatment of the bacterial or rodent liver extracts with PB mimicked the effects of in vivo PB treatment on the retardation patterns [43]. This implies that: 1) the effects of the PB on the binding protein involved in these interasctions do not require a new protein synthesis; and 2) the putative PB receptor protein may be present in the nucleus in the case of the rat liver extracts. Further study is needed to c o n f i i the validity and generality of these conclusions, and also may reveal the extent to which other mechanistic aspects of the PB induction process are conserved in prokatyotic and eukarytic systems. As described in the section on Biochemical approach, we have found that Cyt P-450sca is induced not only by its substrate, ML-236BNa, but also by PB, a typical inducer of many Cyt P-450s. We are intersted in the mechanism of Cyt P-450sca induction from S. carbophilus since it may serve as a model for mammalian systems. In this section, we have characterized induction of PB and ML-236BNa-inducible Cyt P450sca-2 from S. carbophilus, an industrial pravastatin-producing strain. Transcriptional activation by ML-236BNa and PB in S. carbophilus To determine the induction mechanism of Cyt P-450sca-2 by PB and ML-236BNa, we determined the nucleotide sequence belonging to the 5'-noncoding region of Cyt P450sca-2. In the upstream region of the ATG codon, a parindromic sequence (23bp) was found. A similar palindromic sequence was also described in the 5'noncoding region of Cyts P-45OSUZ and P-45OSU2 derived from S. griseolus [35]. Pate1 et al. [38] described a protein which binds to the palindromic sequence in S. griseolus. A putative RNA polymerase binding site was present adjacent to the palindromic sequence, suggesting that the palindromic sequence might regulate transcription of Cyt P-450sca-2. ML-236BNa, a substrate of Cyt P-450sca and PB, is a typical inducer of many . effect of ML-236BNa and PB on the transcription of Cyt P-45Osca-2 Cyt P - 4 5 0 ~The of S. carbophilus SANK 62585 was analysed by northern hybridization. After addition of the inducers, the mRNA amount of Cyt P450sca-2 was gradually elevated. A 27-fold increase was observed in the presence of ML-236BNa at 6 h after addition of the inducer. A 5-fold increase was observed in the presence of PB at 6 h after addition of PB. The amount of mRNA was dependent upon the concentration of the inducers. This finding suggests that induction of Cyt P-450sca-2 in S. carbophilus was regulated at the transcriptional level. Transcriptional activation by ML-236BNa and PB in S. lividans We have reported on expression of Cyt P-450sca-2 in S. lividans TK21, a frequently used strain in the gene manipulation of Streptomyces [37]. S. lividans TK21/ pSCA205, harboring Cyt P450sca-2 and 1 kbp of its 5'-noncoding region, converted ML-236BNa to pravastatin at the same level as that of S. carbophilus SANK 62585
384 [29]. This stimulated us to resolve whether induction of Cyt P-450sca-2 by ML236BNa or PB was also observed in S. lividans TK21/pSCA205. Effect of ML236BNa or PB on the amount of mRNA of the Cyt P-450sca-2 in S . lividans TK2 l/pSCA205 was analysed by northern hybridization. After addition of ML236BNa, immediately the amount of the mRNA was increased. A 14-fold increase was observed in the presence of ML-236BNa at 2 h after addition. The transcriptional enhancement by ML-236BNa depended on its concentration. These findings suggest that the 5’-noncoding region of Cyt P-450sca-2 participate in the regulation of transcription. We also examined transcription of P-450sca-2 in the transformant S. lividans TK2 l/pSCA205. We expected expression of Cyt P-450sca-2 in S . lividans TK21/pSCA205 would not be regulated by ML-236BNa but would be transcribed constantly. On the contrary, as shown in S. carbophilus SANK 62585, transcription was strongly induced in S. lividans TK2 l/pSCA205 by ML-236BNa. Furthermore, the transcription level in S. lividans TK2l/pSCA205 was higher than that in S. carbophilus SANK 62585 from the result of northern hybridization. This phenomenon corresponds with the result that the conversion rate of ML-236BNa to pravastatin in S. lividans TK21/pSCA205 was faster than that of S. carbophilus SANK 62585. This is most likely due to the pSCA205 being a multicopy plasmid, thus the mutigene of Cyt P-450sca-2 was transcribed in S. lividans TK21/pSCA205. These results are indicative of the possibility that the same induction mechanism of Cyt P-450sca-2 exists both in S. carbophilus SANK 62685 and S. lividans TK2l/pSCA205. The results obtained from the induction of Cyt P-450sca-2 were examined in S. lividans TK21/pSCA205 and enable us to construct chimeric genes that contain various lengths of the 5’-upstream region oftthe Cyt P-450sca-2. Moreover, induction of Cyt P-450sca-2 by ML-236BNa could then be analyzed further. Regulation mechanism of Cyt P-450sca-2 expression To examine the region concerning the regulation of the transcriptional induction by ML-236BNa, a series of plasmids containing the deleted 5’-noncoding region of Cyt P-45Osca-2 were constructed and S. lividans TK21 was transformed using these plasmids. These transformants were cultivated in the presence or absence of ML236B.Na. Figure 7 shows relative transcriptional intensity of Cyt P-45Osca-2 in each transformant. Strong transcriptional enhancement was observed in S. lividans TK21/pSCA205 habouring 1,013 bp of the 5’-noncoding region in the presence of ML-236BNa. On the other hand, S. lividans TK2l/pSCA310-A428 and S . lividans TK2l/pSCA310-8320 showed strong mRNA production both in the presence and even in the absence of ML-236BNa. Also, the amount of the transcript was similar to that in S. lividans TK21/pSCA205 in the presence of ML-236BNa. Based on this observation we can infer that the transcription is repressed by the 5’-noncoding region between -1,013 bp and 4 2 8 bp. As a result of the relative production of pravastain by transformants of S. lividans TK21, pravastatin formation by S. lividans TK21/pSCA310-A428 and S . lividans TK2l/pSCA310-A320 was faster than S. lividans TK21/pSCA205 up to 4 h cultivation after addition of ML-236Na. It is most probably because the transcription
385
Fig. 7. A: Stucture of the deleted 5‘-noncoding region of Cyt P-450sca-2 in a series of the plasmids. B: Effect of the length of 5’-noncoding region of Cyt P-450sca-2 on the transcription of its gene.
of Cyt P-450scu-2 is enhanced even in the absence of ML-236BNa in S. lividuns TK2l/pSCA3 10-A428 and S. lividuns TK2l/pSCA310-A320 (Fig. 7) which in turn enables the conversion of ML-236BNa to pravastatin to be faster than for the same conversion occurring in S. lividuns TK2 l/pSCA205. This pravastatin formation correlates to the transcriptional enhancement of the mRNA in Cyt P-450sca-2. A palindromic sequence between -46bp and -24bp in the 5’-noncoding region is probably bound to a protein produced by S. curbophilus and S . lividans TK21/ pSCA205 which harbors 1,013 bp of the 5‘-noncoding region. It should be noted that its binding was inhibited in the presence of ML-236BNa. These findings suggest that induction of Cyt P-450sca-2 was negatively regulated at the transcriptional level and that the palindromic sequence possibly functions as an operator in regulating the transcription. In the light of the above observations we proposed a mechanism of Cyt P-450sca induction by ML-236BNa as depicted in Fig. 8. A DNA binding protein possibly exists in the cell and interacts with the palindromic sequence (in the domain of 4 6 and -24 bp) in the 5’-noncoding region. In the absence of ML-236BNa, the
386
Fig. 8. A proposed mechanism of Cyt P-450sca induction by ML-236BNa.
protein possibly represses the transcription. In the presence of ML,-236BNa, the protein is separated from the palindromic sequence by changing its affinity to the sequence. As a result, RNA polymerase can transcribe the Cyt P-45Osca-2. This triggers an increase in the amount of mRNA of Cyt P-450sca-2 and thus many Cyt P-450sca are produced in the cell. The gel mobility shift assay showed the presence of a protein in S. lividans TK21/pSCA205 and in S. carbophilus SANK 62585, suggesting a gene of the protein might be encoded in the 5'-noncoding region of Cyt P-450sca-2. As described above, PB is one of the well-known inducers in Cyt P-450s. It is interesting that although between PB and ML-236BNa there exists no obvious structural relationship, the mRNA enhancement was observed in S. carbophilus. However, PB induction was not apparent in S. lividans TK21/pSCA205. Presently we can offer no explanation for this disparity. Fulco et al. [44]described the presence of a repressor (Bm3R1) which interacts with an operator (Barbie box: -ATCAAAAGCTGGAGG-) in the 5'-noncoding region of CYPI02 in B. megaterium. Two similar sequences resembling that of the Barbie box in S. carbophilus were detected. We searched a DNA binding protein to locate these sequence by means of a gel shift mobilily assay, however, a gel shift band was not detected. It is likely therefore that there might be a different protein possibly regulating the Cyt P-450sca induction by PB in S. carbophilus. Further effort will be needed to clarify the whole induction mechanism. In addition, a DNA binding protein participating in the induction of the Cyt P45Osca ought to be identified. Cyt P-450sca-2 plays a critical role in the production of pravastatin. We have shown that the Cyt P-45Osca-2 is functional in S. lividans. Our future will concentrate on improving the efficiency of pravastatin production by genetic engineering of Streptomyces.
387
References 1. Page HI, Berrettoni JN, Butkus A, Sones FM Jr. Prediction of coronary heart disease based on clinical suspicion, age, total cholesterol, and triglyceride. Circuration 1970;42:625--645. 2. Dietschy JM, Wilson JD. Regulation of cholesterolmetabolism. N Engl J Med 197Q282:1179-1 183. 3. Endo A, Kuroda M, Tsujita Y, Terahara A, Tamura C. Jpn Pat Kokai 1975;50-155690. 4. Brown AG, Smale TC, King TJ, Hasenkamp R, Thompson RH. Crystal and molecular structure of compactin, a new antifungal metabolite from Penicillium Brevicompactnm. J Chem Soc Perkin Trans 1976;1:1165- 1170. 5. Serizawa N, Nakagawa K, Hamano K, Tsujita Y, Terahara A, Kuwano H. Microbial hydroxylation of ML-236B (compactin) and monacolin K. J Antibiot 1983;36:604-607. 6. Tsujita Y, Kuroda M, Shimada Y, Tanzawa K, Arai M, Kaneko I, Tanaka M, Masuda H, Tarumi C, Watanabe Y, Fujii S. CS-5 14, a competitive inhibitor of 3-hydroxy-3-methylglutarylcoenzyme A reductase: tissue-selective inhibition of sterol synthesis and hypolipidemic effect on various animal species. Biochim Biophys Acta 1986;877:50-60. 7. Matsuoka T, Miyakoshi S, Tanzawa K, Hosobuchi M, Nakahara K, Serizawa N. Purification and characterization of cytochrome P-450sca from Streptomyces carbophilus. Eur J Biochem 1989;184:707-7 13. 8. Serizawa N, Matsuoka T. A two-component-type cytochrome P-450 monooxygenase system in prokaryotes that catalyzes hydroxylation of ML-236B to pravastatin, a tissue-selective inihibitor 3hydroxy-3-methylglutaryl coenzyme A reductase. Biochim Biophys Acta 1991;1084:35--4O. 9. Tsujita Y, Serizawa N, Hosobuchi M, Komai T, Terahara A. HMG-CoA reductase inhibitors as potent cholesterol-lowering drugs. In: Tsujita Y, Serizawa N, Hosobuchi M. Komai T, Terahara A (eds) Japanese Technology Reviews, Section E: Biotechnology, vol 5 , no. 1. Montreaux: Gordon and Breach Science Publishers, 1994. 10. Peterson H, Murray HC. Microbiological oxygenation of steroids at carbon 11. J Am Chem Soc 1952;74:1871. 11. Serizawa N, Serizawa S, Nakagawa K, Furuya K, Okazaki T, Terahara A. Microbial hydroxylation of ML-236B (compactin): Studies on microorganisms capable of 3b-hydroxylation of ML-236B. J Antibiot 1983;36:887-891. 12. Okazaki T, Serizawa N, Enokita R, Torikata A, Terahara A. Taxonomy of actinomycetes capable of hydroxylation of ML-236B (compactin). J Antibiot 1983;36:1176-1183. 13. Okazaki T, Enokita R, Miyaoka H, Otani H, Torikata A. Streptomyces danvinensis sp. nov., S. cactaceus sp. nov., from Australian soils. Ann Rep Sankyo Res Lab 199;41:123-133. 14. Katagiri M, Ganguli BN, Gunsalus ICA soluble cytochrome P-450 functional in methylene hydroxylation. J Biol Chem 1968;243:3543-3546. 15. Peterson J. Camphor binding by Pseudomonos putida cytochrome P-450. Arch Biochem Biophys 1971;144:678493. 16. Poulos TL, Finzel BC, Gunsalus IC, Wagner GC, Kraut J. The 2.6-A crystal structure of Pseudomonas putida cytochrome P-450. J Biol Chem 1985;260:16122-16130. 17. Unger BP, Gunsalus IC, Sliger SG. Nucleotide sequence of the Pseudomonas putida cytochrome P-450cam gene and its expression in Escherichia coli. J Biol Chem 1986;261:1158-1163. 18. Koga H, Aramaki H, Yamaguchi E, Takeuchi K, Horiuchi T, Gunsalus IC. camR, a negative regulator locus of the cytochrome P-450cam hydroxylase operon. J Bacteriol 1986;1089-1095. 19. Narhi LO, Fulco A. Phenobarbital induction of a soluble cytochrome P-450-dependent fatty acid monooxygenase in Bacillus megaterium. J Biol Chem 1982;257:2147-2150. 20. Schwalb H, Narhi LO, Fulco AJ. Purification and characterization of pentobarbital-induced cytochrome P-450BM-1 from Bacillus megaterium ATCC 14581. Biochim Biophys Acta 1985;838:302-311. 21. Narhi LO, Fulco AJ. Characterizationof a catalytically self-sufficient 119,000dalton cytocrome P-450 monooxygenase induced by barbiturates in Bacillus megaterium. J Biol Chem 1986;261:71-7 169. 22. Narhi LO, Fulco AJ. Identification and characterization of two functional domains in cytochrome
23.
24. 25. 26. 27. 28.
29.
30. 31.
32. 33.
34.
35.
36. 37.
38. 39. 40. 41. 42.
P-450BM-3, a catalytically self-sufficient monooxygenase induced by barbiturates in Bacillus megaterium. J Biol Chem 1987;262:6683-6690. Shafiee A, Hutchinson R. Macrolide antibiotics biosynthesis: Isolation and properties of two forms of 6-deoxyerythronolideB hydroxylase from Saccharopolyspora erythraea (Streptomyces erythreus). Biochemstry 1987;26:6204-6210. Romesser J, O’Keefe DP. Induction of cytochrome P-450-dependent sulfonylurea metabolism in Steptomyces griseolus. Biochem Biophys Res Commun 1986140:65+659. Sariaslani FS, Kunt DA. Induction of cytochrome P-450 in Sterptomyces griseus. Biochem Biophys Res Commun 1986;141:405-410. Serizawa N. Cytochrome P-450 of actinomycetes (in Japanese). Nippon Nogeikagaku Kaishi 1992;66:14F153. Sariaslani FS. Microbial cytochrome P-450 and xenobiotics metabolism. Adv Appl Microbiol 1991;36:133-178. Watanabe I, Nara F, Serizawa N. Cloning, characterization and expression of the gene encoding cytochrome P-450sca-2 from Streptomyces carbophilus involved in production of pravastatin, a specific HMG-CoA reductase inhibitor. Gene 1995;163%1-85. Wen LO, Fulco AJ. Cloning of the gene encoding a catalytically self-sufficient cytochrome P-450 fatty acid monooxygenase induced by barbiturates in Bacillus megaterium and its functional expression and regulation in heterologous (Escherichia coli) and homologous (Bacillus megaterium) host. J Biol Chem 1987;262:667&6682. Omer CA, Lenstra R, Litle PJ, Dean C, Tepperman JM, Let0 KJ, O’Keefe DP. Genes for two herbicide-induciblecytochromesP-450 from Streptomyces griseolus. J Bactriol 1990;172:3335-3345. O’Keefe DP, Gibson KJ, Emptage MH, Lenstra R, Romesser JA, Litle PJ, Omer CA. Ferredoxins from two sulfonylurea herbicide monooxygenase systems in Saeptomyces griseolus. Biochemistry 1991;30:47-455. Anderson JF,Hutchinson CR. Characterization of Saccharopolysora erythraea cytochrome P-450 genes and enzymes, including 6-deoxyerythronolide B hydroxylase. J Bacteriol 1992;174:725-735. Serizawa N, Watanabe I, Nara F. Transcriptional regulation of cytochrome P-450sca gene from Streptomyces carbophilus concerning production of pravastatin, a cholesterol-lowering drug. Abstracts from the Annual Meeting of Japanese Biochemical Society, Osaka, Japan 1994;66:742. Shaw GC, Fulco AJ. Inhibition by barbiturates of the binding of Bm3Rl repressor to its operator site on the barbiturate-inducible cytochrome P450BM-3 gene of Bacillus rnegaterium. J Biol Chem 1993;268:2297-3004. Omer CA, Lenstra R, Litle PJ, Dean C, Tepperman JM, Leto KJ, Romesser JA, O’Keefe DP. Genes for two herbicide-inducible cytochromes P-450 from Streptomyces griseolus. J Bacteriol 1990172:3335-3345. Katz E, Thompson CJ, Hopwood DA. Cloning and expression of the tyrosinase gene from Streptomyces antibioticus in Streptomyces lividans. J Gen Microbiol 1983;129:2703-27 14. Hopwood DA, Bibb MJ, Chater KF, Kieser T, Bruton CJ, Kieser HM, Lydiate DJ, Smith CP, Ward JM, Schrempf H. Genetic Manipulation of Streptomyces: a Laboratory Mannual, John Innes Foundation, Norwich, UK, 1985. Pate1 NV, Omer CA. Phenobarbital and sulfonylurea-inducible operons encoding herbicide metabolizing cytochromes P-450 in Streptomyces griseolus. Gene 1992;112:67-76. Waxman DJ, Azaroff L. Phenobarbital induction of cytochrome P-450 gene expression. Biochem J 1992;281:577-592. Shaw GC, Fulco AJ. Barbiturate-medicated regulation of expression of the cytochrome P450BM-3 gene of Bacillus megaterium by Bm3R1 protein. J Biol Chem 1992;267:5515-5526. Fulco AJ. P450BM-3 and other inducible bacterial P450 cytochromes: biochemistry and regulation. Ann Rev Pharmacol Toxicol 1991;31:177-203. Wen L-P, Ruettinger RT, Fulco AJ. Requirement for a 1-kilobase 5’-flanking sequence for barbiturate-inducible expression of the cytochrome P-450BM-3 gene in Bacillus megaterium. J Biol Chem 1989;264:1099&11003.
3 89 43. He J-S, Fulco AJ. A barbiturate-regulated protein binding to a common sequence in the cytochrome P450 genes of rodents and bacteria. J Biol Chem 1991;266:7864-7869. 44. Fulco AJ, He J-S, Liang Q. Induction by barbiturates of P450 cytochromes and other drugmetabolizing enzymes in bacteria and eukaryotes. Abstracts of 10th International Symposium on Microsomes and Drug Oxidations, Toronto, Canada, July 18-21, 1994 (University of Toronto).
This Page Intentionally Left Blank
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R. El-Gewely, editor.
39 1
Novel molecular biology approaches to acellular vaccines Rino Rappuoli and Mariagrazia Pizza IRIS, Chiron Biocine Immunobiological Research Institute Siena, Siena, Italy
Abstract. Bacterial toxins are commonly detoxified by chemical treatment in order to use them in human vaccines. We have used site-directed mutagenesis of toxin genes to obtain bacteria that produce naturally nontoxic mutants of bacterial toxins, such as pertussis toxin (PT), cholera toxin (CT) and Escherichia coli heat-labile enterotoxin (LT). Genetically detoxified PT showed a superior safety and immunogenicity in animal models, phase I and phase II clinical trials, and a superior protective efficacy in the early and late stage of a phase III efficacy trial, proving in a definitive and extensive way that genetic detoxificationof bacterial toxins can, and should, replace chemical treatment. The results obtained with genetically inactivated LT and CT indicate that genetic detoxification of bacteria1 toxins can be used not only to produce vaccines for systemic immunization that are superior to the ones produced by conventional technologies, but suggest that these type of molecules may be the prototype molecules for the design and construction of innovative vaccines with a totally new design, such as mucosally delivered preventive and therapeutic vaccines.
Key words: acellular vaccines, ADP-ribosyltransferase activity, animal model, B. pertussis, cholera toxin, clinical trials, computer modelling, E. coli heat-labile enterotoxin, filamentous hemagglutinin, genetic detoxification, immunogenicity,LT-K63 mutant, PAETA, DT, pertussis toxin, PT-9W129G mutant, sitedirected mutagenesis, V. cholerae, vaccine development.
Introduction Large-scale, safe vaccination was introduced after the discovery of Ramon in 1924 that diphtheria toxin could be detoxified by a simple chemical treatment: exposure to formaldehyde [l]. This process was rapidly applied also for the detoxification of tetanus toxin allowing the development of diphtheria and tetanus vaccines, which were introduced for universal immunization in the subsequent decades. Today, over 60 years after the initial discovery, diphtheria and tetanus vaccines are still produced .by the same process described by Ramon in 1924 [2,3]. Following the initial discovery, formaldehyde was frequently used for the inactivation of other toxins, to kill bacteria and viruses that were used in vaccines. The best examples of vaccines used for large-scale immunization, produced by formalin inactivation of microorganisms are the Salk polio vaccine, and the whole cell pertussis vaccine. Even very recently, one of the latest vaccines introduced for human immunization contains a virus, Hepatitis A, that has been inactivated by formalin treatment [4]. In spite of the great successes obtained by using formalin or other chemical treatments to inactivate bacterial toxins and whole microorganisms, this process is not ideal. The chemical treatment modifies the antigenic properties of the vaccine molecules so that large
392 amounts of antigen are required to induce a low affinity immune response against the native antigen. In addition, the chemical reaction may be reversible and the inactivated microorganisms and toxins may revert and become pathogenic. Several examples of reversion are available, in some cases resulting in a severe intoxication or death of the vaccinated healthy people. Today, thanks to the new technologies, it is possible to inactivate toxins and microorganisms using genetic tools, obtaining vaccines that are safer than the conventional ones and induce an immune response that is qualitatively superior to the one induced by conventional vaccines. In this chapter we will describe the development of a recombinant pertussis vaccine, the first example of a vaccine produced by genetic inactivation of a bacterial toxin. The vaccine resulted safer and superior to the conventional ones in animal models, phase I, phase 11 and phase 111 studies, confirming the superior performance of vaccines obtained by these technologies. The genetic inactivation of cholera and E. cofi heatlabile enterotoxin will be also described as another example of the potential of this field.
Pertussis vaccine Whooping cough: the disease Pertussis, or whooping cough, is a disease caused by Bordetella pertussis, a gramnegative bacterium which adheres to the cilia of the upper respiratory tract of humans, colonizes this tissue and releases a number of substances, which induce local and systemic damages that cause the clinical symptoms [5,6]. B. pertussis infects virtually all children who are not immune. The disease is characterized by longlasting paroxysmal cough, accompanied by whoops, vomiting, cyanosis and apnoea; fever is either modest or absent. The most common complications are pneumonia, seizures, encephalopathy and death. Antibiotic treatment is very effective in clearing the bacteria but has little consequence on the disease, mostly because the disease is usually diagnosed when the bacteria have already released the toxins that are responsible for the local and systemic damages. The disease can occur at any age but is most frequent in children. Mortality is frequent under 1 year of age and occurs mostly in developing countries where it can be as high as 1/100. At a lower frequency, mortality is still present also in developed countries where the frequency is between 1/1,000 and l/lO,OOO. In Western countries, of infants who get the disease in the 1st year of life, 50% are hospitalized and 1% die. After the 1st year of life, only 4% of children with the disease are hospitalized. It has been estimated that worldwide whooping cough causes over 500,000 deaths each year [7,8]. Prevention of the disease by conventional vaccines Vaccination is the only way to control pertussis. Mass vaccination using killed bacteria (cellular vaccine) was introduced in the 1950s and reduced the incidence of
393 diseases in infants by 99%. In the USA, the incidence decreased from 150 cases per 100,OOO population in the prevaccine era, to 0.6 cases per 100,000 population in 1980 [8,91. In spite of its efficacy, the vaccine composed of whole, inactivated B. pertussis cells is not widely used mostly because of the fear of the side effects which have been associated with it. Most of these are mild reactions such as redness, swelling and fever and occur in 3CF70% of the vaccinees. However, the vaccine has also been associated with more severe reactions such as neurological damages and death, which have been reported with a frequency of 1/100,000 and 1/300,000, respectively [lo]. Although never proven to be caused by the vaccine, these reactions have caused a drop in vaccine uptake and stressed the need for a new vaccine, devoid of side effects and possibly composed of well-defined and purified antigens. According to the current immunization practices, vaccination is given only to infants and toddlers. As a consequence, children, adolescents and adults who were fully immunized as infants become susceptible to infection approximately 2-5 years after the last vaccine dose, when immunity wanes [ll-141. The infection in previously immunized individuals usually results in a form of the disease that is less severe than the one observed in infants, and is characterized by a persistent cough lasting 1 month or longer. However, a typical, severe disease consisting of a catarrhal phase followed by painful paroxysms of cough, and vomiting lasting for several weeks is also observed. Therefore, in spite of vaccination of infants, control of B. pertussis has not yet been achieved and the bacterium continues to spread within the population and attacks susceptibles [15-181. For this reason, in the USA alone pertussis is still responsible for over 30,000 cases, 3,300 hospitalizations and 25 deaths per year [8].
The rationale of acellular vaccines B . pertussis produces many molecules which are involved in the virulence of the bacterium, and are candidates for inclusion in a vaccine against whooping cough [5]. These include molecules involved in the adhesion of the bacteria to the eukaryotic cells and to the cilia of the upper respiratory tract, and molecules which cause local and systemic damage of the host. The pioneering work of Yuji and Hiroko Sato, showed that two of these molecules, pertussis toxin (PT) and filamentous hemagglutinin (FHA), purified from culture Supernatants of B. pertussis and detoxified with 'formaldehyde, can protect mice from the challenge with virulent B. pertussis and children from disease [19]. Using a combination of these molecules, they developed a new vaccine, which has been used in Japan since 1981 in children over 2 years of age. Following the Japanese example, many laboratories developed acellular vaccines containing pertussis toxin and filamentous hemagglutinin and also other purified B. pertussis antigens, such as the fimbriae or agglutinogens, and an outer membrane protein named pertactin or 69K [20]. In order to be included in vaccines, pertussis toxin needs to be detoxified. Y. Sat0 used formalin treatment to make the toxin inactive, a procedure developed by Ramon
394 and Glenny for diphtheria in 1924 [1,21]. This procedure, although effective for diphtheria and tetanus, was not entirely satisfactory for pertussis toxin that, in some cases, showed some reversion to toxicity [22]. To obtain completely detoxified molecules, a number of other chemical methods were used; these included hydrogen peroxide [23,24], tetranitromethane [25] and glutaraldehyde. In our laboratory we used genetic tools instead of chemical reagents to detoxify pertussis toxin.
Pertussis toxin Pertussis toxin is a complex bacterial protein toxin composed of five noncovalently linked subunits named S1 through S5 [26]. The subunit S1 is an enzymatic protein of 26,220 daltons with ADP-ribosyltransferase activity. The S1 subunit binds NAD and transfers the ADP-ribose group to a family of GTP-binding proteins, such as G l i and Go, that are involved in signal transduction in eukaryotic cells [27,28], thus altering their response to extracellular stimuli [29-311. B is a nontoxic oligomer formed by four distinct subunits, named S2, S3, S4 and S5, of 21,920,21,860, 12,060 and 11,770 daltons, respectively, that are present in a 1:1:2:1 ratio [26]. The B oligomer binds the receptor on the surface of eukaryotic cells and facilitates the translocation of the enzymatically active subunit across the cell membrane so that it can reach the target G proteins. The genes coding for the five subunits of pertussis toxin are clustered in a 3.2 kb fragment of the chromosomal DNA and have the typical organization of a bacterial operon, that was sequenced by two different research groups in 1986 [32,33]. Downstream of the PT operon there is another operon coding for the secretion apparatus of PT [34,35]. The crystal structure of the molecule has also been recently solved [36].
Genetic detoxification of pertussis toxin Following the cloning and sequencing of the 3.5 kb DNA fragment containing the five PT genes [32,33], the attempt to express the operon in E. coli failed. The individual subunits of the pertussis toxin were successfully expressed in E. coli and Bacillus subtilis, however, these recombinant molecules were unable to induce protective immunity, because protection was found to be mediated by conformational epitopes which are present in the entire, assembled pertussis toxin but absent in the individual recombinant subunits. The failure to induce a good protective immunity with recombinant subunits 'expressed in E. coli and B. subtilis, suggested that the ideal vaccine should be a PT molecule whose toxicity had been eliminated by the genetic manipulation of the gene coding for the S1 subunit. Therefore, to obtain a molecule suitable for vaccine use, we decided to modify the PT genes in the chromosome of B. pertussis in order to obtain a molecule fully assembled but devoid of toxicity. For this purpose, we and other investigators generated a number of recombinant S 1 molecules containing amino acid substitutions and tested their enzymatic activity. Substitutions of Arg9, Asp1 1, Argl3, Trp26, His35 and Glu129 were found to reduce
395 the enzymatic activity of the S1 subunit to undetectable levels [2,37-391). Each of the above mutations was then introduced into the chromosome of B. pertussis using homologous recombination between DNA sequences cloned in E. coli, and those present in the chromosome of a B. pertussis strain whose wild-type PT gene had been deleted. These new B. pertussis strains were found to produce molecules indistinguishable from PT in an SDS-PAGE, but that had a toxicity ranging from 0.1 to 10% of wild-type PT. Since even 0.1% of the toxicity is by far too high for a molecule to be used in a vaccine, we combined some of the above mutations and obtained several PT double mutants. One of these (PT-9K/129G) containing the mutations (Arg9 Lys and Glu129 Gly), had lost all toxic properties of PT and represented an ideal candidate for vaccine development [40]. Biochemical and antigenic properties of the genetically detoxified PT Extensive studies were performed to establish that PT-9K/129G was identical in structure to the wild-type pertussis toxin, but had lost all the toxic properties of the molecule. Biochemically, the molecule showed an identical behavior in SDS-PAGE. The presence of the two amino acid changes in the protein were confmed by mass spectrometry of peptides derived from the proteolytic treatment of the purified S1 subunit. The absence of toxicity was confirmed in vitro by the CHO cell assay that showed that toxicity could not be revealed even using a million-fold the amount of toxin inducing toxicity [40] (Table l), and by a number of in vivo assays such as the histamine sensitivity assay, the enhancement of insulin secretion, the induction of lymphocytosis, and the potentiation of anaphylaxis; in all instances, complete absence of toxicity was observed [41] (Table 1). Recently, new assays have shown that PT9K/129G has lost also other toxic properties such as the ability to induce IgE [42,43], to modify the permeability of intestinal epithelium, a phenomenon that is observed for many months after exposure to a few nanograms of active pertussis toxin [42]. In summary, the molecule showed an excellent safety profile that encouraged us to use it to develop a new vaccine against pertussis. While all the toxic properties disappeared, the new molecule retained all the nontoxic biological properties of pertussis toxin, which are mainly due to the receptor binding domain of pertussis toxin. These are the ability to agglutinate red blood cells, to induce praliferation of T cells [40,44], and to activate platelets [45] (Table 1). The antigenic properties of the molecule were found to be identical to those of the wild-type toxin both from the B and T cell point of view; studies with a panel of mono- and polyclonal antibodies showed that all the B cell epitopes tested were present and recognized with the same affinity. Similarly, a panel of T cell clones raised against the wild-type toxin were not able to distinguish the mutant from the wild-type [40] (Table 1). After the biochemical and antigenic characterization, the new molecule was tested for its ability to induce an appropriate immune response. Mice, guinea pigs, rabbits were immunized with PT-9K/129G, and the antibody response was measured. In all instances, the recombinant molecule was found to be from 10- to 20-fold more
396 Table 1. Toxic and nontoxic properties of PT and PT-9W129G mutant. Property Toxic properties of PT CHO cell-clustered growth (ng/ml) Histamine-sensitization (pg/mouse) Leukocytosis stimulation (&mouse) Anaphylaxis potentiation (pg/mouse) Enhanced insulin secretion (&mouse) IgE induction (in vitro) (ng/ml) IgE induction (in vivo) (@rat) Long-lasting enhancement of nerve-mediated intestinal permeabilization of antigen uptake (@rat) Inhibition of IL1-induced IL2 release in EL4 6.1 cells (pg/ml) Lethal dose (pglkg) ADP-nbosylation (ng) Nontoxic properties of PT T cell mitogenicity (pg/ml) Hemagglutination (pg/well) Mitogenicity for PT-specific T cells (pg/ml) Platelet activation (pg/ml) Mucosal adjuvanicity (&mouse) Affinity constant (monoclonal 1B7,anti4 1) [K(l/mol)l Affinity constant (polyclonal anti-PT) [K(l/mol)]
T Native F
PT-9WI29G mutant
Ref.
>5000 >50 >50 >1.5 >25 >loo >200 >200
40 41 41 41 41 43 42 42
0.1
>I00
89
15.0 1.o
> 2 m
40
0.1-0.3 0.1-0.5 3.0 5.0 3.0 2.4 x 10.08
0.1-0.3 0.1-0.5 3.0 5.0 3.0 6.1 x 10.08
44 44 40 90 91 41
2.0 x 10.O'o
9.8 x 10.0~
44
0.005 0.1-0.5 0.02 0.04 <1 0.8 10.0 1.o
>1500
U.D."
Note: < means no effect was observed at the highest dose reported that was used in the assay. "U.D. = unpublished data.
immunogenic than any toxin form that had been detoxified by chemical treatment [44]. Finally, the new molecule was tested for its ability to induce protection from experimental bacterial infection in mice. Dose-dependent, efficient protection was observed (see Table 2) [40,41,44]. The above studies indicated that PT-9W129G represented a safe and very immunogenic molecule that could be used either alone or combined with other B. pertussis antigens to develop a human vaccine. The molecule was then produced in large scale, formulated in antidiphtheria, tetanus and pertussis vaccine (DTaF'), where the antipertussis component was either made by the mutant PT alone (10 pg), or by the mutant PT, combined with FHA and 69K (5, 2.5 and 2.5 pg each, respectively). The vaccines were tested for safety and immunogenicity in phase I and phase I1 studies [46-50] with optimal results before being used for larger comparative international studies.
Clinical trials Following the Japanese experience, an efficacy study was organized in 1986 in
397
Table 2. Mouse protective activity of PT-9K/129G against IC infection. Vaccine Dosea NIH standard cellular vaccine (l@.$ljD 0.008 0.0010 0.00032 30.00 12.00 4.80
-_
No. of survivors/total no. tested 16/16 13/16 9/16 1/16 16/16 16/16 12/16
b a E t E w k r r l m s d m ~ ~ ~ n tEPl5XqEGd F e in pg/mouse. 1. Vaccine contains eight i n t e m a t w protectia&ts per ml. Sweden to test two vaccines: one containing chemically (formalin-treated) detoxified PT, and the other containing formalin-treated PT and FHA. Although infants were immunized only with two doses, both vaccines were able to protect infants from disease (54 and 69%, respectively) [51]. At the end of this trial, acellular vaccines were not licensed because the observed efficacy was not considered enough, and the absence in the study of an arm immunized with the whole cell vaccine did not allow to compare the efficacy of acellular vaccines with that of the vaccine already in use. In addition, some safety concerns were raised. Finally, during the study, the formalin detoxified pertussis toxin showed some reversion to toxicity and this suggested that further development of acellular vaccines was still necessary [22]. During the period 198E1989, many acellular vaccines were developed by vaccine manufacturers. All of them contained either detoxified pertussis toxin alone or combined with FHA, with 69K, and with the agglutinogens. Some of the vaccines developed are reported in Table 3. As shown, the pertussis toxin included in the vaccines has been detoxified with a variety of chemical methods, including formaldehyde, glutaraldehyde, tetranitromethane and hydrogen peroxide. Our vaccines were the only ones containing genetically detoxified pertussis toxin. In 1990, the National Institute of Allergy and Infectious Diseases, performed a large scale phase I1 trial in the USA to compare the safety and the immunogenicity of most of the acellular vaccines available at that time, in order to select the vaccines to be subsequently used in new efficacy studies. Thirteen acellular vaccines were tested and compared with two whole cell vaccines. One hundred and twenty infants were immunized with three doses of each vaccine. Eleven of the acellular vaccines contained chemically detoxified PT, and two vaccines contained our genetically detoxified PT. The results showed that all acellular vaccines were much safer [52] and more immunogenic than the whole cell vaccines [53]. A comparison of the immunogenicity of the pertussis toxin used in the vaccines of the study, demonstrated unequivocally that the genetically detoxified pertussis
398 Table 3. List of acellular vaccines and their composition. Manufacturer
Vaccine composition
Chiron Biocine (B-I) North American Vaccine
Antigen quantity (Pi%)
FT detoxification
10 40
50
Genetic Hydrogen peroxide Tetranitromethane
23.4, 23.4
Formaldehyde
25.0, 25.0
Formaldehyde
25.0, 25.0 25.0, 25.0
Glutaraldehyde Formaldehyde + Glutaraldehyde
("4V Swiss Serum 8z Vaccine Inst. (SSV-I) Connaught (US)/Biken (CB-2) Michigan Dept. Public Health (Mich-2) Pasteur Merieux (PM-2) SmithKline Beecham Biologicals (SKB-2) Chiron Biocine (B-3) Lederle-Praxis Biologicals (LPB-3) SmithKline Beecham Biologicals (SKB-3)
FT,FHA,69K PT,FHA,69K
5.0, 2.5, 2.5 10.0, 20.0, 5.0
Genetic Formaldehyde
PT,FHA,69K
25.0, 25.0, 8.0
Formaldehyde + Glutaraldehyde
Connaught Laboratories (Canada) (CLL-3) Porton International (Por-3) Jxderle F'raxisflakeda (LPT-4) Connaught Labs. (Canada) (CLL-4)
PTT,FHA,FIM2,FIM3
10.0, 5.0, - 5.0"
Glutaraldehyde
FT,FHA,FIMZ,FIM3
10.0, 10.0,- 10.O8
Formaldehyde
PTT,FHA,69K,FIM2
3.5, 35.0, 2.0, 0.8
Formaldehyde
10.0, 5.0, 3.0, 5.0"
Glutaraldehyde
PT,FHA,69K,FIM2,FIM3
"FIM2+FIM3.
toxin induced anti-PT levels that in ELISA and toxin neutralization were 5- to 20-fold higher than those induced by chemically detoxified forms of PT (see Table 4). Following the comparative phase II study described above, four vaccines were selected to be tested in efficacy trials. Two of them, the Connaught five-component vaccine and the SmithKline Beecham Biologicals two-component vaccine, were tested in an efficacy trial performed in Sweden, while our vaccine was tested in Italy, in parallel with the three-component vaccine produced by SmithKline Beecham Biologicals. We therefore had the unique opportunity to test two vaccines containing exactly the same components (PT, FHA and 69K), but differing in the method used to detoxify F T and in the amount of antigen present (25, 25 and 8 pg in the SmithKline Beecham Biologicals vaccine, and 5, 2.5 and 2.5 pg in our vaccine, respectively). It is clear that although the two vaccines contained the same three antigens, since our vaccine contained a lower amount of them, any advantage that was found in our vaccine had to be attributed to the superior quality of the
399 Table 4. Immunogenicity of genetically (Chiron Biocine) vs. chemically (all others) inactivated PT in phase II and phase III trials [53-55]. Vaccine
B-1 B-3 CB-2 SKB-2 SKB-3 SSVI-1 PM-2 Mich-2 LPB CLL-3 CLL-4 Por-3 LPT-4
Immunogenicity
Immunogenicity (phase 111 efficacy studies)
Absolute value
Units/pg protection
Absolute value
Units/pg protection
ELISA
CHO
ELISA
CHO
ELISA
CHO
ELISA
CHO
180 99 127 104 54 99 68 66 39 38 36 29 14
1035 487 841 530 205 259 432 327 163 158 142 118 116
18.0 19.8
103.5 97.4 36.0 21.2 8.2 5.1 17.2 13.0 16.3 15.8 14.2 11.8 33.1
-
-
59.9 51.3
787 NA 230
18.8 -
-
94.4
-
-
5.4
4.1 2.1 1.9 2.7 2.6 3.9 3.8 3.6 2.9 4.0
-
2.3 2.0 -
157 -
NA 9.2 -
-
4.9 -
Note: NA = not applicable.
genetically detoxified toxin. The results of the Italian and Swedish trials showed unequivocally that all acellular vaccines had a greatly superior safety than the whole cell vaccines (fever alone occurred with a frequency of 40% in infants vaccinated with whole cell vaccines, and with a 5-8% in children vaccinated with acellular vaccines). All acellular vaccines showed efficacy, however, the most efficacious vaccines were the three-component vaccine of Chiron Biocine, containing the genetically inactivated PT (84.2%), the SmithKline Beecham Biologicals vaccine containing the formalin detoxified PT (83.9%), and the five-component of Connaught Laboratories, also containing formalin inactivated PT (85%). The SmithKline Beecham Biologicals vaccine containing the PT and FHA alone showed a surprisingly low efficacy (58%). The whole cell vaccine performed very poorly (only 3 6 4 8 % protection) [54,55]. As previously observed in phase 11 studies, the genetically detoxified PT was the most immunogenic in these studies, inducing a superior ELISA and toxin neutralizing antibody titer ([54] and Table 4). This superior immunogenicity may be responsible for other important features observed in the study: the vaccine containing genetically detoxified PT was the only one able to protect, starting from the first vaccination dose, and showed a longer lasting protective immunity. In addition, it showed a lower reactogenicity. The early protection observed with the Chiron Biocine vaccine may be the result of the superior immunogenicity of PT, while the low reactogenicity is likely to be due to the low antigenic content of this vaccine. Although the efficacy of the whole cell vaccine was unexpectedly low, and other studies suggest that it can be much higher than the one observed in the Swedish and
400 Italian trials, there is no doubt that the results of these two trials define the end of the whole cell vaccines and the beginning of the era of acellular vaccines. While we expect all efficacious acellular vaccines to be licensed, a close look at the results of the trials suggests that a vaccine containing the genetically detoxified pertussis toxin has the following advantages: 1. The native conformation of the genetically detoxified pertussis toxin results in a superior immunogenicity. In addition to the superior antibody titers achieved with a very low dose of PT, in the phase I11 trial in Italy, the vaccine containing the genetically detoxified PT was the only one able to confer protective immunity starting after the first vaccine dose, allowing the protection of infants in the first few months of life, when the disease is most dangerous. 2. The native conformation of the genetically pertussis toxin induced a better priming of the immune system against the natural molecule that resulted in a longer lasting protective immunity. 3. The superior immunogenicity of genetically detoxified PT allowed to use a lower antigen dose in the vaccine, which resulted in a lower frequency of common side effects such as fever, redness and swelling. 4. Only genetic detoxification can guarantee the absolute absence of active pertussis toxin. The absence, even in minimal amounts, of active pertussis toxin is crucial in novel vaccines, because this toxin has been shown to cause anaphylaxis and permanent modification of the nerve-mediated permeability of the intestine [42,56]. Chemical detoxification is less reliable, and active pertussis toxin has been reported in several acellular vaccine preparations [57], while on another occasion, reversion to toxicity has been reported [22].
Conclusion Genetically detoxified PT showed superior safety and immunogenicity in animal models, phase I and phase I1 clinical trials, and a superior protective efficacy in the early and late stage of the phase 111 efficacy trial. These results prove in a definitive and extensive way that genetic detoxification of bacterial toxins can, and should, replace the chemical treatment developed by Ramon in 1924 because it produces vaccines with a superior safety and efficacy. This new technology that has been proven so effective in pertussis, opens also the future to the development of new vaccines of this type.
Cholera and E. coli vaccines The diseases Vibrio cholerae and enterotoxigenic E . coli are noninvasive gram-negative bacteria that infect the small bowel, where they release potent toxins which cause fluid accumulation and diarrhea. Cholera is still an important problem for public health in
40 1 many developing countries. It is estimated that more than 150,000 people, including children and adults, die each year from cholera. Enterotoxigenic E. coli cause the socalled travellers' diarrhea, which is a milder but frequent and fastidious disease. Both diseases are caused by two related toxins, cholera toxin (CT) and E. coli heat-labile enterotoxin (LT). The first vaccine developed against cholera was an injectable vaccine containing killed bacteria cells. This vaccine has never been extensively used, and has been recently abandoned, because of the low efficacy and the high reactogenicity. Very recently two new oral vaccines have been developed against cholera. One is a liveattenuated strain (CVD 103-HgR) [58-601 of V. cholerue containing a deletion of the A subunit of cholera toxin, which has been shown to protect volunteers from challenge with V. cholerue. The other is a vaccine containing 1 mg of purified B subunit of cholera toxin and 10" killed bacteria [61-631. This vaccine showed a protective efficacy against cholera and also against E. coli-induced diarrhea [64,65]. Although the above two vaccines are not the final solution to the problem, because they are unable to cover against the new strain (Bengal or 0139) of V. cholerue that has recently emerged, they prove that effective vaccination against the above diarrheal diseases can be achieved, and pave the way to development of optimal vaccines. Cholera and LT toxins Cholera and the E. coli heat-labile enterotoxins are two homologous proteins that belong to the family of bacterial ADP-ribosylating toxins. As all members of the group, they are composed of two subunits with enzymatic and binding activity, respectively. In CT and LT, the A subunit is formed by a 239-polypeptide chain bearing ADP-ribosyltransferase activity that, as in the case of pertussis toxin, binds NAD and transfers the ADP-ribose group to GTP-binding proteins involved in transmembrane signaling. The main target of CT and LT is G,, a protein that activates the adenylate cyclase thus inducing the synthesis of the CAMP second messenger [27,28]. The B oligomer is formed by five identical subunits of 103 amino acids that assemble into a pentameric structure; this structure binds the GM1 receptor ganglioside. LT, in addition to GM1, also binds other receptors containing a terminal galactosyl moiety. The genes coding for CT and LT are highly homologous and are organized in a 1.6 kb operon, located on the chromosome of V. cholerue and on a plasmid of E. coli. The amino acid sequences of both toxins and the nucleotide sequences of their genes are known [67,68]. A review of the different versions of the primary amino acid sequences of these toxins has been recently published [69]. The crystal structure of LT, with and without the bound galactose, has been solved [70,7 11. Protein structure and computer modelling The crystal structure of ADP-ribosylating toxins [36,71-731 has shown that, in spite of the absence of obvious similarities in the primary structure, the active site of these
402 toxins is remarkably conserved [74,75]. PT, CT, LT, diphtheria toxin (DT) and pseudomonas exotoxin A (PAETA), share a common structure of the NAD-binding and catalytic site. This can be described as a cavity formed by an a-helix bent over a p-strand that form the ceiling and the floor of the NAD-binding cavity, respectively. Two amino acids that are essential for catalysis are conserved in all toxins and are located in the same position, at the two sides of the cavity. These amino acids are a glutamic acid that is common to all toxins and that had been shown to be essential for catalysis by biochemical and genetic studies before the structure was known [76-791, and a second residue that can be either a histidine [80] in DT and PAETA, or an arginine in PT, CT and LT [81-831. Site-directed mutagenesis and genetic detoxification Encouraged by the results obtained with the PT and on the basis of this experience, we started the mutagenesis work on heat-labile and cholera toxins. Initially, the same amino acids found to be important for PT were mutagenized. Arg7 (equivalent to Arg9 of PT) and Glu112 (equivalent to Glu129 of PT) were changed into Lys and Ala, respectively. In contrast to the finding observed with pertussis toxin, where single mutations reduced but did not eliminate the toxicity, in LT and CT the single mutations were able to eliminate completely the enzymatic activity of the A subunit and the in vitro toxicity of the molecule. However, the amino acid mutations that were optimal in pertussis, turned out not to be very useful in LT and CT. In fact, the mutants containing the above mutations were very sensitive to protease digestion and very unstable to storage and manipulation [82,84]. Therefore, new amino acid mutations were designed using computer modelling of the LT structure that had become available in the meantime [85]. Of the many mutants tested LT-K63 (containing a Ser63 + Lys substitution, designed to fill the active site with the bulky side chain of the Lys residue) was found to be devoid of enzymatic activity, nontoxic both in vitro and in vivo, and very stable to protease treatment. This molecule was then purified and tested in immunogenicity and adjuvanticity studies. A CT molecule containing the same mutation was also developed and shown to have similar properties [86,87]. Perspectives for new vaccines The nontoxic derivatives df LT and CT were used in a number of preclinical studies to test whether these molecules had an immunogencity superior to the ones conventionally used, as we had observed in the case of pertussis. When used to immunize systematically mice and rabbits, CT-K63 gave an excellent antibody titer that was superior to that induced by the fragment B of CT, that is used in conventional vaccines. This result confirmed that, as in the case of pertussis, genetically detoxified toxins are the best immunogens that can be achieved. In addition to inducing a superior antibody titer, CT-K63 induced neutralizing antibodies against the fragment A of the toxin, thus giving also an immune response which is different in
403 quality to the one induced by conventional vaccines. Therefore, LT-K63 and CT-K63 represent nontoxic immunogens which are excellent candidates for new live and subunit vaccines against cholera and enterotoxigenic E. coli. LT-K63, in addition to being tested as antigen, was also tested for its ability to induce an immune response against antigens that were coadministered at the mucosal surface. This experiment was performed because LT and CT are well known to be excellent mucosal adjuvants that, however, cannot be used in humans because of their toxicity. Surprisingly, we found that LT-K63 in mice was able to induce an immune response against mucosally co-administered antigens [88]. The mucosal adjuvanticity of LT-K63 has now been proven for many different antigens, including ovalbumin, KLH, fragment C of tetanus toxin, the HIV antigen, gp120 and gp24, and Helicobucter pylori antigens such as native and recombinant vacuolating cytotoxin, the urease, CagA and whole cell bacteria. In all instances, LTK63 acted as mucosal adjuvant inducing both a mucosal and systemic immune response. In the case of H . pylori, LT-K63 induced a response that was able to protect mice from infection and to eradicate an already established infection. These results suggest that the mucosal adjuvanticity of LT-K63 can be used to develop mucosally delivered preventive and therapeutic vaccines.
Conclusion The results obtained with LT-K63 indicate that genetic detoxification of bacterial toxins can be used not only to produce vaccines for systemic immunization that are superior to the ones produced by conventional technologies, but suggest that these type of molecules may be the prototype molecules for the design and construction of innovative vaccines with totally new design, such as mucosally delivered preventive and therapeutic vaccines.
References 1. 2.
Ramon G. Sur la toxine et sur I’anatoxine diphtheriques. AM Inst Pasteur 1924;38:1-10. Pappenheimer AM. Diphtheria. In: Germanier R (ed) Bacterial Vaccines. Orlando: Academic Press, 1984; 1-36.
Rappuoli R. New and improved vaccines against diphtheria and tetanus. In: Woodrow GC,Levine MM (eds) New Generation Vaccines. New York and Basel: Marcel Dekker Inc., 1990;251-268. ‘4. Andr6 FE. Approaches to a vaccine against hepatitis A: development and manufacture of an inactivated vaccine. J Infect Dis 1995;171:S33-S39. 5. Rappuoli R. Pathogenicity mechanisms of Bordetella. In: Dangl JL (ed) Bacterial Pathogenesis of Plants and Animals. Berlin: Springer-Verlag, 19943 1F336. 6. Weiss A, Hewlett EL. Virulence factors of Borderella pertussis. Ann Rev Microbiol 1986;40:6613.
686. 7.
Muller AS, Leeuwenburg J, Pratt DS. Pertussis: epidemiology and control. Bull WHO 1986;64: 321-33 1.
8. 9.
Sutter RW, Cochi SL. Pertussis hospitalizations and mortality in the United States, 1985-1988. Evaluation of the completeness of national reporting. JAMA 1992;267:386391. Farizo KM, Cochi SL, Zell ER, Brink EW, Wassilak SG, Patriarca PA. Epidemiological features
404 of pertussis in United States, 198E1989. Clin Infect Dis 1992;14:708-719. 10. Miller DL, Ross EM, Alderslade R, Bellman MH, Rawson NSB. Pertussis immunization and serious acute neurological illness in children. Br Med J 1981;282:1595-1599. 11. He QS, Viljanen MK, Nikkari S, Lyytikainen R, Mertsola J. Outcomes of Bordetella pertussis infection in different age groups of an immunized population. J Infect Dis 1994;170873-877. 12. Mink CM, Sirota NM, Nugent S. Outbreak of Pertussis in a fully immunized adolescent and adult population. Arch Pediatr Adolesc Med 1994;148:153-157. 13. Wilhelm J, Kenyon T, Mihalek E, Brusealas K, Shulman S, Bergman E, Daum R, Francis BJ, Robinson D, Adcock M, Daniels J, Wells V, Christie C, Reising S, Halpin TJ, Finger R. Resurgence of Pertussis - United States, 1993 (reprinted from 1993;42:952) JAMA 1994;271:340. 14. Wright SW, Edwards KM, Decker MD, Zeldin MH. Pertussis infection in adults with persistent cough. JAMA 1995;273:1044-1046. 15. Addiss DG, Davis JP, Meade BD, Burstyn DG, Meissner M, Zastrow JA, Berg JL, Drinka P, Phillips R. A pertussis outbreak in a Wisconsin nursing home. J Infect Dis 1991;164:704-710. 16. Doebbeling BN, Feilmeier ML, Herwaldt LA. Pertussis in an adult man infected with the human immunodeficiency virus. J Infect Dis 1990;161:1296-1298. 17. Herwaldt LA. Pertussis in adults. What physicians need to know. Arch Int Med 1991;151:151(t 1512. 18. Herwaldt LA. Pertussis and pertussis vaccines in adults (editorial; comment). JAMA 1993;269:9394. 19. Sat0 Y, Kimura M, Fukumi H. Development of a pertussis component vaccine in Japan. Lancet 1984;i:122-126. 20. Moxon ER, Rappuoli R. Haemophilus injluenzae infections and whooping cough. Lancet 1990;335:1324-1329. 21. Glenny AT, Hopkins BE. Diphtheria toxoid as an immunizing agent. Br J Exp Path01 1923;4:283288. 22. Storsaeter J, Hallander H, Farrington CP, Olin P, Mollby R, Miller E. Secondary analyses of the efficacy of two acellular pertussis vaccines evaluated in a Swedish phase In trial. Vaccine 1990;8:457-461. 23. Krantz I, Sekura R, Trollfors B, Taranger J, Zackrisson G, Lagergard T, Schneerson R, Robbins J. Immunogenicity and safety of a pertussis vaccine composed of pertussis toxin inactivated by hydrogen peroxide, in 18 to 23-month-old children. J Pediatr 199@116:53+543. 24. Sekura RD, Zhang YL, Roberson R, Acton B, Trollfors B, Tolson N, Shiloach J, Bryla D, Muir-Nash J, Koeller D et al. Clinical, metabolic, and antibody responses of adult volunteers to an investigational vaccine composed of pertussis toxin inactivated by hydrogen peroxide. J Pediatr 1988;113:8068 13. 25. Winbeny L, Walker R, Cohen N, Todd C, Sentissi A, Siber G. Evaluation of a new method for inactivating pertussis toxin with tetranitromethane. In: Keith J (ed)Abstracts from the International Workshop of B. pertussis. Montana, Hamilton: Rocky Mountains Laboratory, 1988. 26. Tamura M, Nogimori K, Murai S, Yajima M, Ito K, Katada T, Ui M, Ishii S. Subunit structure of the islet-activating protein, pertussis toxin, in conformity with the A-B model. Biochemistry 1982;21:551&5522. 27. Hepler JR,Gilman AG. G'proteins. Trends Biochem Sci 1992; 17:383-387. 28. Linder ME, Gilman AG. G proteins. Sci Am 1992;267(1):56-61,64-65. 29. Bokoch GM, Katada T, Northup JK, Hewlett EL, Gilman AG. Identification of the predominant substrate for ADP-ribosylation by islet-activating protein. J Biol Chem 1983;258;2072-2075. 30. Katada T, Ui M. ADP ribosylation of the specific membrane protein of C6 cells by islet-activating protein associated with modification of adenylate cyclase activity. J Biol Chem 1982;257:7210-7-16. 31. Rappuoli R, Pizza M. Structure and evolutionary aspects of ADP-ribosylating toxins. In: Alouf J, Freer J (eds) Sourcebook of Bacterial Protein Toxins. Orlando: Academic Press, 1991;l-20. 32. Locht C, Keith JM. Pertussis toxin gene: nucleotide sequence and genetic organization. Science 1986:232: 1258-1264.
405 33.
34. 35. 36. 37. 38.
39. 40.
41.
42.
43.
44. 45. 46.
47.
48.
49.
50.
Nicosia A, Perugini M, Franzini C, Casagli MC, Bom MG, Antoni G, Almoni M, Neri P, Ratti G, Rappuoli R. Cloning and sequencing of the pertussis toxin genes: operon structure and gene duplication. Proc Natl Acad Sci USA 1986;83:4631-4635. Covacci A, Rappuoli R. Pertussis toxin export requires accessory genes located downstream from the pertussis toxin operon. Molec Microbiol 1993;8:42+434. Weiss AA, Johnson FD, Bums DL. Molecular characterization of an operon required for pertussis toxin secretion. Proc Natl Acad Sci USA 1993;90:2970-2974. Stein PE, Boodhoo A, Armstrong GD, Cockle SA, Klein MH, Read RJ. The crystal structure of pertussis toxin. Structure 1994;2:45-57. Barbieri JT, Cortina G. ADP-ribosyltransferasemutations in the catalytic S-1 subunit of pertussis toxin. Infect Immun 1988;56:1934-1941. Cieplak W, Bumette WN, Mar VL, Kaljot KT, Moms CF, Chen KK, Sat0 H, Keith JM. Identification of a region in the S1 subunit of pertussis toxin that is required for enzymatic activity and that contributes to the formation of a neutralizing antigenic determinant. Proc Natl Acad Sci USA 1988;85:4667-4671. Locht C, Capiau C, Feron C. Identification of amino acid residues essential for the enzymatic activities of pertussis toxin. Proc Natl Acad Sci USA 1989;86:3075-3079. Pizza M, Covacci A, Bartoloni A, Perugini M, Nencioni L, de Magistris MT, Villa L, Nucci D, Manetti R, Bugnoli M, Giovannoni F, Olivieri R, Barbieri JT, Sat0 H, Rappuoli R. Mutants of pertussis toxin suitable for vaccine development. Science 1989;246:497-500. Nencioni L, Pizza M, Bugnoli M, De Magistris T, Di Tommaso A, Giovannoni F, Manetti R, Marsili I, Matteucci G, Nucci D et al. Characterization of genetically inactivated pertussis toxin mutants: candidates for a new vaccine against whooping cough. Infect Immun 1990,58:1308-1315. Kosecka U, Marshall JS, Crowe SE, Bienenstock J, Perdue MH. Pertussis toxin stimulates hypersensitivity and enhances nerve-mediated antigen uptake in rat intestine. Am J Physiol Gastrointest Liver 1994;30:G745-G753. Vanderpouwkraan CTM, Rensink HJAM, Rappuoli R, Aarden LA. Co-stimulation of T cells via CD28 inhibits human IgE production; Reversal by pertussis toxin. Clin Exp Immunol 1995;99:473478. Nencioni L, Volpini G, Peppoloni S, de Magistris MT, Marsili I, Rappuoli R. Properties of the pertussis toxin mutant PT-9W129G after formaldehyde treatment. Infect Immun 1991;59:625;-630. Sindt KA, Hewlett EL, Redpath GT, Rappuoli R, Gray LS, Vandenberg SR. Pertussis toxin activates platelets through an interaction with platelet glycoprotein Ib. Infect Immun 1994;62:3108~3114. Podda A, Nencioni L, de Magistris MT, Di Tommaso A, Bossh P, Nuti S, Pileri P, Peppoloni S, Bugnoli M, Ruggiero P, Marsili I, D’Emco A, Tagliabue A, Rappuoli R. Metabolic,.humoral and cellular responses in adult volunteers immunized with the genetically inactivated pertussis toxin mutant PT-9W129G. J Exp Med 1990;172:861-868. Podda A, Nencioni L, Marsili I, Peppoloni S, Volpini G, Donati D, Di Tommaso A, de Magistris MT, Rappuoli R. Phase I clinical trial of an acellular pertussis vaccine composed of genetically detoxified pertussis toxin combined with FHA and 69 kDa. Vaccine 1991;9:741-745. Podda A, De Luca EC, Titone L, Casadei AM, Cascio A, Peppoloni S, Volpini G, Marsili I, Nencioni L, Rappuoli R. Acellular pertussis vaccine composed of genetically inactivated pertussis toxin: safety and immunogenicity in 12- to 24- and 2- to 4-month-old children. J Pediatr 1992;120:680-685. Podda A, Deluca EC, Titone L, Casadei AM, Cascio A, Bartalini M, Volpini G, Peppoloni S, Marsili I, Nencioni L, Rappuoli R. Immunogenicity of an acellular pertussis vaccine composed of genetically inactivated pertussis toxin combined with filamentous hemagglutinin and pertactin in infants and children. J Pediatr 1993;123:81-84. Podda A, Deluca EC, Contu B, Furlan R, Maida A, Moiraghi A, Stramare D, Titone L, Uxa F, Dipisa F, Peppoloni S, Nencioni L, Rappuoli R, Bartalini M, Bona G, Budroni M, Pistilli AMC, Cascio A, Cascio G, Cossu M, Dallorto P, Dileo G, Furlan A, Macagno F, Marsili I, Meloni T, Regoli D, Rig0 G, Trappan A, Vargiu G, Volpini G. Comparative study of a whole-cell pertussis
406 51. 52.
53.
54.
55.
56. 57.
58.
59.
60.
61.
62.
63.
64.
65.
66. 67.
vaccine and a recombinant acellular pertussis vaccine. J Pediatr 1994;124:921-926. Ad hoc group for the study of pertussis vaccine: placebo-controlled trial of two acellular pertussis vaccines in Sweden. Protective efficacy and adverse events. Lancet 1988;i:955-960. Decker MD, Edwards KM, Steinhoff MC, Rennels MB, Pichichero ME, Englund JA, Anderson EL, Deloria MA, Reed GF. Comparison of 13 acellular pertussis vaccines: adverse reactions. Pediatrics 1995;96(3):557-566. Edwards KE, Meade BD, Decker MD, Reed GF, Rennels MB, Steinhoff MC, Anderson EL, Englund JA, Pichichero ME, Deloria MA, Deforest A. Comparison of 13 acellular pertussis vaccines: overview and serologic response. Pediatrics 1995;96(3):548-557. Greco D, Salmaso S, Mastrantonio P, Giuliano M, Tozzi AE, Ciofi ML, Giammanco A, Panei P, Blackwelder WC, Klein DL, Wassilack SGF, The Progetto Pertosse Working Group. Clinical efficacy, immunogenicity and safety of two acellular and one whole-cell pertussis vaccines: results from the Italian trial. N Engl J Med 1995;(In press). Gustafsson L, Hallander HO, Olin P, Reizenstein E, Storsaeter J. Marked differences in efficacy, immunogenicity and reactogenicity between a 5-component acellular, a 2-component acellular, and a U.S. licenced whole-cell pertussis vaccine. N Engl J Med (1995);(In press). Munoz JJ, Peacock MG, Hadlow WJ. Anaphylaxis or so-called encephalopathy in mice sensitized to an antigen with the aid of pertussigen (pertussis toxin). Infect Immun 1987;55:1004-1008. Miller E, Waight P, Ashworth E, Thomton C, Redhead K. Summary of clinical and laboratory data for acellular and whole cell pertussis vaccines from UK studies. In: Salisbury M (ed) Compatibility of Acellular Pertussis with Other Vaccines in UK Primary Immunization and Boosting Schedules. London: PHLS, CAMR and NIBSC, 199%-16. Gotuzzo E, Butron B, Seas C, Penny M, Ruiz R, Losonsky G, Lanata CF, Wasserman S S , Salazar E, Kaper JB,Cryz S , Levine MM. Safety, immunogenicity,and excretion pattern of single-dose live oral cholera vaccine CVD 102-HgR in Peruvian adults of high and low socioeconomic levels. Infect Immun 1993;6133994-3997. Levine MM, Hemngton D, Losonsky G, Tall B, Kaper JB, Ketley J, Tacket CO, Cryz S. Safety, immunogenicity and efficacy of recombinant live oral cholera vaccines, CVD 103 and CVD 103HgR. Lancet 1988;i:467-470. Simanjuntak CH, Ohanley P,Punjabi NH,Noriega F, Pazzaglia G, Dykstra P, Kay B, Suharyono, Budiarso A, Rifai AR, Wasserman S S , Losonsky G, Kaper J, Cryz S, Levine MM. Safety, immunogenicity,and transmissibility of single-dose live oral cholera vaccine strain CVD- 103-HgR in 24-month-old to 59-month-old Indonesian children. J Infect Dis 1993;168;1169-1176. Clemens J, Sack DA, Harris JR,Van Loon F, Chakraborty J, Ahmed F, Rao MR, Khan MR, Yunus M, Huda N, Stanton BF, Kay BA, Walter S , Eeckels R, Svennerholm AM, Holmgren J. Field trial of oral cholera vaccines in Bangladesh: results from 3-year follow-up. Lancet 1990335:27&273. Clemens J, Sack D, Rao M, Chakraborty J, Kay B, Ahmed F, Khan MR, Vanloon FPL, Svennerholm AM, Holmgren J. The design and analysis of cholera vaccine trials - recent lessons from Bangladesh. Int J Epidemiol 1993;22:724-730. Holmgren J, Svennerholm AM, Jertbom M, Clemens J, Sack DA, Salenstedt R, Wigzell H. An oral B subunit: whole cell vaccine against cholera. Vaccine 1992;10:911-994. Clemens JD, Sack D, Harris JR, C h h b o r t y J, Neogy PK, Stanton B, Huda N, Khan MU, Kay BA, Khan MR, Ansaruzzaman M, Yunus M, Rao MR, Svennerholm AM, Holmgren J. Cross protection by B subunit-whole cell cholera vaccine against diarrhea associated with heat-labile toxin-producing enterotoxigenic escherichia coli: results of a large-scale field trial. J Infect Dis 1988;158:372-377. Mindell JA, Silverman JA, Collier RJ, Finkelstein A. Structure function relationships in diphtheria toxin channels. 2. A residue responsible for the channels dependence on trans pH. J Membr Biol 1994;137:2%44. Dallas WS, Falkow S. Homology between cholera toxin and Escherichia coli heat-labile toxin. Nature 1980;288:499-501. Mekalanos JJ, Swartz DJ, Pearson GD, Harford N, Groyne F, de Wilde M. Cholera toxin genes: nucleotide sequence, deletion analysis and vaccine development. Nature 1983;306:551-557.
407 68. Spicer EK, Noble JA. Escherichia coli heat-labile enterotoxin: nucleotide sequence of the A subunit gene. J Biol Chem 1982;257:5716-5721. 69. Domenighini M, Pizza M, Jobling MG, Holmes RK, Rappuoli R. Identification of errors among database sequence entries and comparison of correct amino acid sequences for the heat-labile enterotoxins of Escherichia coli and Vibrio cholerae. Molec Microbiol 1995;15(6):1165-1 167. 70. Sixma TK, Pronk SE, Kalk KH, Wartna ES, van Zanten BA, Witholt B, Hol WG. Crystal structure of a cholera toxin-related heat-labile enterotoxin from E. coli. Nature 1991;35 1:371-377. 71. Sixma TK, Kalk KH, Vanzanten BAM, Dauter Z, Kingma J, Witholt B, Hol WGJ. Refined structure of Escherichia coli heat labile enterotoxin, a close relative of cholera toxin. J Molec Biol 1993;230:89O-918. 72. Allured VS, Collier RJ, Carroll SF, McKay DB. Structure of exotoxin A of Pseudomonas aeruginosa at 3.0-Angstrom resolution. Proc Natl Acad Sci USA 1986;83:132@1324. 73. Choe S, Bennett MJ, Fuji G, Curmi PM, Kantardjieff KA, Collier RJ, Eisenberg D. The crystal structure of diphtheria toxin. Nature 1992;357:216222. 74. Domenighini M, Montecucco C, Ripka WC, Rappuoli R. Computer modelling of the NAD binding site of ADP-ribosylating toxins: active-site structure and mechanism of NAD binding. Molec Microbiol 1991;5:23-3 1. 75. Domenighini M, Magagnoli C, Pizza M, Rappuoli R. Common features of the NAD-binding and catalytic site of ADP-ribosylating toxins. Molec Microbiol 1994;14(1):41-50. 76. Barbieri JT, Mende-Meuller LM, Rappuoli R, Collier RJ. Photolabeling of Glu129 of the S-1 subunit of pertussis toxin with NAD. Infect Immun 1989;57:354+3554. 77. Carroll SF, Collier RJ. NAD binding site of diphtheria toxin: identification of a residue within the nicotinamide subsite by photochemical modification with NAD. Proc Natl Acad Sci USA 1984;81:3307-33 1 1. 78. Douglas CM, Collier RJ. Exotoxin A of Pseudomonas aeruginosa: substitution of glutamic acid 553 with aspartic acid drastically reduces toxicity and enzymatic activity. J Bacteriol 1987;169:4967497 1. 79. Pizza M, Bartoloni A, Prugnola A, Silvestri S, Rappuoli R. Subunit S1 of pertussis toxin: mapping of the regions essential for ADP- ribosyltransferase activity. Proc Natl Acad Sci USA 1988;85: 7521-7525. 80. Papini E, Schiavo G, Rappuoli R, Montecucco C. Histidine-21 is involved in diphtheria toxin NADbinding. Toxicon 1990;28:63 1-635. 81. Bumette WN, Cieplak W, Mar VL, Kaljot KT, Sat0 H, Keith JM. Pertussis toxin S1 mutant with reduced enzyme activity and a conserved protective epitope. Science 1988;242:72-74. 82. Bumette WN, Mar VL, Platler BW, Schlotterbeck JD, McGinley MD, Stoney KS, Rohde MF, Kaslow HR. Site-specific mutagenesis of the catalytic subunit of cholera toxin: substituting lysine for arginine 7 causes loss of activity. Infect Immun 1991;59:4266-4270. 83. Lobet Y, Cluff CW, Cieplak W Jr. Effect of site-directed mutagenic alterations on ADPribosyltransferase activity of the A subunit of Escherichia coli heat-labile enterotoxin. Infect Immun 1991;59:287@2879. 84. Hase CC, Thai LS, Boesmanfinkelstein M, Mar VL, Bumette WN, Kaslow HR, Stevens LA, Moss J, Finkelstein RA. Construction and characterization of recombinant Vibrio cholerae inactive cholera toxin analogs. Infect Immun 1994;62:3051-3057. 85. Pizza M, Domenighini M, Hol W, Giannelli V, Fontana MR, Giuliani MM, Magagnoli C, Peppoloni S, Manetti R, Rappuoli R. Probing the structure-activity relationship of Escherichia coli LT-A by site-directed mutagenesis. Molec Microbiol 1994a;145-60. 86. Fontana MR, Manetti R, Giannelli V, Magagnoli C, Marchini A, Domenighini M, Rappuoli R, Pizza M. Construction of nontoxic derivatives of cholera toxin and characterization of the immunological response against the A subunit. Infect Immun 1995;63:235&2360. 87. Pizza M, Fontana MR, Giuliani MM, Domenighini M, Magagnoli C, Giannelli V, Nucci D, Hol W, Manetti R, Rappuoli R. A genetically detoxified derivative of heat-labile E. coli enterotoxin induces neutralizing antibodies against the A subunit. J Exp Med 1994b;6:2147-2153.
408 Douce G, Turcotte C, Cropley I, Roberts M, Pizza M, Domenghini M, Rappuoli R, Dougan G. Mutants of Escherichiu coli heat-labiletoxin lacking ADP-ribosyltransferase activity act as nontoxic, mucosal adjuvants. Proc Natl Acad Sci USA 1995;92:1644-1648, 89. Zumbihl R, Dornand J, Fischer T, Cabane S, Rappuoli R, Bouaboula M, Casellas P, Rouot B. IL-1 stimulates a diverging signaling pathway in EL4 6.1 thymoma cells. J Immunol 1995;155:181-189. 90. Sindt KA, Hewlett EL, Redpath GT, Rappuoli R, Gray LS, Vandenberg SR. Pertussis toxin activates platelets through an interaction with platelet glycoprotein Ib. Infect Immun 1994;62:3108-31 14. 91. Roberts M, Bacon A, Rappuoli R, Pizza M, Cropley I, Douce G, Dougan G, Marinaro M, McGhee J, Chatfield S. A mutant pertussis toxin molecule that lacks ADP-ribosyltransferase activity, PT9K/129G, is an effective mucosal adjuvant for intranasally delivered proteins. Infect Immun 1995;63(6):21OO-2108. 88.
01996 Elsevier Science B.V. All rights reserved Biotechnology Annual Review Volume 2. M.R. El-Gewely, editor.
409
Strategies and applications of DNA level diagnosis in genetic diseases: past experiences and future directions S.M. Singh, D.I. Rodenhiser, R.N. Ott, J.H. Jung and P.J. Ainsworth Molecular Genetics Unit, Department of Zoology and Paediatrics, Divsion of Medical Genetics, University of Western Ontario and Molecular Medical Genetics Program, Child Health Research Institute, Children's Hospital of Western Ontario and Victoria Hospital, London, Ontario, Canada
Abstract. The development of technologies towards the detection of mutations represents one of the most important areas of molecular biology. It has played a pivotal role in the tremendous success of the elucidation of complex biological problems, including genetic diseases. Today, these proven and emerging technologies have become the basis of successful biological investigations. More importantly, they are expected to play a central role in medicine, particularly the diagnosis and prognosis of genetic diseases including genetic predispositions, the assessment of treatments including transplants and decisions on reproductive choices. In addition, these technologies hold the key to future breakthroughs. This review provides an up-to-date examination of the principles of genetic diseases, the theories behind current methods of genetic diagnosis and detection of mutations including strategies for modification and the development of future technologies as they impact on the practice of medicine and on society as a whole.
Key words: ASA, ASOH, automation for mutation detection, DGGE, DNA testing, genetic predisposition, genetic diagnosis, genetic diseases, haplotype analysis, heteroduplex analysis, linkage, mismatch cleavage, mutation, PCR, F'TT, RED, silica chip technology, SSCP.
Introduction There is an increasing awareness of the importance of genetic determinants of diseases within clinical medicine. Recent surveys demonstrate that 2-3% of livebirths have recognizable problems attributable to either chromosomal, monogenic or more complex polygenic/multifactorial mutations. It is also estimated that over one-half of all childhood deaths probably have a genetic cause. Later in life, over 10% of individuals are at risk for diseases in which there is a significant genetic component. The economic costs of genetic diseases for the health care system are extensive and significant. In terms of hospital utilization data alone, large surveys show that the average prevalence of genetic disease in pediatric and adult referral hospitals is 50% and 12-16%, respectively. It is not surprising, therefore, that considerable research has been directed at understanding the genetic causes of disease. The results provide a better and informed management of the impact of genetic diseases on society.
Address for correspondence: Prof Shiva M. Singh, Department of Zoology and Division of Medical Genetics, 307 Western Science Center, University of Western Ontario, London, Ontario, Canada, N6A 5B7. Tel.: +1-519-661-3135. E-mail:
[email protected]
Our understanding of the determinants of diseases has benefited immensely from recent developments in molecular genetic technology. These developments are directly responsible for a number of advancements in medicine, including characterization of disease causing genes using functional cloning, positional cloning, candidate genes and positional candidate approaches [ 13. The extraordinary successes associated with this field have also provided the basis and incentive for international genome projects involving human and other model genomes. The road that leads from detecting the familial nature of a condition to identifying not only the genes, but specific mutations which cause that condition, is becoming more frequently travelled. With this journey has come the refinement of technologies and the development of affordable and reliable short cuts. This is an exciting time to be working in the area of molecular genetics since the characterization of complete DNA sequences from individual chromosomes is becoming a reality for a number of organisms, including humans, and discoveries of genes related to one disease after another are made almost on a weekly basis. Such results have two important implications. The first deals with the probable function of the gene in question which is initially based on information concerning related sequences in the ever increasing sequence data bases. This offers strategies for novel experimentation, including the development of therapeutics. The second permits the immediate application of information concerning genetic defects toward the informed management and prevention of the disease in question. This relies on the detection of causal mutations in individuals, families and populations, and forms the primary focus of this review. A number of excellent reviews on genetic diagnosis have been published during the last several years [2-61. All of them have become outdated, given the rapid development in the field and the successful characterization of mutational events in some of the more common diseases with complex and heterogeneous etiologies. Before we discuss technological developments and their specific applications, it is important to outline the principles underlying genetic diseases and the circumstances where specific genetic information about a disease is informative and useful.
Principles of genetic disease A role for heredity in disease was recognized before the advent of Mendel's experiments with peas and' the gene concept. As molecular genetics developed, the actual mechanisms involved in particular disorders were identified and elucidated. This development relied heavily on principles of genetic transmission, organization, *recombination,and expression, combined with new tools and techniques. The role of genes in disease is probably best explained using a simple one gene mutation - one disease concept. There are several disorders where the phenotype is attributed to a mutation in a single gene. As an example, sickle cell disease is caused by one base change at the DNA level (A + T in codon six) resulting in an amino acid substitution (Glu + Val)
41 1 in the P-globin chain. This mutant P-globin forms hemoglobin that precipitates at low oxygen concentrations, sickling the red blood cells and causing blockage of the capillaries, thus impeding the flow of oxygen to the cells. One mutation in one gene is responsible for one disorder. Other mutations at the DNA level of this gene generally result in a variety of manifestations, including the much more severe pthalassemia. The detection of the sickle cell mutation was initially based on protein electrophoretic differences between the normal and mutant polypeptide, and is now being carried out at the level of the DNA sequence (see [7] for review). Accurate, rapid and inexpensive approaches are now available to detect this single base substitution at the DNA level in individuals, families, and populations. Most hereditary diseases are not this straightforward in their etiology and genetic causation. They all, however, share the conceptual hallmark that alterations at the level of DNA sequence are directly or indirectly responsible for the development of the disease and its phenotype. The understanding of these relationships has realized extraordinary achievements during the last decade, that are beyond the scope of this review. The following examples are intended only to provide a spectrum illustrating the complexity associated with the genetic causation of diseases. Most single gene diseases can be attributed to a panel of mutations, often resulting in the variable severity (expressivity) of the disorder. As an example, cystic fibrosis (CF) is caused by mutations in the gene encoding for transmembrane conductance regulator protein (CFTR) (see [8] for review). CF is one of the most common fatal autosomal recessive diseases, affecting European populations with an incidence of 1 in 2,000 to 3,000 births. The major mutation for this disease is a three base pair deletion causing loss of a phenylalanine (AF508) in the transmembrane conductance regulator protein. At least 350 other pathogenic mutations and numerous benign variations exist in the CFTR gene, most at relatively low frequencies in most populations. These mutations can be grouped on the basis of the nature of the defect: production, processing, regulation, or conductance (function). More importantly, genotype-phenotype information can often be of significant prognostic value. The identification of the mutations at this disease locus is primarily based on DNA and occasionally on RNA. Another example, familial hypercholesterolemia, is caused by mutations in. the gene encoding the low density lipoprotein (LDL) receptor (see [9]for review). There are over 150 known mutations in this gene. Based on the phenotypic behavior of the mutant protein they have been grouped into five classes, each class being further subdivided into multiple alleles. The effect of these mutations can be classified as null, transport defective, LDL-binding defective, internalization defective, or recycling defective. All these mutations follow an autosomal dominant familial pattern with a gene dosage effect leading to allelic heterogeneity and variable severity of the disease. The detection of the DNA mutation causing familial hypercholesterolemia in an individual, therefore, requires a very different approach than sickle cell disease. It must either take into consideration the familial nature of the mutation (if known), attempt to look for all 150 mutations known to date, or narrow the search based on the phenotypic manifestations. In the latter case, a biochemical approach to assess the
412 functional defect of the receptor protein is the first step towards characterization of the specific mutation. Whether or not different mutations of a given gene gives rise to the same or different disease phenotypes, depends on the function of the gene product. As an example, unlike the mutations in the LDL receptor gene that lead to familial hypercholesterolemia or mutations in the CFTR gene that causes cystic fibrosis, variability in the expression of allelic mutations of a gene could result in different syndromes. For example, two distinct diseases, Duchenne muscular dystrophy (DMD) and Becker muscular dystrophy (BMD) are caused by mutations in the dystrophin gene, but with different sets of mutations [lo]. These two diseases represent two extremes of manifestation, DMD being severe while BMD is a milder form of muscular dystrophy. A similar but not identical analogy could be offered for the two mucopolysaccharidosis disorders, Hurler and Scheie syndromes (see [ l l ] for review). Both diseases are the result of mutations in the a-L-iduronidase gene, the difference being in the nature of the mutation and its consequence on protein function. Such allelic complications of genetic heterogeneity are now well recognized in genetic diseases. Often, one gene can give rise to a number of seemingly unrelated diseases. The proto-oncogene RET codes for a receptor-like tyrosine kinase. Mutations of this gene have been associated not only with a number of inherited cancer syndromes (for example, multiple endocrine neoplasia type 2A (MEN2A), and 2B (MEN2B), and familial medullary thyroid carcinoma (FMTC)), but also with Hirschsprung disease, a disorder of neural crest development [12]. A missense mutation at one of five cysteine residues is reported in 97% of MEN2A patients and 86% of FMTC patients [13,14], while MEN2B is almost exclusively (98%) related to a missense mutation (Met918 + Thr) at a highly conserved amino acid site [15,16]. On the other hand, Hirschsprung disease is the result of a variety of mutations, including deletions, throughout the gene [17,18], indicating a different etiology for this disorder. The unique mutations associated with FMTC, MEN2A, and MEN2B suggest a gain of function as the cause of these disorders, while the broad range of mutations found in Hirschsprung disease patients suggest a loss of function as the possible pathogenic mechanism. A large set of human genes represent multigene families. They may have a common origin and even related or identical function. The presence of mutations in different genes of such a family may represent yet another complication in characterizing the determinants of genetic diseases. For example, the most recently cloned candidate gene for Alzheimer’s disease is located on chromosome 1 (STM2) and is highly homologous to another candidate gene for this disease localized to chromosome 14 (S182). Molecular results support the hypothesis that mutations in these two genes, which are 84% similar in their seven transmembrane domains, could cause familial Alzheimer’s disease by affecting the anchoring of these proteins in cell membranes and thereby altering cytoplasmic partitioning and intracellular protein trafficking of the amyloid precursor protein [19]. It is also likely that mutations in other yet unknown gene(s) may contribute to the development of this common disease in some families and populations.
413 The fibroblast growth factor receptors (FGFR) represent another family of genes having a common ancestry and related functions. Mutations identified in some of these genes appear to be involved in disorders of abnormal bone formation. Of the four FGFR genes characterized to date, mutations in FGFRl, FGFR2, and FGFR3 are responsible for six distinct syndromes (see [20] for review) [21-301. For example, Apert syndrome is caused by one of two mutations affecting a highly conserved region of the ligand binding domain of FGFR2 [30] and mutations in an Ig-like domain of this gene which results in structural changes to the ligand binding region may cause Pfeiffer [29], Crouzon [24] or Jackson-Weiss syndromes [22]. In fact, a single transition which results in a Cys342 + Arg substitution may lead to either Pfeiffer or Crouzon syndrome [22]. The actual mechanisms responsible for different phenotypes are yet to be elucidated. Also, Pfeiffer syndrome could be due to yet another mutation (Pro252 + Arg) in FGFRl [23]. It seems that mutations in two different genes can lead to a single syndrome and the same mutation in one gene may lead to more than one disorder. If this is the case, other factors, genetic or otherwise, may affect the expression of these genes and their associated alleles. Complexity in causations is the rule rather than the exception for most diseases. This complexity may be the result of defects in a number of genes, all leading to diseases with similar or closely related phenotypes. For example, in the mucopolysaccharidoses (MPS) (see [ 111 for review) there are 10 known enzyme deficiencies that give rise to six distinct forms of MPS. Such multiple locus heterogeneities requires that any diagnosis at the DNA or RNA level relies on the known and familial mutations, directly or indirectly. Thus, the type of mutational event, the number of genes involved and the nature of the contribution made by different mutations in the causation of a disease, need not be the same or even similar. The search for causal mutations must take all likely possibilities into consideration. The concept of genetic diseases outlined above relies on a cause and effect association between a mutational event in the DNA sequence of a nuclear gene and the distinct disease phenotypes. A diagnostic protocol for such diseases, therefore, is based on the identification of the causal mutation at the DNA or RNA level. It is easy to look for a single mutation in a single nuclear gene as in sickle cell anemia. However, there must be an increase in the complexity of the genetic approach for the detection of causal mutations, where a large number of heterogeneous mutations in a gene contribute to a disease phenotype or, unlike CF, there is no single mutation to account for a large proportion of the patients in the population. This complexity is further increased if the causal mutations belong to a number of related or unrelated genes. All genetic determinants are not present in the nucleus; some are encoded in the mitochondrial genome. Compared to nuclear genes and sequences, however, the mitochondrial genes behave in a more complex fashion with respect to transmission, expression and mutational events. Mutations in the mitochondrial DNA (mtDNA) may cause or contribute to the development of genetic diseases [31]. This adds yet another complication to the genetic dissection of human diseases. A detailed account of mitochondrial based diseases is beyond the scope of this review but has been
414 explored elsewhere [32]. We will, however, mention aspects of mitochondrial diseases that are relevant in the context of their genetic diagnosis. The mitochondrial genome is an independent, small, circular DNA with many copies in each mitochondrion and numerous mitochondria per cell. Special features of mtDNA include maternal inheritance [33] and heteroplasmy [34]. Heteroplasmy refers to a mixture of mutant and normal mtDNA (genomes) in each cell. As mitochondrial genomes segregate during cell division, daughter cells acquire varying proportions of mutant mtDNA. Therefore, a negative result of mtDNA testing on one tissue sample does not rule out the presence of mtDNA mutations in other tissues. Functionally, mitochondria are responsible for generating the basic energy unit of the cell, ATP. This is done by oxidative phosphorylation (OXPHOS), which requires five enzyme complexes encoded by 13 genes from the mitochondrial genome and over 50 nuclear genes [35]. The reliance on mitochondrial and nuclear genomes for proper OXPHOS functioning is one cause of genetic heterogeneity in mitochondrial diseases. Also, different cells and tissues have different thresholds, or minimum levels of ATP production required for normal functioning [34]. Organs and tissues with high thresholds, such as the central nervous system, are more vulnerable to mitochondrial diseases. Mutations accumulate in mtDNA over time, causing the degeneration of oxidative phosphorylation and ATP production. Because of multiple copies of the genome in each mitochondria, mutations which lead to defective gene products can accumulate, provided there are a sufficient number of normal genomes to keep the cell functional. This leads to the decline in OXPHOS function with aging. The high mutation rate in mtDNA is likely due to the lack of protective histones, and poor repair mechanisms in combination with the proximity to free radicals produced by OXPHOS [36]. Such unique features of mtDNA result in the tremendous heterogeneity of mitochondrial diseases, and a special set of requirements for diagnosis and genetic testing of diseases such as Leber’s hereditary optic neuropathy, myoclonic epilepsy and ragged-red fiber disease, chronic progressive external ophthalmoplegia syndromes, and other mitochondrial disorders. Most often, a collection of mutant mtDNA is passed from mother to child, providing a predisposition to diseases which is dependent on the accumulation of mutant mtDNA beyond the threshold level in the appropriate tissues before the disease manifests. Disease manifestation could also come about as a result of a certain level of mutations interacting with additional contributing factors which affect OXPHOS threshold limits.
Genetic disease and genetic predispositions In principle, DNA based diagnosis assumes a major role for gene mutation in the causation of disease. Although several thousand diseases fall into this category, they do not affect most individuals in a population. The genetic diseases which do affect most individuals are not caused by a mutation in a single gene, rather gene mutations provide a predisposition. By themselves these mutations are not enough to cause the disease. Other factors, either genetic, environmental, or both, are necessary for the
415
development of such diseases. Most genes leading to genetic predispositions do not follow traditional recessive and dominant modes of inheritance, rather they act additively. These are quantitative trait loci (QTLs), a new challenge in the context of genetic diseases. The identification of contributing QTLs will be a major focus of research in the next decade. It will rely on the “positional candidate” approach for the identification of mutations in QTLs, and will depend heavily on the results of the human genome sequencing project. Genetic predispositions for the development of a multifactorial and complex disease may include a number of QTLs in an interacting genetic system. Such a predisposition would be expected to act in a specific environment towards the determination of the penetrance and expressivity of a multifactorial disease. Although a number of common diseases are now recognized as multifactorial with genetic predispositions, a relatively small number of genes or QTLs that contribute to the predisposition of such diseases have been identified. The recently characterized Obese gene in mice [37], which appears to act as a determinant of body weight [38], may represent one such example. A role for such a protein in humans will likely be demonstrated soon. Tumor formation is another complex, multifactorial process which has several genetic components. One gene involved is p53, a key mediator of the cellular response to DNA damage. The p53 gene encodes a 393 amino acid phosphoprotein that exists as a tetramer and functions as a transcription regulatory factor (see [39] for review). Mutant p53 affects wild-type p53 in a dominant negative manner and mutations in p53 are associated with predispositions for the development of a variety of tumors, including colon, brain, lung, breast, skin and bladder [40].Although p53 function is beginning to be understood at the biochemical level, its role at the cellular and organismal level remains elusive. An individual heterozygous for a p53 mutation has a predisposition for tumor formation. The functional copy of the gene is sufficient for normal growth and development but, if this normal gene undergoes a somatic mutation, the cell may become cancerous and begin to multiply. This two-hit phenomenon of tumorigenesis [411 remains the hallmark of all tumor-suppressor (TS) genes. Interestingly, while almost all tumors show somatic mutational events associated with a variety of such genes, some genes (such as p53) show such an effect more often and in more tumor types than other known TS genes. Thus, the role of individual TS genes in tumorigenesis may be viewed as offering a predisposition for the development of tumors. The probability with which an individual with a germline mutation in such a gene will develop the disease, has not been explored. A “second hit” or mutation is required, which may be induced by modifications or depends on environmental factors available. In such a case, the role of environmental factors in magnifying the predisposition for the disease is now recognized but remains poorly understood. Knowledge of the specific genotype of an individual may therefore have important implications for an individual’s lifestyle, prognosis for the disease, strategies for treatment and options for the future.
416
Application of DNA level diagnosis and genetic testing Genetic testing is the analysis of human DNA, RNA and protein, including chromosome structure, to detect disease-related mutations, genotypes or karyotypes. In this discussion we will deal with mutations at the level of DNA and not karyotypic abnormalities. Results of genetic testing may be used for the prediction of disease risks, carrier detection, confirmation of diagnosis, and prediction of prognosis. The applications of genetic testing are divided into three categories: 1) population screening for the presence and frequency of disease mutations; 2) identification of a mutation in an individual and family (including prenatal and preimplantation) in order to undertake preventive or predictive measures; and 3) research to facilitate the identification of an elusive mutation. Genetic testing is different from other testing (physiological, serological, biochemical, or physical imaging, including X-rays) used during the diagnosis or predictive testing of a disease, in that its implications are far reaching for an individual and the family as a whole. As a result, a number of factors must be considered in any program of genetic testing. These include reliability and predictive value of the test, laboratory practices and quality control, informed consent, education and proper interpretation of results, and social and ethical considerations which may vary from society to society and change over time. Population screening for disease mutations is usually implemented on a specific group of individuals defined by founder effect, sex, age or other considerations. Given the very low frequency of most mutations in the population and the resources necessary for population screening, it is prohibitive to implement such a program across an entire population. The population to be screened must be restrictive, informed and benefit directly from such a program. It is obvious that for each genetic disease any population screening program will also consider the disease frequency, impact, confidentiality, social and ethical considerations, feasibility, and reliability of the predictive value of the test. Despite discussions on a number of diseases (cystic fibrosis, breast cancer, Alzheimer’s disease, Tay-Sachs, etc.) in a number of forums, these questions have not been acceptably answered to date. As a result, most proposals for population screening are still under consideration and have not been implemented. In spite of these difficulties, some screening programs have been successfully executed. Identification of a mutation in an individual or family in order to undertake preventive or predictive measures, probably represents the most common application of genetic testing today. It is now possible to test DNA or RNA from cells and prenatal tissues (chorionic villi samples and amniotic cells) for a number of rare genetic disorders. In a few cases such testing may be carried out on a single cell obtained from a dividing fertilized embryo in vitro. In the near future, it will be possible to avoid invasive procedures and routinely sort and select fetal cells from the maternal blood for genetic testing. In general, the DNA tests for genetic diagnosis used today are reliable and their predictive value is recognized. With few exceptions, most of these tests are carried out by hospital associated laboratories, some under the guidelines of a quality
417
controlling body; most laboratories are also involved in direct research and development into aspects of diseases or technologies. Usually they are able to undertake either disease-specific tests or use related tests for a number of diseases. In Canada, most such laboratories are associated with a genetic counselling center which interprets and communicates the results to the individual and family and provides resources to undertake follow-ups; confidentiality is maintained and the wishes of the individual and family are taken into consideration. Most other Western countries appear to deal with genetic testing in the health care system in a similar fashion. The ever changing technology used for most of these diseases is evolving to ensure improved reliability, predictive value, feasibility, economy and quality control. The search for disease causing mutations is usually the first focus of research following the identification and cloning of a gene implicated in human genetic disease. Invariably, this search becomes an ongoing and painstaking process, as many family-specific mutations are identified. For example, the search for mutations in the CFTR gene causing cystic fibrosis continues following the identification and cloning of the CFTR gene in 1989, with 350 mutations now identified. The search for pathologic and polymorphic mutations associated with a disease gene usually involves evaluation of related individuals and populations, since mutations are expected to follow a familial pattern. It is important to emphasize that no single technology (molecular, cytological or biochemical) is either effective, realistic or feasible for the detection of all disease causing mutations. A molecular strategy to detect pathological mutations is based on the type of DNA level alterations in the normal gene sequence. For example, finding mutations in a disease showing anticipation may involve searching for expanding repeats, as seen in such diseases as fragile-X, myotonic dystrophy and Huntington’s disease. Some of these will form the focus of discussion in a later section which will deal with ever evolving molecular technologies for detection of DNA mutations. In the following section, mutational events that are implicated in most genetic diseases are summarized. This summary is not extensive, rather it represents our experience to date in dealing with genetic diseases.
Nature of mutations causing genetic diseases The basis of all genetic diseases is at least one defective gene that is responsible for a qualitative or quantitative abnormality associated with the gene product. Some mutations affect a number of genes and may lead to contiguous gene syndromes [42]. The basic mechanisms for such abnormalities include deletions and duplications of a chromosomal region. In theory they may include epigenetic modifications that will turn off normal functioning of a region of the chromosome. It is, however, far more common for a mutational event to alter the sequence of a single gene. Such mutations may be divided into four categories: base substitutions, deletions, duplications and amplifications. In theory, base substitution involves replacement of a base by one of the other three bases in a DNA sequence. For a long time it was believed that substitutions
418 were a random event in a DNA sequence. Information on substitutional events analyzed and evaluated from observed mutations in a number of genetic diseases, however, present a different picture. Every base in a sequence is not subject to the same probability of substitution and, when a substitution takes place, it is not usually replaced by any of the other three bases with equal probability. Such results argue for the effect of DNA sequentiality on the occurrence of DNA mutations, including substitutions. Among the most mutable subsequences in humans is the CpG dinucleotide [43-461, which often leads to TpG dinucleotides [47]. It has been suggested that the biochemical basis for this involves CpG methylation in mammalian DNA sequences [43,47,48]. Such substitutional events may represent 40-80% of all substitutions analyzed for a given gene [49], where the rate of transitions is estimated to be significantly (more than 20-fold) higher than transitions at non-CpG sites [50]. The transition rate at the CpG sites are estimated for the factor IX gene to be 3.6 x lo4 per base pair per generation [51]. The CpG transition may lead to the TGA stop codon and a truncated protein, with major implications in the development of strategies for the search for new mutations associated with a gene leading to a genetic disease. DNA methylation may have yet another implication in the prediction of occurrence and recurrence of mutations. Ongoing studies of the common autosomal dominant disease, neurofibromatosis type 1 (NFl), where up to 50% of the patients in a population may represent new mutations, have suggested that most new mutations take place in the paternal genome [52-541. Given that sperm DNA is heavily methylated, an epigenetic involvement in parent of origin dependent mutation frequency is hypothesized. This parent of origin effect has been realized in a number of diseases, including Angelman/Prader-Willi syndromes [42] and the gene-specific loss of heterozygosity (LOH) apparent during tumorigeneses. Such results have implications in the prediction of the affected haplotype in linkage analysis. Deletions associated with a mutant gene range from deletions of one or few bases to an entire gene. Thus, any attempt to detect such mutational events must ascertain the type of deletion involved. In any survey of human mutations causing genetic diseases, most deletions involve small sequences. For instance, the rate of small deletion mutations in the factor IX gene is estimated at 3.7 x lo-'' per base pair per generation [5 11. Of particular interest are the frameshift mutations, involving single or dinucleotides, that alter the reading frame or cause an abnormality in intron splicing possibly resulting in early protein truncation. For example, mutations of the breast cancer gene, BRCA1, [55] are often small deletions or additions. These mutations are difficult to detect and identify in a general search for mutational events using classic techniques such as Southern blotting. Although rare, a number of deletions may involve loss of part or the whole gene. Most large deletions are detected by traditional Southern blot or pulse field gel electrophoresis (PFGE), although PCR based VNTR analysis (see linkage analysis) is being used more frequently to identify loss of heterozygosity and other large deletions. In some genetic diseases, however, gene deletions may represent the pre-
419 dominant mechanism of mutational events. For example, in Duchenne and Becker’s muscular dystrophies, the frequency of detectable deletions approximates 60%; these deletions are highly heterogeneous with respect to both size and location [lo]. Unlike deletions, duplication events are not well characterized as mutational mechanisms causing genetic diseases. Duplication may, however, be generated by mechanisms similar to small deletions, such as DNA repair deficiencies and replication slippage. Such small duplications are also expected to generate frameshifts and affect the gene product in a similar manner to small deletions. Where and when such mutations take place and what are the contributing factors remains poorly understood and must await a better understanding of the phenomenon of human mutagenesis. Amplification is another type of mutation and involves the expansion of small repeats in a DNA sequence. The expansion evolves across generations arguing for a dynamic rather than stable nature for the DNA sequence in a genome. If this review had been written 5 years ago, it would not have included the aspect of triplet repeat expansion as a common mutational mechanism leading to genetic diseases, even though di-, tri- and tetra-nucleotide repeats have been known to be a common feature in the organization of the human genome. Today, amplification is thought of as a general mechanism for a group of diseases that show anticipation, or increasing severity of expression and earlier age of onset in successive generations. Amplification of a triplet repeat is involved in several genetic diseases with anticipation, including fragile-X syndrome [56], myotonic dystrophy [57], Huntington’s disease [58], Machado-Joseph disease [59], spinocerebellar ataxia type 1 [60] and others. It is expected that a number of other complex diseases that show anticipation in at least some families may have a similar mechanism of mutagenesis. These may include schizophrenia and bipolar affective disorders [61,621. It is interesting to note that the documented triplet repeat amplification in diseases is not restricted to the coding region of the gene, but may take place almost anywhere in the gene sequence, including the 3‘ untranslated region of the resulting mRNA. Also, it appears that not all possible triplets are involved in these expansions. Rather, it is limited to CGG/GCC and CAG/CTG, a fact which may be related to hairpin stability created by these triplets. Although the mechanism responsible for expansion of these triplets is not established, it may involve strand slippage during DNA replication including the direction of DNA replication [63]. The understanding of the triplet repeats involved in the diseases listed above was brought about,by traditional methods involving Southern blots and PCR that are not always informative. Novel methods, therefore, must be developed in order to increase the efficiency and reliability of testing for the involvement of triplet repeat expansions in the mutagenesis of disease.
Methods of identification of disease mutations All DNA-based methods of diagnosis have their grounding in basic genetic principles. Disease causing mutations amenable to DNA evaluation usually fall into three
420 categories: 1) the disease causing gene or its chromosomal location is known, but causal mutations are highly heterogeneous or unknown; 2) the DNA sequence of the causal mutations in the gene is known; and 3) the gene sequence is known, but disease mutations are unknown. The DNA based testing of a disease causing mutation in an individual may be accomplished either by analysis of genetic linkage and familial haplotypes or by direct detection of DNA mutations. The linkage approach deals with prediction of the presence of a mutation in a given individual; methods dealing with direct detection can be used for diagnosis or screening. Most diagnostic methods are well established and highly reliable for the detection of a specific mutation in a given individual, while screening methods are experimental and aimed at searching for the presence of mutation in a specific region of the DNA sequence in a family or population. We will attempt to explain the foundations behind the methods available in these categories. Linkage-based DNA diagnosis Haplotype transmission across generations allows a mutation to be traced through a pedigree. This approach to the detection of DNA mutation is robust and reliable. It does not require any knowledge of the actual gene involved if the disease gene has been localized to a specific region of a chromosome. A linkage-based prediction for the presence of a causal mutation in a given individual is indirect, based on cotransmission of linked markers and is assigned a probability of accuracy. Almost all genetic diseases (cystic fibrosis, Huntington's disease, fragile-X syndrome, etc.) have used this approach of genetic diagnosis at one time or another, until the final cloning of the causal genes and identification of pathogenic mutations. In some other diseases, where most mutations may be family-specific, linkage will continue to form the basis for genetic diagnosis until the family-specific mutational events are identified. In general, a haplotype is developed by analysis of a number of linked polymorphic markers which lead to the precise localization of a gene by positional cloning. Inevitably, the gene is characterized by DNA sequencing and causal mutations are identified. If a single or small number of mutations cause the disease, they may be directly detected in genomic DNA. If, on the other hand, all or most mutations are unique, diagnosis would require identification of family-specific mutations. In a number of cases, however, even complete sequencing of the gene has not permitted identification of the causal' mutation in a family by direct detection methods. For example, in the common autosomal dominant disease, neurofibromatosis type 1 (NFl), no single common mutation or high frequency mutation has been detected and linkage and haplotype analysis is still an important component of NF1 molecular diagnosis (Fig. 1) [64]. A DNA diagnostic protocol in NF1, therefore, requires DNA samples from appropriate family members and starts with identification of polymorphic linked markers in order to identify the familial haplotype that carries the disease mutation. The probability of the pathogenic haplotype being present in an individual (including a fetus) is then established by following familial transmission.
42 1
x4-2 TH1719 Evi
1 1 1 9 2
2 2 1 9 2
1 2 2 2 P9.8 Taq F~.~RM 2 EW207 Bgl 2 ~~207nind1
2 1 1 1 1 1 1
A1 Y ex 4
N F + Cl l-7 EWw6 ~ 9 . ~8 g i
A
I
t
1
r a
2 1 1 1 9
2
l i i 1 2 1 1 1 2 1
1 2 2 1 2 1 1 1 9 9 2 2 + N r 1 1 2 2 2 2 2 2 2 2 1 2 1 1
Fig. I. Linkage analysis in a family with neurofibromatosis type 1 (NFI).The chromosome-17-linked markers used are based on Southern (TH1719, Evi, ex 4, (211-7, F9.8 Bgl, F9.8 Taq) or PCR analysis (HHH202, 3C4-2, A h , EW206, F9.8 Rsa, EW207 Bgl, EW207 Hind). The markers yielded definitive haplotypes which clearly showed the NFI-linked haplotype being passed from the father to the affected sons. It was also possible to identify a double cross over (F9.8 site) which does not affect disease transmission. (Taken from Rodenhiser et al., 1993, [a].)
Selection of molecular markers to be used in the development of haplotypes is based on the degree of polymorphism of linked markers. Initially, restriction fragment length polymorphisms (RFLPs) were used. The presence or absence of a restriction site would result in DNA fragments of different lengths which could be visualized by Southern blot analysis. This involves separation of genomic DNA on an agarose gel, transfer to a solid support, and hybridization with a probe from the region of interest and detection of DNA bands of specific size(s). Most such markers have been converted to polymerase chain reaction (PCR) based methods. This involves amplification of the region around the polymorphic restriction site, then performing the restriction digest and visualizing the DNA directly in an agarose or polyacrylamide gel. Another type of informative marker commonly used in such an analysis is based on a variable number of tandem repeats (VNTR). Such markers are highly polymorphic with a large number of alleles at each site and are interspersed throughout the human genome. VNTRs are commonly detected by PCR and denaturing polyacrylamide gel electrophoresis. The number of copies of the repeat sequence is directly related to the size of the amplification product. Given the large number of alleles (repeats) in populations at such sites (high polymorphic index), a VNTR marker is usually informative and can be assigned to an unique haplotype and an individual.
422
Direct detection of mutations causing a disease
The ultimate goal of almost every molecular study on genetic diseases is to identify the gene and pathogenic mutation(s) causing a disease. This information is then used for reliable detection of the mutation in an individual using the appropriate molecular technology. Molecular methods of detection of DNA mutations have been reviewed previously [3]. A comprehensive review on mutation detection by Cotton [6] included most methods available until 1993. A number of these methods have been modified, while other approaches have been added to the growing list of methods for the detection of mutations in molecular diagnosis. Most methods of detection rely on the altered physical, chemical or biological properties of the mutated DNA molecule. Commonly, the assessment of these changes includes hybridization with a sequence specific probe and measurement of the differential behavior of the mutated molecule (double- or single-stranded) during electrophoresis under different conditions. At times, these differences are brought to the forefront by an appropriate sequence-specific modification, cleavage, replication or amplification. Methods and approaches used to recognize such DNA sequence differences will form the focus of our discussion in this section. The polymerase chain reaction The polymerase chain reaction (PCR) is the most sensitive and efficient molecular technique for genetic investigations [65]. Today PCR is one of the most commonly used methods in genetic diagnosis. This simple and elegant technology permits production of a large quantity of DNA representing a defined region of a known sequence from a minimal amount of test DNA. The resulting PCR product can be evaluated for mutations by a variety of means, including gel electrophoresis, to assess differences in size (as in CF for detection of a three base pair deletion representing AF508) or to detect conformational changes to the variants in the amplified DNA sequences. At times, electrophoresis is performed following sequence specific restriction enzyme digestion of the PCR product, which will differentiate between normal and mutant alleles. More recently, capillary electrophoresis has been used in a number of situations where precise determination of DNA fragment size for the PCR products is desirable [66]. Capillary electrophoresismay allow direct observation of the mutant sequence or linked marker(s) in a variety of common diseases [67,68]. Of course, PCR products may be sequenced directly or following cloning in order to assess all known and unknown sequence differences among different alleles. Because of its sensitivity, reliability and ease of operation, PCR based detection of mutations has almost replaced traditional Southern blotting in a large number of situations. In addition, the complete procedure can be performed in a single day, as compared to several days for Southern blotting. New systems are constantly being developed to reduce the time of reaction, costs, contamination potential, and amount of source DNA required while at the same time maintaining a high level of specific amplification. One novel approach has been the development of the PCRchip that permits PCR diagnosis to be miniaturized using
423 silicon chip technology [69]. This state-of-the-art technology holds promise for the integration of rapid, thorough and cost-effective molecular diagnosis and will provide a common approach for the assessment of mutations involved in a number of common diseases. Allele or sequence specific oligonucleotide hybridization This method, usually referred to as AS0 or SSO, relies on the hybridization of sequence-specific synthetic oligonucleotides representing the normal and mutant alleles to an individuals DNA sample [70]. The original protocol used radiolabelled oligonucleotides specific for the normal and sickle cell alleles of the P-globin gene to probe Southern blots of genomic DNA [71]. Autoradiography showed which of the two probes were specifically hybridized to a genomic DNA band. The genotype of an individual was established as heterozygous if both probes produced signals, and as homozygous for a given allele if only one probe resulted in the signal. The procedure was simplified with the use of dot blots of genomic or PCRamplified DNA containing the region of such a mutation, followed by hybridization with the two probes (Fig. 2). For genes with several mutations in a limited area (such
-clnomasomewith mutation
normal duomomm
PCR-amplifed DNA from test individual
J
s p a om0 solid sup-
(da
blot) forudl probe used
loo
0 0 0 0 0
l o o
0 0 0 0 01
Fig. 2. An illustration of allele-specific oligonucleotide hybridization. The genomic DNA of a test individual is amplified at the region of a suspected mutation. A pair of dot blots are made using the PCR product. DNA known to carry the mutation and normal alleles are also blotted as controls, on each blot. The normal oligonucleotide probe is hybridized to one blot, and the mutant oligonucleotide probe to the other. The blots are exposed to X-ray film and the results can be analysed. Individual C is homozygous for the mutation tested, and individual B is heterozygous. Individuals A, D and E do not carry the mutant allele.
424 as P-globin), this procedure can be reversed [72]. In this reversed dot blot, specific probes for both mutant and normal sequences are applied to a membrane support and hybridized with PCR-amplified DNA from an individual with biotinylated primers. A colorimetric detection of the biotin allows identification of the mutation present in a sample. This assay has now matured as a reliable, highly sensitive technique for the detection of a large number of specific mutations in a region of the DNA. Also, it is particularly well suited for evaluation of highly polymorphic genetic systems such as HLA and P-globin where a number of alleles need to be assessed and screened simultaneously [72-741. The present push to analyze more oligomers has encouraged the development of highly efficient techniques (see silica chip technology) which have the potential to develop into a common approach for assessing the heterogeneous nature of mutations in genes involved in a number of common diseases. Allele-specific amplification Allele specific oligonucleotides have another use as primers for PCR amplification [75-801. Oligonucleotides, which differ from the target sequence by even one base at the extreme 3’ end, are inefficient primers for DNA polymerases. By setting up a pair of amplification reactions with either the mutant or normal sequence as one primer and a second primer which is common to both reactions, amplification will indicate the presence or absence of the two alleles. Amplification can easily be assessed by agarose gel electrophoresis and ethidium bromide staining (Fig. 3). The use of an unrelated primer pair in the reactions will provide an useful internal positive control for the amplification reactions, since the control fragment should amplify in all the reactions. Occasionally it is helpful to introduce extra base mismatches near the 3’ end of the primers to increase the instability of unmatched 3’ ends. Sommer et al. [81] used PCR amplification “to perform population screening, haplotype analysis, patient screening and carrier testing for 69 different single base alleles”. They provided optimized conditions to evaluate detection of mutations (alleles) for five diseases and were able to analyze 400 alleles in 1 day. By exploiting the possibility of multiplex PCR, several different alleles at different loci can be analyzed at one time. Alternatively, different fluorescent tags can be attached to the two allele-specific primers to be used in a competition assay [79,80]. Fluorometry of the sample allows identification of one or both of the tags towards the interpretation of homozygosity or heterozygosity. This modification minimizes the number of reactions required in order to analyze an individual, but requires further manipulations to determine the outcome of the reactions. The ability to perform allele specific amplification and screen for the presence of fragments using gel electrophoresis is simple, inexpensive and very rapid. Increased sensitivity for mutation detection is often accomplished by combining two or more techniques. For example, 14 different HLA-DQB 1 alleles can be identified using a combination of allele specific amplification and SSCP [82]. The different subgroups of HLA-DQB1 are amplified using allele specific primers and these fragments can then be assigned to different alleles on the basis of their SSCP patterns. This modification has the advantage of being able to detect new mutations as well as known mutations in the amplified sequence.
425
Fig. 3. An illustration of allele-specific amplification. The genomic DNA of a test individual is PCR amplified in two separate reactions, one of which has a primer specific for the normal sequence, and the other a mutation-specific primer. After electrophoresis and ethidium bromide staining the results are analysed directly from the gel. For each individual tested, there are two lanes in the gel: one for the normal allele primer (N) and the other for the mutant allele primer (M). Individual A does not carry the mutation, individual B is homozygous for the mutation, and individual C is heterozygous. Each reaction contains an internal control; a primer pair which amplifies an unrelated sequence of known size.
Primer extension Primer extension mutation detection also uses synthetic oligonucleotides as primers. The original form of the method [83] used an oligonucleotide Complementary to the DNA sequence immediately 5' to the nucleotide which differs between the mutant and normal allele. The primer was annealed to genomic DNA in two different tubes, each containing a single radiolabelled (32P)base complementary either to the mutant or normal allele. The reaction in the tubes permitted extension of the 3' end of the oligonucleotide, and this was assessed by electrophoresis and autoradiography for incorporation of the radiolabelled base. Comparison of the two tubes permitted assessment of the genotype of the individual as heterozygous or homozygous (Fig. 4). The sensitivity of the protocol has since been improved by using PCR-amplified genomic DNA [84], biotinylated primers [85,86], and labelled primers on sequencing gels [87] or mini sequencing gels [86].
-c
c4 h d c h r a m a m s
fmmchmmumr wul molllion
double-stranded genomic DNA hom test individual is PCR amplified using ow end-labelled (0)primer lified IJNA u dcruuusd, rnpnmd by
I 1 1 1 1 1 1 1 1 1 1
exua nucleotides a n washed off. and thc p m n w or absence of label is determined appropriately (auioradiography. fluorescence. etc.)
Fig. 4. An illustration of primer extension analysis. The genomic DNA of a test individual is PCR amplified using one primer which is end-labelled with biotin. The PCR product is denatured. The labelled strand is captured for reactions with both normal and mutant nucleotide using a 96 well plate coated with streptavidin. A primer immediately 5' to the mutation being tested is annealed to the captured DNA and a labelled nucleotide specific for the normal or mutant sequence is added to each reaction. After (potential) elongation of the primer, the unincorporated nucleotides are washed off. Signal from the label is apparent if the nucleotide was incorporated.
The method of analysis depends on the nature of the label used, but lends itself well to fluorometric and colorimetric assays. If two different colorimetric tags are used for normal and mutant nucleotides in one reaction, the detection of either one or both is a simple matter, and quickly gives a complete genotype. These highly sensitive quantitative methods permit evaluation of pooled samples in a primer extension assay for the assessment of the population frequency of a mutation [84]. This method of mutation detection is useful in clinical as well as population screening protocols. Artificial introduction of restriction sites Often a mutation is not present in all cells of an individual, but rather in a subpopulation of cells. This is the case for nuclear mutations in a number of tumors. Mitochondria1 mutations have a similar characteristic due to heteroplasmy of the mitochondria1 genome. In order to screen for such mutations it is necessary to detect the mutation, even in the presence of many-fold more normal sequences. The artificial introduction of restriction sites is one approach "$891 which can detect rare mutations in a DNA sample. This technique is based on the creation of oligonucleo-
427
tide primers with one or two mismatched bases resulting in the presence of a restriction site only if there is no point mutation at the immediate 3' end of the primer. After several rounds of PCR amplification, digestion with the appropriate enzyme cleaves normal alleles, leaving only mutant alleles. To enhance the sensitivity of the assay, a second round of amplification can be done. The primers of the second round are designed such that they amplify only those fragments which were not cleaved by the restriction enzyme. This allows detection of the mutant sequence by simple gel electrophoresis and ethidium bromide staining. This technique can also be used for the screening of mutations in populations. Under these circumstances, pools of DNA samples could be used as a template and the pools with positive results for mutations are further screened to identify the individual with the mutation. The ease and speed of this technology is seductive, but it is often difficult to design primers that will initiate PCR while mismatches are present which incorporate restriction sites. Careful designing of PCR primers and the establishment of optimal amplification conditions can make this a very useful technique. Screening methods for unknown mutations
Most screening methods are aimed at detection of unknown mutations apparent in a given stretch of DNA. The usual strategy is to cover as long a stretch of DNA as possible or to screen as many individuals as possible, since it is not realistic to generate complete DNA sequences for each individual during the hunt for new mutations. Thus, these methods rely on altered physical and chemical properties of the DNA containing the mutation without being able to identify the site or nature of the mutation. In general, such protocols are followed by DNA sequencing for characterization of the mutation. This discussion will concentrate on the principles and strategies involved in screening for unknown mutations along with possible advantages and disadvantages associated with each of the methods. Enzymatic and chemical mismatch cleavage One of the earliest methods used to screen for small unknown mutations was the cleavage of mismatched heteroduplexes [90]. This property was exploited for the detection of P-thalassemia mutations using RNase A to cleave mismatches in RNA:DNA heteroduplexes [911. Labelled RNA probes were transcribed from wildtype DNA and annealed to the test DNA. Deletions or substitutions of even a single base in the test DNA, will result in a bubble of single-stranded RNA that is vulnerable to RNase A cleavage. Deletions within the gene are easily detected while identification of substitutions is dependent on the particular mismatch involved. The actual cleavage is detected by autoradiography after denaturing polyacrylamide gel electrophoresis. A variation of this technique is to end label a wild-type cDNA probe and hybridize with the mRNA from the test individual [92]. Cleavage is done using a combination of RNase A and S1 nuclease. This modification is usually considered more sensitive than the original protocol [92].
428 An alternative to these endonucleases with varying specificities is chemical cleavage of mismatches [93]. Osmium tetroxide reacts with mismatched thymines and hydroxylamine with mismatched cytosines, leading to cleavage upon the addition of piperidine. These chemicals can thus detect all possible point mutations. The initial work showed that 34 mismatches were all cleaved using these chemicals [93]. This is a vast improvement on the enzymatic cleavage but the technique involves several steps and the handling of mutagenic and explosive compounds. The most recent advance in mismatch cleavage technology (Fig. 5 ) is the use of bacteriophage resolvases which cleave DNA:DNA mismatches [94,95]. These enzymes are believed to be involved in the modification of single-stranded DNA during packaging of the phage. T4 endonuclease VII and T7 endonuclease I, have both been used to cleave mismatches between PCR-amplified genomic DNA. Mashal et al. [94] used this approach to detect mutations in the APC, p53 and CFTR genes in fragments as large as 940 base pairs, while others have used this technique to evaluate fragments of up to 1500 base pairs with a high degree of success. Six of
Fig.5. An illustration of the mismatch cleavage method of mutation detection. The genomic DNA of a test individual is amplified. This DNA is mixed with control DNA. The mixed DNA is denatured and allowed to anneal forming homo- and heteroduplexes. Sites of mutation in heteroduplexes are cleaved by specific enzymes or chemicals. Electrophoresis of the DNA allows identification of mutations. Individuals A, B and E have no cleavage products, thus they do not carry any mutations in the region tested. Individuals D and F likely have the same mutations and individual C has a different mutation in the region analysed.
429 seven small deletions tested were identified, and 30 of 32 point mutations were detected. This gives a mutation detection rate of over 92%. This method also allows for the approximate location of the mutation to be established, since the size of the cleavage products reflects the site of cleavage in the DNA. This is a fast and easy technique which will no doubt become one of the more commonly used mutation screening tests of the future. As reliable sources of resolvases become available, standard procedures will be established and this technology will improve. This procedure has the potential to recognize nearly all small mutations present in a gene, in a reliable manner. Also, it may be possible to make use of labelled primers for PCR on pooled DNA samples in order to screen large numbers of individuals from different populations for novel mutations. A positive result in a DNA pool would merit further analysis of the individual DNAs belonging to that pool. A negative result would mean that there are no mutations in that sample pool. DNA:DNA mismatch cleavage using resolvases is amenable to automation with the potential for large scale and efficient population screening. Single-strand conformation polymorphism analysis A widely used technique to detect novel mutations is single strand conformation polymorphism (SSCP) analysis [96]. Single-stranded DNA migrates through a gel at differing rates, dependent on its conformation which is a consequence of the nucleotide sequence. Even minor changes in base composition (such as a base substitution) may affect the conformation of the single-stranded DNA and, therefore, its electrophoretic mobility. The presence of mutations may be detected by utilizing PCR to amplify the region of interest, thermally interrupting the DNA’s double helical structure and allowing the single-stranded DNA to adopt a secondary structure, then separating the conformers on a polyacrylamide gel. For an individual homozygous at the amplified region, two DNA bands will be detected, representing the sense and antisense single strands of a double helix. If an individual is heterozygous, four such bands will be expected (Fig. 6). The SSCP results for unknowns are usually compared with those of known controls towards identification of mutations in a given DNA region. Although this technique is very powerful in its ability to detect minor changes in DNA, and is fast and relatively simple to perform, it has several drawbacks. These include unknown and variable fidelity of mutation detection, the small size of DNA fragment which can be effectively analyzed, the variable conditions required for differential conformations by different single strands, and difficulties in resolving separated strands. Some of the problems of DNA based SSCP can be remedied by using simplified methods [97,98]. Another modification of this technique includes RNA-SSCP [99,100]. Since RNA has a greater degree of secondary structure, more genetic variation can be detected based on conformational changes. In a comparative study, fragments of up to 497 bases yielded detectable changes in almost all of the rSSCP samples, but only three of eight mutations were detected using conventional SSCP when the DNA fragment was 300 bases or more [loo]. Mutations identified by this method are not necessarily pathological. The only way to determine whether a
--
430
chnnmomo with mutation
Mmylrhnmaom
F'CR-amplified double-strandedDNA
A
B
C
D
E
F
Fig. 6. An illustration of single-strand conformation polymorphism analysis. The genomic DNA of a test individual is amplified. The DNA is denatured and electrophoresis of the single strands allows identification of different conformers. Individuals A and F have only the sense and antisense strands of the normal sequence. Individuals B, D and E are heterozygous, and individual C is homozygous for a mutation. It should be noted that although all mutation strands are shown as having a slower electrophoretic mobility, they can have faster mobilities than the normal strands.
conformational polymorphism represents a disease mutation is to sequence the region and determine the nature of the mutation, including its effect on the gene product.
Heteroduplex analysis This technique involves denaturation and annealing of PCR-amplified DNA [ 101,1021. Complementary DNA strands with small differences in base sequence will form a heteroduplex with a region of nonhomology. This forms a bubble (base mismatch) or bulge (deletions or additions) that may have an effect on the electrophoretic mobility of the DNA. Of course, homoduplexes will form from the pairing of perfectly matched complementary strands. When subjected to electrophoresis, the heteroduplexes migrate at a slower rate than the homoduplexes. Heteroduplex analysis will detect any mismatched bases or deletions/additions in a heterozygous sample (Fig. 7). The advantage of heteroduplex analysis over SSCP is that longer stretches of DNA can be analyzed with less optimization of electrophoretic conditions. In addition, SSCP will not detect polymorphisms which do not have an effect on the conformation of the single-stranded DNA [ 1021. Consequently, the efficiency of mutation detection by heteroduplex analysis is high (approximately 80%),and in combination with SSCP analysis, it is close to 100% [6].
43 1
Fig. 7. An illustration of heteroduplex analysis. The genomic DNA of a test individual is amplified. This DNA is mixed with control DNA. The mixed DNA is denatured and allowed to anneal forming homoand heteroduplexes. Electrophoresis'of the duplexes allows identification of mutations. Individuals A, C, E, and F have only the expected fragment of normal electrophoretic mobility. Individuals B and D have mutations resulting in heteroduplexes with impeded mobility.
Denaturing gradient gel electrophoresis The denaturation of DNA strands is highly dependent on its base pair composition [103]. This property of DNA is exploited in order to identify unknown mutations in a gene using denaturing gel gradient electrophoresis (DGGE). Variations in the base composition of a known sequence results in different melting (denaturation) points. Electrophoresis of double-stranded DNA through a gel with increasing concentrations of denaturants (urea and formamide), permits discrimination between two molecules differing by just one base pair. DNA will eventually reach a denaturant concentration which is sufficient to cause portions of the double helix to become single stranded. This impedes the electrophoretic mobility of the DNA in the gel, resulting in a characteristic band position which is essentially unaltered by further electrophoresis. Fisher and Lerman [lo31 have shown that only mutations in the early melting domains are detectable. Double-stranded DNA does not melt evenly. Usually, different domains melt independent of one another. If the highest melting domain is
432 the site of mutation, the lower melting domains will melt at the same denaturant concentration in normal and mutant molecules, and the two sequences will be indistinguishable in the gel. This problem can be eliminated by using PCR to introduce 30 to 40 cytosine and guanine nucleotides at the beginning of the sequence to be analyzed [I041 (Fig. 8). This GC-clamp acts as an extremely high melting domain for the sequence, thus allowing all other melting domains a chance to exert their effect on the electrophoretic mobility of the PCR fragment. A further advantage of this technique is that amplification of the gene allows detection by ethidium bromide staining, rather than radioisotope labelling and autoradiography. DGGE has been used effectively for the detection of novel cystic fibrosis (CF) mutations [105]. Using this technique to evaluate 109 non-AF508 CF chromosomes, 25 were assigned one of eight mutations. Eighty-four of the chromosomes studied, however, gave no information regarding their mutation. The study was done using only five exons, so the unknown errors are presumably within regions of the gene which were not tested. GC-clamped denaturing gradient gel electrophoresis is a safe and quick method which can detect nearly all small mutations in DNA sequences up to 500 base pairs in length [104], and possibly longer, although with less sensitivity. The major
I
I.
Fig.8. An illustration of denaturing gradient gel electrophoresis. The genomic DNA of a test individual is amplified using one primer preceded by 35-40 guanines and cytosines. This produces PCR fragments with a GC-clamp on one end. Electrophoresis of the DNA on a denaturing gradient gel allows identification of mutations. Individuals A and F have only the expected fragment, melting at a specific point. Individuals B, D and E are heterozygous for the normal allele, and mutations which affect the melting point of the DNA fragment. Individual C is heterozygous for two different mutations. It is possible for mutations to cause increases in melting point resulting in bands lower in the gel than the normal allele.
433 disadvantage of DGGE is the special equipment required for pouring and running the gradient gels. A constant temperature must be maintained for accurate results and this can be done only with a specialized apparatus. Preliminary study must be carried out to determine denaturing domains, in order to design optimal PCR primers, and to predict appropriate gradients for each sequence [ 1061. Electrophoretic separation can be further maximized using constant denaturant gel electrophoresis (CDGE) [ 1071. First, a perpendicular DGGE is run with GC-clamps to determine the optimum denaturant concentration for a PCR product. Electrophoresis at this denaturant concentration can be performed for extended periods of time, increasing the visual difference caused by the differential mobility. This allows better detection of mutations that have a small effect on the melting of the double-stranded DNA. When a change in denaturation condition is observed using CDGE or DGGE, that fragment can be excised from the gel and eluted for further analysis to determine the actual mutation. Protein truncation test Identification of a mutation using the different methods summarized above may not be equally informative, particularly in a screening protocol for pathogenic mutations. An alteration detected in a DNA molecule may represent a polymorphism with no biological effect. Differentiation between silent and pathogenic mutations remains a major challenge in molecular diagnosis of diseases. Recent observations on the nature of mutations, which argue for a predominance of CpG to TpG base substitutions resulting in premature stop codons [50,51], and small deletions and additions leading to frameshifts as common disease mutations, have made the protein truncation test (PTT) [lo81 an attractive method for detection of mutations. This test detects abnormal (shortened) peptides which are generally pathogenic. For PTT the mRNA of a patient is isolated and amplified using RT-PCR. A second PCR reaction is performed using nested primers, one of which is prefaced by a transcription primer and the ATG translation start codon. The secondary PCR product is transcribed and translated, then the protein products are separated by discontinuous SDS-PAGE. The presence of a truncated protein is apparent due to its increased electrophoretic mobility (Fig. 9). The size of the truncated protein gives an indication as to the general location of the nonsense mutation, facilitating sequence analysis for confirmation of the mutation at the DNA level. For large reading frames, such as the dystrophin gene in Duchenne muscular dystrophy [ 1091 or the APC gene in familial adenomatous polyposis (FAP) [ 1lo], several reactions are performed to produce overlapping amplification products of the mRNA. Although this technique can be performed with genomic DNA [ 1101, the use of mRNA for RT-PCR and PTT provides a highly sensitive procedure for the detection of novel mutations. The combined techniques allow for the rapid screening of an entire reading frame for large deletions and additions as well as mutations which result in protein truncation. These include point mutations, frameshifts, and splice site alterations. This is a powerful technique to screen for unknown pathological mutations. For example, evaluation of the APC gene in individuals with familial
434
Fig. 9. An illustration of the protein truncation test. The mRNA of a test individual is isolated. The cDNA of the gene of interest is produced by reverse transcription, then PCR amplified. One primer is preceded by a transcription promoter, allowing in vitro transcription and translation. Electrophoresis of the protein product allows identification of truncation mutations. Individuals A and E have only the expected normal size peptide. Individuals B, C and D are heterozygous for the normal allele, and for mutations causing truncated proteins. The size of the truncated peptide is an indication of location of mutation in the gene, facilitating further analysis.
adenomatous polyposis resulted in the detection of mutations in 82% of the patients [ 1101 and analysis of the dystrophin gene revealed the presence of mutation in 77% of Duchenne muscular dystrophy patients [ 1091. More recently, application of this technology for detection of heterogeneous mutations of the BRCAl gene has begun to yield valuable information concerning the nature of mutational events in familial breast cancer [ 1111. Recently, our laboratory has applied this technology for mutation screening in both BRCAl and NFl patients with encouraging results. The protein truncation test is a valuable tool for the detection of pathological mutations for genes where a large proportion of mutations result in truncated proteins, as is the case for many tumor suppressor genes.
435 DNA sequencing The most comprehensive method for mutation detection is DNA sequencing [ 1 12,1131.Originally, high-percentage denaturing polyacrylamide gels were used to separate the varying length fragments at a resolution of one base difference in size, allowing determination of the nucleotide sequence. The Maxam and Gilbert chemical method [ 1 121 uses dimethyl sulfate, formic acid, hydrazine, and hydrazine with sodium chloride to modify guanine, guanine and adenine, thymine and cytosine, and cytosine, respectively. Piperidine is added to cleave the modified bases, and the DNA which has been selectively end-labelled, is separated on a sequencing gel and exposed for autoradiography. Special plasmids have been created which aid in the selective end-labelling of a DNA strand [ 1141 making this a relatively simple, if prolonged, process. This technique is less popular than the technically easier dideoxy method of sequencing [ 1 131. Dideoxy sequencing [ 1131 uses four reactions, each of which carries a labelled primer, a set of the four deoxynucleotides, and one of the dideoxynucleotides. The dideoxynucleotides do not allow addition of other nucleotides during replication, thus causing premature termination of chain elongation whenever they are incorporated into the DNA sequence. Each reaction will produce fragments of various lengths, each ending with the dideoxynucleotide that was used in the reaction. The order of nucleotides can be easily read following electrophoresis of the four reactions and autoradiography.A number of modifications to these techniques have been developed. These include the utilization of 7-deaza-2'-deoxyguanosine-5'-triphosphate (c'dGTP) for dGTP which allows better resolution on the sequencing gels [ 1151, and direct sequencing of PCR products [ 1161.It is no longer necessary to clone the DNA into a sequencing vector. After PCR amplification, a labelled primer is added with dNTPS, a ddNTP and a thermostable DNA polymerase, and several cycles of replication are performed using a thermal cycler to regulate the denaturing and elongation steps. This has been termed cycle sequencing. A major development in direct sequencing is the use of fluorescent ddNTPs rather than labelled primers [ 1 171,thereby permitting automation of sequencing [ 1 17- 1 191 by using different fluorescent tags for each of the ddNTPs. Reactions are now set up and performed using a robotic work station. After the replication reactions are completed, the four tubes from each of the adenine, cytosine, guanine, and thymine ddNTP reactions are combined and separated by size using capillary electrophoresis. The fluorescent tag is read to identify the ddNTP labels of each fragment as it passes a detector. The ease and efficiency of automated direct dideoxy sequencing permits the processing of a large number of samples. It is relatively expensive compared to other methods of mutation detection. It is usually necessary to perform many sets of reactions to cover an entire gene. For these reasons, sequencing alone is not used to screen for unknown mutations in a gene. The use of previously discussed techniques for screening often identify regions of a gene which carry a genetic alteration and this region can be sequenced to determine the actual mutation present. With the human genome project's goal of sequencing the entire genome by the year 2005, there is
436 much work currently being done to find less expensive methods of DNA sequencing (see “Miniaturization and automation” below). These methods could eventually allow sequencing to be used as a quick and efficient screening method to detect new mutations in a gene. Repeat expansion detection Expansion of triplet repeats is now recognized as the major genetic cause of diseases showing anticipation. Here the mutational mechanism is expected to increase the size of the repeat as an allele is passed on to the next generation. This mutational mechanism, when responsible for diseases, appears to be associated with CAG/CTG or GCC/GGC motifs, only. The reason for the restrictive nature of these trinucleotides in any expansion phenomenon is thought to be due to the variable stability of the hairpins formed by different triplet repeats and their replication properties. The search for such dynamic mutations as the cause of genetic diseases of unknown etiology represents a novel area of research in molecular genetics. Given the possibility of only a few trinucleotide sequences being involved in such mutational mechanisms, it is possible to search for such expanding mutations in families with diseases exhibiting anticipation of unknown etiology. The repeat expansion detection assay [120] is a modification of ligation amplification reactions [ 1211. Using oligonucleotides which are specific for a defined number of trinucleotide repeats, several rounds of annealing, ligation and denaturation are performed. In regions of a large number of trinucleotide repeats, the oligonucleotides will anneal next to each other and be ligated together in a ligase chain reaction. Additional oligonucleotides may be added at successive reactions, resulting in a long chain which correlates roughly to repeat size. Since this reaction is not exponential, a large amount of genomic DNA is needed for a template, and detection of ligation is performed using electrophoresis, transfer and hybridization with a labelled complementary probe. Using oligonucleotides consisting of different numbers of different repeats ((CGG),,, (CTG),,, (CGT),,, etc.) several repeats can be assayed at the same time since the ultimate chain length is always a multiple of the oligonucleotide used. Identification of the expanded trinucleotide repeat, correlating with disease in a family would form the basis for the identification of the gene involved. Recently, this approach has been applied towards the search for a causal mutation in schizophrenia and bipolar disorders [61]. Preliminary results are encouraging but need further experimentation and verification on appropriate families. Miniaturization and automation The creation of specific oligonucleotide sequences on a glass support [ 1221 initiated a new era in oligonucleotide hybridization technology. Using walls to separate the individual cells of an array, oligonucleotides were systematically built to produce a discrete set of octamers. This method requires a rather large template: all octamers (4* = 65,536) would result in a 25.6 cm2 array. An alternative is a silica microchip which can be easily encoded with synthetic, single-stranded DNA sequences [123,124]. Silica chips are optically transparent, allowing the use of light for both
437
reactions and result analysis. By covering (priming) the chip with photoprotected hydroxyl groups, desired regions can be activated by illumination. Single-stranded DNA is built by masking areas of the chip to allow only the light-exposed regions to add on the nucleotide which is provided. The amount of light scatter, diffraction and reflection must be kept at a low level in order to ensure that activation is not occumng in undesired regions of the chip. By using Gray code masks, this can be achieved [125]. The nucleotide added to each strand is thus carefully controlled. Nucleotides have photolabile hydroxyl groups, thus allowing the controlled addition of second, third, fourth, etc. nucleotides to the oligonucleotide matrix. This technology has already been used to produce arrays of 256 octamers which are only 0.64 cm2 [124]. The key advantage of this technology is the potential for complete automation, from oligonucleotide chip manufacturing to hybridization, detection and computer analysis of results. It has been suggested that an alternative to standard sequencing protocols might be sequencing by hybridization [ 126,1271. Hybridization to a complete set of oligonucleotides of a given length, using a labelled DNA fragment, allows the detection of all oligonucleotides which bind this fragment. In a set of octamers, one octamer would hybridize with nucleotides 1-8 of the nucleic acid sequence, another with nucleotides 2-9, the next with nucleotides 3-10, and so on. Computer analysis can piece together the overlapping sequences of the oligomers to give the complete sequence of the test fragment. It is now possible to generate all the possible decamers (41°= 1,048,576) on a silica chip 2.56 cm2 [124]. This would allow the efficient sequencing of nucleic acids up to a kilobase in size [127]. The details of such experiments, including the computer analysis necessary, are progressing rapidly and this technology will no doubt become a reality for sequencing in the near future [128]. Silicon chip technology can also be applied to genetic testing. Preliminary studies have illustrated and begun to correct many of the problems inherent in the technology [ 129,1301. These include construction of computer algorithims to analyze hybridization patterns, and optimizations of oligonucleotide density, target-to-probe ratios-and hybridization times. Although there is still much work to be done before this is a reliable and efficient test procedure, it has the potential to revolutionize genetic testing. Eventually, DNA chips with all the known (or possible) mutatiops for a gene or a disease will be screened with amplified DNA from the test individual, the sites of hybridization can be determined by a laser reader, and the complementary DNA sequence can be identified. If this technology proves reliable, the implications for genetic diagnosis are profound. Mass produced custom chips for different genetic diseases may allow diagnosis within a few hours. It is even possible that the analysis of a handful of DNA chips will produce a summary of the morbid anatomy of an individual’s genome. At the current pace of research and development, it is likely that within the next few years such chips will be routinely utilized for detection of a set of mutations associated with a disease. Assessment of the complete morbid anatomy of a genome, however, may still be many years away.
438
An overview and future directions Hereditary diseases have always attracted our deep personal interest. Familial, biochemical and molecular studies from the beginning of the century have contributed to the fundamental understanding of the relationship between the genes we inherit and phenotypes we manifest during life. The progression of these studies necessitated the development of powerful molecular technologies and tests, which have become the armaments in diagnosis of genetic diseases and the practice of medicine. Molecular diagnosis is usually undertaken on DNA isolated from blood, the most accessible cell type available in adequate quantities. The sensitivity of molecular technologies, including PCR, has provided an incentive to consider DNA tests on very small amounts of biological material. Today, the use of very small cell samples has become a common procedure in most DNA-based genetic testing. It has led to miniaturization of most protocols to the point where multiple PCR reactions could be performed on silica chips. This type of miniaturization is also compatible with automation. It permits large scale population screening for the presence of mutations or screening for a large number of mutations in individual samples. Automated, miniaturized procedures of the future will probably rely on robotics, in order to minimize human error and maintain consistency of results at relatively low cost. In specialized cases, it may be more proficient if only a few centers performed specific genetic tests for large geographical areas. The sensitivity of PCR and related technologies in genetic testing now permit the use of small quantities of cells, either fresh or stored. The stored material may include dried blood spots, fixed-tissue sections or other sources. It may also include a single cell, whether haploid or diploid. The cells for testing may come from abundant sources like sperm and blood or from rare sources such as dividing embryos in vitro to be transferred for implantation. They may also represent rare fetal cells circulating in maternal blood, which would avoid the trepidation associated with invasive procedures such as chorionic villi sampling and amniocentesis. The protocols for use of maternal blood in fetal testing have advanced during the last few years with mixed results. It is possible that in the future, fetal cells in the maternal blood may become the routine source of fetal DNA for prenatal diagnosis. As indicated earlier, all genetic testing need not rely on genomic DNA. In some cases it may be preferable to use mRNA as the starting biological material for identification of uncharacterized mutations. Usually, the mRNA is isolated from fresh cell samples and used to generate cDNA or polyIjeptides towards evaluation of the presence of specific mutations. Genetic testing is here to stay as an element of the practice of medicine. It does not rely on the manifestation of disease symptoms, but identifies a pathogenic mutation in mitochondria1 or nuclear genomes. The DNA sample may be fetal in origin, it may represent a living person with or without any manifestation of the disease, or it may be from a deceased individual. Unlike most other diagnostic methods which are based on the manifestation of disease symptoms, the implications of genetic testing are not limited to the proband. Rather, the results can impact on a
439 still undiagnosed individual who may develop the disease at some future time, and every member of the extended family. The implications of such results are far reaching and necessitate appropriate and sensitive handling of diagnostic information. Our ability to undertake genetic testing has created a number of social and ethical dilemmas for the practice of medicine. In particular, these dilemmas primarily arise from concerns pertaining to genetic privacy. Recognition of these sensitive social and ethical concerns has prompted a number of organizations, including the Human Genome Project, to undertake the challenge of developing guidelines for the use and application of genetic testing. Although recognized and appreciated, a major discussion of the subject is beyond the scope of this limited review. It must be emphasized, however, that the complications are enormous, the implications are far reaching and the damage done by any breach of confidentiality and insensitive treatment of such information is usually irreversible. Genetic testing allows not just an accurate diagnosis, but also the prognosis for an individual, including lifestyle adaptations to ameliorate or even prevent symptoms. Consequently, the results of genetic testing have implications in such diverse areas as health insurance, employee hiring practices, and personal decisions like career choice, marriage and reproductive options. The development of guidelines for the application of genetic testing will have to be an ongoing process. These must adapt as the technologies evolve, as the genetic determinants of superficial and aesthetic traits are established, and as society comes to appreciate the incredible impact genetic testing has on our lives. Also, there may not be an absolute rule; rather, a set of principles deeply rooted in respect for the individual. As stated by a number of groups, including the Royal Commission on New Reproductive Technologies (Canada) [131], the most logical approach must be “to proceed with care”.
Acknowledgements This manuscript originated from experiences gained in diagnosis of genetic diseases through the Molecular Diagnostic Laboratory, Molecular Medical Genetics Program and Division of Medical Genetics at the Children’s Hospital of Western Ontario. This review was financially supported by grants from the National Science and Engineering Research Council of Canada and the Canadian Genome Analysis and Technology Programs to SMS. Valuable comments on the original draft of this manuscript by Dr D.B. McMillan and members of our laboratories are gratefully acknowledged.
References Collins FS. Positional cloning moves from perditional to traditional.Nature Genet 1995;9:347-350. Cooper DN, Schmidtke J. Diagnosis of genetic disease using recombinant DNA. Hum Genet 1986;73:1-1 1. 3. Caskey CT. Disease diagnosis by recombinant DNA methods. Science 1987;236:1223-1229.
1. 2.
440 4. 5. 6. 7.
8.
9.
10.
11.
12. 13.
14.
15.
16.
17.
18.
19.
20. 21. 22.
Cotton RGH. Detection of single base changes in nucleic acid. In: Verma R (ed) Advances in Genome Biology: Unfolding the Genome, 1st edn. London: JAI Press, 1991;253-300. Cotton RGH. Detection of mutations in DNA. Curr Opin Biotech 1992;3:24-30. Cotton RGH. Current methods of mutation detection. Mutat Res 1993;285:125-144. Weatherall DJ, Clegg JB, Higgs DR, Wood WG. The hemoglobinopathies. In: Scriver CR, Beaudet AL, Sly WS, Valle D (eds) The Metabolic and Molecular Bases of Inherited Disease III, 7th edn. Toronto: McGraw-Hill Inc., 1995;3417-3484. Welsh MJ, Tsui L-C, Boat TF, Beaudet AL. Cystic fibrosis. In: Scriver CR, Beaudet AL, Sly WS, Valle D (eds) The Metabolic and Molecular Bases of Inherited Disease III, 7th edn. Toronto: McGraw-Hill Inc., 1995;379+3876. Goldstein JL, Hobbs HH, Brown MS. Familial hypercholesterolemia. In: Scriver CR, Beaudet AL, Sly WS, Valle D (eds) The Metabolic and Molecular Bases of Inherited Disease 11, 7th edn. Toronto: McGraw-Hill Inc., 1995;1981-2030. Worton RG, Brooke MH. The X-linked muscular dystrophies. In: Scriver CR, Beaudet AL, Sly WS, Valle D (eds) The Metabolic and Molecular Bases of Inherited Disease 111, 7th edn. Toronto: McGraw -Hill Inc., 19954195-4226. Neufeld EF, Muenzer J. The mucopolysaccharidoses. In: Scriver CR, Beaudet AL, Sly WS, Valle D (eds) The Metabolic and Molecular Bases of Inherited Disease II, 7th edn. Toronto: McGraw-Hill Inc., 1995;2465-2494. Van Heyningen V. One gene - four syndromes. Nature 1994;367:319-320. Mulligan LM, Kwok JBJ, Healey CS, Elsdon MJ, Eng C, Gardner E, Love DR, Mole SE, Moore JK, Papi L, Ponder MA, Telenius H, Tunnacliffe A, Ponder BAJ. Germ-line mutations of the RET proto-oncogene in multiple endocrine neoplasia type 2A. Nature 1993;363:458-560. Mulligan LM, Eng C, Healey CS, Clayton D, Kwok JBJ, Gardner E, Ponder MA, Frilling A, Jackson CE, Lehnert H, Neumann PH, Thibodeau SN, Ponder BAJ. Specific mutations of the RET proto-oncogene are related to disease phenotype in MEN 2A and FMTC. Nature Genet 1994;6:7074. Carlson KM, Dou S , Chi D, Scavarda N, Toshima K, Jackson CE, Wells SAJ, Goodfellow PJ, Donis-Keller H. Single missense mutation in the tyrosine catalytic domain of the RET protooncogene is associated with multiple endocrine neoplasia type 2B. Proc Natl Acad Sci USA 1993;91:157+1583. Hofstra RMW, Landsvater RM, Ceccherini I, Stulp RP, Stelwagen T, Luo Y, Pasini B, Hoppener JWM, van Amstel HKP, Romeo G, Lips CJM, Buys CHCM. A mutation in the RET proto-oncogene associated with multiple endocrine neoplasia type 2B and sporadic medullary thyroid carcinoma. Nature 1994;367:375-376. Romeo G, Ronchetto P, Luo Y, Barone V, Seri M, Ceccherini I, Pasini B, Bocciardi R, Lerone M, K*linen H,Martucciello G. Point mutations affecting the tryosine kinase domain of the RET proto-oncogene in Hirschsprung’s disease. Nature 1994;367:377-378. Edery P, Lyonnet S, Mulligan LM, Pelet A, Dow E, Abel L, Holder S , Nihoul-FCkete C, Ponder BAJ, Munnich A. Mutations in the RET proto-oncogene in Hirschsprung’s disease. Nature 1994;367:378-380. Levy-Lahad E, Wasco W, Poorkaj P, Romano DM, Oshima J, Pettingell WH, Yu C, Jondro PD, Schmidt SD, Wang K, Crowley AC, Fu Y-H, Guenette SY, Galas D, Nemens E, Wisjman EM, Bird TD, SchellenbergGD, Tanzi RE. Candidate gene for the chromosome 1 familial Alzheimer’s disease locus. Science 1995;269:973-977. Yamaguchi TP, Rossant J. Fibroblast growth factors in mammalian development. Curr Opin Genet Devel 1995;5:485-491. Lajeunie E, Ma H W ,Bonaveneture J, Munnich A, Le Merrer M, Renier D. FGFR2 mutations in Pfieffer syndrome. Nature Genet 1995;9:108. Jabs EW, Li X, Scott AF, Meyers G, Chen W, Eccles M, Mao J, Charnas LR, Jackson CE, Jaye M. Jackson-Weiss and Crouzon syndromes are allelic with mutations in fibroblast growth factor receptor 2. Nature Genet 1994;8:275-279.
441 23.
24. 25.
26.
27.
28.
29.
30.
31. 32. 33. 34. 35. 36.
37. 38.
39. 40. 41. 42.
43.
Muenke M, Schell U, Hehr A, Robin NH, Losken HW, Schinzel A, Pulleyn LJ, Rutland P, Reardon W, Malcolm S, Winter RM. A common mutation in the fibroblast growth factor receptor 1 gene in Pfieffer syndrome. Nature Genet 1994;8:26!3-274. Reardon W, Winter RM, Rutland P, Pulleyn LJ, Jones BM, Malcolm S. Mutations in the fibroblast growth factor receptor 2 gene cause Crouzon syndrome. Nature Genet 1994;8:98-103. Rousseau F, Bonaventure J, Legeal-Mallet L, Pelet A, Rozet J-M, Maroteaux P, Le Merrer M, Munnich A. Mutations in the gene encoding fibroblast growth factor receptor-3 in achondroplasia. Nature 1994;371:252-254. Rutland P, Pulleyn LJ, Reardon W, Baraitser M, Hayward R, Jones B, Malcolm S, Winter RM, Oldridge M, Slaney SF, Poole MD, Wilkie AOM. Identical mutations in the FGFR2 gene cause both Pfieffer and Crouzon syndrome phenotypes. Nature Genet 1995;9:173-176. Shiang A, Thompson LM, Zhu Y-Z, Church DM, Fielder TJ, Bocian M, Winokur ST, Wasmuth JJ. Mutations in the transmembrane domain of FGFR3 cause the most common genetic form of dwarfism, achondroplasia. Cell 1994;78:335-342. Tavormina PL, Shiang R, Thompson LM, Zhu Y-Z, Wilkin DJ, Lacman RS, Wilcox WR, Rimoin DL, Cohn DH, Wasmuth JJ. Thanatophoric dysplasia (types I and 11) caused by distinct mutations in fibroblast growth factor receptor 3. Nature Genet 1995;9:321-328. Schell U, Hehr A, Feldman GJ, Robin NH,Zackai EH, De Die-Smulders C, Viskochil DH, Stewart JM, Wolff G, Ohashi H. Mutations in FGFRl and FGFR2 cause familial and sporadic Pfeiffer syndrome. Hum Molec Genet 1995;4:323-328. Wilkie AOM, Slaney SF, Oldridge M, Poole MD, Ashworth GJ, Hockley AD, Hayward RD, David DJ, Pulleyn U,Rutlend P, Malcolm S, Winter RM, Reardon W. Apert syndrome results from localized mutations of FGFR2 and is allelic with Crouzon syndrome. Nature Genet 1995;9:165- 172. Wallace DC. Mitochondria1 genetics: a paradigm for aging and degenerative diseases? Science 1992;256:628-632. Shoffner JM, Wallace DC. Mitochondria1 genetics: principles and practice. Am J Hum Genet 199231: 1179-1 186. Giles RE, Blanc H, Cann HM, Wallace DC.Maternal inheritance of mitochondrial DNA. Proc Natl Acad Sci USA 198077:6715-6719. Wallace DC. Mitotic segregation of mitochondrial DNAs in human cell hybrids and expression of chloramphenicol resistance. Somat Cell Molec Genet 1986;12:41-49. ShoffnerJM, Wallace DC. Oxidative phosphorylation diseases: disorders of two genomes. Adv Hum Genet 1990 19:267-330. Wallace DC, Ye J, Neckelmann N, Singh G, Webster KA, Greenberg BD. Sequence analysis of cDNAs for the human and bovine ATP synthase b subunit: mitochondrial DNA genes sustain seventeen times more mutations. Curr Genet 1987;12:81-90. Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, Friedman JM. Positional cloning of the mouse obese gene and its human homologue. Nature 1994;372:425-432. Halaas JL, Gajiwala KS, Maffei M, Cohen SL, Chait BT, Rabinowitz D, Lallone RL,Burley SK, Friedman JM. Weight-reducing effects of the plasma protein encoded by the obese gene. Science 1995;269:543-546. Haffner R, Oren M. Biochemical properties and biological effects of p53. Curr Opin Genet Dev 1995584-90. Chang F, Syrjiinen S, Syrjhen K. Implications of the p53 tumour-suppressor gene in clinical oncology. J Clin Oncol 1995;13:1009-1022. Knudson AG. Mutations and cancer: statistical study of retinoblastoma. Proc Natl Acad Sci USA 1971;85: 159G1594. Ledbetter DH, Ballabio A. Molecular cytogenetics of contiguous gene syndromes: mechanisms and consequences of gene dosage imbalance. In: Scriver CR, Beaudet AL, Sly WS, Valle D (eds) The Metabolic and Molecular Bases of Inherited Disease I, 7th edn. Toronto: McGraw-Hill Inc., 1995;811-839. Bird AP. CpG islands as gene markers in the vertebrate nucleus. Trends Genet 1987;3:342-347.
442 44. Cooper DN, Youssoufian H. The CpG dinucleotide and human genetic disease. Hum Genet 45.
46. 47. 48. 49.
50.
51.
52.
53. 54. 55.
56.
57.
58.
59.
60.
61. 62. 63.
1988;78:15 1-155. Koeberl DD, Bottema CDK, Ketterling RP, Bridge PJ, Lillicrap DP, Sommer SS. Mutations causing hemophilia B: direct estimate of the underlying rates of spontaneous germ-line transitions, transversions, and deletions in a human gene. Am J Hum Genet 1990;47:202-217. Rideout WM, Coetzee GA, Olumi AF, Jones PA. 5-methylcytosine as an endogenous mutagen in the human LDL receptor and p53 genes. Science 1990;249:1288-1290. Rodenhiser DI, Coulter-Mackie MB, Singh SM. Evidence of DNA methylation in the neurofibromatosis type 1 (NFI) gene region of 17ql1.2. Hum Molec Genet 1993;2:439-444. Laird PW, Jaenisch R. DNA methylation and cancer. Hum Molec Genet 1994;3:1487-1495. Upadhyaya M, Shaw DJ, Harper PS. Molecular basis of neurofibromatosis type 1 (NF1): mutation analysis and polymorphisms in the NFI gene. J Hum Mutat 1994;4:83-101. Bottema CDK, Ketterling RP, Vielhaber E, Yoon H-S, Gostout B, Jacobson DP, Shapiro A, Sommer SS. The pattern of spontaneous germ-line mutation: relative rates of mutation at or near CpG dinucleotides in the factor IX gene. Hum Genet 1993;91:496-503. Sommer S S , Ketterling RP. How precisely can data from transgenic mouse mutation-detection systems be extrapolated to humans?: lessons from the human factor IX gene. Mutat Res 1994;307:517-531. Jadayel D, Fain P, Upadhyaya M, Ponder MA, Huson SM, Carey J, Fryer A, Mathew CGP, Barker DF, Ponder BAJ. Paternal origin of new mutations in Von Recklinghausen neurofibromatosis. Nature 1990;343:558-559. Stephens K, Kayes L, Riccardi VM, Rising M, Sybert VP, Pagon RA. Preferential mutation of the neurofibromatosis type 1 gene in paternally derived chromosomes. Hum Genet 1992883279-282. Lkaro C, Ravella A, Gaona A, Volpini V, Estivill X. Neurofibromatosis type 1 due to germ-line mosaicism in a clinically normal father. N Engl J Med 1994;331:1403-1407. Shattuck-Eidens D, McClure M, S h a r d J, Labrie F, Narod S , Couch F, Hoskins K, Weber B, Castilla L, Erdos M, Brody L, Ostermeyer E, Szabo C, King M-C, Jhanwar S , Offit K, Norton L, Gilewski T, Lubin M, Osbome M, Black D, Boyd M, Steel M, Ingles S , Haile R, Lindblom A, Olsson H, Borg A, Bishop DT, Solomon E, Radice P, Spatti G, Gayther S , Ponder B, Warren W, Stratton M, Liu Q, Fujimura F, Lewis C, Skolnick MH, Goldgar DE. A collaborative survey of 80 mutations in the BRCAl breast and ovarian cancer susceptibility gene. Implications for presymptomatic testing and screening. JAMA 1995;273:535-541. Nussbaum RL, Ledbetter DH. The fragile-X syndrome. In: Scriver CR, Beaudet AL, Sly WS, Valle D (eds) The Metabolic and Molecular Bases of Inherited Disease I, 7th edn. Toronto: McGraw-Hill Inc., 1995;795-810. Harper PS. Myotonic dystrophy and other autosomal muscular dystrophies. In: Scriver CR, Beaudet AL, Sly WS, Valle D (eds) The Metabolic and Molecular Bases of Inherited Disease 111, 7th edn. Toronto: McGraw-Hill Inc., 1995;4227-425 1. Hayden MR, Kremer B. Huntington disease. In: Scriver CR, Beaudet AL, Sly WS, Valle D (eds) The Metabolic and Molecular Bases of Inherited Disease 111, 7th edn. Toronto: McGraw-Hill Inc., 1995;448345 10. Kawaguchi Y, Okamoto T, Taniwaki M, Aizawa M, Inoue M, Katayama S , Kawakami H, Nakamura S , Nishimura M, Akiguchi I, Kimura J, Narumiya S , Kakizuka A. CAG expansions in a novel gene for Machado-Joseph disease at chromosome 14q32.1. Nature Genet 1994;8:221-228. Zoghbi HY, Ballabio A. Spinocerebellarataxia type 1. In: Scriver CR, Beaudet AL, Sly WS, Valle D (eds) The Metabolic and Molecular Bases of Inherited Disease III,7th edn. Toronto: McGraw-Hill Inc., 1995;4559-4567. O’Donovan MC, Guy C, Craddock N, Murphy KC, Cardno AG, Jones LA, Owen MJ, McGuffin P. Expanded CAG repeats in schizophrenia and bipolar disorder. Nature Genet 1995;10:38+38 1. Bassett AS, Honer WG. Evidence for anticipation in schizophrenia. Am J Hum Genet 1994;54:864870. Kang S , Jaworski A, Ohshimas K, Wells RD. Expansion and deletion of CTG repeats from human
443
64.
65. 66. 67.
68. 69. 70.
71.
72. 73.
74.
75.
76. 77.
78. 79. 80.
81. 82. 83. 84.
disease genes are determined by the direction of replication in E. coli.Nature Genet 1995;10:213218. Rodenhiser DI, Ainsworth PJ, Coulter-Mackie MB, Singh SM, Jung JH. A genetic study of neurofibromatosis type I (NFl) in south-westem Ontario. 11. A PCR based approach to molecular and prenatal diagnosis using linkage. J Med Genet 1993;30363-368. Mullis KB, Ferre F, Gibbs RA (eds): The Polymerase Chain Reaction. Boston: Birkhauser, 1994. Baba Y, Tomisaki R, Sumita C, Tsuhako M, Miki T, Ogihara T. High-resolution separation of PCR product and gene diagnosis by capillary gel electrophoresis. Biomed Chrom 1994;8:291-293. Corder EH, Saunders AM, Strittmatter WJ, Schmechel DE, Gaskell PC, Small GW, Roses AD, Haines JL, Pericak-Vance MA. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer's disease in late onset families. Science 1993;261:921-923. Hixon JE,Vernier DT. Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with Hhu I. J Lipid Res 199031:545-548. Wilding P, Shoffner MA, Kricka LJ. PCR in a silicon microstructure. Clin Chem 1994;40:18151818. Wallace RB, Johnson MJ, Hirose T, Miyake T, Kawashima EH, Itakura K. The use of synthetic oligonucleotides as hybridization probes. 11. Hybridization of oligonucleotides of mixed sequence to rabbit P-globin DNA. Nucl Acids Res 1981;9:879-895. Conner BJ, Reyes AA, Morin C, Itakura K, Teplitz RL, Wallace RB. Detection of sickle cell 0"globin allele by hybridization with synthetic oligonucleotides. Proc Natl Acad Sci USA 1983;80278-282. Saiki RK, Walsh PS, Levenson CH, Erlich HA. Genetic analysis of amplified DNA with immobilized sequence-specificoligonucleotideprobes. Proc Natl Acad Sci USA 1989;86:6230-6234. Bugawan TL, Begovich AB, Erlich HA. Rapid HLA-DPP typing using enzymatically amplified DNA and nonradioactive sequence specific oligonucleotide probes: application to tissue typing for transplantation. Immunogenetics 1990;32:231-241. Tiercy J-M, Sanchez-Mazas A, Excoffier L, Shi-Isaac X, Jeannet M, Mach B, Langaney A. HLADR polymorphism in a Sengalese Mandenka population: DNA oligotyping and population genetics of DRPl specificities. Am J Hum Genet 1992;51:592-608. Newton CR, Graham A, Heptinstall LE, Powell SJ, Summers C, Kalsheker N, Smith JC, Markham AF. Analysis of any point mutation in DNA. The amplification refractory mutation system (ARMS). Nucl Acids Res 1989;17:2503-25 16. Okayama H, Curie1 DT, Brantly ML, Holmes MD, Crystal RG. Rapid, nonradioactive detection of mutations in the human genome by allele-specificamplification. J Lab Clin Med 1989;114:105- 1 13. Sommer SS, Cassady JD, Sobell JL, Bottema CDK. A novel method for detecting point mutations or polymorphisms and its application to population screening for carriers of phenylketonuria. Mayo Clin Proc 1989;64:1361-1372. Wu DY, Ugozzoli L, Pal BK, Wallace RB. Allele-specific enzymatic amplification of P-globin genomic DNA for diagnosis of sickle cell anemia. Proc Natl Acad Sci USA 1989;2757-2760. Gibbs RA, Nguyen P-N Caskey CT. Detection of single DNA base differences by competitive oligonucleotide priming. Nucl Acids Res 1989;17:2437-2448. Chehab FF, Kan YW. Detection of specific DNA sequences by fluorescenceamplification: A colour complementation assay. Proc Natl Acad Sci USA 1989;86:9178-9182. Sommer SS, Groszbach AR, Bottema CDK. PCR amplification of specific alleles (PASA) is a general method for rapidly detecting known single-base changes. Biotechniques 1992;1232-87. Lo Y-MD, Pate1 P, Mehal WZ, Fleming KA, Bell JI, Wainscoat JS. Analysis of complex genetic systems by ARMS-SSCP: application to HLA genotyping. Nucl Acids Res 1992;201005-1009. Sokolov BP. Primer extension technique for the detection of single nucleotide genomic DNA. Nucl Acids Res 1989;18:3671. Kuppuswamy MN, Hoffman JW,Kasper CK, Spitzer SG, Groce SL, Bajaj SP. Single nucleotide primer extension to detect genetic disease: experimental application to hemophilia B (factor LX)and cystic fibrosis genes. Proc Natl Acad Sci USA 1991;88:1143-1147.
444 85. SyvSnen A-C, Aalto-Setala K, Harju L, Kontula K, Soderlund H. A primer-guided nucleotide incorporation assay in the genotyping of apolipoprotein E. Genomics 1990;8:684-692. 86. Syvinen A-C, Ikonen E, Manninen T, Bengtstrom M, Siiderlund H, Aula P, Peltonen L. Convenient and quantitative determination of the frequency of a mutant allele using solid-phaseminisequencing: application to aspartylglucosaminuria in Finland. Genomics 1992;12:59(t595. 87. Lee J-S, Anvret M. Identification of the most common mutation within the porphobilinogen deaminase gene in Swedish patients with acute intermittent porphyria. Proc Natl Acad Sci USA 1991;88:10912-10915. 88. Cohen JB,Levinson AD. A point mutation in the last intron responsible for increased expression and transforming activity of the c-Ha-rus oncogene. Nature 1988;334:119-121. 89. Chen J, Viola MV. A method to detect rus point mutation in a small subpopulation of cells. Anal Biochem 1991;195:51-56. 90. Freeman GJ, Huang AS. Mapping temperature-sensitive mutants of vesicular stomatitis virus by RNA heteroduplex formation. J Gen Virol 1981;57:103-117. 91. Myers RM, Larin Z, Maniatis T. Detection of single base substitutions by ribonuclease cleavage at mismatches in RNADNA duplexes. Science 1985;2301242-1246. 92. Atweh GF, Baserga SJ, Brickner HE. Detecting small mutations in expressed genes by a combination of S1 nuclease and RNase A. Nucl Acids Res 1988;16:8709. 93. Cotton RGH, Rodrigues NR,Campbell RD. Reactivity of cytosine and thymine in single-base-pair mismatches with hydroxylamine and osmium tetroxide and its application to the study of mutations. Proc Natl Acad Sci USA 1988;85:4397-4401. 94. Mashal RD, Koontz J, Sklar J. Detection of mutations by cleavage of DNA heteroduplexes with bacteriophage resolvases. Nature Genet 1995;9:177-183. 95. Youil R, Kemper BW, Cotton RGH. Screening for mutations by enzymatic mismatch cleavage with T4 endonuclease VII. Proc Natl Acad Sci USA 1995;92:87-91. 96. Orita M, Suzuki Y, Sekiya T, Hayashi K. Rapid and sensitive detection of point mutations and DNA polymorphisms using the polymerase chain reaction. Genomics 1989;5:874-879. 97. Ainsworth PJ, Surh LC, Coulter-Mackie MB. Diagnostic single strand conformational polymorphism, (SSCP): a simplified non-radioisotopic method as applied to Tay-Sachs B1 variant. Nucl Acids Res 1991;19:405-406. 98. Ravnik-Glavac M, Glavac D, Dean M. Sensitivity of single-strand conformation polymorphism and heteroduplex method for. mutation detection in the cystic fibrosis gene. Hum Molec Genet 1994;3:801-807. 99. Danenberg PV, Horikoshi T, Volkenandt M, Dananberg K, Lenz H-J, Shea LCC, Dicker AP, Simoneau A, Jones PA, Bertino JR. Detection of point mutations in human DNA by analysis of RNA conformation polymorphism(s). Nucl Acid Res 1992;20573-579. 100. Sarkar G, Yoon H-S, Sommer SS. Screening for mutations by RNA single-strand conformation polymorphism (rSSCP): comparison with DNA-SSCP. Nucl Acids Res 1992;2087 1-878. 101. Keen J, Lester D, Ingleheam C, Curtis A, Bhattacharya S. Rapid detection of single base mismatches as heteroduplexes on hydrolink gels. Trends Genet 1991;7:5. 102. White MB, Carvalho M, Derse D, O’Brien SJ, Dean M. Detecting single base substitutions as heteroduplex polymorphisms. Genomics 1992;12:301-306. 103. Fischer SG, Lerman LS. DNA fragments differing by single base-pair substitutions are separated in denaturing gradient gels: Correspondence with melting theory. Proc Natl Acad Sci USA 1983;80:1579-1583. 104. Sheffield VC, Cox DR, Lerman LS, Myers RM. Attachment of a 40-base-pair G + C-rich sequence (GC-clamp) to genomic DNA fragments by the polymerase chain reaction results in improved detection of single-base changes. Proc Natl Acad Sci USA 1989;86:232-236. 105. Devoto M, Ronchetto P, Fanen P, Orriols JJT, Romeo G, Goosens M, Ferrari M, Magnani C, Seia M, Cremonesi L. Screening for non-deltaF508 mutations in five exons of the cystic fibrosis transmembrane conductance regulator (CFTR) gene in Italy. Am J Hum Genet 1991;48:1127-1132. 106. Sheffield VC, Fishman GA, Beck JS, Kimura AE, Stone EM. Identification of novel rhodopsin
445 mutations associated with retinitis pigmentosa by GC-clamped denaturing gradient gel electrophoresis. Am J Hum Genet 1991;49:699-706. 107. Hovig E, Smith-S~rensonB, Brogger A, B~rresenA-L. Constant denaturant gel electrophoresis, a modification of denaturing gradient gel electrophoresis, in mutation detection. Mutat Res
1991;262:63-71. 108. Roest PAM, Roberts RG, Sugino S, van Ommen G-JB, den Dunnen JT. Protein truncation test (PIT) for rapid detection of translation-terminating mutations. Hum Molec Genet 1993;2:171% 1721. 109. Gardner RJ, Bobrow M, Roberts RG. The identification of point mutations in Duchenne muscular dystrophy patients by using reverse-transcription PCR and the protein truncation test. Am J Hum Genet 1995;57:311-320. 110. Powell SM, Petersen GM, Krush AJ, Booker S, Jen J, Giardiello FM, Hamilton SR,Vogelstein B, Kinzler KW. Molecular diagnosis of familial adenomatous polyposis. N Engl J Med 1993;329:
1982- 1987. 111. Hogervorst FBL, Cornelis RS, Bout M, van Vliet M, Oosterwijk JC, Olmer R, Bakker B, Klijn JGM, Vasen HFA, Meijers-Heijboer H, Menko FH, Cornelisse CJ, den Dunnen JT, Devilee P, van Ommen G-JB. Rapid detection of BRCAl mutations by the protein truncation test. Nature Genet 1995;10:208-212. 112. Maxam AM, Gilbert W. A new method for sequencing DNA. Proc Natl Acad Sci USA 1977;74:560-564. 113. Sanger F, Nicklen S, Coulson AR. DNA sequencing with chain-terminating inhibitors. Proc Natl Acad Sci USA 1977;74:5463-5467. 114. Eckert R. New vectors for rapid sequencing of DNA fragments by chemical degradation. Gene 1987;51~245-252. 115. Mizusawa S, Nishimura S, Seela F. Improvement of the dideoxy chain termination method of DNA sequencing by use of deoxy-7-deazaguanosine triphosphate in place of dGTP. Nucl Acids Res 1986;14:1319-1324. 116. Reynolds TR,Uliana SR, Floeter-Winter LM, Buck GA. Optimization of coupled PCR amplification and cycle sequencing of cloned and genomic DNA. Biotechniques 1993;15:462-467. 117. Prober JM, Trainor GL, Dam RJ, Hobbs FW,Robertson CW, Zagursky RJ, Cacuzza AJ, Jensen MA, Baumeister K. A system for rapid DNA sequencing with fluorescent chain-terminating deoxynucleotides. Science 1987;238:336-341. 118. Smith LM, Sanders JZ, Kaiser RJ, Hughes P, Dodd C, Connell CR, Heiner C, Kent SBH, Hood LE. Fluorescence detection in automated DNA sequence analysis. Nature 1986321:674-679. 119. Wilson RK, Chen C, Avdalovic N, Burns J, Hood L. Development of an automated procedure for fluorescent DNA sequencing. Genomics 1990;6:626-634. 120. Schalling M, Hudson TJ, Buetow KH, Housman DE. Direct detection of novel expanded trinucleotide repeats in the human genome. Nature Genet 1993;4:135-139. 121. Barany F. Genetic disease detection and DNA amplification using cloned thermostable ligase. Proc Natl Acad Sci USA 1991;88:18+193. 122. Southern EM, Maskos U, Elder JK. Analyzing and comparing nucleic acid sequences by hybridization to arrays of oligonucleotides: evaluation using experimental models. Genomics 1992;13:1008- 1017. 123. Fodor SP, Rava RP, Huang XC, Pease AC, Holmes CP, Adams CL. Multiplexed biochemical assays with biological chips. Nature 1993;364:555-556. 124. Pease AC, Solas D, Sullivan EJ, Cronin MT, Holmes CP, Fodor SPA. Light-generated oligonucleotide arrays for rapid DNA sequence analysis. Proc Natl Acad Sci USA 1994;91:5022-5026. 125. Feldman W, Pevzner P. Gray code masks for sequencingby hybridization. Genomics 1994;23:233-
235. 126. Bains W, Smith GC. A novel method for nucleic acid sequence determination. J Theor Biol 1988;135:303-307. 127. Khrapko KR, Lysov YP, Khorlyn AA, Shick W, Florentiev VL, Mirzabekov AD. An oligonucleo-
446 tide hybridization approach to DNA sequencing. FEBS Lett 1989;256:118-122. 128. Chetverin AB, Kramer FR. Sequencing of pools of nucleic acids on oligonucleotide arrays. BioSystems 1993;30:215-231. 129. Guo Z, Guilfoyle RA, Thiel AJ, Wang R, Smith LM. Direct fluorescence analysis of genetic polymorphism by hybridization with oligonucleotide arrays on glass supports. Nucl Acids Res 199q22:5456-5465. 130. Matson RS, Rampal J, Pentoney SL, Anderson PD, Coassin P. Biopolymer synthesis on polypropylene supports: oligonucleotide arrays. Anal Biochem 1995;224:11CF116. 131. Royal Commission on New Reproductive Technologies. Proceed with Care. Ottawa: Canada Communications Group, 1993.
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R. El-Gewely. editor.
447
Molecular genetics as a diagnostic tool in farm animals Gerald Stranzinger and Dirk F. Went Institute of Animal Science, Swiss Federal Institute of Technology (ETH), Zurich, Switzerland
Abstract. In this review, the importance of molecular genetics for diagnostic applications in animal production and breeding is underlined. Recently, several new techniques and methods based on gene technology have been developed, such as the polymerase chain reaction, fluorescencein situ hybridization, and the use of microsatellitepolymorphism. The examples include detection of favourable alleles of genes coding for milk proteins, recognition of negative recessive alleles in hereditary syndromes, the use of microsatellite variants for breeding purposes and parentage control, and application of specific DNAprobes for identification of Y-chromosome-bearing spermatozoa and the sex of embryos. It is to be understood that this list is not complete and more applications will undoubtedly show up in the future. For this review, the authors have mainly selected areas where they themselves or their co-workers have gained experience.
Key words: allelic variation, biopsy, calcium release channel (CRC) protein, DNA primer, embryo transfer (ET), fluorescent in situ hybridization (FISH), gene mapping, halothane test, haplotyping, Kcasein, malignant hyperthermia (MH), microsatellite, pig oedema disease, pale, soft and exudative (PSE) pork, parentage control, polymerase chain reaction (PCR), polymorphic DNA marker, porcine stress syndrome (PSS), qualitative trait locus, restriction enzyme, ryanodine receptor (RYR), sex diagnosis, sex preselection, sperm separation, tandem repeat nucleotide sequence, transgenesis,X-/Y-chromosome-bearing spermatozoa, Y-chromosome-specific DNA-probe.
Milk and qualitative trait loci analysis by molecular techniques Milk and milk products play an important role in human and animal nutrition. Differences in the content of milk have been evolved naturally during evolution and by breeding strategies following domestication. The concept of modem genetics in animal breeding includes gene technology in attempts to change milk proteins and other traits to the desired properties [l]. It is important to say that the environmental situation and efficient use of natural resources asks for a consequent reduction of the number of animals used for the production of goods like milk and meat. At the same time, the productivity of animals should be increased in order to solve the nutrition problems worldwide. This is one of the reasons to intensify the study of genetics of milk proteins as inherited traits, as well as of other components which are regulated by genetic interactions.
Address for correspondence: Gerald Stranzinger, Institute of Animal Science, Swiss Federal Institute of Technology (ETH), 8092 Zurich, Switzerland.
The K-CaSein diagnosis In this chapter the case of the bovine K-casein will be used as an example for the application of molecular genetics, since this trait is well documented, investigated and easy to understand for people not active in the field. Already in 1984, Stewart et al. [2j described the nucleotide sequences of bovine as,- and K-casein cDNAs. Beforehand, Scherbon et al. [3] had documented the variants of milk proteins and their possible relationship to milk properties. This research was established without the help of the PCR technique (polymerase chain reaction) first described by Mullis et al. [4],which since then revolutionised diagnostic procedures in molecular biology. But only a combination of these scientific developments could lead to an understanding of the different K-casein properties being due to allelic variation. In Europe, especially in Italy, Switzerland, Germany, Austria, France and The Netherlands, cheese making is an old tradition and of economic importance. In this respect it should be mentioned that the casein fractions contribute to more than 80% of the protein fraction in the milk and their allelic properties are essential for cheese production. The allele frequencies in the various breeds are quite different and more or less correspond to the cheese making priorities in the different countries [5j. Due to an exchange of genetic material by artificial insemination and embryo transfer (mainly by introducing American breeds into European breeds) a shifting of the allele frequencies can be observed. The casein fraction of the milk consists of a group of calcium-binding phosphoproteins. These can be divided in calcium-sensitive variants like the as-(asl, as2, asl,s2), p- and y-variants, and calcium-insensitive K-variants. The latter play an important role for producing the casein micelles and, in this way, are responsible for the irreversible rennetability of the milk. To date, four different alleles of the K-casein have been described for different breeds [6], the main alleles being A and B, the less frequent being C and E. Several investigations have shown that milk of the K-casein type BB is of advantage for cheese making as compared to the AA type [7j. They also showed that the BB type produces a different size of micelles, the protein content is slightly increased, the heat-resisting and freezing stability of the milk proteins is better and the clotting time is shorter. Most important is the enhancement of the curd firmness which causes a 5-10% higher cheese yield [8j. The single locus theory is supported by the inheritance pattern and studies have shown that AB types respond intermediately. C and E variants are very rare and therefore have not been investigated regarding theh properties. Breakthroughs in diagnostics have been the description of the amino acid components of the K-casein variants A and B by Grosclaude and his co-workers as early as 1972 (see [9]), the description of the nucleotide sequence of the cDNA of the A variant by Stewart et al. [2j and that of the cDNA of the B variant by Gorodetsky and Kaledin [lo]. Rando et al. [l 13 found different restriction fragment length polymorphisms (RFLPs), the number of which has been extended by Eggen [12j. The introdexon structure of the bovine K-casein gene was described by Alexander et al. [13] and is shown in Fig. 1. These investigations and the sequencing of the different exons have been mandatory to
449 Exon 1 65 bp #
Exon 2
Exon9 33bp
62 bp
Exon4
Exon5
516bp
173bp
3’
5 lntron 1 2.5 kb
lntron 2 5.8 kb
lntron 3 2.0 kb
lntron 4 1.8 MI
Fig. 1. Introdexon structure of the bovine K-CaSein gene (according to Alexander et al. [13]). Differences in alleles A and B are mainly based on point mutations in exon 4. (bp, base pairs; kb, kilobase = 1,OOO base pairs).
evaluate important aspects of the allelic properties. It was found that exon 4 with 5 16 bp carries most of the significant mutations causing the allelic variations. As soon as a nucleotide in a codon (unit of three nucleotides coding for an amino acid) of exon 4 is mutated, a change in the amino acid composition of the protein may take place. The change of the amino acid sequence may cause a different folding, and, consequently, also a different functioning of the protein. These point mutations within the gene are used for the detection of the different alleles in the K-casein gene. The diagnosis procedure Since the mutations are present in the genome and can be detected in any cell material carrying a nucleus, it is possible to investigate animals of any age and sex, without the need for examining milk or casein fractions. This advantage of the use of molecular techniques is unique and has been further extended by the PCR technique, as this technique allows the amplification of the DNA of a single cell to such an amount, that testing can be carried out. Cell material such as sperm, blood samples, skin biopsies or other body or embryonic tissues can be used in small amounts of only a few cells. The samples are washed and treated with proteinase K to remove other material interfering with the PCR reaction. The extracted DNA solution can be used for further testing by the PCR method and for different applications as described by White et al. [14].Since the flanking regions of the introns from exon 4 in the K-casein gene and also the base sequence of exon 4 are known [13], it is possible to determine complementary sequences for the construction of primers necessary for amplification (approximately 40 cycles) of the DNA to examine. The amplified DNA piece of 583 bp can be cut with restriction enzymes specific for detection of a mutation. In the case of K-casein variant A, there is a recognition site in exon 4 - near the nucleotide triplet coding for amino acid 144 - which is cut by Hinfl. HinfI is a restriction enzyme extracted from the bacterium Haemophilus infuenzae Rf. In the B variant this recognition site is missing and therefore the enzyme does not cut that region, producing one fragment less than with variant A. Since HinfI also cuts in two non-allele-specific regions, in the case of an A allele it will produce three cutting sites giving four fragments of 326, 129, 69 and 59 bp length. In the B allele, only the two unspecific cuts will
450 occur, resulting in three fragments of 455,69 and 59 bp. In contrast to allele A, allele B can be cut by another enzyme HindIII (Huernophilus influenzae Rd), again near the nucleotide triplet coding for amino acid 144. This will lead to two fragments sized 452 and 131 bp. These DNA fragments can be run in a gel-electrophoretic field where they are separated according to their size (Fig. 2). The migration rate is reciprocally proportional to log,, of their molecular weight. Small particles migrate faster than large ones and later can be found separated from each other in the gel. The bands consisting of DNA fragments can be stained and visualised under UV light. The different alleles are documented by photographs. Both homozygotes and heterozygotes can be distinguished. PCR testing of casein types, especially of K-casein variants, indeed has become a routine method; yet, problems may still arise from contamination or enzyme failure. Experience and control setups are needed to keep a high standard. One should be aware of the relative value of these findings in genetic terms since an investigation of one genetic locus does not show any possible genetic interactions. It would be advisable to carry out haplotype investigations on the casein chromosome region, examining the complete casein gene family in addition to other gene loci (possibly located on other chromosomes) affecting milk parameters and milk components.
Alleles
I
AA
AB
BB
bp
Fig. 2. Schematic representationof electrophoretic migration of K-CaSein PCR-amplified fragments after digestion with restriction enzymes. Numbers represent bp, black bars show undigested fragments, bars with stripes indicate fragments specific for either A or B alleles after digestion with HindIII or HinfI. In the case of the AB genotype, all fragments of A and B alleles can be seen, resulting in a superimposed banding pattern of the homozygotes AA and BB.
45 1
Testing the porcine stress syndrome by molecular genetic techniques Genetic variants of proteins are commonly observed in farm animals. Since they contribute to the normal make-up and variation of genotypes or phenotypes, they are not regarded as mutants. In genetic terms, however, every allelic variation is based on a mutation causing a specific modification in a given gene. Such gene mutations may have a positive or negative influence on a given trait and are inherited. If this trait is economically important and if these mutations cause a reduction of fitness or death of an animal following stress situations, the breeder has an interest to select and eliminate those animals at an early age and use effective selection criteria against those variants. One such trait is known as the porcine stress syndrome (PSS) in pigs. It is associated with a positive reaction to the halothane challenge test and caused by an Arg615-Cys615mutation (C-T mutation) in the skeletal muscle calcium release channel protein (see below). This mutation was unintentionally spread over many pig populations due to an advantage of pigs harbouring the mutation, concerning body conformation (more meat and less fat on the carcass). This fact alone would have offered an advantage for producers and consumers if no disadvantage would have been linked to it. Stress susceptibility, which may lead to death of an animal and to pale, soft and exudative (PSE) pork in large segments of the carcass, is the result of PSS. This mutation is therefore a genetic defect and inherited as an autosomal recessive trait with incomplete penetrance [15]. Resistant animals are denoted as NN or Nn animals, whereas susceptible animals are of the M genotype. Exposure to inhalational anaesthetics with halothane [ 161 (as applied formerly to identify and select the susceptible animals) only did detect the homozygote nn animals but not the heterozygote Nn carriers. For this reason other methods have been developed to substitute for halothane testing. Gene mapping, sequencing and comparative genetics The assignment of the halothane gene to the chromosome 6 in pigs, specifically to segment pll-q21 [ 17-19], has also brought information on other loci linked to this chromosome region, such as the H and S blood group loci, erythrocyte enzyme loci, glucosephosphate isomerase (GPI, formerly PHI), 6-phosphogluconatedehydrogenase (PGD) and serum protein locus a-1-B-glycoprotein (AlBG). With this information, haplotyping can be applied [2&22] to distinguish between carriers and noncarriers of this mutation. In man there exists a comparable situation for malignant hyperthennia (MH) causing symptoms similar to PSS and also anaesthesia sensitivity. In addition, some enzyme loci linked to human chromosome region 17q, to which the MH susceptibility locus is assigned, are analogous for human and pigs in the comparable chromosome regions [23]. This information emphasises the suitability of the pig as an experimental model in human research [24]. In association with the mapping of the halothane gene, haplotyping of a larger segment of chromosome 6 in pigs has the advantage that new favourable recombinations may be found and used
452 for specific matings and breeding programmes by applying the linkage and expression information of the additional loci. A few years ago, a Canadian research group [25,26] identified the calcium release channel (CRC) protein as identical to the ryanodine receptor (RYR); the alkaloid ryanodine has the ability to bind to the CRC protein. When the gene for the CRC protein was sequenced, a point mutation was found in position 1843 where the base cytosine (C) was replaced by the base thymine (T). This causes a corresponding alteration in the protein of the amino acid arginine (Arg) to cysteine (Cys). In several investigations it could be shown that in pigs a mutation C to T is found in the recessive halothane locus allele defined as n [27--291. This information can now be used for testing pigs on the presence or absence of the C-T mutation using the PCR technique. The test has been patented by a Canadian group and licenses must be obtained to make commercial use of it. To demonstrate concordance of the results achieved by halothane testing and linkage studies on the one hand and the PCR method on the other hand, Table 1 gives a summary of tests on different pig breeds carried out in our laboratory.
Diagnostic procedure with microsatellite polymorphism for breeding purposes In addition to the findings regarding the point mutation in the CRC gene, Bolt et al. [28] found a microsatellite in the same gene. This microsatellite consists of sequence units according to the formula (CA(GA)m)n. As polymorphic microsatellites are optimal tools for animal identification and selection purposes, the microsatellite was characterized and examined in some detail. The microsatellite is located in an intron at the 3' end of the CRC gene. Using the PCR system and choosing the primers PB25 and PB33, five alleles with different microsatellite sizes of 148, 142, 136, 128 and 96 base pairs were found (Fig. 3). The microsatellite can be used for selection of appropriate recombinants carrying the normal CRC gene together with closely linked gene@) regulating the oedema disease in piglets [30]. Associations between the genes for the H blood group system, the GPI red-cell enzyme system and a locus coding for receptors for an Escherichiu coli (bacterial) strain responsible for the oedema disease have been found. Under experimental conditions and challenge of piglets with oral inoculation of E. coli strain 124/76 (serotype 0139: K12 (B): Hl:F(107)), Bertschinger et al. [31] were able to develop a resource family' for studying the linkage situation. According to this investigation, there is a linkage disequilibrium between specific alleles of the receptor locus for susceptibility to bacterial colonisation, certain H and GPI genotypes and alleles of the halothane locus. It is unknown, however, if this linkage disequilibrium also holds for other pig breeds in other countries. The frequency of halothane-positive animals varies considerably between breeds, populations and countries, and comparable linkage studies for these populations are still lacking. For the present we therefore presume that the noninvasive DNA-based test for the C to T mutation at nucleotide 1843 is very accurate for the diagnosis of the PSS status in the pigs, but
I . A c o m p a r i s o n of h a l o t h a n e - g e n o t y p i n g a n d PCR a l l e l e - t y p i n g f o r t h e C - T b a s e m u t a t i o n i n t h e C R C in d i f f e r e n t p i g b r c c d s a n d c r o s s i n g s . T h e g e n o t y p e w a s d e t e r m i n e d by t h e h a l o t h a n e t e s t . p a r t l y in c o m b i n a t i o n w i t h l i n k a g e s t u d i e s . N N a n d N n : h a l o t h a n e - r e s i s t a n t a n i m a l s . nn: s u s c e p t i b l e a n i m a l s . All a n i m a l s c h a r a c t e r i z e d as n n s h o w t h e m u t a t i o n C ( y t o s i n e ) t o T ( h y m i n e ) in b o t h a l l e l e s o f t h e C R C - g e n e ( m u t a t i o n T T ) . P i = P i t t r a i n . V L S = S w i s s Landrace. N L = Norwegian L a n d r a c e . B L = B e l g i a n L a n d r a c e : m = mak. f = female. n.d. = mdelcrmined. Table gene
TT
BF 380
Pi
m
nn
CH 1153
VLS
f
N"
CT
CH 3576
PiXVLS
m
N"
CT
CH 3571
PlXVLS
m
M
TT
CH 3578
PiXVLS
m
N"
CT
CH 3579
FiXVLS
m
Nn
CT
CH 3580
Pi iVLS
m
M
TT
CH 3581
PiXVLS
m
N"
CT
CH 3582
PiXVLS
m
M
TT
CH 3583
Pi x VLS
m
Nn
CT
CH 3586
Pi
VLS
f
N"
CT
CH 3587
Pi I VLS
f
M
IT
CH 3588
Pi IVLS
f
M
IT
CH 3589
PiXVLS
f
M
TT
CH 3590
PI I: VLS
f
M
TT
88
NL
m
N"
CT
89
NL
f
M
TT
90
NL
n.d.
NN
Nn
CT
91
NL
n.d.
NN M Nn
CT
92
NL
n.d.
M
TT
93
NL
n.d.
M
IT
94
NL
n.d.
NN or Nn
CT
95
NL
n.d.
M
TT
I
M
Nn
CT
%
NL
n.d.
NN
91
NL
n.d.
nn
TT
98
NL
n.d.
NN or Nn
CT
99
NL
n.d.
M
IT
IW
NL
nd
NN
101
NL
n.d.
NN M Nn
CT
I 02
NL
n.d.
nn
TT
183
BL
f
N"
CT
184
BL
m
""
TT
350
BL
m
M
TT
351
BL
m
NN or Nn
CT
352
BL
m
M
TT
353
BL
m
M
IT
354
BL
m
NN
Nn
CT
355
BL
f
NN or Nn
CT
356
BL
f
NN M Nn
CT
357
BL
f
M
TT
358
BL
f
NNMN~
CT
359
BL
f
M
IT
3w
BL
f
NNaNn
CT
361
BL
f
NN or Nn
CT
M
M
M
Nn
CT
454
5 ' C C C A c ; c A c n r r r c ; A A T A ~ ~ ~ ~ ~ ~ C ( ~ ~ ~ G A G T G A C ~ ~ A ~ ~ ~ ~ t i
TCG l'TC Cl7j TAC CTG GGC TGG TAC A K GTT GAT TGT CCC TCC TCC
Ser Phe Leu Tyr Leu Gly Trp Tyr Met Val Asp Cys Pro Ser Trp GTC ACT ACA ACA ACT TCT TCT TNN ClT GCC CAC CTC ClG GAC KIT Val Thr Thr Thr Thr Ser Ser ??? Leu A h His Leu Leu Asp Re
GCC A n i GGG GTC Mti ACG Cl'G CGT ACC ATC CTC TCG TCC GTC ACC A h Met Gly Val Lys Tbr Leu Arg Tbr Re Leu Ser Ser Val Tbr Rimer PB25 CAC AAT GGC AAA
CAGGTGTGGAGAGGACCMKCTWGCCAGTGGACGT
His Asn Gly Lys ( j G G C G G C G G G C A ~A~ ~ ~ ~ A~ ~
~
~
C
GGTGAGGGGTGAGCCGAGCGAGGCGClGACCKKTCXCCGCCCCGCCCC CAG
~ 1
(-TG
1
~
GTG
Gln Leu Val Rimer PB33 GTG G W CTC (JTcr GCG GTT tiW GTC TAC CTG TAC ACC GTC Met Thr Val Gly Leu Leu Ala Val Val Val Tyr Leu Tyr Thr Val
ATG ACC
GTG GCC TTC AAC TlX lTC C W AAG .rrC TAC M C AAG AGC GAU GAC Val Ala We Asn Phe Phe Arg Lya Phe Tyr Asn Lya Ser Glu Asp GAG
ccr
GAC A T C ~ AAG TGT GAT GAC ATG ATG ACG GTGAGCCCCTGCCCCCA~'
Glu Pro Asp Met L p Cys Asp Asp Met Met Thr Fig.3. Representation of part of the porcine CRC gene, including two exons and the intermediate intron with a microsatellite (box, 96 bp allele) containing repetitive sequences. The DNA sequence of the exons is given in base triplets each coding for an amino acid (bold type, abbreviated). The primers used to amplify the microsatellite are given in the figure; their DNA sequences are complementary to those indicated by the lines.
does not include useful information with regard to the oedema disease. Effective diagnostic tests such as the DNA-based ones have to be combined with other test methods to take care of ambiguous linkage situations in breeding animals.
Parentage control with DNA markers In animal breeding, parentage control is an integral part of the procedures and considerations applied to selecting the most desirable animals for breeding purposes. Long production periods, extended generation intervals and high costs of investment in the rearing or purchase of large farm animals make it mandatory to deal with correct pedigrees.
~
455 Since the early findings of Landsteiner concerning blood group inheritance around 1900, this knowledge has led to the use of blood typing in most mammalian and bird species for parentage control. The disadvantage of this test is occasionally the exclusion probability. Recent molecular genetic research has brought new techniques on the DNA level to investigate the inheritance and segregation of markers, which can also be used for parentage control. Highly polymorphic DNA markers can detect genetic variants and can be used to describe individual animals and their offspring very precisely. Tandem repeat nucleotide sequences, such as CA-repeats, occur all over the genome. They usually consist of up to several hundred kb, but the optimal length to be used is between 100 and 250 kb [32]. Since for most species such microsatellites have been found [32-341 and can be selected as markers evenly distributed on every chromosome of the karyotype, they are well suited for use in parentage control (Fig. 4). For each microsatellite, the 5'- and 3'-ending sequences are known and, therefore, optimal primers for the PCR technique can be chosen. The allele size of the PCR-amplified microsatellites can be detected automatically by instruments such as a sequencer along with applied software. The 5' primers can be marked by different fluorochromes so that simultaneous electrophoretic analysis of several microsatellite loci can be carried out on one gel. Such a microsatellite panel must be tested for different breeds and should be compared with a standard. This test method cannot completely substitute for blood typing techniques, but it is able to decide most of the cases unsolved by conventional techniques [35]. At the same time these markers can also be used for recombination studies, haplotyping and marker assisted selection. Statistical and biometrical methods have yet to be developed to use these new techniques in a constructive way for animal breeding purposes and breeding value estimates.
Sperm separation The separation of X- and Y-chromosome-bearing spermatozoa has been attempted already for several decades by many scientific groups. A successful separation method might have a profound impact on breeding programmes for farm animals, especially for cattle. By artificial insemination with either X- or Y-spenn, the sex of the fertilised eggs and, thus, of the developing progeny would be predetermined or preselected, respectively. In the course of the years, several research groups have claimed success in separation of sperm. As many of these groups did work with fixed and killed spermatozoa, a verification of the separation procedure was not possible; the sperm could not be used anymore in fertilisation experiments. In other cases where the viability of the spermatozoa during the separation procedure was apparently not affected, the number of separated spermatozoa was too low for artificial insemination or the number of progeny obtained with separated sperm was insufficient for statistical evaluation. Thus, a method showing if a separation procedure of living or dead spermatozoa indeed leads to enrichment of either X- or
456
148.23 200150-
100, 50, 0 ,
Boar 166
,176
186
196
a76
186
,196
200150-
10050-
sow 400300-
200-
-
1000
7
-
.
n-r
Progeny 1: ,126
136
a46
1156
166
bp
148.25
250200-
15010050,
0
rHIr
Progeny 2:
Fig.4. Individual identification and parentage control in pigs using PCR-amplified,fluorochrome-labelled microsatellite fragments (microsatelliteS0088: 148-165 bp) and an automatic DNA-sequencer. Ordinate: fluorescence index; abscissa: number of bp. The diagrams show (from top) four electrophoretic patterns for a boar, sow and two of their offspring, respectively. Three microsatellite variants (alleles with 148, 150 or 164 bp) can be seen. Piglet 1 has obtained variant 148 from its father and variant 150 from the mother, piglet 2 variant 148 from the mother and variant 164 from the father. Similar electrophoretograms can be used in case of parentage exclusion.
457 Y-sperm would be very valuable and could save much time, effort and money for research in this area. A few years ago, an effective method to separate X- and Y-spermatozoa based on the difference in DNA-amount of X- and Y-chromosomes - and thus of X- and Yspermatozoa - was elaborated and published [36]. The separation method makes use of a vital fluorochrome which binds quantitatively to the DNA present in the spermatozoa. The technique uses flow cytometry and cell sorter and has plausibly shown that altered sex ratios in offspring can be obtained for rabbits and swine [36,37]. For several reasons (such as the cost of the apparatus, the low number of separated spermatozoa and the possibility of damage to the sorted sperm) the method does not seem to be suited for general use in artificial insemination programmes. On the other hand, the method has been employed to verify sperm separation results claimed by several research groups (review in [38]). This examination indicated that practically all of these procedures failed to separate X- and Y-sperm. The apparatus used for this verification is expensive, complicated and delicate in handling, so that worldwide only a few well-equipped laboratories are in a position to carry out this kind of research. Fortunately, a convenient method using chromosome-specific DNAprobes and in situ hybridization (FISH = fluorescent in situ hybridization) is now at hand to determine if spermatozoa are containing an X- or Y-chromosome. In the following, we will describe this procedure for bull spermatozoa in some detail. Decondensation of bull spermatozoa As the chromatin in spermatozoa is very densely packed [39], the deep-frozen or fresh spermatozoa have to be decondensed first. To this aim, the sperm is washed and purified in distilled water and transferred to slides for air-drying (1 h), fixation in alcohol (overnight, 4°C) and again air-drying (>1 h) [40]. In the following steps the spermatozoa remain fixed on the slides. Decondensation takes place in SDSDTT for 40-80 min, after which the spermatozoa again are fixed in alcohol (15 min) and airdried. In this stage the slides can be stored in -20°C for several weeks. The slides are then treated with RNAse to digest RNA (1 h, 37°C) and dehydrated with alcohol. For denaturation of the DNA the slides are kept in formamide-solution (2 min, 70°C) and then dehydrated in alcohol at -20°C to keep the DNA single-stranded.,At this point the denatured DNA-probe should be added (see below). DNA-probe and hybridization Some characteristics of the DNA-probe we use are given by Schwerin et al. [41]. It is highly repetitive on the Y-chromosome and 562-bp long. Prior to use, the probe is biotin-labelled during PCR-amplification. The probe is mixed (in alcohol, overnight, -80°C) with herring or salmon DNA, which has the function to saturate (later on) all unspecific DNA attachment sides on the slides. The volume of the mixture is then reduced by centrifugation and shedding/evaporation of the alcohol, and a standardised hybridization solution (including formamide) is added. After
458 denaturation (15 min, 70°C) this DNA-formamide mixture is given immediately to the denatured and dehydrated sperm on the slides and covered by a coverslip. Hybridization takes place overnight (37°C) in a moist chamber. For further handling, the coverslip has to be rinsed off cautiously using SSC-solution. Now, unbound DNA is removed by formamide washing and formamide, in its turn, by washing with distilled water. Immunological evidence of the Y-chromosome To prevent unspecific binding of antibodies at a later stage, the preparations are treated with a blocking solution (including bovine serum albumin and Tween 20, 1 h, 37°C). Then, the slides are incubated in avidin-FITC to react with the biotin (40 min, 37°C) and washed to remove unbound avidin. If necessary, amplification of the reaction is possible by the use of biotin-labelled avidin-antibodies and another incubation in avidin-FITC. Finally, the slides are stained with propidium iodide to visualise the sperm, washed, air-dried and covered with antifade and coverslip. An example of the result is given in Fig. 5. In case of staining of the Y-chromosome (as above), normal ejaculates should show fluorescence in approximately half of the number of spermatozoa. This is of course, easy to verify and, at the same time provides a suitable control [42]. If, however, spermatozoa are examined following tests for separation, it is advisable to use a second DNA-probe specific for the X-chromosome. This is, for example, practised by Johnson et al. [43] in separation of human spermatozoa.
Control of sex of offspring Separation of X- and Y-spermatozoa in combination with artificial insemination is expected to be the technique of the future for sex preselection. Because of the abovementioned shortcomings of sperm separation by flow cytometry/cell sorter, this
Fig. 5. Several fixed, decondensed und tailless bull spermatozoa on a slide after in situ hybridization with a Y-chromosome-specific DNA-probe. Some of the sperm show a bright spot indicating the Ychromosome visualized by the reaction of the fluorochrome FITC/avidinsystem with the biotin. (The length of the head of a living sperm is about 10 pm, after decondensation about 20 pm.)
459 method is at present only used in in vitro fertilisation test programmes and in some other special cases. There are, however, also other methods dealing with the control of sex of livestock offspring. Such a method is, for example, diagnosis of the sex of bovine embryos mainly practised at embryo transfer (ET). (Diagnosis of the sex of embryos is sometimes also called “sex determination”. This term has been reserved, however, by biologists already for more than 70 years to describe the genetic and environmental factors determining the sexual fate of animal and plant eggs and embryos and thus, should be avoided in cases where recognition of the sex is meant.) In the next chapter we will review the possibilities of this recent technique in embryo sexing.
Sex diagnosis of embryos
Before amplification of DNA-sequences by the polymerase chain reaction was introduced [44] and bovine Y-chromosome-specific DNA-sequences were becoming available, already several methods for the diagnosis of the embryonic sex had been developed (review in [45]). Obviously, such methods should not seriously affect the viability of the embryos, because the latter were to be used for embryo transfer to foster mothers. A method developed already many years ago depends on the cytological analysis of metaphases (review in [46]). This method is highly accurate with respect to diagnosis of the sex. On the other hand, it is invasive (the need for biopsy reduces the rate of pregnancies), rather demanding and takes at least 6 h. Two noninvasive methods are based on the colourimetric demonstration of the amount of X-chromosome-linked enzymes (e.g., [47]) and the detection of H-Y antigen by monoclonal antibodies [48], respectively. Both methods have the advantage of taking a few hours at most, but they lack accurateness and have other disadvantages. A method appropriate for commercial use in ET-practice should be rapid, highly accurate, noninvasive or rely on minute biopsies only, uncomplicated and inexpensive. Since the above-mentioned methods do fulfil only part of these needs, the search for a practicable method has been continued. Availability of Y-specific DNA-probes for several livestock species and development of the PCR reaction have now opened up new possibilities and here we will briefly report our own results. DNA-probes and biopsies To be independent of any patents on Y-chromosome-specific DNA-sequences, we have developed a sexing test based on sequence information from the bovine satellite DYZ-1 [49]. This satellite represents approximately 5% of the bovine Y-chromosome and the basic motif is about 60,000 times repeated. The two primers we use (PSI and PS2) defiie a 51 1 bp long DNA-sequence from the DYZ-1 satellite. In tests for sex diagnosis it is important that proper controls are carried out. It should be prevented, for example, that absence of a male marker (indicating female sex) might be based on loss of DNA or similar limitations. Therefore, we also make use of sequence
460 information of the bovine satellite 1.709 [50], which is of autosomal origin. The basic motif of this satellite is about 36,000 times repeated. The two primers (PS3 and PS4) we use in this case define a 263 bp long DNA-sequence. Since the satellite lies on an autosome, the amplified sequence should show up in the male as well as in the female sex, giving evidence for a correct functioning of the method. For PCR-amplification of specific DNA-sequences, only very small amounts of DNA are necessary and, in principle, DNA from one cell is sufficient (e.g., see Handyside et al. [51] who worked with human embryos). In our laboratory we took biopsies from stages between morula (16-cell embryos) and late blastocyst, containing between five cells and 30% of the total cell mass. Of 20 tested embryos 19 could be sexed unambiguously (Fig. 6). A later comparison with the sex of four developing foetuses confirmed the correctness of our sexing. These examinations had been carried out in connection with an investigation on the disease of arachnomely. This disease may have affected the viability of the embryos, which, in its turn, may have lead to a rather low pregnancy rate after embryo transfer. In summary, it can be said that by using PCR-reaction and chromosome-specific DNA-probes the most important criteria for the suitability of a method to sex embryos are fulfilled. This is also manifested by the worldwide marketing of embryo sexing by PCR-amplification of male-specific DNA-sequences in ET-programmes by several commercial groups.
Detection of transgenesis in embryos In the past few years molecular genetics has facilitated or even made possible diagnosis or detection of traits, alleles, DNA-sequences, sex, hereditary diseases, etc., in farm animals, and we do not doubt that genetic research will continue to yield new
1
2
3
4
Fig. 6. Sexing of cattle embryos by simultaneous PCR-amplification of Y-specific and autosomal satellite DNA-sequences. The left panel gives the DNA size marker. Panels 1,2, 3 and 4 all show the lower band from the autosomal satellite, panels 1 and 2 also the upper band from the Y-specific satellite DYZ-I. Consequently, 1 and 2 are male embryos, and 3 and 4 are female.
46 1 information and methods to be employed in animal science and breeding. On the other hand, it is also quite natural that molecular tools sometimes may fall short to the expectations. Such hopes were set on the PCR-method for detection of transgenic embryos in the generation of transgenic farm animals. Transgenic animals are animals with foreign DNA (or transgenes) experimentally introduced into their genome. In the case of transgenic livestock current research is aimed at obtaining genetically modified farm animals with improved productivity traits, enhanced health features or as suppliers of pharmaceuticals (reviews in [52,53]). It is obvious that the generation of transgenic livestock is a laborious, exacting and financially demanding undertaking. This is due, for example, to the long gestation periods, the great expenses, the need for qualified personnel, but also to the uncertainties of achieving the goal, that is transgenic offspring. The technique most frequently used to produce transgenic farm animals, is by microinjecting foreign DNA-sequences into the pronucleus of the egg cell [54].Some of the eggs treated in this way do incorporate the transgene into their genome. The eggs are cultured in vitro through a short period of embryonic development and then transferred to foster mothers. Usually, this results only in a few percent of transgenic newborns. It is, of course, possible to determine the state of the foetus (transgenic or nontransgenic) before birth through analysis of the amniotic fluid [55].This may, however, have unwanted consequences (abortion) and does not increase the efficiency of the method per se. Therefore, several projects examining the possibility of selecting transgenic embryos prior to embryo transfer have been carried out. As in sex diagnosis (see above), this requires the taking of a small biopsy (one or a few cells of the embryo) and PCR-analysis for detection of the transgene. Unfortunately, this promising technique did not increase the percentage of transgenic newborns due to determination of a high number of false-positive embryos [56]. This unexpected finding cannot yet be explained completely, but one interpretation is that the transgene is sometimes maintained extrachromosomally during the first cell divisions before degradation or removal from the embryo. Other possibilities are now explored to overcome these difficulties.
Conclusions In this article, we have stressed the importance of molecular genetics and gene technology for modern animal science and breeding. Several new techniques, such as the polymerase chain reaction and fluorescence in situ hybridization, have been developed recently and can now be applied as diagnostic tools. These techniques have facilitated or even enabled diagnosis or detection of traits, alleles, DNA-sequences, hereditary diseases, sex and so forth. Our examples include detection of favourable alleles of genes coding for milk proteins, recognition of negative recessive alleles in hereditary syndromes, the use of microsatellite variants for breeding purposes and parentage control, and application of specific DNA-probes for identification of Ychromosome-bearing spermatozoa and the sex of embryos. It is obvious that molecular genetics cannot solve all problems of animal breeding, and drawbacks and
462 shortcomings will be encountered. We have included such cases where the expectations were not yet realised. It is also clear that several newly developed methods still have to be improved and simplified before they can be applied in practice on a large scale. Yet, we think that in the future, gene technology in combination with conventional and approved methods will represent a valuable support in diagnostics and, by affecting selection in many ways, will have a great impact on progress in animal breeding.
Acknowledgements A substantial part of the data in this review was elaborated and compiled by coworkers of the Group “Breeding Biology” of the Institute of Animal Science (ETH Zurich). We want to thank these colleagues, in particular Roger Bolt, Jutta Hohenhorst, Johannes Kaiser, Andreina Schoberlein and Paul Steffen, for permission to publish these data. The help of Manfred Schwerin (Rostock) with hybridization of DNA-probes to the bovine Y-chromosome is also greatly acknowledged. We thank Regula Jenny for help with the manuscript.
References 1. Geldermann H, Pieper V, Roth B. Effects of marked chromosome sections on milk performance in cattle. Theoret Appl Genet 1985;70:138-146. 2. Stewart AF, Willis IM, Mackinlay AG. Nucleotide sequences of bovine 4,- and K-casein cDNAs. Nucl Acids Res 1984;12:3895-3907. 3. Scherbon IW, Ledford RA, Regenstein I, Thompson MP. Variants of milk proteins and their possible relation to milk properties. J Dairy Sci 1967;50:951. 4. Mullis K, Faloona F, Scharf S, Snikl R, Horn G, Ehrlich H. Specific enzymatic amplification of DNA in vitro: the polymerase chain reaction. Cold Spring Harbor Symp Quant Biol 1986;51:263273. 5 . Jakob E. Beziehungen zwischen dern genetischen Polymorphismus der Milchproteine und der Labf&gkeit von Milch. Dissertation ETH Nr. 10224, 1993:1-219. in verschiedenen 6. Erhardt G. K-Kaseine in Rindermilch - Nachweis eines weiteren Allels (K-CN~) Rassen. J Anim Breed Genet 1989$l6:225-23 1. 7. Losi G, Castagnetti GB, Morini D. Le varianti genetiche della caseina K e attitudine del latte alla coagulazione presamica. I1 Latte 1979;4:1062-1068. 8. Damiani G, Ferretti L, Rognoni G, Sgaramella V. Restriction fragment length polymorphism analysis of the K-Casein locus in cattle. Anim Gen 1990,21:107-114. 9. Levtziel H, Mtttnier L, Mght M-F, Choplain J, Furet J-P, Paboeuf G, Mercier JC, Grosclaude F. Identificationof two common alleles of the bovine K-casein locus by the RFLP technique, using the enzyme HindIII. Gtntt Stl Evol 1988;20248-253. 10. Gorodetsky SI, Kaledin AS. Nucleotide sequence analysis of cow K-CaSein cDNA. Genetika 1987;23:596-604. 11. Rando A, di Gregorio P, Masina P. Identification of bovine K-casein genotypes at the DNA level. Anim Genet 1988;19:51-54. 12. Eggen A. Mise en tvidence de RFLPs dans le locus de la casCine bovine. Diplomarbeit, Institut fur Nutztienvissenschaften,Gruppe Zuchtungsbiologie, ETH Zurich, 1988. 13. Alexander LJ,Stewart AF, Mackinlay AG, Kapelinskaya TV, Tkach TM, Gorodetsky SI. Isolation and characterization of the bovine K-casein gene. Eur J Biochern 1988;178:395-401.
463 14. White TJ, Arnheim N, Erlich HA. The polymerase chain reaction. Trends Genet 1989;5:185-189. 15. Smith C, Bampton PR. Inheritance of reaction to halothane anaesthesia in pigs. Genet Res 1977;29:287-292. 16. Mitchell G, Heffron JJA. Porcine stress syndromes. Adv Food Res 1982;28:167-230. 17. Davies W, Harbitz I, Fries R, Stranzinger G, Hauge JG. Porcine malignant hyperthermia carrier detection and chromosomal assignment using a linked probe. Anim Genet 1988;19:203-212. 18. Harbitz I, Chowdhary B, Thomsen P, Davies W, Kaufmann U, Kran S, Gustavsson I, Christensen K, Hauge J. Assignment of the porcine calcium-release channel gene, a candidate for the malignant hyperthermia locus, to the 6pll-q21 segment of chromosome 6. Genomics 1990;9:243-248. 19. Yang H, Jung HR, Solinas Toldo S, Lang E, Bolt R, Fries R, StranzingerG. A reciprocal whole-am translocation, rcp (1;6) (lp6p;lq6q) in a boar, localization of the breakpoints, 'and reassignment of the genes for glucose phosphate isomerase (GPI) and calcium-releasechannel (CRC). Cytogen Cell Genet 199261:67-74. 20. Gahne B, Juneja RK. Prediction of the halothane (Hal) genotypes of pigs by deducing Hal, Phi, P02, Pgd haplotypes of parents and offsprings: results from a large-scale practice in Swedish breeds. Anim Blood Groups Biochem Genet 1985;16:265-283. 21. Hojnf J, Cepica S, Hradeclj J. Gene order and recombination rates in the linkage group S-Phi-HalH-(PoZ)-Pgd in pigs. Anim Blood Groups Biochem Genet 1985;16:307-318. 22. Vogeli P, Kiihne R, Gerwig C, Kaufmann A, Wysshaar M, Stranzinger G. Prediction of the halothane genotypes with the aid of the S, PHI, HAL, H, P02, PGD haplotypes of parents and offspring in Swiss Landrace pigs. Ziichtungskunde 1988;60:24-37. 23. Levitt RC, Olckers A, Meyers S, Fletcher JE,Rosenberg H, Isaacs H, Meyers DA. Evidence for the localization of a malignant hyperthermia susceptibility locus (MHS2) to human chromosome 17q. Genomics 1992; 14562-566. 24. Otsu K, Phillips MS, Khanna VK, DeLeon S, MacLennan DH. Refinement of diagnostic assays for a probable causal mutation for porcine and human malignant hyperthermia. Genomics 1992;135335837. 25. MacLennan DH, Duff C, Zonato F, Fujii J, Phillips MS, Korneluk RG, Frodis W, Britt BA, Worton RG. Ryanodine receptor gene is a candidate for predisposition to malignant hyperthermia. Nature 1990;343:55!+-561. 26. Fujii J, Otsu K, Zorzato F, DeLeon S, Khanna VK, Weiler J, O'Brien PJ, Machnnan DH. Identification of a mutation in the porcine ryanodine receptor associated with malignant hyperthermia. Science 1991;253:44-45 1. 27. Hughes IP,Moran C, Nicholas FW. PCR genotyping of the ryanodine receptor gene for a putative causal mutation for malignant hyperthermia in Australian pigs. J Anim Breed Genet 1992;109:465 476. 28. Bolt R, Vogeli P, Fries R. A polymorphic microsatellite at the RYRl locus in swine. Anim Genet 1993;24:72. 29. Vogeli P, Bolt R, Fries R, Stranzinger G. Co-segregation of the malignant hyperthermia and the Arg6'5-Cys6'5mutation in the skeletal muscle calcium release channel protein in five European Landrace and Pietrain pig breeds. Anim Genet 1994;25(Suppl 1):59-66. 30. Vogeli P, Fries R, Bolt R, Gerwig C, Affentranger P, Kiinzi N, Bertschinger HU, Stranzinger G. Genetics of porcine stress syndrome and association with oedema disease. In: Puolanne E, Demeyer DI (eds) Pork Quality: Genetic and Metabolic Factors. Wallingford, UK: CAB International, 199222-36. 31. Bertschinger HU, Munz-Miiller M, ffirter HP, Schneider A. Genetic resistance against Escherichiu coli enterotoxaemia in the pig. J Anim Breed Genet 1986;103:255-264. 32. Fries R, Vogeli P, Stranzinger G. Gene mapping in the pig. In: McFeely RA (ed) Advances in Veterinary Science and Comparative Medicine. Domestic Animal Cytogenetics. New York: Academic Press Inc., 1990;34:273-303. 33. Bishop MD, Kappes SA, Keele JW,Stone RT, Sunden SLF, Hawkins GA, Solinas Toldo S, Fries R, Grosz MD, Yoo J, Cundiff LW, Beattie CW. A genetic linkage map for cattle. Genetics
34.
35. 36. 37. 38. 39. 40. 41. 42. 43.
44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55.
56.
1994;136:61+639. Barendse W, Armitage SM, Kossarek LM, Shalom A, Kirkpatrick BW, Ryan AM, Clayton D, Li L, Neibergs HL, Zhang N, Grosse WM, Weiss J, Creighton P, McCarthy F, Ron M, Teal kl, Fries R, McGraw RA, Moore SS, Georges M, Soller M, Womack JE, Hetzel DJS. A genetic map of the bovine genome. Nature Genet 1994;6:227-235. Glowatzki-Mullis M-L, Gaillard C, Wigger G, Fries R. Microsatellite-based parentage control in cattle. Anim Genet 1995;26:7-12. Johnson LA, Flook JP, Hawk HW. Sex preselection in rabbits: live births from X and Y sperm separated by DNA and cell sorting. Biol Reprod 1989;41:19+203. Johnson LA. Sex preselection in swine: altered sex ratios in offspring following surgical insemination of flow sorted X- and Y-bearing sperm. Reprod Dom Anim 1991;26:30+314. Gledhill BL. Selection and separation of X- and Y-chromosome-bearing mammalian sperm. Gamete Res 1988;20:377-395. Ward WS, Coffey DS. DNA packaging and organization in mammalian spermatozoa: comparison with somatic cells. Biol Reprod 1991;44:56+574. Schwerin M, Blottner S, Thomsen PD, Roschlau D, Brockmann G. Quantificationof Y chromosome bearing spermatozoa of cattle using in situ hybridization. Molec Reprod Devel 1991;30:3+43. Schwerin M, Gallagher DS, Miller JR,Thomsen PD, Mapping of repetitive bovine DNA sequences on cattle Y chromosomes. Cytogent Cell Genet 1992;61:18+194. Went DF, Kaiser J, StranzingerG. Geschlechtspraselektiondurch Trennung von X- und Y-Spermien. Landwirtsch Schweiz 1993;6:445-449. Johnson LA, Welch GR, Keyvanfar K, Dorfmann A, Fugger EF, Schulman JD. Gender preselection in humans? Flow cytometric separation of X and Y spermatozoa for the prevention of X-linked diseases. Human Reprod 1993;8:1733-1739. Mullis KB, Faloona FA. Specific synthesis of DNA in vitro via a polymerase-catalyzed chain reaction. Meth Enzymol 1987;155:335-350. van Vliet RA, Gibbins AMV, Walton JS. Livestock embryo sexing: a review of current methods, with emphasis on Y-specific DNA probes. Theriogenology 1989;32:421-438. King WA. Sexing embryos by cytological methods. Theriogenology 1984;21:7-17. Williams TJ. A technique for sexing mouse embryos by a visual colorimetric assay of the X-linked enzyme, glucose 6-phosphate dehydrogenase. Theriogenology 1986;25:733-739. Wachtel S, Nakamura D, Wachtel G, Felton W, Kent M, Jaswaney V. Sex selection with monoclonal H-Y antibody. Fertil Steril 1988;50:355-360. Perret J, Shia Y-C, Fries R, Vassart G, Georges M. A polymorphic satellite sequence maps to the pencentric region of the bovine Y chromosome. Genomics 1990;6:482-490. Skowronski J, Plucienniczak A, Bednarek A, Jaworski J. Bovine 1.709 satellite: recombination hot spots and dispersed repeated sequences. J Molec Biol 1984;177:39+416. Handyside AH, Kontogianni EH, Hardy K, Winston RML. Pregnancies from biopsied human preimplantation embryos sexed by Y-specific DNA amplification. Nature 1990;344:768-770. Ebert KM, Schindler JES. Transgenic farm animals: progress report. Theriogenology 1993;39:121135. Pursel VG, Rexroad CE Jr. Status of research with transgenic f a m animals. J Anim Sci 1993;71(Suppl 3):l(t19. Hammer RE, Pursel VG, Rexroad CE, Wall RJ, Bolt DJ, Ebert KM, Palmiter RD, Brinster RL. Production of transgenic rabbits, sheep and pigs by microinjection. Nature 1985;315:680-683. Hyttinen JM, Peura T, Tolvanen M, Aalto J, Alhonen L, Sinervirta R, Halmekyto M, Myohhen S, Jinne J. Generation of transgenic dairy cattle from transgene-analyzed and sexed embryos produced in vitro. Biotechnology 1994;12:60&608. Bowen RA, Reed ML, Schnieke A, Seidel GE Jr, Stacey A, Thomas WK, Kajikawa 0. Transgenic cattle resulting from biopsied embryos: expression of c-ski in a transgenic calf. Biol Reprod 1994;50:664-668.
01996 Elsevier Science B.V. All rights reserved. Biotechnology Annual Review Volume 2. M.R.El-Gewely, editor.
465
Biotechnology in developing countries: critical issues of technological capability building Rohini Acharya‘ and John Mugabe’ ‘International Economics Programme, Royal Institute of International Affairs,London, UK; and 2African Centre for Technology Studies, Nairobi, Kenya
Abstract. It is now widely recognised by industrialised and developing countries alike that technology and the successful adaptation of new technologies is crucial for economic growth and sustainable development. The wide range of techniques presented by biotechnology provide an important opportunity for developing countries to adopt technologies which can be tailored to their own environmental and societal needs in a sustainable manner. The successful adaptation of a science-intensive technology such as biotechnology, however, involves a number of layers of management, including the development of the basic sciences, institutional capabilities and the establishment of a macroenvironment which is conducive to technological change and production. The basic task of this paper is to assess national technological capabilities for biotechnology management in developing countries. The paper is concerned with what constitutes a nation’s capabilities for managing the development and application of biotechnology as well as guiding its evolution in local socioeconomic systems. We examine national policies to develop biotechnology in a number of countries in south and south-east Asia and Africa and compare their experiences in an attempt to determine what their particular strengths and weaknesses are. Due to the different levels of capability found in these parts of the world as well as different industrial structures across sectors, the tendency has been to concentrate on developing different areas of specialisation. Thus, technological change has been incremental rather than radical in both these areas. However, Asia and especially the newly industrialising countries of Southeast Asia have a tendency to race ahead of many other countries, as would be expected, because of their previous accumulation of technological capabilities. Nevertheless, the fact that even these countries have had differing degrees of success in commercialising biotechnology in different sectors, shows that possessing scientific capability to adapt technology is not sufficient. Instead, capabilitiesat different levels, to innovate and also to efficiently manage the technology must be developed in order for developing countries to register success in harnessing biotechnology for their own needs and priorities.
Key words: Africa, biodiversity, biotechnology management, government policies on science and institutional capability, South Asia, Southeast Asia, sustainable development, technology.
‘Introduction The capability to develop and utilize technology is one of the most important factors accounting for differences in nations’ economic performance. The development of the capability involves: the creation and management of institutions that are charged with the specific activities of generating and utilizing scientific and technological
Address for correspondence: Rohini Acharya, Senior Research Fellow, International Economics Programme, Royal Institute of International Affairs, 10 St. James’s Square, London, SWlY 4LE, UK.
466 knowledge; the creation of human capital in specific areas of research and development (R&D); the establishment of an economic and political environment supportive of technological change; and the institutionalization of specific policies for promoting technological innovation and learning. The basic task of this paper is to assess national technological capabilities for biotechnology management in developing countries. The paper is concerned with what constitutes a nation’s capabilities for managing the development and application of biotechnology as well as guiding its evolution in local socioeconomic systems. There are several reasons as to why attention is placed on national capabilities for biotechnology management. First, in order to determine what governments should do to promote the development and application of biotechnology, it is useful to know the specific context in which a particular government intervenes. Some government policies may be hostile to local technological learning and may perpetuate the erosion of technological capabilities through poor incentive schemes and rigid institutional structures. Others may deliberately promote the accumulation of indigenous capabilities for innovation. Secondly, during the past decade a large corpus of literature has been generated on the socioeconomic impacts of biotechnology on developing economies. Most of this literature is generic in focus and tells us very little about specific efforts and limitations of particular developing countries in biotechnology management. It does not guide policy reforms aimed at promoting the development of national capabilities for biotechnology management. The long-term benefits that developing countries are likely to draw from biotechnology are intimately linked to the ability of countries to effectively use existing biotechnological techniques and engage in R&D activities to generate new innovations to meet their specific demands and increase the welfare of local populations. In other words, it is the extent to which the countries absorb old and new biotechnological techniques and integrate them into local socioeconomic systems that provides a measure of their competence to manage biotechnology. Third, biotechnology management activities are undertaken by specific institutional organs. It is within institutions that scientific skills and knowledge are mobilized and directed to the management of biotechnology. In order for an institution to engage in the development and application of biotechnology, it must know certain scientific procedures and should also possess some basic capabilities in terms of the human capital, adequate R&D resources, and technical infrastructure. The understanding of institutional systems is crucial in guiding efforts aimed at strengthening national capabilities for biotechnology management. This paper is based on the premise that programmes aimed at contributing to the development of capabilities for biotechnology R&D in developing countries should be dependent upon an adequate understanding of the nature of institutional systems as well as national policies pertaining to technological change and economic development. The first section of the paper provides an overview of the nature of biotechnology; essentially the salient features (characteristics) of the technology. It also examines international developments in biotechnology and the role of biotechnology in sustainable development. The main objective of the section is to outline those aspects
467 of biotechnology development that require critical understanding for capability building. The second section of the paper assesses national technological capabilities for biotechnology R&D in Asia. The third examines the status of biotechnology in Africa and the capabilities of the countries to engage in biotechnology R&D. The fourth section deals with issues of policy reform, and development of institutional and national capabilities to develop and manage biotechnology. We end with some concluding observations on the nature of technological capabilities for biotechnology that have been developed in these two parts of the world and the task ahead.
1Global trends in biotechnology Biotechnology is herein defined as a body of knowledge and techniques involving the integrated application of biological sciences such as genetics, molecular biology, microbiology and engineering to produce goods and/or services from living organisms or parts thereof. This definition of biotechnology is broad to encompass both new techniques, such as genetic engineering, and older techniques which tend to be used more extensively in developing countries because of issues of access to technology and patented information. The evolution of biotechnology is characterized by major institutional changes and scientific developments. The technology has gone through three generations: from household traditional fermentation to a large-scale fermentation industry using “lowlevel” techniques of mobilizing enzymes with relatively little scientific knowledge to the now established modem biotechnology characterized by “high-level” techniques of genetic engineering, embryo transfer, tissue culture and others. Modem biotechnology emerged from university laboratory research in the USA in the early 1970s’ followed soon after by other countries of the OECD such as Westem Europe and Japan. The emergence of the biotechnology industry was closely associated with rapid scientific breakthroughs in areas such as genetic engineering and tissue culture. Scientific research was the fundamental source of knowledge and opened opportunities for innovation. The institutional basis for scientific research was university laboratories. US universities invested considerable resources in basic scientific research in areas such as molecular biology, plant physiology and genetics. Most of their research was based on the exploitation of their knowledge of living matter. The involvement of universities in biotechnology and the fundamental role of scientific research have determined and guided the evolution of the technology in various ways. First, the growth of biotechnology as an industry has been characterized by a strong element of uncertainty. There has been, and still is, uncertainty about the potential benefits of the emerging developments. That uncertainty is extended to the nature of the benefits and how such benefits will be shared among a wide range of actors from different entrepreneurial interests and resources. The uncertainty feature continues to determine the levels and nature of investment devoted to R&D in biotechnology.
468 Second, biotechnology is a science- or knowledge-intensive technology. The growth of activities in biotechnology followed major scientific breakthroughs. R&D activities leading to the establishment of the biotechnology industry were spurred by the motivations and interests of scientists and scientific institutions and evolved on trajectories that initially had little to do with industrial demands [2]. Thus, biotechnology has its origins in a “scientific-push” which is by and large divorced from economic motivations but has crucially contributed to determining the emergence of particular technological opportunities and the subsequent trajectories of technological innovation [ 11. Though the application of various biological techniques to produce goods from living organisms has been there for a long time, ’scientific breakthroughs in genetic engineering have revitalised some of the traditional techniques. Genetic engineering “had to be integrated with other technologies, most notably process technologies related to large-scale production”.’ The growth of biotechnology epitomizes the nature of the process of technological change in general: the cumulativeness of technological change. Different biotechnological trajectories are evolving from classical techniques in chemistry, biology and engineering. The role of scientific knowledge cannot be ignored in biotechnology. Our ability to use genetic engineering to produce disease-resistant varieties of crops to no small measure depends on the existing knowledge, breeding techniques and production systems. Biotechnology is a pervasive technology. Its impact is spread across various industries (ranging from pharmaceuticals to agriculture) and socioeconomic groups. It is also multidisciplinary in nature in the sense that it encompasses various techniques which are used in an integrated way. Particular biotechnological techniques can be applied across a wide range of sectors and therefore different actors with different economic and social interests are involved. This not only raises the complexity inherent in the process of technical change but also determines, again, the nature and levels of investment in R&D. Furthermore, the technical feasibility of the potential applications of biotechnology, the time required to develop them and, above all, the economic revenues which could be generated by innovations [l] in the domain of biotechnology remain often uncertain during the early stages of investment in R&D. The potential impacts of some of the applications of biotechnology are also, to a large extent, uncertain. A related issue of uncertainty concerns the appropriation of the benefits or revenues arising from biotechnology innovations. Most of the scientific information and knowledge, as well as some of the raw materials underlying innovations of biotechnology, are in the public domain. The information is accessible through scientific publications or can be acquired easily by firms or persons that have the requisite capabilities. The protection of bioinnovations (biotechnological innovations), in this context, depends mainly on patents and on secrecy [5]. Various problems arise which make patenting of bioinnovations complex and controversial. First, biotechnology involves the use and transformation of genetic material which have often been obtained from developing countries without payment. Second, the technology deals with living matter and so ethical questions have been raised on whether living matter
469 is patentable [6]. The two issues have been extensively discussed in various regional and international fora. They formed the most debatable part of the negotiations for the Convention on Biological Diversity. The first issue - relating to the sharing of the benefits of biotechnological innovations from genetic material acquired from developing countries - has partly been resolved. The resolution is deposited in Article 19 (Handling of Biotechnology and Distribution of its Benefits) of the Convention on Biological Diversity signed in 1992 in Rio de Janeiro. The second issue is still being discussed in the context of the debates on patenting life forms. The debates are conducted under the aegis of the European Parliament.
1.2 International trends in the development of biotechnology The international development of biotechnology has been associated with the growth of corporate involvement and investment in R&D. Corporate interest and involvement in biotechnology has grown in the USA, Germany, France, the UK, Japan and a number of other industrialised European countries [7].Private sector interest in biotechnology has been stimulated by an awareness of the potentials of new techniques such as genetic engineering and the prospects of getting profitable new products onto the market. The anticipated potentials of new biotechnological techniques have driven firms in the industrialised countries to invest in R&D and some have established inhouse research facilities and expertise. In other cases, particularly in the agrwhemical sector, some companies have acquired in-house expertise through takeovers of smaller firms with considerable scientific expertise [2,7]. This has been accompanied by an increasing tendency for companies to acquire vertical integration through R&D, production and marketing alliances with each other.3 In the early 1980s biotechnology firms in the USA and, to a lesser extent, Europe were relatively small and specialised in R&D, focusing on only one product or process. However, towards the late 1980s most of the firms were diversifying their R&D activities to capture the benefits associated with the multidisciplinary and pervasive nature of biotechnology. Those that have succeeded in diversifying their R&D activities and engaging in the production of a wide range of products have done so for one major reason: they have been able to mobilise scientific capabilities in a wide range of areas and have also been able to mobilise and invest considerable financial resources in biotechnology R&D. Furthermore, the policy environments in which they operate have been supportive of their efforts. The development of innovative activities in the area of biotechnology has over the years been uneven across sectors and Countries. Most of the biotechnology R&D has been in the pharmaceutical sector in particular and human health sector in general. In the USA a large share of R&D investment and activity has focused on human therapeutics, followed by diagnostics, chemicals, plant agriculture, animal agriculture and reagents. In Germany, pharmaceutical R&D and biochemical processing have received more priority." With increasing scientific research various technological opportunities are emerging. New areas of scientific research interest include the application of genetic
470
engineering in environmental management as well as mining. But on the whole, a large share of global biotechnology R&D effort is directed to medical biotechnology and the pharmaceutical sector in particular. The possession of strong core scientific capabilities in areas underlying modem biotechnology has been a vital precondition of the engagement in the biotechnology enterprise and the attainment of high levels of technological performance. The US companies which for a long time have been leaders in the domain of biotechnology have maintained their hegemony through the accumulation of scientific knowledge. They have mobilised scientific knowledge and expertise through various institutional arrangements. American universities and medical colleges have been the source of scientific knowledge and information to the companies. The companies have mobilised and utilised the scientific information and knowledge of these institutions through strategic alliances. Such strategic alliances have also become more important over the years for small biotechnology companies which do not have the required vertical integration (a notable exception is the American company, Amgen) to compete with the larger multinationals, especially in later stages of product development.’ It should be noted that American scientific superiority has been the basis for an enormous technological gap, but it has also been used by its competitors as a pool of knowledge on which to build new technological capabilities [ 11. Japanese companies have been at the forefront of searching for and acquiring American scientific knowledge and information. They have done so through strategic institutional alliances [ 11. International developments in biotechnology have been associated with, and in many cases have stimulated, major changes in government policies. The USA has established policy regimes that are supportive of efforts of local biotechnology companies and has also given more financial support to basic research. The UK and Germany, as well as other industrialised European countries, have tended to provide more support to applied research.6 German policies have encouraged private investment in biotechnology R&D. Industry in Germany invests about 58% of the national total [12]. Developments in biotechnology are reordering the world economy and changing patterns of international trade and relations. In dynamic sectors such as biochemical processing and pharmaceuticals, economic competitiveness and technological performance are gained and sustained through the strategic hamessing of scientific knowledge and application of new biotechnological techniques. It is the application of new techniques and the creation of new technological capabilities that are the key to competitiveness of biotechnology companies in the USA, Japan and Germany [13]. As noted above, there are apparent differences across countries in the extent to which they command biotechnology. These differences are largely determined by the levels of technological capabilities countries possess. While access to new scientific and technological knowledge on biotechnology is relatively open, it is only those countries that make strategic investments in terms of accumulating the requisite capabilities to absorb and master new scientific information that exploit technological opportunities emerging from the global growth of biotechnology.
47 1 In contrast, the most striking feature of developing countries, particularly those of Africa, is the limited and uneven nature of their technological capabilities in biotechnology. These countries have low absorptive capacities and their abilities to engage in technological learning to use new techniques of biotechnology are limited.7 For these countries to make significant steps in building up their capabilities and successfully engage in the application of biotechnology to enhance their economic growth and international competitiveness, they need to establish programmes aimed at creating core scientific and technological capabilities in various areas of biotechnology. This is a process involving investment in training as well as the establishment of specific biotechnology promotion institutions. It is only after the developing countries have accumulated capabilities for R&D, established institutions and formulated policies that promote local development of the technology will they be able to move to higher planes of technological performance and harness its potentials.
1.3 Biotechnology and sustainable development Sustainable development, which is “the ability of present generations to meet their obligations without compromising the ability of future generations to meet their own needs”, involves the generation and application of knowledge for efficient management of ecological and human resources for the long-term improvement of the social and economic welfare of populations [ 161. Technology and the associated scientific information and knowledge form a major source of economic growth, social change and ecological governance. The development and application of biotechnology have extended the scope and scale of global, national and local economic activities. They have brought immense changes both in human life and on the ecological base. For the past 20 years or so biotechnological techniques as wide ranging as genetic engineering and tissue culture have been applied to improve and manage the status of health and agricultural production. For this reason the development and application of biotechnology are of major concern in public policy discourse. Over the past 30 years or so numerous articles and books have been generated on the topic of the impact of biotechnology on developing economies. In the domain of economics, issues such as the impact of biotechnological advances on international trade and global economic change have come to the forefront as research topics. Other important and commonly discussed issues include the potential dislocation of Third World economies as exporters of raw agricultural commodities such as coffee and cocoa from the international market as a result of biotechnological advances in the industrialised countries [6]. The role of biotechnology in the economic transformation of developing countries and how the countries can harness the potentials of the technology in dealing with their socioeconomic and ecological problems have also been accorded significant attention by some technology policy scientists [ 171. The role of biotechnology as a motor for economic change and growth in developing countries - through improved agricultural production, genetic resource conservation, increased industrial performance and improved health status - can no
472 longer be disputed. However, associated with the potential benefits are various negative impacts of the technology on developing economies. In the industrialised world public awareness and concerns have been raised on the impacts of the various biotechnological developments on human health and the ecological base. Indeed concern over the potential negative impacts of genetic engineering on the environment (through the release of genetically modified organisms into the atmosphere), the potential erosion of the earth’s genetic resources through the spread of monocultures and the ethical questions associated with displacement of Third World traditional exports from global markets consequent upon expansive application of biotechnology have raised public consciousness to the fact that, if not properly managed, the development and application of the technology can cause irreversible negative socioeconomic consequences. The concept of sustainable development articulates the fact that technology may have harmful environmental and social effects. This fact has provided impetus for the concerns demanding the development and utilisation of environmentally sound technologies. There is global concern for the establishment of regulatory (legal, institutional and policy) measures to control the generation and use of biotechnologies that have negative ecological and socioeconomic effects. For example, most counties of the European Union (EU) have adopted regulations which ban the application of certain biotechnological processes used in the manipulation of life forms. At the international level concerns on managing technology to avoid ecological and socioeconomic catastrophes are reflected in Agenda 21. Specific chapters articulating concerns of regulating the development and use of technology are: chapter 16 on environmentally sound management of biotechnology and chapter 34 on transfer of environmentally sound technology, cooperation and capacity building. The principles of sustainable development also encompass issues of equity in the allocation of the benefits of technological development. In many developing countries the results of technological development in the form of improved health facilities, increased nutrition, enhanced agricultural production and improvements in the quality of food, easy and fast transportation, telephones, radios etc. are accessible to only a small percentage of the people. The proper application of biotechnology should enhance national economic growth, improve the socioeconomic welfare of the people, increase national competitiveness and improve efficiency in the use of natural resources. It should increase the range and volume of a country’s exports by making it possible for the country to respond quickly and more effectively to changing patterns of global trade and opportunities created by technological advances. It should also make it possible to effectively address social and environmental problems, particularly those of the poor and disadvantaged. The technology offers developing counties new opportunities for increasing their food production as well as dealing with the health and ecological problems [ 181. The WCED has observed: Biotechnology will have major implications for the environment. The products of genetic engineering could dramatically improve human and animal health.
473 Researchers are finding new drugs, new therapies, and new ways of controlling disease vectors. Energy derived from plants could increasingly substitute for the nonrenewable fossil fuels. New high-yield crop varieties and those resistant to unfavourable weather conditions and pests could revolutionise agriculture..... Biotechnology could also yield cleaner and more efficient alternatives to many wasteful processes and polluting products. New techniques to treat solid and liquid wastes could solve the pressing problem of hazardous waste disposal.8 However, by itself, biotechnology cannot resolve all the problems of ecological degradation and economic development. The expected or anticipated potentials of biotechnology should be realistic and be based on certain priority areas of sustainable development. What biotechnology is likely to do is to contribute to the enhancement of human welfare through, for example, better health care, enhanced food security through sustainable agricultural practices, improved supplies of potable water, more efficient industrial development processes for transforming raw materials, support for sustainable methods of afforestation and reforestation, and detoxification of hazardous wastes? But for countries to harness the potentials of biotechnology and move onto the paths of sustainable development they need to create “national systems of innovation” [20]. This means that they should establish policy and institutional systems that promote indigenous technological learning and the accumulation of capabilities to master imported biotechnological techniques, improve on them and integrate them with local technological knowledge. It is essentially the capacities or abilities of the countries to make technological decisions and effectively apply suitable techniques, that determine their success in the global process of biotechnology development. The need to develop these capabilities in biotechnology becomes especially important because of the need of these countries to confront their own regional and local problems. In health, for example, the R&D done at present in the industrialised countries tends to focus on problems predominant in these countries such as cancer and heart disease. In developing countries the priorities are to focus on tropical and regional diseases such as malaria, hepatitis and sleeping sickness. Similarly, agricultural priorities tend to be different in industrialised and developing countries and indeed between developing countries, making it necessary for developing countries to acquire technological capabilities for adopting and adapting new biotechnologies for their needs and priorities. There are certain basic factors that enable us to gauge a country’s technological performance. These may be outlined as: a country’s status of economic development, the capacity of its productive sector to manage technology (including the nature and level of technology in use and the extent of the involvement of local people in the use and management of the technology), the availability of funds for investment in technological activities both in the productive sector and for R&D, scientific and technological infrastructure (institutions and their research and service potentials), availability of local human capital at all levels of scientific and technological training, and the state of science and technology policy (its nature and the kinds of institutions,
474
including their capacities, for implementing it). It should be noted that the potential benefits of biotechnology to a country’s economic system are determined by the broad policies of the country towards income distribution, employment, health, basic needs and by specific technological interventions for specific problems and opportunities affecting the majority of poor people. Indeed the challenge to developing countries lies in the establishment of suitable policies and institutions that promote the harnessing of appropriate biotechnological techniques and the application of these to solving local problems. The challenge extends to reforming the existing economic and political institutions and policies which allocate the benefits of development to a small population of society. It is mainly through well-informed policy and institutional reforms that the imperatives of sustainable development will be addressed through proper application of a new technology such as biotechnology. The United Nations Conference on Environment and Development (UNCED) through Agenda 2 1 has articulated the importance of building national capabilities for biotechnology development. Chapter 16 para. 16.43 observes: The accelerated development and application of biotechnologies, particularly in developing countries, will require a major effort to build up institutional capacities at the national and regional levels. In developing countries, enabling factors such as training capacity, know-how, research and development facilities and funds, industrial building capacity, capital (including venture capital), protection of intellectual property rights, and expertise in areas such as marketing, research, technology assessment, socioeconomic assessment and safety assessment are frequently inadequate. Efforts will therefore need to be made to build up capacities in these and other areas and to match such efforts with appropriate levels of financial support. There is therefore a need to strengthen the endogenous capacities of developing countries by means of new international initiatives to support research in order to speed up the development and application of both new and conventional biotechnologies to serve the needs of sustainable development at the local, national and regional levels. On the whole, the technology offers ample technological opportunities for improved agricultural production, enhanced health care and industrial innovation. But given its science intensiveness and multidisciplinary nature, the application of biotechnology requires capabilities for mastering and creating or accumulating many differentiated and complimentary skills and knowledge. In other words, while the technology offers opportunities for economic diversification, the ability of countries to harness the potentials and opportunities associated with its evolution and growth depends on the nature and levels of capabilities that they create or accumulate in a wide range of scientific and industrial disciplines. This becomes clear in the next two sections where we discuss the adoption and adaptation of biotechnology in two parts of the developing world to address their needs: Asia, especially south and south-east Asian countries, and Africa. The former, as one would expect, has been more successful at
475
establishing biotechnology, especially the more sophisticated techniques of modem biotechnology, than the latter. This is largely due to a number of factors such as technical skills and training, availability of financial capital in the local and foreign markets, as well as better coherence of government policy and coordination between government departments responsible for the implementation of biotechnology policies.
2 Biotechnology in Asia” The Asian region, and in particular, Southeast Asia is one of the fastest growing areas in the world today. The populous giants, India and China are also showing signs of opening up, and China has recorded growth rates of over 9% during the last decade. This growth has been accompanied by higher rates of investment in capital infrastructure and “human capital”, biotechnology being one of the new technologies that has attracted attention and funding. Although relatively recent, the effort by policy makers in south and south-east Asia to promote biotechnology has spilled over into the private sector, accelerating commercial application and diffusion of biotechnology. This section examines, in particular, six of these countries with respect to their efforts to promote biotechnology research and development through policy making and institution building: China, India, the Philippines, South Korea, Taiwan and Thailand. Many of them have tried to emulate the structure established in the USA and other industrialised countries, of providing opportunities and environments in which the private sector can thrive. We fist look at the basic R&D infrastructure and the role of government policy which remains important despite an emerging private sector.
2.1 The role of government policy Government policies, with the possible exception of South Korea, have formed the most important component of national programmes of biotechnology diffusion in this part of the world. Indeed, in many industrialised countries the government has also had to play a major role in encouraging public and private investment in biotechnology. In developing countries, this component tends to be all the more important given the absence of private sector activities, especially during the early stages of R&D. Table 1 provides some statistics on R&D capabilities of selected developing and industrialising countries. For the newly industrialising countries and the more advanced developing countries such as Taiwan and Korea, there is already evidence that overall government spending on R&D forms a relatively high percentage of Gross National Product, especially in Korea where spending has almost reached the 2% level. Countries such as China and India, where investment in R&D is also quite high are, however, still lagging behind and in absolute terms the gap in expenditure between industrialised and developing countries is still greater. Nevertheless, that increasing importance is being placed on investment in both research and training, is clearly
476 Table I . R&D expenditures across selected countries (US $ millions). (Source: [21], Korea (1993), Dr Tae
Ik Mheen, Director Genetic Engineering Research Institute (personal communication).) Country
Year
Total R&D expenditure
% of GNP
Researchers per R&D per researcher loo00 population (US$ 1OOO)
USA
1989 1991 1989 1991 1989 1990 1989 1991 1987 1988 1988
140486 151600 76049 101557 3980 448 1 2094 3175 3413 3873 2494
2.69 2.67 2.69 2.77 1.92 1.91 1.38 1.70 0.96 1.01 0.7
39 41 37 41 16 16 20 23 3 4 3.12
Japan Korea Taiwan China India
142 -
165 186 60 64 63 69 9 9 20.79
evident from these positive trends both in industrialised countries, and increasingly in lagging countries such as India and China where the number of researchers as a percentage of the labour force is still quite low, indicating a need for a large injection of funds in this direction. In the countries surveyed here, biotechnology policies were implemented relatively early among developing countries, namely in the early to late 1980s. This closely followed the pattern established in industrialised countries, i.e., national guidelines outlining priority areas of research as well as funding for basic research and training. National programmes on biotechnology have by and large been coordinated by national government departments or centres and laboratories of excellence. The former, as in the case of India, Thailand, China and the Philippines, are largely semiautonomous public sector bodies set up for coordinating national biotechnology policies, and disseminating funding for priority projects. These national coordinating agencies, the National Centre for Genetic Engineering and Biotechnology (NCGEB) in Thailand, the China National Centre for Biotechnology Development (CNCBD) in China, and the Department of Biotechnology (DBT) in India, were all set up in the 1980s to perform the important task of implementing national policies on biotechnology and are in charge of overseeing national projects in priority areas. The Philippines' National Plan on biotechnology which outlines its priority areas of agriculture, aquaculture, health, industry and the environment, was approved in 1990 by the Science and Technology Coordinating Council (STCC), which implements and coordinates national science and technology policies. In the case of Taiwan and Korea, the task of developing capabilities in biotechnology have been left up to National Centres of Excellence. The Development Centre for Biotechnology (DCB) in Taiwan and the Genetic Engineering Research Institute (GERI) in Korea, attract some of the best scientists in the country and along with their own R&D also provide assistance, especially to small companies with insufficient facilities to do their own R&D."
477 The importance of involving the private sector, especially through cooperation with basic research strengths, was recognised at an early stage and for some countries, especially Taiwan and Korea, the USA was the role model. Most, with the possible exception of Korea, whose previous experience with industrialisation led to the formation of large companies who were well equipped to deal with the large sums of up-front investment needed for modem biotechnology, recognised the need for investing in applied research. Public sector institutions, while possessing the capabilities for basic research, had in many cases very little experience with commercialization [23]. On the other hand, the relatively small size of the private sector, when considering the size of investments that had to be made, made it difficult for private companies to operate alone in many cases. This recognition has led to efforts to try and pool the resources of these two sectors. In Korea, the Korean Genetic Engineering Research Association (KOGERA) and in India, the Biotechnology Consortium India Limited (BCIL) were formed to deal specifically with this task. KOGERA, to date, has been relatively successful in harnessing the potential of the private sector for large national projects, especially concerning environmental biotechnologies. The BCIL was only formed in 1991 and it remains to be seen how successful it will be in fulfilling its task. At present its activities are confined to providing information to the private sector on investment possibilities in India. Online facilities which provide information on technical matters and patents, as well as market size for individual technologies and products, are also available at the BCIL and also through a new on-line service set up by the Technology Information and Forecasting System which provides sectoral information.” Other countries, rather than setting up new organisations, have used a previously established infrastructure to tackle this problem. In Taiwan, for example, the DCB’s role is also to encourage and improve cooperative research between national universities and research institutions such as the Academia Sinica, and private companies, and has been instrumental in commercialising a number of products developed at the DCB through the private s e c t ~ r . Similarly, ’~ in China the torch and spark programmes which were established to commercialise technology and diffuse it amongst the rural population, respectively, have been used for biotechnology as well. Thailand’s NCGEB has also been instrumental in performing this task, while in the Philippines the National Institutes of Biotechnology and Applied Microbiology (BIOTECH), is not only a national laboratory for biotechnology, with a large percentage of its staff seconded from academic institutions around the country, but has also been rather successful in commercialising a number of products in recent years.
2.2 Research and training One of the most uphill tasks faced by many developing countries is that of acquiring qualified personnel. Biotechnology, although science based, straddles a number of different research areas, and breaks down traditional boundaries between natural science subjects. Most countries in Asia, until very recently, did not include specific
478 biotechnology courses in their national curricula. A major portion of government funding for biotechnology in recent years, therefore, has gone into designing and setting up courses in biotechnology. As a result, a number of universities in the region offer training courses or degrees in biotechnology. In Korea, other than the GERI, of the 103 or so universities and colleges in the country, 17 universities have initiated new departments of biotechnology, both at the graduate and undergraduate levels. In addition, 15 universities and colleges had established biotechnology or genetic engineering centres on their campuses by 1992 [24]. Taiwan, on the other hand, has relied more on sending students abroad for training, at least in the short term. The growth of biotechnology in the country in the last decade or so has, however, lured a number of these researchers back and their expertise is being used to set up courses and a research network in T a i ~ a n . ' ~ Most of the other countries surveyed, however, have been unable to overcome the problem of training as quickly, for two main reasons. First, many of them have not built up sufficient strength previously in science and technology, especially in terms of human capital. Second, a lack of financial resources have created major blocks to further investment in training or in collaborative projects with foreign universities or training institutions. There are other complications added to this problem, e.g., in China, while there is a concerted effort to improve training in biotechnology, especially through an exchange of researchers and collaborative work, the major problem still remains that of labour mobility, where opportunities for scientists to move from one research institution or university to another are limited. This creates obvious bottlenecks and reduces incentives for research and further development of biotechnology. However, the new reform programmes are attempting to address these problems along with many others also relating to training. Those with some degree of previous capability, e.g., India, have been somewhat more successful than Thailand or the Philippines, where a major shortage of skilled staff continues. The DBT in India has thus far helped over 20 universities across the country to establish postgraduate training programmes in biotechnology. Short-term training courses (2-4 weeks each), especially for those already working in the field and requiring an update on new biotechnology techniques as well as fellowships to study abroad, are offered each year. Thailand and the Philippines still tend to rely largely on foreign training for their researchers, although the longer term goal is to develop local programmes of a similar nature. The Philippines at present makes efficient use of its few human resources through a system of exchanging and sharing researchers between research institutions and universities, while developing training programmes in biotechnology. Access to knowledge, especially information about regional R&D programmes and research on matters of common interest, such as disease control and prevention, is vitally important in a fast-moving field such as biotechnology. Although still in a stage of infancy, this method is also being pursued by a number of countries in the region, firstly through the establishment of centres to promote collaboration between researchers at an international level and also to exchange information and technologies between countries. One recent example of this is the China-EC Biotech-
479 nology Centre which aims to improve research in agricultural and medical biotechnologies through collaborative research between Chinese and European scientists [25]. Similarly, Thailand set up the US-Thailand Commercialisation of Science and Technology Programme (UST/COST) in 1990 whose participants include the US Agency for International Development, the Board of Science and Technology for International Development (BOSTID) of the US National Research Council and Biotechnology International, a programme at the University of Maryland in the USA [26]; secondly, to overcome the problem of access to information, a number of countries in the region have also established biotechnology databases and computerised networks which can be accessed by researchers and industrialists across the nation. In India, for example, the BCIL has established the Bio-Informatics Network in India which has contributed greatly to improving the flow of information to Indian researcher^.'^ Similarly, in Thailand the NCGEB ’s documentation centre includes international newsletters about developments in the field of biotechnology, both in terms of scientific breakthroughs as well as policy changes [29].
2.3 Industrial biotechnology in the region In most industrialised countries today, biotechnology, especially applied biotechnology R&D, is dominated by the private sector, where either the small biotechnology company, as in the US, or the large multinational firms dominate, as is the case especially in Europe. In developing countries, as argued below, the private sector tends to be smaller and the technology less sophisticated. This, combined with little or no access to investment capital, has meant that the private sector has been slower to respond to the potential of biotechnology. Nevertheless, a number of successes in recent years, especially in simpler technologies, have gradually revealed the nature of the relative strengths and weaknesses of the private sector and its potentials and problems, and have also resulted in a greater emphasis on encouraging private entrepreneurs to develop biotechnology-based products. This is not to say that the public sector has not had minor successes as well. However, the problems of the public sector tend to be compounded by political considerations and often miscommunications between different government departments which tend to have different goals. Increasingly, therefore, the private sector, where it is able to invest and add to technological capabilities, has shown that it is more efficient in a number of sectors.16 The goal of this section is to examine the technological capabilities of these countries in bringing a product onto the market. Hence, much of the analysis concentrates on the private sector as the decision-making or -breaking process in the public sector is far more complex and beyond the scope of this present paper.” Of the countries in the region, the private sector is perhaps most well established in Korea and Taiwan. The importance of Korea’s private sector is apparent from the size of its total contribution to biotechnology R&D which now exceeds similar government contributions (Table 2). Private sector investment grew at an average annual rate of 44.5%, as compared to the government contribution which, at 29.4%, was almost half that of the private sector during the 1980s.
480 Table 2. R&D funding for biotechnologyin Korea (US $ OOO). (Source:[24] and Tae Ik Mheen, Director, GEM (personal communication).) Source
1983
1984
1985
1986
1987
1988
1989
Growth per year (%)
Government Private Total
2860 6857 9717
4143 16571 20714
5636 22429 28065
6143 22714 28857
8143 27430 35573
11552 33871 44014
17469.4 n.a. n.a.
35.92 44.5 40.22
Note: n.a. = not applicable.
Korea has a relatively well-established industrial structure which was established to help the country industrialise in the 1970s and 80s. Biotechnology has been incorporated within this industrial structure. The two kinds of large firms which dominate industrial production of biotechnology products are the chaebols or the large industrial houses and the pharmaceutical companies. Many of these companies have reached an advanced stage of technological and scientific capability and are now investing in industrialised countries, either through direct buy outs or collaboration at the industrial and the research level. For example, Cheil Sugar, which is part of the Samsung Group, established Eugenetech in New Jersey which does research in cell biology and develops cell lines for the production of new drugs (e.g., alpha interferon). Lucky Limited, owned by the Lucky Goldstar Group, set up Lucky Biotech in California in collaboration with the US biotechnology company, Chiron. The research skills, primarily in cloning and growth hormones, are concentrated in the subsidiary company which probably benefits immensely from collaboration with researchers in the USA, while the results are then sent back to the parent company in Korea, for further development and scale-up where production costs are lower [30]. In the pharmaceutical sector the largest company, Dong-A, controls 10% of the Korean market, while Chong Kun Dang and Yuhan have a market share of about 5% each [30]. The sector most active in biotechnology research in terms of sales is fermentation, where Korea has a comparative advantage arising out of traditional biotechnology applications in this industry. Output in the pharmaceutical sector, although less phenomenal, is also growing extremely rapidly. Vaccines and diagnostics especially have shown high average growth rates and it is estimated that by the year 2000 Korea will produce 2% of the world’s biologically produced pharmaceuticals [ 101. Although comparable figures on government and private sector investment in biotechnology are more difficult to obtain for the other countries surveyed, with the possible exception of Taiwan, the private sector is in general less active in biotechnology R&D in terms of absolute size in the region. Even in Taiwan, it appears that funding is dominated by the government, especially in the area of applied research, demonstrating perhaps a reluctance on the part of industry dominated by small- and medium-sized enterprises, to invest in biotechnology research [31]. Size must also play a role in investment in biotechnology because of
48 1 the complex nature of the technology itself and the up front investment required for R&D. This is especially true of the pharmaceuticals, where technology life cycles are becoming shorter and size more important as a result. Countries like Taiwan, which have consistently emphasised small size, have had a problem with innovation in the industry, which tends to be low. In contrast, the country has a very high skill level. The ability to compete internationally in cutting edge technologies therefore exists in the country, but size in this sector appears to have been a major obstacle. Therefore, in order to encourage investment in biotechnology R&D, the private sector in Taiwan has needed a significant amount of support either from the government or foreign companies to raise capital for investment (Table 3). Table 3. Breakdown of biotechnology R&D in Taiwan in NT$ in 1991. (Source: [31].) Total
Basic research
Applied research
Experimental
2765 (% of total)
1116 (40.4%)
1423 (51.5%)
226 (8.1%)
In terms of total sales, Taiwanese biotechnology companies recorded US $22 million in 1987. In 1989, annual sales totalled US $44.67 million for pharmaceutical products, US $279 million for foodstuffs, US $65.96 million for agricultural supplies and US $13.83 million for servicesI8.By the year 2000 Taiwan is aiming to produce about 2% of the world market in biotechnological products [lo]. Projections made by the National Science Council in Taiwan (Table 4) also show the market value of a number of sectors into the mid-1990s. For example, in the case of antibiotics and diagnostic reagents, the size of the market has expanded considerably, resulting also in an interest in developing capabilities in these areas."
Table 4. Market projections of products in Taiwan NT$ (US$) millions. (Source: [32].)
Product
1990
1991
1992
1993
1994
1995"
1996"
Tissue culture
4000 (151) 700 (26.3) 120 (4.5) 300 (1 1.3) 10 (0.38) 20 (0.75) 120 (4.51)
5000 (185) 1000 (36.9) 200 (7.38) 400 (14.7) 30 (1.11) 40 (1.48) 160 (5.9)
6000 (232.6) 1200 (46.52) 300 (11.63) 580 (22.48) 60 (2.33) 80 (3.1) 200 (7.75)
8000 (305.3) 1300 (49.62)
loo00 (371.8) 1400 (52.05) 600 (22.31) 800 (29.74) 300 (11.15) 400 (14.87) 300 (11.15)
12000 (446.1) 1500' (55.76) 1000 (37.18) 900 (33.46) 500 (18.59) 600 (22.31) 360 (13.38)
16000 (594.8) 1600 (59.5) 1500 (55.76) 1000 (37.18) 800 (29.74) 800 (29.74) 400 (14.89)
HepatitisB andvaccines Medical instruments Waste water treatment Speciality chemicals Antibiotics Diagnostic reagents
"Exchange rate for 1994 (US$ 1 = NT$ 26.9) used.
400 (15.27) 700 (26.72) 120 (4.58) 160 (6.11) 250 (9.54)
482
There are today a number of small biotechnology companies in the country which are doing research in diagnostics, especially relating to hepatitis. It remains to be seen, however, whether their success will be sufficient to encourage established pharmaceutical companies to increase investment in R&D and make the sector innovative. The success story in agricultural exports in this region has been that of orchids from Thailand. Although the technology is relatively low-tech in the traditional hierarchy of techniques in biotechnology, namely tissue culture and cloning, the success of Thai orchid growers and exporters is phenomenal. A large number of these private companies are concentrated around Bangkok with an estimated annual average turnover of about US $20 million. To date, indications are that this market, especially in industrialized countries, is likely to continue growing for a period. Moreover, since Thailand’s food-processing industry has been expanding at the rate of about 20% per year, it is likely that government support, as well as private investment in agricultural biotechnology, will grow in order to take advantage of this value-added market [33]. Present government policy, in addition to emphasising other export crops such as rubber and rattan, recognizes this success by encouraging R&D in temperate flowers for which there is a large regional market. In India, private firms have also been most successful in the agricultural sector. Indian producers of ornamental plants and flowers have largely followed Thai exporters by concentrating their efforts in ornamental plants and orchids. India’s local market, however, islarge enough to absorb much of this surge in production and with a growing middle class, local companies are finding tremendous success in India itself (see Table 5 for a list of some of these companies). There are a number of companies which have built up their export markets as Table 5. Important tissue culture companies in India. (Source: [34].)
Name
Investment Foreign (US $ mil) collaboration
A.V. Thomas, Cochin 1.35 Indo-American Hybrid 2.9 Seeds, Bangalore Unicorn Biotech 0.697 Hyderabad
Bio Tissue laboratory 0.232 Hyderabad Harrison Malayalam, 7.74 Bangalore SPIC Madras/ Coimbatore ITC Agrotech, Hyderabad
7.74 1.94
Crops
Turnover Capacity (US $ mil.) plantdyear
Phyto Nova Shell, Cardamom, Banana, 9.675 6 million UK Lillies, Ornamental Sunkee, Australia Banana, Ornamental 1.51 10 million plants Godrej, Hindustan Banana, Strawberry, 0.58 total 6 million Lever Ornamental order from Holland, Italy APIDC, IDBI, Banana, Orchids, CVF Roses Semundo Saatzucht Vegetable seeds, and Agro Saten orchids, banana (Germany) Ornamental plants Continental of Rains Australia
-
5 million
-
10 million
-
10 million
Ornamental plants, cash crops, oil seeds
483
well. This is largely in response to the incentive system built up by the Indian Government, including export processing zones and tax incentives to companies to export. Two of the most successful tissue culture companies in India have now established separate biotechnology divisions or laboratories where tissue culture techniques are being applied to a large number of indoor plants and commercial plants and although the size of the initial investment and sales is still small relative to industrialized countries, considerable interest has been generated as a result of the initial successes of tissue culture companies. Another sector where private activity is increasing in India is in the pharmaceutical sector. While most of this research is geared to the local market, such as diagnostic kits, which have been developed for a number of diseases, there appear to have been minor successes on the export front as well. Genei Limited, for example, is the country’s first manufacturer of indigenously designed recombinant DNA research tools. It presently exports some of its products to the USA. The Indian pharmaceutical company Cipla is presently producing and exporting the AIDS drug AZT, to other developing countries in the region.20Inward investment as a result of liberalisation policies has also increased. A notable example is the Astra Research Centre which was established in Bangalore in the late 1980s. The Research Centre is wholly owned by the Swedish multinational Astra and its main goal is new drug discovery. The Centre has, however, produced a number of products and technologies over the years, many of which have been transferred to local companies. The main attraction for Astra in investing in India appears to have been the presence of a highly skilled pool of labour, which is relatively cheaper than in most industrialised countries: The location was also a deciding factor for a company, with the proximity to the Indian Institute of Sciences being a key reason for setting up a research centre in Bangalore. In China, the government has attempted private sector investment in biotechnology through its two programmes, namely the “torch programme” whose goal is the commercialization and industrialization of biotechnology largely through the private sector and foreign collaboration and the “spark programme” which aims to bring biotechnology to its vast rural population. Overall funding and coordination of biotechnology research in China is carried out by the China National Centre for Biotechnology Development (CNCBD). All public funding for biotechnology is initially channeled through the CNCBD which following the advice of its reviewing panels, which include scientists and policy analysts, allocates the funds to appropriate and priority projects across the country. At present there are over 100 research institutions across the country that are funded in this way by the CNCBD. The priority areas at present include agriculture, pharmaceuticals and protein engineering for industrial use. In the pharmaceutical sector, regional diseases such as hepatitis B are a major worry and research centres are actively collaborating with local and foreign companies to develop vaccines and diagnostic products. Foreign support has also been sought for the production of vaccines for a range of diseases and the World Bank has recently funded a project for the production and large-scale commercialisation of vaccines. In agriculture, there is less of a tendency for private sector involvement and the
484 government funds research projects mainly in genetic engineering applications. Rice, being one of the main staples of the country receives a proportionately high level of funding and there is close cooperation with the International Rice Research Institute (IRRI), based in the Philippines and responsible for some of the region’s most radical innovations in rice, through the Rockefeller foundation. On a more general level, the government’s priority areas include disease resistance, nitrogen fixation, animal genetic engineering especially in pigs and fish, and research projects at a more fundamental scientific level are also funded. The major problem the country is facing today is the commercialisation of biotechnology which has been less successful than the basic research. This indeed is a common problem to most other countries in the region, and a dynamic private sector has often been the necessary ingredient to fill the gap left by the public sector. In both Thailand and the Philippines, agricultural biotechnology is likely to continue to dominate production and exports. There is also a large potential market in the industrialized world for exotic fruits such as star fruit, rambutan, durian, shiitake mushrooms and Amaryllis. Thailand is presently doing tissue culture research on a number of other agricultural products such as palm oil, rubber, rattan, bananas and rice. Similarly, the Philippines is encouraging tissue culture research in agricultural products which have a large potential export market, such as coconut, rattan and bamboo, the first two forming major exports. Biotechnology can be used to enhance production and strengthen already strong markets or to exploit potential ones. There is also considerable demand in south Asia for a number of temperate climate crops such as strawberries, asparagus, Carnations and roses, some of which are already being produced in high altitude areas in this region. However, biotechnology research which enables the adaptation of some of these varieties to certain climatic regions, can potentially ensure their all-year-round growth. Thus, especially for those countries with a large agricultural base and a diverse climate, the potentials for biotechnology-based agricultural exports are enormous. However, the key to successful development is matching production and marketing capabilities [35]and Thailand made effective use of this in the 1980s. Another area where there is great scope, especially for China, Taiwan, the Philippines and Thailand, is in aquaculture. Until recently, aquaculture in Southeast Asia and other countries included in this survey, such as India, was largely dependent on traditional techniques such as controlling alkalinity, oxygen content and traditional breeding methods. Recently, however, biotechnology methods have increased their attraction by improving productivity and product quality. Since the first international symposium on marine biotechnology held in Japan in 1989, there has been growing interest in the use of modern biotechnology techniques to improve marine output. The market for shrimp and other aquaculture products, it has been estimated, has been growing since 1970 and is expected to continue to grow at about 5% per year, placing pressure on existing shrimp farmers to improve productivity. With the exception of India, most of the large producers and exporters of aquaculture-based products in the region, have registered high growth rates in aquaculture production (Table 6), and the Indian government hopes to catch up through biotechnology,
485 Table 6. Aquaculture production in selected Asian countries (lo00 tons). (Source: [36].) Country
1980
1986
1987
1988
1989
Average annual growth rate 1980-89 (8)
China India Philippines Rep of Korea Thailand
2552.9 413.0 289.2 541.6 94.6
5048.1 416.3 470.9 993.6 128.4
5705.2 427.3 561.0 876.8 174.5
6658.7 437.1 599.5 900.3 219.1
6557.8 490.0 629.3 859.8 215.8
11.1 1.9 9.0 5.3 9.6
especially to produce shrimp, prawn and carp. The steadily growing aquaculture market has increased competition between a number of countries, with many of them stressing the use of biotechnology in aquaculture to improve productivity. Countries more advanced in biotechnology, such as Taiwan, will probably be quicker to benefit from this expanding market. Environmental biotechnologies have become important in recent years as concern about the environment rises and the need to reduce pollution grows. Developing countries are increasingly looking to biotechnologies for cheap and effective ways of reducing pollution. Of the countries surveyed here, a number are using biotechnology for this. In Thailand, the NCGEB has funded projects in bioleaching, biogas production and pollution-combatting biotechnology at KMITT in Bangkok where the facilities include a biogas plant which will shortly be ready for commercialization and scale-up. In Korea, years of intensive farming combined with extensive use of chemical fertilizers has severely polluted the groundwater and soil. Research in biofertilizers is being encouraged by the government to reduce this pressure on the land. In addition, current environmental research emphasizes the development of environmental effect analysis. Spirulina is presently being used at a commercial level in Thailand to clean waste produced by the country’s many starch factories. The National Institutes of Biotechnology and Applied Microbiology (BIOTECH) in the Philippines has also successfully commercialized some of its products, such as inoculants used for fertilizers in reforestation projects and the development of thermophilic and mesophilic anaerobic fermentators to produce biogas from distillery slops thereby reducing pollution. Biogas plants have been in use in China and in India for some.years. In both countries, but especially in China, the plants have helped to reduce pollution while providing energy for electricity and gas, and fertilizer to the farmer. India’s other major projects on environmental biotechnology include the biotechnological conversion of methane to methanol and microbial desulphurization of fossil fuels. In addition, the rapidly deteriorating quality of water supplies has encouraged the development of DNA probes for detecting viruses and bacteria in sewage waters and water supplies. Reforestation has also become a priority in recent years, especially for developing countries, where pressure on the land is reducing the forest cover drastically. A number of the countries surveyed here are presently developing a wide range of
486 indigenous tree species through tissue culture. In India, the Tata Energy Research Institute (TERI) and the National Chemical Laboratory have been funded by the Department of Biotechnology to develop improved tissue culture varieties of bamboo, eucalyptus and acacia. The Philippines has identified reforestation as one of its priority research areas. Unlike other sectors, however, the market for environmental technologies still appears to be dominated by the public sector, perhaps because the returns to investment are lower than in other sectors, notably pharmaceuticals. Thus it appears that the private sector has largely determined its own priorities and often the public sector or policy making body has had to follow by offering funding for further research. This is especially so in agriculture and moreover in commercial agriculture, where private companies have been quicker to take advantage of the opportunities offered by export markets. This is perhaps not surprising, especially since the returns on subsistence agriculture, pharmaceuticals and industrial biotechnologies are longer term. Tax holidays and the creation of exclusive export processing zones have also played a role in encouraging investment from the private sector, especially in particular industries. In Taiwan for example, companies investing in biotechnology are provided a 5-year tax exemption as well as tax credit for R&D. In Korea, foreign investment is encouraged by providing tax exemption on technology and also on customs duty. Local investment, especially for small companies, is provided in both countries by venture capital funding which, as discussed earlier, has been more successful in Taiwan, perhaps because of an older emphasis on small- and medium-sized enterprises. In India and China also, both local and' foreign private investment is being encouraged by the formation of exclusive export zones, where 100% of production for a certain number of years is exported. Lower customs tariffs on raw material imports such a? basic enzymes have also played a role in encouraging biotechnology research, especially in India where economic liberalization policies are reducing restrictions on international trade and investment. The setting up of regions or areas of excellence, most notably the phenomenon of the Science Park has also recently found favour in many developing countries. Taiwan, for example, has set up the Hsinchu Science Park which is located 80 miles south of Taipei and contains many of the new biotechnology companies which were formed in the 1980s [37]. Similarly, Daeduck (or Taejon) Science and Industrial Park in Korea, located in the centre of the country, is home to many of the large companies involved in modem biotechnology research. In India, although a number of export zones have been created by the government, the city of Bangalore in the south has emerged as a centre for scientific research, and a number of biotechnology companies, both old and new, have established laboratories and offices in and around the city.*' The government of the southern Indian state of Kerala has also proposed the development of a Science Park in the state. Thus, governments are increasingly taking a back seat, as applied research becomes the domain of the industrial sector, whether it be private or public. Increasing financial returns from the simpler biotechnologies such as tissue culture are also encouraging companies to invest in more expensive and advanced techniques in agriculture, and increasingly in industry
487 and the true potential of these companies may only become apparent in a few years from now. Where the government still has an important function to fulfill is in funding basic research and training, information acquisition and dissemination and providing a healthy economic environment to enable the rapid growth and diffusion of biotechnology.
3 Biotechnology in Africa 3.1 Status of biotechnology
Biotechnology research in the region is scattered across the institutional terrain with varying levels of activity, funding, expertise and experience. The work is carried out mainly in universities, national research centres and international research institutions located or operating in the region. There is also evidence that the private sector (including parastatal organizations) is becoming interested in the prospects of biotechnology and is starting to fund public research institutions to conduct biotechnology research. In most African countries there are no institutions expressly charged with the mandate of coordinating research on biotechnology. Zimbabwe is the country in the region that has established a unit to coordinate and oversee biotechnology R&D. However, feasibility studies have been carried out in countries such as Kenya and Burundi to explore the possibility of setting up biotechnology centres. There are other efforts aimed at consolidating biotechnology R&D under existing institutions. Despite these efforts most of the biotechnology activities are undertaken by individual scientists. Most of the current biotechnology R&D activities are focused on improving productivity in the agricultural sector. The direction of biotechnology research in Africa is influenced by the traditional research agenda. The research is conducted in public universities and in some of the national agricultural research centres. A number of international research organizations located in the region are also engaged in biotechnology R&D. Biotechnology R&D activities are still in their infancy in Cameroon. The application of biotechnology is largely in the agricultural (crop production and improvement) and livestock sectors. In the area of crop production and improvement research is being done on the application of tissue culture techniques to breed for high-yielding and disease-free varieties of coffee at the Institute of Agriculture and Forestry Research (IRA). At the Institute of Zootechnical Research work on the development of diagnostic kits for livestock diseases has been initiated. Embryo transplantation research is carried out at the same institute to enhance the productivity of local cattle. In the domain of medical biotechnology, the Institute of Medical Research and Medical Plant Studies has developed research to screen for medicinal chemicals and identify medicinal properties of some of the herbs using biotechnological techniques
488
such as the ELISA techniques. While the country is engaged in a number of biotechnology R&D activities, it has not formulated a specific coherent body of policy to guide the evolution of the technology in the socioeconomic system. The country established a Biotechnology Center at the University of Yaounde whose main responsibility includes the identification of the country’s research priorities and needs in the domain of biotechnology and the coordination of all biotechnology R&D activities. Currently, the centre is inactive because of lack of funds and a specific coherent policy to guide its activities. Other institutions engaged in biotechnology related research activities include: - The Institute of Zootechnical Research; - The Institute of Medical Research and Medicinal Plant Sciences; - The Institute of Agriculture and Forestry Research; - The University of Yaounde (Faculty of Agriculture and the School of Medicine); and - The Institute of Human Sciences (dealing with policy research). In Kenya most of the agricultural biotechnology R&D activities focus on improving the yield potential of cereals and some of the export crops such as coffee and pyrethrum. Institutions engaged in agricultural biotechnology R&D in Kenya include the Kenya Agricultural Research Institute (KARI), the Department of Crop Science of the University of Nairobi, the National Potato Research Centre (NPRC), the Genebank of Kenya and the Jomo Kenyatta University of Agriculture and Technology (JKUAT). Agriculture biotechnology R&D in Kenya focuses on the application of tissue culture and clonal techniques on crops such as potatoes, pyrethrum and tea. In vitro mass propagation of potatoes, ornamentals, bananas, pyrethrum, sugarcane, pawpaws, coffee and citrus fruits has been undertaken at KARL Research on viral disease eradication in potato and cassava using meristem culture has been undertaken at the National Plant Quarantine Station (NPQS) while the eradication of citrus-greening disease by ovule and nuclear embryo culture is carried out at the Department of Crop Science at the University of Nairobi. Kenya is the largest exporter of pyrethrum extract in the world. The country currently produces about 7,000 tonnes of dry pyrethrum flowers per year while the world demand is about 16,000 tonnes. On the production side, there is an annual demand of 10 million plants for planting while the nurseries are able to supply only about 2 million plants. The reason for this lies in the slow production process for the planting material. It has been established that tissue culture biotechnology can resolve problems of slow growth and disease susceptibility. It is also reported that farmers’ attitudes to growing pyrethrum are likely to be improved because of the better returns they get by growing superior clones developed through tissue culture. For these reasons, KARI is collaborating with the Pyrethrum Board of Kenya to meet demand for high-quality planting material through tissue culture work such as cloning and seed propagation and hence meet the worldwide demand for pyrethrum products [17]. In Kenya biotechnology R&D has also focused on the improvement of livestock
489
with an emphasis on cattle breeding, development of vaccines and diagnosis of diseases. KARI has developed a specific DNA probe for the diagnosis of heartwater disease. There are also research projects to develop vaccines for rinderpest, anaplasmosis and babesiosis through biotechnological techniques. The projects have provided Kenyan scientists with valuable experience in molecular biology. There have been attempts to engage in embryo transfer techniques in cattle breeding. Though these efforts are at infancy, they offer great potential for the improvement of local cattle herds. Some of these local breeds have shown some resistance to the Huemonchus contorlus worm. Related developments in the area of human medicine have been immunological research at KEMRI in the tropical diseases of bilharzia and malaria. As mentioned above, biotechnology R&D is more established in Zimbabwe than in most of the sub-Saharan African countries. Zimbabwe has made significant efforts to define target areas of biotechnology. The Department of Crop Sciences at the University of Zimbabwe has been applying tissue culture to develop disease-free varieties of coffee, potatoes and tomatoes. Elite coffee bushes have been cloned using the leaf disc technique of Staritsky. The Tobacco Research Institute in Zimbabwe has over the last decade been using pollen culture to incorporate resistance to two troublesome diseases in a new variety of tobacco. Research is underway to introduce resistance to other diseases in tobacco using somaclonal variation. It is notable that tobacco has been a model plant for biotechnology research and Zimbabwean scientists have had access to the latest techniques. This is also the most important export crop for the country and therefore it has received special research attention. Biotechnology developed in Zimbabwe is of considerable regional interest not only because the country has a well-developed research infrastructure, but also because Zimbabwe has been given the responsibility for food security in the Southern African Development Co-ordination Conference (SADCC). This responsibility provides a framework for transferring technology developed in Zimbabwe to other SADCC countries. There is already a tradition of technology transfer from Zimbabwe to other countries in the region. In Ethiopia biotechnology-related R&D activities are limited to a number of projects focusing on the improvement of crops and conservation of plant genetic material. Research on tissue culture was started at the Plant Genetic Resources Centre (PGRC) in 1989. This research was, however, discontinued due to lack of funding. The application of tissue culture techniques to forestry is one of the main areas of research of the Department of Biology of Addis Ababa University, which has started working on various indigenous tree species. The main aim of the research is to generate seedlings of tree species that are difficult to generate under natural conditions. Efforts are also underway to multiply indigenous tree species for subsequent planting in appropriate agro-ecological conditions. In Uganda tissue culture is underway at Makerere University and at Namulonge and Kawanda research stations. The Department of Crop Science at Makerere University is developing tissue culture protocols in order to derive high-yielding, stress-tolerant and disease-resistant plants from calli induced from explants of various
490 local crops. There are four Ugandan postgraduate candidates training in tissue culture in Europe [ 171. One of the most important sources of biotechnology in the region has been international research centres such as the International Laboratory Research on Animal Diseases (ILRAD). The centres have acquired considerable expertise in specialized areas of biotechnology research because of their access to international R&D. Given the technical and financial resources available to them, these centres have maintained access to some of the latest techniques in the biotechnology. This has been enhanced by their commitment to specific areas of research which allows them to accumulate experience and institutional memory. All the CGIAR Centres are engaged in certain types of biotechnology. ILRAD’s mandate is to work on East Coast Fever (ECF) and typanosomiasis, which are reported to account for an annual loss of 3 million cattle a year. These diseases have prevented livestock from being established in some 7 million km2 of land, which could support over 120 million cattle and an equal number of small small ruminants. In dealing with these diseases, ILRAD has embarked on a number of biotechnology research programmes aimed at developing vaccines. In order to carry out research effectively, ILRAD has had to develop diagnostic methods. It has developed species-specific monoclonal antibodies against three African trypanosomes species that infect livestock. These antibodies are now being used to identify, isolate, and purify the corresponding antigens used in serodiagnosis. ILRAD has also used antibody technology to type bovine lymphocyte populations and subpopulations. The reagents have been vital in characterising the kinds of cells infected by Theileria parva (T.parva). In addition, monoclonal antibodies against T. parva intralymphocytic developmental stages have shown that distinctive strains of T. parva exist. Similar advances have been made in the in vitro propagation of African trypanosomes. ILRAD has developed and standardized an in vitro system to support the production of Trypanosoma congolense (T. congolense) metacyclics. By 1985 eight lines of cultures for producing the metacyclics had been established of which six were from cloned stocks. In its work, ILRAD has had to use genetic engineering techniques, especially to develop molecular hybridization reagents for epidemiological studies of T. congolense and Trypanosoma vivax (T. vivax). ILRAD has synthesized complementary DNA (cDNA) copies of mRNA derived from four clones of T . congolense for a library. ILRAD has pushed its research further by looking at the genetic basis for resistance against the African trypanosome. The aim of the research is to identify genes that code for resistance and transfer them to the cattle germ line. ILRAD is therefore screening local cattle breeds and has acquired the resistant West Africa NDama through embryo transfer. To facilitate research, the Centre has used superovulation techniques to reproduce the N-Dama. One of the most significant achievements of ILRAD is the mapping of the genome of T. parva parva, which is organized in four chromosomes. A restriction map of the genome is nearing completion. There are a number of nongovernmental research institutions working on
49 1 biotechnology in the region. Some NGOs, e.g., the Manor House Agricultural College in western Kenya, have plans to set up a tissue culture laboratory. Others, such as ACTS, have restricted their work to policy studies and information dissemination. The African Academy of Sciences (AAS) has in the past carried out fact-finding missions. Other NGOs with projects in the region include the African Biosciences Network (ABN).
3.2 National biotechnology policy and priorities The brief sketch of biotechnology activities above shows that in most African countries there are no explicit policies and plans governing biotechnology R&D activities. Most of the ongoing activities are treated in the broader framework of science and technology policy. Kenya and Zimbabwe have engaged in processes that are meant to articulate national priorities in biotechnology. However, these processes do not clearly reflect the needs of the majority of the population. The policies governing science and technology are both implicit and explicit. Countries such as Kenya and Tanzania have developed and published policy papers which articulate the direction of research in the countries. Only Ethiopia has recently prepared a detailed science and technology plan. The rest of the countries deal with issues of science and technology as part of overall national policy and planning. Policies on biotechnology are therefore treated as part of the overall research goals. Ethiopia has began considering the integration of biotechnology in its national science and technology policy regime. The country has made significant attempts to develop a detailed science and technology policy. The draft policy document identified the following as priority areas: agriculture, environmental management, developing alternative energy sources, health, industry, transport and communication and emerging technologies. “Emerging technologies” dealt with in the document are biotechnology and electronics. Cameroon’s science and technology policy is relatively underdeveloped. The country’s Sixth Five-Year Development Plan 19861991 is more explicit on issues of science and technology. The Plan calls for the development of a national research infrastructure that enhances the acquisition and utilization of technological knowledge in the following areas: chemical engineering and biotechnology, food processing and preservation, metallurgy and the preparation of alloys. However, virtually all the science and technology issues raised by the plan are yet to get political legitimacy. Forje remarks, “[tlhat grandiose plan has yet to leave the dusty drawers of the Ministry of Planning and Development. On the whole, the country lacks a coherent corpus of policy to guide the evolution of biotechnology. What the country needs is an articulate or explicit corpus of policy on science and technology: a science and technology policy that clearly articulates national socioeconomic aspirations”.” Zimbabwe has prepared a science and technology policy document which identifies a wide range of areas where science and technology will play significant roles so as to transform the economy. The areas identified include agriculture, forestry, wildlife, environmental conservation, communication, health, mineral
492 resources, energy and industry. The document notes that the goals will not be met just by formulating the policy and creating institutions but that an appropriate instrument for planning, coordination and management is required so as to realize the potential of these research institutions. However, the draft document does not make efforts to suggest an institutional framework that will enhance the implementation of the policy measures. On the whole, most countries of the region have insufficient plans and policies to guide the evolution of biotechnology. There are several obstacles to the process of formulating biotechnology plans and policy in Africa. First, the level of expertise in issues of science and technology policy in particular or public policy in general is relatively low. In Cameroon only two people have acquired postgraduate (doctoral) training in science and technology policy. There are eight for Zimbabwe and four for Uganda. All the Ugandans who are pursuing the course are based in Europe. Secondly, there is limited exchange among those qualified in scientific and technological fields. This problem could be addressed by the establishment of academies of sciences or similar scientific and technological think-tanks. Moreover, the scientific community has not made biotechnology accessible to its end-users or those most likely to be affected by its developments such as farmers, healers and the fishermen. Third, the success of the process of science and technology policy formulation in any African country largely depends on the nature of the political system in place. A favourable political system should allow and guide a harmonious interaction between science and technology activities and institutions on the one hand and the regulatory system on the other. Political space, broadly speaking, is required if coherent science and technology policies are to be formulated and implemented. Unless the countries have science and technology policies it may be difficult for them to define biotechnology plans. Unfortunately the bureaucratic nature of most African political regimes has divorced science and technology from national development activities.
3.3 National technological capabilities Despite the fact that many of the countries in the region have recognized the importance of biotechnology, only a few have established specific training programmes in biotechnology. The University of Zimbabwe has established a masters course in biotechnology. Other universities considering similar courses include the University of Nairobi, Moi University and the Jomo Kenyatta University College of Agriculture and Technology. Although many of the countries in the region have had a long tradition in research, especially in agriculture, their manpower base is still weak compared to the tasks that need to be performed. Since biotechnology is science-intensive, the quality of training and level of technical competence needs to be high. Using the national agricultural research institutes as indicators of the available capability, the region’s institutions are staffed by undertrained people. Less than 8% of the local scientists in most sub-
493 Saharan African countries have doctoral training and some 57% have not undergone postgraduate training at all. The share of research activities in the local research institutions is relatively small and most of the resources are used for general administration. It is estimated that only 15% of finances in the institutions is spent on research in Kenya. The same situation applies to the utilization of manpower in the research institutions. For example, the national research institutions in Kenya devote only 9.6% of their manpower to research, the rest is devoted to support activities. It should be noted that institutions require certain levels of support staff to be functional. It is normally easier to focus on administrative expansion than to develop programmes that will increase the ratio of researchers. This is a situation that needs to be examined carefully in order to identify viable ways of rationalizing these institutions and increasing the ratio of researchers. Zimbabwe is another country in the region with a well-established institutional network for agricultural research. The role of coordinating technological research in Zimbabwe is performed by the Zimbabwe Research Council which operates through standing committees. Agricultural research is coordinated through the Zimbabwe Agricultural Research Council (ARC) which advises the Ministry of Lands, Agriculture and Rural Resettlement (MLARR) as well as the Department Research and Specialised Services (DR&SS). The ARC operates through five subcommittees dealing with livestock and pastures, grains, crop research, horticulture, and research services. The most important crop and livestock research body in Zimbabwe is the DR&SS which extends its services to the agricultural industry through seed certification, pesticide registration and advisory services. The Department operates through 11 institutes and has a staff of over 170 graduate scientists. The DR&SS, however, focuses most of its attention to the 8,000 large-scale farmers in the country, although it now has the mandate to serve nearly 1 million small holder farmers. Further agricultural services, especially focussing on soil conservation and agricultural technology, are provided by Agritex. The development of biotechnology in most African countries is hampered by the lack of basic equipment and expertise. For example, Sokoine University at Morogoro, has plans to undertake a wide range of biotechnology research and establish a tissue culture centre. The Veterinary Department of the university would like to venture into embryo transplant but cannot do so because of limited expertise and lack of basic equipment including an ultracentrifuge for preparing antigen samples and diagnostic ELISA kits. They also need HPLC equipment for drug research. It should be noted that one of the main limitations of biotechnology R&D in Africa is the lack of scientific information or mechanisms for acquiring information from the industrialized countries. Most African countries seem to have downplayed the role of scientific information in the promotion of biotechnology. Government policies in some of the countries (e.g.. Kenya) have hindered the acquisition of scientific information. Rigid fiscal (monetary exchange) policies constrain efforts of some research institutions to acquire scientific publications. There are taxes imposed on the importation of scientific equipment and publications. Most of the policies in
494
the countries have failed to encourage the establishment of suitable infrastructure to cater for training, joint research, and information acquisition and exchange.
4 Options for capability building 4.1 Institutional reforms As noted above, most of the biotechnology R&D in Africa and also in parts of Asia, is focused on agriculture. However, it is not necessarily true that significant biotechnology breakthroughs will also result from this sector. Furthermore, it is not necessarily true that the highest economic returns to research investment will also result from agriculture. On the contrary, it can be argued that in most agricultural research areas, traditional breeding techniques have already pushed yields to relatively high levels and the returns on research investment are not likely to be as high as in areas that have hitherto received less attention. Indeed in most industrialised countries, the biopharmaceutical sector leads in terms of R&D, number of f m s and pr~fitability.’~In developing countries, however, dependency on agriculture is proportionally greater than in industrialised countries and there is also a wellestablished tradition of research on agriculture, as was shown in sections 2 and 3 above. It should also be noted that the risks in nonagricultural biotechnology in Africa are relatively higher than in agriculture and this may explain, to a certain extent, the current emphasis on agricultural biotechnology. The competitive advantage of nations in international trade is currently being defined by the technological competitiveness of the leading firms in these countries. While Asia, especially countries in Southeast Asia are making some progress in narrowing the technology gap between themselves and industrialised countries, Africa is still at a disadvantage on this front and therefore any choice of technology or priorities for technological development needs to take into account these long-term competitive considerations. The current focus in Africa is not in itself a disadvantage given the fact that Africa has genuine agricultural problems that can be solved by biotechnology. The problem, however, is that Africa may be pursuing a route that has more limited opportunities in terms of overall biotechnology development. By paying less attention to other areas of biotechnology, Africa may be limiting its ability to draw from fundamental advances in these sectors. In many Asian countries, while the dependency on agriculture and therefore the emphasis of research is still high, considerable investment has taken place in other cutting-edge technologies. Thus countries as diverse as Taiwan and India have emphasised both natural resource based products such as agriculture and aquaculture, and also medical technologies and the development of vaccines and diagnostic technologies which address the particular needs of their regions. Advances in tissue culture have been of particular interest to the African countries because of their potential to solve some of the persistent problems of African agriculture. The potential benefits of tissue culture include rapid plant multiplication,
495
development of disease-free plantlets, production of uniform plants, year-round propagation of plants, rapid development of improved varieties, and better conditions for exchange and storage of genetic material. One of the key features of agricultural biotechnology research is the growing collaboration between the international agricultural research centres and institutions in the industrialized countries. This collaboration has made it possible for the centres to have access to the latest techniques developed by specialized research institutes. The building of capacity in the field of biotechnology cannot be supported without being considered in the context of specific activities. It should be noted that numerous donors have supported a wide range of projects in the name of “capacity building”, but such support has not resulted in any important change in the configuration of institutions in the region. The term “capacity building” is increasingly becoming a general term that is used to mean the same types of isolated projects that donors have supported over the last 3 decades in Africa. In many cases, such projects have contributed to the erosion of institutional capacity, or the projects were designed under the assumption that the institutional capacity to implement them already existed. Over the last 3 decades much of the support for genuine institutional development has gone to international institutions conceived elsewhere but located in Africa. While these institutions have played an important role in their areas of competence, they have often been isolated from the policy-making process, and their impact has been minimal. In some countries, policy makers have tended to avoid obtaining policy advice from such institutions because they view them as external and pursuing their own agenda. International institutions located in Africa are often viewed by host governments as places which provide local employment and bring foreign exchange into their countries, not necessarily as potential sources of ideas for development and policy making. The strengthening of policy-making capacity in these countries should be linked to the development of specialized institutions which focus on research, training and information dissemination and promote interactions and policy dialogue among representatives from governmental agencies, NGOs, research institutes and the private sector. Such institutions could enter into collaborative arrangements and organize joint activities. However, networking should be seen as an activity that emerges from the strengthening of certain activities, and not created to merely coordinate ongoing activities.
4.2 The need for cooperation The development of biotechnology represents a convergence of a wider range of skills and knowledge than any single institution or country is likely to possess. Thus, institutional cooperation will be critical to the success of biodiversity prospecting and biotechnology development. Clearly, the coordination of biotechnology activities, as part of the larger enterprise of science and technology, needs to be given legitimacy and impetus at the highest government levels. Unfortunately, the growing awareness
496 of the role of science and technology in many African countries is seldom accompanied by measures aimed at putting this vital issue at the core of national development planning and economic liberalization activities. Many African countries have set up stand-alone institutions to promote biotechnology or biodiversity conservation. Such units - often established in anticipation of funding rather than out of genuine interest in promoting the use and conservation of biodiversity - are unlikely to yield long-term benefits unless they are part of a broader institutional and policy framework. New institutional arrangements aimed at promoting biotechnology development should be constructed with national goals in mind rather than for donor politics and aspirations. There are a number of reasons for why it makes sense for developing countries as widely dispersed and as different from each other as Asian and African countries to cooperate with each other on biotechnology. First, in areas such as agriculture, developing countries in Asia such as India and the Philippines are using technologies which are also being widely applied in many African countries. The potential for cooperation and mutual gain for institutions using techniques such as tissue culture is therefore present. Second, in sectors such as agriculture and pharmaceuticals the welfare needs of developing country populations become very important. The need for feeding local populations as well as giving them adequate access to health care has often driven developing country governments to pursue self-sufficiency and develop technological capabilities in these sectors. In addition, the priorities of developing countries are often different from those of industrialised countries. For example, while the modem biopharmaceutical sector in industrialised countries tends to concentrate on diseases such as cancer and cardiovascular illnesses, the needs of developing countries relate much more to tropical diseases predominant in their regions. The potential for cooperation is therefore much greater between developing countries whose priorities and research thrusts tend to be more similar than in industrialised countries. Similarly, in agriculture, African countries have much more in common with countries in Asia with similar climatic conditions and local crops rather than in temperate regions. Thus with regard to both the technologies as well as the priorities, developing countries are much better placed to cooperate with one another. Technological capabilities are often most successfully built incrementally and technologies adapted to conditions in south Asia are more likely to be suitable to African skills and conditions. In addition, R&D on a number of agricultural crops and diseases tends to be often duplicated in developing countries and there may be a possibility to pool resources and knowledge together to develop technologies jointly. This may be useful from the point of view of resource and skill shortages in both continents. Indeed, this has been shown to be the case in the service provided by the international agricultural centres in Africa and in Asia. The Centres which have access to both finances and also some highly skilled research are able to use these services to their advantage to adapt technologies to local conditions and needs and transfer the technology to these countries.
497 4.3 Human resource development The key elements of a strategy to foster biotechnology transfer are training and access to information. This is often the major deficiency in most developing countries. Some of the Asian countries surveyed have made the acquisition of information an important part of their national programmes on biotechnology and a number of them have established electronic facilities for accessing international databases. Despite this, there are continued gaps in information and often communication with industrialised countries is better developed than regionally, even though regional capabilities may often be more appropriate. Many African countries have severely restricted the international flow of technical information or have failed to give the local scientific community incentives to exploit the information available internationally. If the communications infrastructure is not improved and the research environment liberalized, African countries will be hard-pressed to enter the field of biotechnology. Developing countries will not be able to develop biotechnology industries until they have built up a critical minimum level of biotechnological competence. Among other things, “technological capability” means being able to manage a new technology deployed in the economy. The acquisition of technological production capacity is associated with the flow of different kinds of knowledge and expertise. The first category includes the know-how needed to transfer and set up production facilities and various operational services. The second category includes expertise needed to operate and maintain the new system once it has been installed - both the codified knowledge in manuals, schedules, charts and diagrams and the “peopleembodied know-how” fostered through training, information services and on-the-job learning. The third category includes the knowledge and expertise needed to implement technical change: an understanding of how the technological system itself works and the techno-managerial capabilities needed to evaluate and transform plants already operating to meet new conditions. The most important factor in enhancing competitiveness in biotechnology is a country’s ability to bring available knowledge and expertise to bear on the development of specific products and processes. The entry barriers for biotechnologymastering traditional techniques, such as tissue culture, are lower than in other frontier technologies, such as microelectronics, so developing countries have unique opportunities to enter the field. Moreover, such precedents as the development of diagnostic kits for tropical diseases in Africa and work on developing vaccines for regional diseases such as hepatitis in Asia, confirm that a small group of well-trained scientists in the south can contribute heavily to biotechnology development.
4.4 Policy reforms Governments play a fundamental part in the promotion of technological change for sustainable development. The role of government is crucial, particularly in countries with weak economies and where the private sector’s abilities to promote technological
498
innovation are constrained by the fragmented nature of markets. In these conditions governments face problems of scarce resources for allocation to technological activities in general and biotechnology R&D in particular. They face problems in implementing certain national and international policies for promoting biotechnology R&D in the context of economic uncertainty and conflicting interests. Some governments have set national goals and priorities in the area of biotechnology. Public policy discourse is also marked by frequent references to the role of biotechnology in sustainable development in general and in national economic progress in particular. On the whole, there seems to be a broad international acceptance that biotechnology is one of the technologies that offer potentials for dealing with complex economic, health and ecological problems confronting developing countries. We do not have grounds to write off national goals and interests of biotechnology development. But we find some of the efforts of, and rhetoric on, national policies for biotechnology development to be zero-sum. There are many cases or examples where governments’ policies nurture trade-offs between short-term economic growth and long-term imperatives of institutional development and capability building [39]. The value premises of national policies must therefore be confronted when addressing issues of capability building and sustainable development. Government policies and programmes to promote the building of innovative capabilities for biotechnology development should support or promote institutional change and institutional capacity building. Indeed institutional building and change is a crucial facet of creating national capabilities for biotechnology R&D: “Institutional upgrading or other appropriate measures (are) needed to build up technical, managerial, planning and administrative capacities at the national level.”” But in order to develop systemic policies to promote institutional building or upgrading there must be some basic understanding of the nature of institutions involved in biotechnology R&D, what they are, their activities and how they behave and change over time. A central objective of national policies geared towards enhancing biotechnological capabilities should be to explicitly guide and strengthen institutions involved in specific activities of biotechnology R&D. Policies that are supportive of creating innovative capabilities for biotechnology development provide authority, flexibility and autonomy to research institutions to utilize available resources in solving specific problems through technological learning and application of specific techniques of biotechnology. On the whole, government policies are important in promoting the creation of technological capabilities for biotechnology R&D. Government policies can hamper or accelerate efforts to acquire and apply specific scientific information and knowledge. For instance, rigid fiscal (monetary exchange) policies can curtail efforts of firms or research institutions to acquire scientific publications. On a more sectoral level, Government efforts to promote small scale industry may leave biotechnology industries without the critical mass needed to develop commercial biotechnology capabilities. Government policies that promote the accumulation of innovative capabilities for biotechnology development are comprehensive by addressing national
499 priorities in the domain of biotechnology R&D and provide incentives to firms and institutions that invest their resources to address priority areas of biotechnology. In general, such policies promote technological development directly by stimulating scientific research, setting up a scientific infrastructure and giving preference to the output of indigenous technological capabilities. Specific policy measures include significant or adequate allocation of national resources to institutions for biotechnology R&D. They also encourage the establishment of suitable infrastructure to cater for training, joint research, and information acquisition and exchange.
Conclusion The brief survey of a number of countries in Asia and Africa has demonstrated the following: 1. While both regions are currently developing their technological and institutional capacity in biotechnology, this is much more coherent in Asia and especially among the newly industrialising counties of Southeast Asia. However, even in Asian countries considerable obstacles are present, especially in moving from public research stages to commercial production. 2. Institution building to support and promote biotechnology activities has largely taken place in a combination of old and new institutions in many Asian countries, whereas in Africa there tends to be fewer new institutions, much of the current research in agriculture, for example, being supported by older established agricultural research institutions. The obvious advantage of this is that institutional memory and history can provide major benefits to the research infrastructure in the country as a whole. This, however, places even greater pressure on these institutions and their ability to provide adequate attention to the new technology becomes crucial. It is not clear from the above evidence that sufficient skills and funding are available for these older institutions in many African and some Asian countries. 3. Policy coherence, especially cooperation between different ministries or departments within the same government, is crucial and one often finds that ministries or institutions work at cross purposes with each other. Much work has to be done by policy makers in ensuring that this does not become a common occurrence. . While south and especially south-east Asian countries have been relatively more successful, both in basic and applied R&D, as is evident from their private sector activities, countries in Africa are also slowly forming policy and institutions to facilitate the diffusion of biotechnology. The problems of coordination and translation of research results from basic R&D to commercialisation and management that are evident from the Asian experience are, however, magnified in Africa partly because of a scarcity of resources and partly because of the weaker institutional base that biotechnology is taking off from. An additional problem is that of the privatisation of knowledge. The science-based nature of biotechnology and the rapid rate of technological change is resulting in a
500
furious race among industrialised nations especially to innovate and patent each generation of technology. Developing countries are increasingly being left out of this equation and it is unclear whether the conclusion of the Uruguay Round Multilateral Trade Negotiations will bring positive returns for producers in developing countries, who tend to be imitators and adaptors of technology rather than innovators. For all these reasons, it is crucial for developing countries to collaborate and pool their resources to monitor technologies and adapt them for their particular needs and environments. This is as relevant in areas of knowledge exchange as it is in training and collaboration on specific research projects that target common goals and needs.
Acknowledgements This paper draws heavily from our discussions with a number of researchers in different parts of the world. We would like to thank a number of people without whose help we could not have successfully completed our research activities and without whose organizational skills we could not have visited all the countries covered during the research period. Rohini Acharya travelled to India, Thailand, Taiwan, The Philippines, Republic of Korea and China. John Mugabe travelled to Cameroon, Ghana, Nigeria, Tanzania, Zambia and Zimbabwe in Africa; and Brazil and Argentina in South America. During our trips to these countries we held discussions with researchers working on different aspects of biotechnology.We would like to thank them all for sharing their work and opinions with us. A number of persons deserve a special word of thanks: Dr N.K. Jain in India; Drs Yongyuth Yuthavong and Sakarindr Bhumiratana in Thailand; Dr William Padolina in the Philippines; Drs Shaw and Leah Lo in Taiwan; Dr Hong-Ik Chung and Prof Kong Deyong in the Republic of Korea; Dr Yonghui-Liu in China; Prof Joseph Gopo, Dr Ian Robertson, Mrs Wendy Martins, Dr Joseph Mwandazwile, Mr Rodrigues Mpande and Prof J. Chesanga; Dr John Forje and Prof Joseph Ngu in Cameroon; Dr S. Rugumamu and Mr P. Mweka in Tanzania; Dr J. Misuli and Ms Mary Olika in Zambia; Prof Laing in Ghana; Prof S.N. Okonkwo and Dr D. Oke in Nigeria; Dr B. Sorje in Brazil and Prof P. Williamson in Argentina. Financial assistance from the ACTS Biopolicy Institute and Finnida is gratefully acknowledged. Our greatest debt is to Dr Calestous Juma and Dr Norman Clark for their guidance over the years. Any errors remaining, however, are our own responsibility.
Notes See for example [I], [2] and [3] on biotechnology in the USA and Europe. A recent paper [4], also looks at biotechnology in Japan. 2. [l], op. cit. p. 2. 3. See for example, the industry surveys of Emst and Young in the USA [8] and in Europe [9]. 4. [lo]. This was partly as a result of traditional German industrial strength which had been built up by industry since the Second World War. 1.
501 5. 6.
7. 8. 9. 10.
11. 12.
13. 14.
15.
16.
17. 18. 19. 20.
21.
For further discussion of this point, please refer to [ 111. [lo]. This is also the subject of ongoing research by one of the authors on the subject of competitiveness and interlinkages in European biotechnology. See 14, [15]. [16], op. cit. pp. 217-218. [19], Chapter 16 para. 16.1. Most of this section is based on a survey of biotechnology in six countries in Asia. The authors are grateful to those officials, academics and industrialists who were consulted and provided details of their activities and their opinions on the development of biotechnology in their country. Based on a survey conducted in 1991 in six Asian countries. For more details please refer to [22] and [ l l ] for a more detailed comparison between India and Taiwan. Technology forecasting and improving these facilities in developing countries is the subject of an UNIDO volume based on a workshop on Technology monitoring, held in New Delhi, India in December 1994, presently being edited by one of the authors and due to be published shortly. Personal communication with members of the DCB, December, 1991. The National Science Council in Taiwan is the body in charge of funding public sector institutions. This funding includes both research programmes at prestigious institutions such as the Academia Sinica, and also programmes geared specifically for training young scholars in the biological sciences. It is estimated (personal communication with officials at the National Science Council in Taipei, August, 1994) that as many as 2,000 Taiwanese researchers return from the USA each year. [27]. Although the network is relatively small and young, BCIL has had considerable success in a short period of time in providing this service to industrialists and also researchers in the country. Another, larger, database, but which as yet only includes some information on biotech, is that maintained by the Technology Information, Forecasting and Assessment Council (TIFAC). TIFACLINE, a database which provides relevant information such as patent status and major producers and markets for energy, environmental and food technologies, has recently been established and is also accessible from nodal centres which have been established across the country (see also [28]). In the case of “essential” sectors or products, e.g., vaccines, the private sector has been less willing to invest because the sector is basically controlled, with the government becoming a monopsonist. Price ceilings are often placed on these products, especially in developing countries and as a result companies are less willing to go into sectors such as these. The government, similarly, is wary of allowing the private sector to completely dominate the sector and a certain, often quite large, part of investment, if in the form of joint ventures or other technology transfer and production alliances, remains public sector dominated (for more information on this, please refer to the comparative study of vaccines and diagnostics in India and Taiwan in Acharya (1995a)). For those interested in a discussion of public sector enterprise activities in commercial applications of biotechnology in selected countries please refer to [ 111. Personal communication with members of the Development Centre for Biotechnology in Taipei in October, 1991. The pharmaceutical sector in India and Taiwan is the subject of a forthcoming study [ 111. At present the company is able to manufacture the drug without paying royalties to the patent holdier, Wellcome, because India does not recognise product patents. The technology used by Cipla is based on an alternative process and hence does not violate the patent under Indian law. It is not clear what the position of the product will be once India changes its laws to recognise product patents, as required by the recent changes made in the Uruguay Round trade negotiations. It has been argued (see for example [8]), that the presence of a large pool of scientists, which graduate from one of the foremost scientific institutions in India, the Indian Institute of Sciences in Bangalore, has prompted many R&D-based ventures, both domestic and foreign to locate to this part of the country. In 1993 it was estimated that of a total of 337 biotechnology firms and public R&D institutions across the nation, the greatest concentration were in the south (1 13) of which Bangalore is a major base [27]. Bangalore is also becoming a centre for Electronics R&D where the
502 Government is building a software technology park (Far Eastern Economic Review, December 1992). The government of the state of Kerala is also proposing a science park for biotechnology in that state (Personal communication, BCIL, August 1994). 22. Forje J. Biotechnology in Cameroon. In: [14], 1995;66. 23. This is the subject of ongoing research by one of the authors (Acharya) on the competitiveness of European industrial biotechnology. 24. [19], Chapter 16 para. 16.11.
References 1. Orsenigo L. The Emergence of Biotechnology: Institutions and Markets in Industrial Innovations. London: Pinter Publications, 1989. 2. Kenney M. Biotechnology: The University-Industrial Complex. New Haven, CT: Yale University Press, 1989. 3. Sharp M. The New Biotechnology: European Governments in Search of a Strategy, Sussex European Paper, no. 15. Brighton: SPRU, University of Sussex, 1985. 4. Fransman M, Tanaka S. Government, Globalisation and Universities in Japanese Biotechnology, Research Policy, 1995. 5. Krimsky S. Genetic Alchemy: The Social History of the Recombinant DNA Controversy. Boston: MIT Press, 1982. 6. Juma C. The Gene Hunters: Biotechnology and the Scramble for Seeds. London, Princeton, NJ: Zed Books and Princeton University Press, 1989. 7. Farrington J. Agricultural Biotechnology in a Global Economy. Washington, DC: US Office of Technology Assessment, 1989. 8. Ernst, Young. Biotech 1994: Long-Term Value, Short-Term Hurdles. Ernst and Young’s Eighth Annual Report on the Biotech Industry. San Francisco: Ernst and Young, 1993. 9. Ernst, Young. European Biotech 95: Gathering Momentum. San Francisco: Ernst and Young, 1994. 10. OTA. Biotechnology in a Global Economy. Washington, DC: Office of Technology Assessment, Congress of the United States, 1992. 11. Acharya R. Competition, innovation and competitivenessin biotechnology in India and Taiwan: the case of pharmaceuticals. In: Mytelka L (ed) Competition, Innovation and Competitiveness in Developing Countries. Paris: OECD (Forthcoming), 1995a. 12. O’Connor K. Biotechnology: an international survey. In: United Nations Biotechnology and Development: Expanding the Capacity to Produce Food. New York: United Nations, 1992. 13. OECD. Technology and Global Competition: The Challenge for Newly Industrialising Countries. Paris: Organisation for Economic Co-operation and Development, 1989. 14. Juma C, Mugabe J, Mbote K (eds). Coming to Life: Biotechnology in African Economic Recovery. Nairobi: Acts Press, 1994. 15. Fransman M. Biotechnology: Generation, Diffusion and Policy. United Nations University, Institute of New Technologies (UNU/INTECH), Working Paper, no. I , 1991. 16. World Commission on Environment and Development. Our Common Future. Oxford: Oxford University Press, 1987. 17. Clark N, Juma C. Biotechnology for Sustainable Development: Policy Options for Developing Countries. Nairobi: Acts Press, 1991. 18. Sasson A. Biotechnologies and Development. Paris: UNESCO, 1988. 19. United Nations. Agenda 21. New York: United Nations, 1992a. 20. Freeman C. Technology Policy and Economic Performance: Lessons from Japan. London: Frances Pinter, 1987. 21. National Science Council, Taiwan. Report on the Survey of R&D in Science and Technology, 1993. 22. Acharya R. The Impact of New Technologies on Economic Growth and Trade: a Case Study of Biotechnology. PhD manuscript at MERIT, University of Limburg, Maastricht, The Netherlands.
503 23.
24.
25. 26. 27. 28. 29. 30. 31. 32.
33.
34. 35. 36. 37. 38. 39. 40. 41. 42. 43.
44.
45. &6. 47. 48. 49. 50.
51.
Maastricht: Universitaire Pers, 1995b. Chandrasekhar. Role of Biotech Consortium India Ltd. in the Commercialisation of Biotechnology ProductsProcesses - Status Paper. Paper presented at the Workshop on the Commercialisation of Biotechnologies in Agriculture and Aquaculture at IIM, Ahmedabad, India, 23-24 April, 1993. Han MH. Present Status of R&D Endeavour of Genetic Engineering and Biotechnology in Korea. Proceedings of an International Conference on Biotechnology, held in Kyungju, Korea, 22-25 April 1990, 12-17. Ebis. Various Issues. Brussels: Commission of the European Communities, 1991. Biotechnology and Development Monitor, no. 12, September, 1992;ll-13. BCIL. Directory of Biotechnology Industries and Institutions in India. New Delhi: BCIL, 1993. TIFAC. Role of TIFAC in Accelerating Technological Development. Paper Presented by TIFAC at ASSOCHAM, New Delhi, 2 September, 1994. NCGEB. National Centre for Genetic Engineering and Biotechnology: Annual Report 1993. NCGEB: Bangkok, 1991. Yuan R, Hsu M. Biostrategies in the Pacific Rim. Genetic Engineering News, May 1991. National Science Council, Government of Taiwan. Yuan R. Biotechnology in Singapore, South Korea and Taiwan. Report Prepared for the International Trade Administration, US Department of Commerce. Washington, DC: US Department of Commerce, 1986. Singh RB. Current status and future prospects of plant biotechnologies in developing countries in Asia. In: Sasson, Costarini (eds) Plant Biotechnologies for Developing Countries, Proceedings of an International Symposium organized by CTA and FAO, Luxembourg, 2 6 3 0 June, 1989. Department of Biotechnology. Annual Report. New Delhi: Department of Biotechnology, 1992. Klausmeier WH, Hall P, Chase-Lansdale C, Parra R. Emerging biotechnology enterprises in Southeast Asia: prospects and challenges. In: Flegel, Tharun, Yuthavong (eds), 1990;170-184. FAO. Selected Indicators of Food and Agriculture in Asia Pacific Region. Rome: FAO, 1991. Science Based Information Park (SBIP) information package, 1994. Reddy ASP, Sigurdson J. Strategic Location of R&D and Emerging Patterns of Globalisation: The Case of Astra Research Centre in India, Research Policy Institute, University of Lund, Sweden 1994. Mugabe J. Technological Capab es for Environmental Management: The Case of Biodiversity Conservation in Kenya. PhD Thesis, University of Amsterdam, 1994. Alper J. R&D alliances give biggest bang for buck. Bio/Technology 1993;11:150. Aubert J-E. What evolution for science and technology policies? OECD Observer 1992;2/3:5. Davis C et al. Biotechnology in Thailand: ScientificCapacity and Technological Change”, Biopolicy International no. 10. Nairobi: Acts Press, 1993. Development Centre for Biotechnology Biotechnology Information Service Report. Tapei, Taiwan: Development Centre for Biotechnology, 1994. Direcorate General International Cooperation. Biotechnology and Development Cooperation: Priorities and Organization of the Special Programme. The Hague: Directorate General International Cooperation, Ministry of Foreign Affairs, 1992. de Groot C. China and Europe together in biotechnology. Biotechnol Devel Monit 1992;13:20. Mody A. Staying in the Loop: International Alliance for Sharing Technology, World Bank Discussion Paper, no. 61. Washington, DC: World Bank, 1989. OECD. Biotechnology and the Changing Role of Government. Paris: Organization for Economic Cooperation and Development, 1988. OTA. DevelopmentAssistance, Export Promotion, and EnvironmentalTechnology. Washington, DC: Office of Technology Assessment, Congress of the United States, 1993. Reid, Walt et al. (eds). Biodiversity Prospecting: Using Genetic Resources for Sustainable Development. Washington DC: World Resources Institute, 1993. Sanchez V, Juma C (eds). Biodiplomacy: Genetic Resources and International Relations. Nairobi: Acts Press, 1994. Sasson A. Biotechnology and Natural Products: Prospects for Commercial Application. Nairobi: Acts
504 52.
53. 54. 55. 56. 57.
Press, 1992. Soong, Tai-Sen. Current Industrial Biotechnology Development in Taiwan. Agro-Industry, Hi-Tech: International Journal for Food Chemicals, Pharmaceuticals, Cosmetics, As Linked to Agriculture through Advanced Technology, 11-17, 1991. Spalding BJ. Biopharmaceutical firms up R&D spending 71%. Bioflechnology, 1993;11(2):768. UNEP. Conventionon Biological Diversity. Nairobi: United Nations Environment Programme, 1992. UNIDO. Industry and Development: Global Report. Vienna: United Nations Industrial Development Organization, 1991. United Nations. Biotechnology and Development: Expanding the Capacity to Produce Food. Advanced Technology Assessment System Issue 9. New York: United Nations, 1992b. Walsh V. Demand, public markets and innovation in biotechnology. Sci Pub1 Pol 1993;20(3):138156.
505
Index of authors Acharya, R. 465 Ainsworth, P.J. 409 Baptista, A. 315 Barrios-Gonzdez, J. 85 Bisseling, T. 151 Burdette, D.S. 1 Chen, T.T. 205 Dunham, R.A. 205
Paiva, C.L.A. 293 Panek, A.D. 293 Pawlowski, K. 151 Petersen, S.B. 315 Pizza, M. 391 Preiss, J. 259 Rappuoli, R. 391 Reimschuessel, R. 205 Ribeiro, A. 151 Robaglia, C. 185 Rodenhiser, D.I. 409
Ehrmann,M. 123 Hirano, S. 237 Hollingsworth, R.I. 281
Serizawa, N. 373 Singh, S.M.409 Stranzinger, G. 447
Jung, J.H. 409
Tepfer, M. 185
Lin, C.-M. 205 Lu, J.-K. 205
Vieille, C., 1 Vogel, R.F. 123 Vrolijk, N. 205
Martel, P.J. 315 Mejia, A. 85 Mugabe, J. 465
Went, D.F. 447 Zeikus, J.G. 1
Ott, R.N. 409
This Page Intentionally Left Blank
507
Keyword index 2'-5' oligoadenylate synthase 185 absorbable materials 237 acellular vaccines 391 actinomycete 373 actinorhiza 151 ADP -ribosyltransferase activity 391 glucose pyrophosphorylase 259 activation by 3-P-glycerate 259 activation by fructose-1,6-bis-P 259 allosteric mutants 259 inhibition by AMP 259 inhibition by Pi 259 regulation 259 transformation in plants 259 affinity chromatographic media 237 Africa 465 agricultural materials 237 alkaline chitin 237 alkaloids 85 allelic variation 447 amylose 281 anhydrobiotic organisms 293 animal model 391 antibacterial agents 237 antibiotics 85 antibodies 185 anticoagulant materials 237 antiprotease 185 antithrombogenic materials 237 antitumor agents 237 Archaea 1 'ASA 409 ASOH 409 automation for mutation detection 409 B. pertussis 391
bacterial glycogen synthesis 259 regulation of 259 bank cells 293 basic principles 85 biocatalysis 1 biodiversity 465
biological self-defense function 237 biomedical materials 237 biopsy 447 biosafety expression 123 biotechnology management 465 branching enzyme 259 different chain transfer functions of isoenzymes 259 properties 259 brominolysis 28 1 calcium release channel (CRC) protein 447 K-casein 447 chelate complexes 237 chiral synthons 281 chitin 237 digestibility 237 xanthate 237 chitin and chitosan films 237 chitinase 237 chitosan 237 digestibility 237 -coated papers 237 chitosanase 237 chlorinolysis 28 1 cholera toxin 391 cholesterol 237 -lowering drug 373 cholestyrarnine resin 237 clinical trials 391 CM-chitin 237 computer modelling 391 control coefficient analysis of starch synthesis 259 of metabolism 85 core model 1 coronary heart disease (CHD) 373 cosmetic ingredients 237 cryoprotectant 293 CYPlAl 205 CYPlA2 205
508 cytochrome P-450 gene 373 P-450 373 degradation 28 1 depolymerisation 281 dextrins 281 DGGE 409 disaccharide 293 DNA primer 447 testing 409 dsRNA-activated protein kinase 185
E. coli heat-labile enterotoxin 391 trehalase 293 effect of environmental and nutritional factors 85 electroporation 205 electrostatic interactions 1, 3 15 embryo transfer (ET) 447 environmental stress 293 enzymatic activity 315 feed additives 237 filamentous hemagglutinin 391 fluorescent in situ hybridization (FISH) 447 food 293 additives 237 fermentation 123 Frankia 151 freeze-drying 293 gel permeation chromatographic media 237 gene cloning 123 mapping 447 genetic detoxification 391 diagnosis 409 diseases 409 predisposition 409 genetics 123 government policies on science and institutional capability 465 growth hormone 205 halogenation 28 1 halothane test 447
haplotype analysis 409 haplotyping 447 HE-chitin 237 heme binding site 373 'hemostatic materials 237 heteroduplex analysis 409 HMG-CoA reductase inhibitor 373 HP-chitin 237 hydrolysis 28 1 hydrophobic interactions 1 hypersensitive reaction 185 hyperthermophiles 1 hypocholesterolemic function 237 identification 123 immobilizing media 237 immunoadjuvant activity 237 immunogenicity 391 industrial enzymes 1 insulin-like growth factor 205 key variables 85 lactic acid bacteria 123 Lactobacillus 123 lactones 281 legume 151 linkage 409 liposomes 293 LT-K63 mutant 391 lysozyme 237 malignant hyperthermia (MH) 447 mechanisms 28 1 membranes proteins 293 microinjection 205 microsatellite 447 mismatch cleavage 409 ML-236B 373 molecular dynamics 3 15 mutation 409 N-acetylchitosan 237 N-deaceylase 237 N-hexanoylchitosan 237 N-methylenechitosan 237 N-octanoylchitosan 237 negative regulation 373 nodulin 151
509 organs 293 oxidative cleavage 281 P450 205 P-450 repressor 373 PAETA 391 pale 447 pantropic viral vector 205 parentage control 447 PCR 409 Penicillum citrinum 373 pertussis toxin 391 pH 315 pharmaceutical materials 237 phenobarbital 373 phytohormones 85 pig oedema disease 447 pigments 85 plasmid vector 123 Poisson-Boltzmann equation 315 polyelectrolyte complexes 237 polymerase chain reaction (PCR) 447 polymorphic DNA marker 447 porcine stress syndrome (PSS) 447 PR-proteins 237 pravastatin 373 production 85 proline zipper 1 protein engineering 3 15 flexibility I folding 315 rigidity 1 stability 315 PT-9K/129G mutant 391 PTT 409 qualitative trait locus 447 radicals 281 radiolysis 28 1 RED 409 resistance gene 185 restriction enzyme 447 Rhizobium 151 ribosome-inactivating protein 185 root hair deformation 151 nodule 151 ryanodine receptor (RYR) 447
(S)-3,4-dihydroxybutyricacid 28 1 S. cerevisiae trehalase 293 Saccharomyces cerevisiae 293 secondary metabolites 85 sex diagnosis 447 preselection 447 silica chip technology 409 simulation methods 3 15 site-directed mutagenesis 39 1 ski genes 185 sludge dewatering agents 237 soft and exudative (PSE) pork 447 soil microbial flora 237 solid-state fermentation (SSF) 85 solvation 315 South Asia 465 Southeast Asia 465 specialty enzymes 1 sperm separation 447 sponge sheets 237 SSCP 409 stabilization dehydration 293 starch 281 synthase 259 isoenzymes 259 properties 259 synthesis 259 increase of starch by transformation 259 regulation of 259 starter organism 123 strain improvement 85 Streptomyces carbophilus 373 lividans 373 structure-function relationship 3 15 substrate induction 373 sustainable development 465 symbiotic nitrogen fixation 151 systemic acquired resistance 185 tandem repeat nucleotide sequence 447 technology 465 textiles 237 thermophiles 1 thermophilicity 1 thermostability 1 thermotolerance 293 transcriptional activation 373 transformation 123
510 transgene expression 205 inheritance 205 integration 205 transgenes 205 transgenesis 447 transgenic fish 205 plants 185 trehalase 293 trehalose 293 purification 293 quantification 293 uranium ion recovery 237
V . cholerae 391 vaccine development 391 vaccines 293 virus resistance 185
waste-water treatment 237 wound dressing 237 -healing materials 237 X-/Y-chromosome-bearing spermatozoa 447 Y-chromosome-specific DNA-probe 447 yeast 293
