Antibody Generation

Review

TRENDS in Biotechnology

Vol.21 No.6 June 2003

275

Antibodies in proteomics I: generating antibodies Andrew Bradbury1, Nileena Velappan1, Vittorio Verzillo2, Milan Ovecka1, Leslie Chasteen1, Daniele Sblattero3, Roberto Marzari3, Jianlong Lou4, Robert Siegel5 and Peter Pavlik1 1

B Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA Institute for Neurodegenerative Diseases, UCSF, San Francisco, CA 94143, USA 3 Dipt. Biologia, University of Trieste, Trieste, Italy 4 Department Anesthiology, University of California, San Francisco, CA 94110, USA 5 Mailstop P7-56, Pacific Northwest National Lab, Richland, WA 99352, USA 2

The explosion in genome sequencing, and in subsequent DNA array experiments, has provided extensive information on gene sequence, organization and expression. This has resulted in a desire to perform similarly broad experiments on all the proteins encoded by a genome. Panels of specific antibodies, or other binding ligands, will be essential tools in this endeavour. Because traditional immunization will be unlikely to generate antibodies in sufficient quantity, and of the required quality and reproducibility, in vitro selection methods will probably be used. This review – the first of two – examines the strategies available for in vitro antibody selection. The second review discusses the adaptation of these methods to high throughput and the uses to which antibodies, once derived, can be put. Full or draft genome sequences are available for increasing numbers of organisms, including human [1,2], yeast [3] and many others (see www.tigr.org/tigr-scripts/CMR2/ CMRHomePage.spl for a list of microbial genome sequences). This has allowed the implementation of genome-wide studies, the most extensive of which have been carried out in yeast, with individual gene knockouts [4], overexpression and proteome chips [5], intracellular localization by tagging [6], protein – protein interaction studies by phage display [7], yeast two-hybrid [8,9], and widespread mass spectrometric (MS) analysis of purified complexes [10,11] providing large amounts of information. One of the reasons for using yeast so extensively is the availability of homologous recombination, which permits the replacement of endogenous genes by modified copies. In fact, most of the above-mentioned studies would not have been possible without this technique, which often involves the genetic fusion of a translated tag to each gene product. However, this powerful technology is not available for most genomes and the only alternative to the fusion of a general tag is the derivation of specific binding ligands for all gene products that can be used for techniques in which antibodies have been traditionally used (e.g. Western blotting flow cytometry, immunoprecipitation, Corresponding author: Andrew Bradbury ([email protected]).

immunofluorescence, immunohistochemistry and purification) but within a proteomic context rather than on a single gene scale. In addition to traditional uses, such binding ligands will also be useful in new proteomic techniques still under development, such as antibody chips, and – potentially – in applications such as biosensors. It is probable that greater understanding of protein function at a genomic level will come when such banks of binding ligands are derived and made generally available, as has been done virtually for DNA chips by the publication of genomic sequence. The traditional binding ligand is the antibody, and polyclonal antibodies are usually produced by immunization with proteins, conjugated peptides [12] or DNA expression vectors [13]. The fact that multiple different antibodies recognize a single target in polyclonal sera is both a strength (polyclonals can be used in all experimental formats) and a weakness (each polyclonal serum, even from the same animal, is unique and not reproducible). With the introduction of hybridoma technology [14], it became possible to avoid polyclonal antibodies, which also have problems of cross-reactivity and background, and produce large amounts of monoclonal antibodies of defined specificity. However, although extremely useful, this technology is not easily amenable to high throughput and is unable to overcome the problems of toxicity or poor immunogenicity. It is hoped that the adoption of a new suite of technologies, generically termed the ‘biomolecular diversity’ or ‘display’ technologies will be useful. In general, these technologies involve the selection of specific binders from large libraries of binding ligands and usually involve several selection cycles, each of which has several common features (Table 1). These cycles are usually carried out two or three times before binding ligands can be screened directly and, if the library is of good quality, binding ligands can usually be obtained against all targets. The first examples of biomolecular diversity focused on peptides [15– 17] and used phage display – in which displayed peptides are fused to one of the coat proteins of filamentous phage – as the platform to identify monoclonal antibody binding epitopes. Specific binders have also been isolated from large naı¨ve antibody libraries

http://tibtec.trends.com 0167-7799/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0167-7799(03)00112-4

Review

276

Table 1. Key technologies

elements

TRENDS in Biotechnology

of

the

biomolecular

diversity

Step

Feature

1

The generation of diversity at the nucleic acid level to create the ‘library’, usually of binding ligands. For example, natural or synthetic V genes, V genes with mutations The coupling of genotype to phenotype, alternatively translated protein with encoding gene, or information to function. This is done using living organisms or in vitro methods The application of selective pressure. For example, binding to a specific target Amplification of selected clones after selection, by growth, infection or PCR

2

3 4

Vol.21 No.6 June 2003

Interact library with selector

Wash

[18 – 24] and the display of hormones [25] and many other proteins (see [26 – 28] for reviews) has been carried out with the predominant goal of optimizing binding affinity. In addition to phage display, bacteria [29] and yeast [30] have been developed as microorganismal display methods, whereas ribosome display [31] and puromycin display [32], in which mRNA (or cDNA) is coupled directly to the polypeptides they encode, have been developed as in vitro methods. All of these are physical selection methods and require significant quantities of the selector to perform selection and screening. Several genetic methods for the selection of binding ligands have also been developed: such methods do not require the physical selector during the selection procedure and rely on the in situ synthesis of, and subsequent interaction between, binding ligand and target, to confer a selectable phenotype. Both the yeast two-hybrid system [33] and the protein complementation assay [34] have been adapted to scFv selections [35 – 37] in model systems, and offer the possibility of selection without the need for antigen synthesis and purification. Different genome projects are in different phases of development and therefore no single method will be appropriate to all genomes. If purified proteins are available, physical selection methods will probably be most effective. If only the sequenced genome is available, genetic methods – in which interaction between binding ligand and target confers survival on the cell containing the interacting pair, or the use of synthetic peptides as selectors – are likely to prove more effective, avoiding the need to produce purified protein. Whichever methods are used, however, validation of the specificity of selected antibodies will be an important and difficult component of the complete process. This review will concentrate on the processes of selecting binding ligands and not on the alternative binding ligands themselves, which were reviewed recently [38]. Because much of this technology was developed within the context of antibody fragments, and of scFvs and Fabs in particular, these will be the most commonly Table 2. Differences between scFvs and Fabs ScFv

Fab

Single protein molecule

Two protein molecules, chains must find each other More stable More difficult to synthesize No dimerization DNA insert .1500 bp Fewer intellectual property constraints

Less stable Better tolerated by bacteria Can form dimers (diabodies) DNA insert 700 bp More intellectual property constraints http://tibtec.trends.com

Amplify

Elute TRENDS in Biotechnology

Fig. 1. Physical selection methods: phage antibodies, or other selectable element, are represented by ‘Cs’. Each ‘C’ can bind only to a target of the same color. After a library of phage antibodies is interacted with a target in solid phase, the nonspecific binders are washed away and the specific binders are eluted and subsequently amplified.

described binding ligands and the differences between them are summarized in Table 2. However, most of the technology described is also directly applicable to alternative scaffolds, and these are mentioned when they might be more appropriate. Physical selection methods To select binding ligands against protein targets using physical selection methods (Fig. 1; Table 3), a selector needs to be available in a purified, synthetic or recombinant form. Depending on the method of selection used, 200– 1000 mg is usually required to carry out selection and screening. There are two general methods of physical selection. In the first, antigen is fixed to a solid support, such as a polystyrene tube or pin, and incubated with the library of antibodies. Those that recognize the antigen bind and can be eluted after non-binding antibodies are washed away. In the second method, antigen is labeled, usually with biotin or fluorescein, and used to separate antibodies that bind from those that do not. In the case of biotin, this is carried out using streptavidin-coated magnetic beads (MACS) [39], whereas with fluorescein, fluorescence-activated cell sorting (FACS) is used [40]. In phage, bacterial and yeast display, living organisms are responsible for amplification, display and the coupling of phenotype and genotype, whereas ribosome- and puromycin-based mRNA display systems rely on PCR for amplification, and in vitro translation of RNA to produce the binding ligand, which is then attached to the encoding RNA (or cDNA) either covalently (puromycin display) or non-covalently (ribosome display). More than one round of selection is usually required for all these methods, although screening thousands of clones can yield a greater diversity of binders after a single round [41], albeit with a far lower percentage of positives.

Review

TRENDS in Biotechnology

Vol.21 No.6 June 2003

277

Table 3. Display formats and their use Display method

Selection method

Recloning required for expression

Ab libraries published

Predominant uses and notes

Phage Bacteria Yeast

Solid support, MACS, in vivo Solid support, MACS, FACS MACS, FACS

Noa Yes Yes

Yes No Yes

Ribosome

Solid support, MACS

Yes

Yes

Puromycin (Profusione)

Solid support, MACS

Yes

No

Selection from naı¨ve libraries, affinity maturation Affinity maturation Affinity/stability maturation. Affinity can be determined from FACS analysis. Selection from naı¨ve libraries Selection from naı¨ve libraries, affinity maturation can also be built into naı¨ve selection Selection from naı¨ve libraries, affinity maturation can also be built into naı¨ve selection

a Recloning is not required for quick experiments but it is recommended to produce large amounts of antibody. FACS, fluorescence-activated cell sorting; MACS, streptavidincoated magnetic beads.

Binders have been isolated from large naı¨ve phage and in vitro display libraries, whereas bacterial [42] and yeast display [43] libraries have been used predominantly for affinity maturation, with yeast display also having been used for the improvement of antibody expression and folding [44]. Recently, a large naı¨ve yeast display library has also been produced, with antibodies against several targets being selected with affinities as good as those from phage libraries [45]. The availability of sufficient quantities of antigen for selection and screening is one of the major bottlenecks in the use of physical selection methods, with screening being the more antigen-intensive process. There are two general approaches to selection and screening: gene based and proteome based. In gene-based selection, the identity of the selector is known in advance and, as a result, the specificity of selected antibodies is known. Synthetic peptides [46,47], polypeptide fragments or recombinant full-length proteins would be examples of suitable selectors. Proteome-based selection uses fractionated natural sources (e.g. tissue or cell extracts). Once selected, the identity of the antigen recognized by such selected antibodies has to be determined subsequently. Selection of phage antibodies on proteins transferred onto PVDF membranes after separation by 2D gel electrophoresis has been described [48]. Although this might be a solution to the use of natural protein sources for selection, the screening and identification of appropriate binders in high throughput remain key issues, and the problems of sufficient amounts of selector remain. For example, some proteins are present in serum at 1 pg ml21, which would require purification of at least 1000 litres to yield 1 mg. In general, the big advantage of gene-based selection methods is that antigen identity is known in advance and sufficient quantities can be produced, whereas proteome-based selection methods have the advantage that the antigen is in its most natural state and will include appropriate post-translational modifications. Phage and yeast display Phage display has been by far the most commonly used method to select antibodies. Most phage antibody libraries have been created by cloning large numbers of different antibody genes upstream of the gene 3 coat protein gene and using phage [19] or phagemid [20 – 24] as the display vehicle. Most of these use scFvs as the antibody format, http://tibtec.trends.com

with one large Fab library also published [22]. The antibody genes are derived either from natural sources (e.g. human peripheral blood lymphocytes) or created synthetically by introducing diversity using oligonucleotides into frameworks with desirable properties. When an antibody gene is cloned upstream of gene 3, the antibody is displayed as a fusion protein with the gene 3 coat protein. A library of such phage antibodies theoretically consists of as many as 1011 different members (the diversity is generally measured by counting the number of independent colonies), with each different specificity being represented by a relatively small number of phage in a library. In general, diversity is limited by the transfection efficiency of bacteria. However, recombinatorial methods of library creation [21,49,50], in which VH and VL genes are shuffled using cre recombinase after an initial cloning step, are capable of creating far larger libraries and – because recombination is associated with amplification – almost unlimited supplies of such libraries, which is important in a genomic context. Although antibodies with subnanomolar affinities can be selected directly [20,23], or affinity matured from these libraries [51 – 53], this usually requires a considerable degree of effort, which would be difficult to marshall at a genomic scale. As a result, the usual affinity range of antibodies selected directly from such libraries is 10 – 1000 nM. However, the effective affinity can be increased significantly by genetically fusing multimerization domains, such as jun/fos for dimerization, to the ends of such selected scFvs [54]. Selection of phage antibodies is usually carried out against single antigens. A potentially high-throughput method using antigen immobilized on polystyrene pins in a microtiter format has recently been described [55] and it is likely that similar methods can also be developed for biotinylated antigens [39] using robotic washing and elution systems. However, selection is usually the least difficult part of the procedure, with the identification of different positive clones being far more time- and antigenintensive (see Screening in the second review of this series). Antibodies selected from naı¨ve phage antibody libraries have very variable expression levels (ranging from 10 mg to 20 mg per liter), and can be evolved to be expressed at higher levels [56– 58]. An alternative to the evolution of individual antibodies is to create libraries in which most

278

Review

TRENDS in Biotechnology

antibodies are well expressed. Several promising libraries [24,59 – 61] have been constructed with one or more stable scaffolds and synthetically introduced diversity. It has been shown [62] – at least in the case of yeast – that scFv stability, expression and display are related, indicating that improvement in any one of these parameters is likely simultaneously to improve the others, and suggesting that the adoption of any of the strategies described below to increase the intracellular stability of scFvs will also increase expression levels. The recent description of a naı¨ve yeast scFv library [45] is an encouraging addition to the repertoire of selection technologies. Antibodies with good affinities were selected from this library using biotinylated antigens and either magnetic or fluorescent selection procedures. It will be interesting to compare the ease of use of this technology with phage display – it is likely that both will have specific advantages and disadvantages. In vitro display systems In in vitro display systems, genes are coupled to the proteins they encode after translation in an in vitro translation mix. In RNA display (or Profusione) [32,63], puromycin covalently links the RNA to the encoded protein, whereas in ribosome display [31,64] the ribosome itself acts as a non-covalent linker between gene and encoded protein. The affinities of scFvs isolated from primary selections are similar to those from phage antibody libraries, although theoretical library sizes are much larger than most phage libraries (no transfection is required). Whereas positive binders can be selected after two or three rounds using phage display, in vitro display systems tend to require many more cycles. An advantage of in vitro display systems is the possibility of incorporating in-built affinity maturation [65,66] by using rounds of error-prone PCR or DNA shuffling [67] between selection rounds. Whereas this requires even more selection rounds, antibodies with picomolar affinities have been achieved using this method, and it is more likely to be amenable to automation for high-throughput selection. Once selected, antibodies or other binding ligands selected by in vitro display systems are usually cloned into bacterial expression systems; in vitro translation systems [68] are a currently unused alternative. This represents a bottleneck in the procedure because not all antibodies selected by in vitro display are well expressed in bacteria. Within this context, other scaffolds, such as the tenth fibronectin type III domain [69], have also been used in ribosome display with the isolation of high-affinity binders against several targets. Genetic selection methods Physical selection methods are appropriate for the selection of binding ligands in cases where a physical selector is available. Although this field is advancing rapidly, with a few groups producing small amounts of proteins on a genomic scale [5,70,71], in general, sufficient quantities of selectors are not available for most proteome projects. One way around this is to avoid the use of physical selectors altogether and to develop genetic selection methods (Fig. 2) that use DNA encoding either the whole or part of the gene http://tibtec.trends.com

Vol.21 No.6 June 2003

of interest. The selection of binders from libraries is essentially a protein – protein interaction problem, with the yeast two-hybrid system [33] being the most widely used genetic selection method to identify such interactions. Under ideal circumstances, it would be possible to clone the gene of interest as the bait and to transfect an antibody library as the prey each time a selection was carried out. However, most current antibody libraries [20 –23] contain more than five billion clones, which far exceeds the transfection capability of yeast. This problem has been overcome by carrying-out a single round of selection on a protein of interest and cloning the selection output into a yeast two-hybrid vector [35,72,73]. Although this reduces the diversity of the library to 105 – 106 (amenable to yeast transfection), permitting the selection of several different scFvs, it suffers from the need for a physical selection before performing the genetic selection, so eliminating some of the advantages of a genetic approach. Furthermore, most scFvs are not functional under the intracellular conditions used in this genetic approach because they contain disulfide bonds, which are required for their stability. These cannot be formed in the reducing environment of the cytoplasm. It might be possible to overcome the need to carry-out physical selection before genetic selection by using either bacterial genetic systems [74–77], relying on transcriptional activation, or ‘protein complementation assays’ [34,78,79], in which an enzyme required for cell survival (e.g. dihydrofolate reductase or b-lactamase) is divided into two parts in such a way that enzyme activity is reconstituted only when the two parts are brought together, by virtue of the presence of two interacting species, such as antigen– antibody. Preliminary experiments [37] using model antibodies and DHFR, showed that specific antigen– antibody pairs conferred survival

Survival

Death

Functional transcription factor, antibiotic resistance or essential enzyme

Non-functional transcription factor, antibiotic resistance or essential enzyme TRENDS in Biotechnology

Fig. 2. Genetic selection methods: an antibody fragment (or other binding ligand) is attached to half of a selection protein; the target to be recognized is attached to the other half of the selection protein. The two halves can come together only when antibody and target interact, so stabilizing the selection protein. Under appropriate selective conditions, the cell cannot survive unless the selection protein is functional. Examples of appropriate selection proteins are given in the figure.

Review

TRENDS in Biotechnology

more than 107 times more effectively than non-specific pairs. This, coupled with the high transfection efficiency of Escherichia coli, might make this method useful for proteomic-scale antibody selections. Most of these bacterial systems also require antibodies to be stable in the cytoplasm because, with the exception of b-lactamase, they all involve cytoplasmic interactions. Although the stability of any single antibody can be increased significantly by appropriate mutation and selection [44,80 – 83], it would be more appropriate to develop libraries of binding ligands, in which a high proportion of binders are functional under intracellular conditions, rather than to optimize individual scFvs. Such libraries are not yet available but a diverse body of work [72,84 – 88] indicates that frameworks based on antibody consensus sequences are generally more stable. By using either stable antibodies or consensus frameworks such as scaffolds, with synthetic oligonucleotides providing diversity, it should be possible to select stable scFvs without the need for further manipulation. Such libraries of antibodies would also be extremely useful as intracellular antibodies [89] in downstream target investigation. However, recent work has shown that many scFvs can be stabilized in the cytoplasm, as well as being expressed at high levels, by fusion to the C-terminus of maltose binding protein [90], perhaps avoiding the need for cytoplasmically stable antibody frameworks altogether. The selection for antibody– antigen interactions using b-lactamase complementation [79] in a fashion similar to that used for DHFR would avoid the need for libraries of antibodies that were stable intracellularly, because this enzyme is active in the periplasm. However, it does require antigens to be secreted into the periplasm, and it is expected that cytoplasmic antigens might not be stable under these conditions. As a result, it is possible that genetic selection approaches might require the use of more than one method to be effective, depending on the identity of the antigen. This caveat also applies to selection by avidity capture [91], an approach that can be best described as a combined genetic/physical method, which relies on co-expression of antigen and phage antibody in the periplasm. After antigen and antibody have interacted in the periplasm, positive clones are screened using a filter-based approach [41] that identifies many different specific antibodies, albeit after one round of traditional phage display selection. Conclusion At present, several different methods are available to select antibodies against gene products in high throughput. Determining which method should be used for a particular genome will depend on whether purified selectors (proteins or protein fragments) are available. If so, physical selection methods (phage display, yeast display and ribosome display) are most suitable, although it is not yet known which of these is most effective. In the absence of purified selectors either peptides or genetic selection methods can be used, although both of these lag far behind physical selection methods in maturity and ease of use. The second review in this series will deal with http://tibtec.trends.com

Vol.21 No.6 June 2003

279

screening and the adoption of these technologies to high throughput. References 1 Venter, J.C. et al. (2001) The sequence of the human genome. Science 291, 1304– 1351 2 Lander, E.S. et al. (2001) Initial sequencing and analysis of the human genome. Nature 409, 860 – 921 3 Goffeau, A. et al. (1996) Life with 6000 genes. Science 274, 546 – 567 4 Ross-Macdonald, P. et al. (1999) Large-scale analysis of the yeast genome by transposon tagging and gene disruption. Nature 402, 413 – 418 5 Zhu, H. et al. (2001) Global analysis of protein activities using proteome chips. Science 293, 2101– 2105 6 Kumar, A. et al. (2002) Subcellular localization of the yeast proteome. Genes Dev. 16, 707– 719 7 Tong, A.H. et al. (2002) A combined experimental and computational strategy to define protein interaction networks for peptide recognition modules. Science 295, 321– 324 8 Uetz, P. et al. (2000) A comprehensive analysis of protein-protein interactions in Saccharomyces cerevisiae. Nature 403, 623 – 627 9 Schwikowski, B. et al. (2000) A network of protein-protein interactions in yeast. Nat. Biotechnol. 18, 1257– 1261 10 Ho, Y. et al. (2002) Systematic identification of protein complexes in Saccharomyces cerevisiae by mass spectrometry. Nature 415, 180 – 183 11 Gavin, A.C. et al. (2002) Functional organization of the yeast proteome by systematic analysis of protein complexes. Nature 415, 141– 147 12 Anderer, F.A. and Schlumberger, H.D. (1966) Cross-reactions of antisera against the terminal amino acid and dipeptide of tobacco mosaic virus. Biochim. Biophys. Acta 115, 222– 224 13 Tang, D.-C. et al. (1992) Genetic immunization is a simple method for eliciting an immune response. Nature 356, 152 – 154 14 Kohler, G. and Milstein, C. (1975) Continuous cultures of fused cells secreting antibody of predefined specificity. Nature 256, 495 – 497 15 Cwirla, S.E. et al. (1990) Peptides on phage: a vast library of peptides for identifying ligands. Proc. Natl. Acad. Sci. U. S. A. 87, 6378– 6382 16 Scott, J.K. and Smith, G.P. (1990) Searching for peptide ligands with an epitope library. Science 249, 386– 390 17 Kay, B.K. et al. (1993) An M13 phage library displaying random 38amino-acid peptides as a source of novel sequences with affinity to selected targets. Gene 128, 59– 65 18 McCafferty, J. et al. (1990) Phage antibodies: filamentous phage displaying antibody variable domains. Nature 348, 552 – 554 19 Huie, M.A. et al. (2001) Antibodies to human fetal erythroid cells from a nonimmune phage antibody library. Proc. Natl. Acad. Sci. U. S. A. 98, 2682– 2687 20 Vaughan, T.J. et al. (1996) Human antibodies with sub-nanomolar affinities isolated from a large non-immunised phage display library. Nat. Biotechnol. 14, 309– 314 21 Sblattero, D. and Bradbury, A. (2000) Exploiting recombination in single bacteria to make large phage antibody libraries. Nat. Biotechnol. 18, 75 – 80 22 de Haard, H.J. et al. (1999) A large non-immunized human Fab fragment phage library that permits rapid isolation and kinetic analysis of high affinity antibodies. J. Biol. Chem. 274, 18218 – 18230 23 Sheets, M.D. et al. (1998) Efficient construction of a large nonimmune phage antibody library; the production of panels of high affinity human single-chain antibodies to protein antigens. Proc. Natl. Acad. Sci. U. S. A. 95, 6157– 6162 24 Knappik, A. et al. (2000) Fully synthetic human combinatorial antibody libraries (HuCAL) based on modular consensus frameworks and CDRs randomized with trinucleotides. J. Mol. Biol. 296, 57 – 86 25 Lowman, H.B. and Wells, J.A. (1993) Affinity maturation of human growth hormone by monovalent phage display. J. Mol. Biol. 234, 564– 578 26 Lowman, H.B. (1997) Bacteriophage display and discovery of peptide leads for drug development. Annu. Rev. Biophys. Biomol. Struct. 26, 401– 424 27 Hoogenboom, H.R. et al. (1998) Antibody phage display technology and its applications. Immunotechnology 4, 1 – 20 28 Bradbury, A. and Cattaneo, A. (1995) The use of phage display in neurobiology. Trends Neurosci. 18, 243 – 249

280

Review

TRENDS in Biotechnology

29 Francisco, J.A. et al. (1992) Transport and anchoring of b-lactamase to the external surfaceof Escherichia coli. Proc. Natl. Acad. Sci. U. S. A. 89, 2713 – 2717 30 Boder, E.T. and Wittrup, K.D. (1997) Yeast surface display for screening combinatorial polypeptide libraries. Nat. Biotechnol. 15, 553 – 557 31 Hanes, J. and Plu¨ckthun, A. (1997) In vitro selection and evolution of functional proteins by using ribosome display. Proc. Natl. Acad. Sci. U. S. A. 94, 4937– 4942 32 Roberts, R.W. and Szostak, J.W. (1997) RNA-peptide fusions for the in vitro selection of peptides and proteins. Proc. Natl. Acad. Sci. U. S. A. 94, 12297 – 12302 33 Fields, S. and Song, O. (1989) A novel genetic system to detect protein – protein interactions. Nature 340, 245 – 246 34 Pelletier, J.N. et al. (1999) An in vivo library-versus-library selection of optimized protein-protein interactions. Nat. Biotechnol. 17, 683 – 690 35 Visintin, M. et al. (1999) Selection of antibodies for intracellular function using a two-hybrid in vivo system. Proc. Natl. Acad. Sci. U. S. A. 96, 11723– 11728 36 Auf-der-Maur, A. et al. (2001) Antigen-independent selection of stable intracellular single-chain antibodies. FEBS Lett. 508, 407– 412 37 Mossner, E. et al. (2001) Fast selection of antibodies without antigen purification: adaptation of the protein fragment complementation assay to select antigen-antibody pairs. J. Mol. Biol. 308, 115– 122 38 Skerra, A. (2000) Engineered protein scaffolds for molecular recognition. J. Mol. Recognit. 13, 167 – 187 39 Hawkins, R.E. et al. (1992) Selection of phage antibodies by binding affinity: mimicking affinity maturation. J. Mol. Biol. 226, 889 – 896 40 Francisco, J.A. et al. (1993) Production and fluorescence-activated cell sorting of Escherichia coli expressing a functional antibody fragment on the external surface. Proc. Natl. Acad. Sci. U. S. A. 90, 10444 – 10448 41 de Wildt, R.M. et al. (2000) Antibody arrays for high-throughput screening of antibody-antigen interactions. Nat. Biotechnol. 18, 989 – 994 42 Daugherty, P.S. et al. (1998) Antibody affinity maturation using bacterial surface display. Protein Eng. 11, 825– 832 43 Boder, E.T. et al. (2000) Directed evolution of antibody fragments with monovalent femtomolar antigen-binding affinity. Proc. Natl. Acad. Sci. U. S. A. 97, 10701 – 10705 44 Shusta, E.V. et al. (1999) Yeast polypeptide fusion surface display levels predict thermal stability and soluble secretion efficiency. J. Mol. Biol. 292, 949– 956 45 Feldhaus, M.J. et al. (2003) Flow-cytometric isolation of human antibodies from a nonimmune Saccharomyces cerevisiae surface display library. Nat. Biotechnol. 21, 163 – 170 46 Persic, L. et al. (1999) Single-chain variable fragments selected on the 57-76 p21Ras neutralising epitope from phage antibody libraries recognise the native protein. FEBS Lett. 443, 112 – 116 47 Van Beijnum, J.R. et al. (2002) Target validation for genomics using peptide-specific phage antibodies: a study of five gene products overexpressed in colorectal cancer. Int. J. Cancer 101, 118– 127 48 Liu, B. et al. (2002) Towards proteome-wide production of monoclonal antibody by phage display. J. Mol. Biol. 315, 1063– 1073 49 Sblattero, D. et al. (2001) In vivo recombination as a tool to generate molecular diversity in phage antibody libraries. J. Biotechnol. 74, 303 – 315 50 Griffiths, A.D. et al. (1994) Isolation of high affinity human antibodies directly from large synthetic repertoires. EMBO J. 13, 3245 – 3260 51 Yang, W.P. et al. (1995) CDR walking mutagenesis for the affinity maturation of a potent human anti-HIV-1 antibody into the picomolar range. J. Mol. Biol. 254, 392 – 403 52 Schier, R. et al. (1996) Isolation of picomolar affinity anti-c-erB-2 single-chain Fv by molecular evolution of the complementarity determining regions in the center of the antibody binding site. J. Mol. Biol. 263, 551 – 567 53 Thompson, J. et al. (1996) Affinity maturation of a high-affinity human monoclonal antibody against the third hypervariable loop of human immunodeficiency virus: use of phage display to improve affinity and broaden strain reactivity. J. Mol. Biol. 256, 77 – 88 54 Pack, P. and Pluckthun, A. (1992) Miniantibodies: use of amphipathic helices to produce functional, flexibly linked dimeric FV fragments with high avidity in Escherichia coli. Biochemistry 31, 1579– 1584 http://tibtec.trends.com

Vol.21 No.6 June 2003

55 Lou, J. et al. (2001) Antibodies in haystacks: how selection strategy influences the outcome of selection from molecular diversity libraries. J. Immunol. Methods 253, 233 – 242 56 Kipriyanov, S.M. et al. (1997) Two amino acid mutations in an antihuman CD3 single chain Fv antibody fragment that affect the yield on bacterial secretion but not the affinity. Protein Eng. 10, 445 – 453 57 Duenas, M. et al. (1995) A point mutation in a murine immunoglobulin V-region strongly influences the antibody yield in Escherichia coli. Gene 158, 61 – 66 58 Coia, G. et al. (1997) Use of mutator cells as a means for increasing production levels of a recombinant antibody directed against hepatitis B. Gene 201, 203 – 209 59 Krebs, B. et al. (2001) High-throughput generation and engineering of recombinant human antibodies. J. Immunol. Methods 254, 67 – 84 60 Desiderio, A. et al. (2001) A semi-synthetic repertoire of intrinsically stable antibody fragments derived from a single-framework scaffold. J. Mol. Biol. 310, 603– 615 61 Holt, L.J. et al. (2000) By-passing selection: direct screening for antibody – antigen interactions using protein arrays. Nucleic Acids Res. 28, E72 62 Boder, E.T. and Wittrup, K.D. (2000) Yeast surface display for directed evolution of protein expression, affinity, and stability. Methods Enzymol. 328, 430– 444 63 Nemoto, N. et al. (1997) In vitro virus: bonding of mRNA bearing puromycin at the 30 -terminal end to the C-terminal end of its encoded protein on the ribosome in vitro. FEBS Lett. 414, 405 – 408 64 He, M. and Taussig, M.J. (1997) Antibody – ribosome-mRNA (ARM) complexes as efficient selection particles for in vitro display and evolution of antibody combining sites. Nucleic Acids Res. 25, 5132– 5134 65 Hanes, J. et al. (2000) Picomolar affinity antibodies from a fully synthetic naive library selected and evolved by ribosome display. Nat. Biotechnol. 18, 1287 – 1292 66 Hanes, J. et al. (2000) Selecting and evolving functional proteins in vitro by ribosome display. Methods Enzymol. 328, 404 – 430 67 Stemmer, W.P. (1994) DNA shuffling by random fragmentation and reassembly: in vitro recombination for molecular evolution. Proc. Natl. Acad. Sci. U. S. A. 91, 10747 – 10751 68 Sawasaki, T. et al. (2002) A bilayer cell-free protein synthesis system for high-throughput screening of gene products. FEBS Lett. 514, 102– 105 69 Xu, L. et al. (2002) Directed evolution of high-affinity antibody mimics using mRNA display. Chem. Biol. 9, 933 – 942 70 Lesley, S.A. (2001) High-throughput proteomics: protein expression and purification in the postgenomic world. Protein Expr. Purif. 22, 159– 164 71 Lesley, S.A. et al. (2002) Structural genomics of the Thermotoga maritima proteome implemented in a high-throughput structure determination pipeline. Proc. Natl. Acad. Sci. U. S. A. 99, 11664– 11669 72 Visintin, M. et al. (2002) The intracellular antibody capture technology (IACT): towards a consensus sequence for intracellular antibodies. J. Mol. Biol. 317, 73– 83 73 Tse, E. et al. (2002) Intracellular antibody capture technology: application to selection of intracellular antibodies recognising the BCR-ABL oncogenic protein. J. Mol. Biol. 317, 85 – 94 74 Joung, J.K. et al. (2000) A bacterial two-hybrid selection system for studying protein – DNA and protein – protein interactions. Proc. Natl. Acad. Sci. U. S. A. 97, 7382 – 7387 75 Karimova, G. et al. (1998) A bacterial two-hybrid system based on a reconstituted signal transduction pathway. Proc. Natl. Acad. Sci. U. S. A. 95, 5752– 5756 76 Dove, S.L. and Hochschild, A. (1998) Conversion of the omega subunit of Escherichia coli RNA polymerase into a transcriptional activator or an activation target. Genes Dev. 12, 745 – 754 77 Dove, S.L. et al. (1997) Activation of prokaryotic transcription through arbitrary protein-protein contacts. Nature 386, 627 – 630 78 Michnick, S.W. (2001) Exploring protein interactions by interactioninduced folding of proteins from complementary peptide fragments. Curr. Opin. Struct. Biol. 11, 472 – 477 79 Galarneau, A. et al. (2002) Beta-lactamase protein fragment complementation assays as in vivo and in vitro sensors of protein – protein interactions. Nat. Biotechnol. 20, 619 – 622

Review

TRENDS in Biotechnology

80 Worn, A. and Pluckthun, A. (2001) Stability engineering of antibody single-chain Fv fragments. J. Mol. Biol. 305, 989 – 1010 81 Proba, K. et al. (1998) Antibody scFv fragments without disulfide bonds made by molecular evolution. J. Mol. Biol. 275, 245 – 253 82 Jung, S. et al. (1999) Selection for improved protein stability by phage display. J. Mol. Biol. 294, 163 – 180 83 Martineau, P. et al. (1998) Expression of an antibody fragment at high levels in the bacterial cytoplasm. J. Mol. Biol. 280, 117 – 127 84 Kabat, E.A. et al. (1991) Sequences of Proteins of Immunological Interest U.S. Department of Health and Human Services, U.S. Government Printing Office 85 Ohage, E.C. et al. (1999) Intrabody construction and expression. II. A synthetic catalytic Fv fragment. J. Mol. Biol. 291, 1129 – 1134

Vol.21 No.6 June 2003

281

86 Steipe, B. et al. (1994) Sequence statistics reliably predict stabilizing mutations in a protein domain. J. Mol. Biol. 240, 188 – 192 87 Ohage, E. and Steipe, B. (1999) Intrabody construction and expression. I. The critical role of VL domain stability. J. Mol. Biol. 291, 1119 – 1128 88 Wirtz, P. and Steipe, B. (1999) Intrabody construction and expression III: engineering hyperstable V(H) domains. Protein Sci. 8, 2245– 2250 89 Cattaneo, A. and Biocca, S. (1997) Intracellular antibodies: developments and applications, Landes Bioscience, Academic Press 90 Bach, H. et al. (2001) Escherichia coli maltose-binding protein as a molecular chaperone for recombinant intracellular cytoplasmic singlechain antibodies. J. Mol. Biol. 312, 79 – 93 91 de Wildt, R.M. et al. (2002) Isolation of receptor-ligand pairs by capture of long-lived multivalent interaction complexes. Proc. Natl. Acad. Sci. U. S. A. 99, 8530– 8535

Do you want to reproduce material from a Trends journal? This publication and the individual contributions within it are protected by the copyright of Elsevier Science. Except as outlined in the terms and conditions (see p. ii), no part of any Trends journal can be reproduced, either in print or electronic form, without written permission from Elsevier Science. Please address any permission requests to: Rights and Permissions, Elsevier Science Ltd, PO Box 800, Oxford, UK OX5 1DX. http://tibtec.trends.com