Engineering proteins to facilitate bioprocessing

184 reviews 48 Hammer, J., Takacs, B. and Sinigaglia, F. (i992)./. Exp. Med. 176, 1007-1013 49 Blond-Elguindi, S. et al. (1993) Cell 75, 717-728 50 Schatz, P.J. (1993) Bio/Teehnology 11, 1138-1143 51 Keller, P. M. et al. (1993) Vi~vloy,y 193, 709-716 52 Devlin, J. J., Panganiban, L. C. and Devlin, P. E. (1990) &ience 249, 404-406 53 Weber, P. C., Pantoliano, M. W. and Thompson, L. D. (1992) Biochemistry 31, 9350-9354 54 Scott, J. K., Loganathan, D., Easley, R. B., Gong, X. and Goldstein, I.J. (1992) Proc. Natl Acad. Sd. USA 89, 5398-5402 55 Oldenburg, K. R., Loganathan, D, Goldstein, I. J., Schultz, P. G. and Gallop, M. A. (1992) Proc. NatI Acad. Sci. USA 89, 5393-5397 56 Hoess, R. H., Brinkmann, U., Handel, T. and Pastan, I. (t993) Gene 128, 43-49 57 Smith, G. P., Schultz, D. A. and Ladbmy, J. E. (1993) Gene 128, 37~42 58 Balass, M., Heldman, Y., Cabilly, S., Givol, D., Katchalski-Katzir, E. and Fuchs, S. (1993) Proc. NatlAcad. &i. USA 90, 10638-10642 59 McLafferty, M. A., Kent, R. B., Ladner, R. C. and Marldand, W. (1993) Gene 128, 29-36 60 O'Neil, K. T., Hoess, R. H., Jackson, S. A., Ramachandran, N. S., Mousa, S. A. and DeGrado, W. F. (1992) Proteins 14, 509-515 61 Luzzago, A., Felici, F., Tramontano, A., Pessi, A. and Cortese, R. (1993) Gene 128, 51-57 62 Barbas, C. F., Languino, L. R. and Smith, J. W. (1993) Proc. Natl Acad. Sci. USA 90, 10003-10007 63 Lain, K. S., Salmon, S. E., Hersh, E. M., Hruby, V.J., Kazmierski, W. M. and Knapp, R.J. (1991) Nature 354, 82-84 64 Houghten, R. A., Pinilla, C., Blondelle, S. E., Appel, J. R., Dooley, C. T. and Cuervo, J. H. (1991) Nature 354, 84-86 65 Fodor, S. P. A., Read, J. L., PilTung, M. C., Stryer, L., Lu, A. T. and Solas, D. (1991) &ience 251,767-773 66 Yu, H., Chen, J. K., Feng, S., Dalgarno, D. C., Brauer, A. W. and Schreiber, S. L. (1994) Cell 76, 933-945 67 Houghten, R. A. and Cooley, C. T. (1993) Bioor2, Med. Chem. Lett. 3, 405 412 68 Salmon, S. E. et al. (1993) Pro& Natl Acad. Sci. USA 90, 11708-11712 69 Simon, R.J. et al. (i992) Proc. Natl Acad. Sd. USA 89, 9367-9371 70 Hagihara, M., Anthony, N.J., Stout, T. J., Clardy, J. and Schreiber, S. L. (1992)./. Am. Chem. Soc. 114, 6568-6570 71 Szostak, J. W. (1992) Trends Biochem. Sci. 17, 89-93 72 Bunin, B. A. and Ellman, J. A. (1992) ft. Am. Chem. Soc. 114, 10997-10998 73 Hirschmann, R. et al. (1992)J. Am. Chem. Soc. 114, 9217-9218 74 Kerr, J. M., Banville, S. C. and Zuckennann, R. N. (1993)J. Am. Chem. Soc. 115, 2529 2531 75 Needels, M. C. et al. (1993) Proc. Natl Acad. Sci. USA 90, 10700-10704 76 Ohlmeyer, M. H. J. et al. (1993) Proc. Natl Acad. Sci. USA 90, 10922-10926 77 Cunningham, B. C. and Wells, J. A. (1993) ./. Mol. Biol. 234, 554-563 78 Olivera, B. M., Rivier, J., Scott, J. K., Hilyard, D. R., and Cruz, L. J. (1991)./. Biol. Chem. 266, 22067-22070 79 Chothia, C. et al. (1992)ff. 3/lol. Biol. 227, 799 817 Engineering proteins to facilitate bioprocessing Per-Ake Nygren, Stefan St hl and Mathias Uhl n Genetic engineering is now being applied to aid the purification of recombinant proteins. The addition of specifically designed tags or the modification of sequences within the target-gene product has enabled the development of novel strategies for downstream processing that can be employed for efficient recovery of both native or modified proteins. This article discusses novel trends in genetic engineering that aid the bioprocessing of recombinant proteins. A decade has passed since the first example of the use of gent, fusions for affinity purification of recombinant proteins was reported 1. In the intervening years, a large number of affinity-fusion systems have been developed 2,3 and literally thousands 0£ gene fusions have been used for affinity purification. However, the use of gene technology to engineer proteins to facilitate bioprocessing has not yet been widely used for the production of recombinant proteins on an industrial P-A. Ny~,ren, S. Stahl and M. Uhl{n are at the Department of Biochemistry and Biotechnology, Royal Institute of Technology, S- 1 O0 44 Stockholm, Sweden. BTECH MAY 1994 (VOL12) scale. The main reason for this is probably due to the demand for the 'first generation' of recombinant proteins to be produced in their native form. Gene-fusion strategies introduce an additional problem into the downstream processing, since a site-specific cleavage is required to remove the affinity tag. This review discusses the application of genetic engineering to bioprocessing, including the use of small affinity tags and the rational design ofgene products in terms of their isoelectric point, solubility or partitioning coefficient to simplify the recovery of recombinant proteins. In addition, recent improvements in enzymic-cleavage methods are addressed. © 1994, Elsevier Science Ltd 185 reviews Table 1. Examples of commercial systems used in the production of recombinant proteins fused to affinity tails a Fusion partner Size Ligand Elution condition Supplier(s) ZZ His tail 14 kDa 6-10 aa IgG Ni2÷ Low pH Imidazole Strep-tag PinPointTM MBP GST FlagTM peptide 10 aa 13 kDab 40 kDa 25 kDa 8 aa Streptavidin Streptavidinc Amylose Glutathione Specific mAb Iminobiotin Biotin Maltose Reducing agent Low calcium Pharmacia, Sweden Novagen, USA; Invitrogen, USA; Qiagen, USA Biometra, Germany Promega, USA New England Biolabs, USA Pharmacia, Sweden IBI Kodak, USA ~Abbreviations:aa, amino acids; His, histidine;GST, glutathione-S-transferase;MBP, maltose-bindingprotein. bEncodes a domain biotiny]atedin vivo in Escherichia coil CMonomericstreptavidin. Despite the initial reluctance to apply this technology on a commercial scale, the demand to reduce capital costs and operating times for downstream processing will motivate wider use of gene-fusion technology in the future, since it will be possible to integrate several unit operations by rational genetic design of the protein being produced. Fusions to facilitate protein purification A large number of affinity tails have been developed to facilitate the downstream purification of recombinant proteins, and these permit several alternative purification protocols. However, factors such as proteolytic stability, solubility, the ability of the protein to be secreted, protein folding and the purification conditions have to be taken into consideration when choosing a suitable expression strategy for a specific protein. In particular, proteins produced as inclusion bodies must be refolded, or at least solubilized, prior to purification via affinity tails. Kits based on several of these tails have been produced commercially, in which expression vectors are manufactured together with the corresponding affinity resin. A selection of these systems is shown in Table 1. One of the best-characterized systems, staphylococcal protein A and its derivative ZZ, has been used successfully in several different hosts such as bacteria <S, yeast<7, Chinese hamster ovary (CHO) cells (E Lind, unpublished) and insect cells 8. The need for a low pH (pH 3 is routinely used) for elution of the protein from the affinity column can be circumvented by the use of competitive elution strategies based on engineered competitor proteins that can be removed efficiently from the eluate mix (J. Nilsson, M. Uhl6n and E A. Nygren, unpublished). The human p olyclonal IgG used as the ligand in this system can be replaced by recombinant Fc fragments, thus avoiding the use of a human serum protein in the purification protocol. The Strep-tag system9 uses a short (10-residue) peptide sequence with affinity for streptavidin which, in contrast with the natural ligand, biotin, can be eluted using mild conditions (11rim iminobiotin). To date, only fusions with this tag placed C-terminally have been described. A related system is PinPont TM, which utilizes the in vivo biotinylation by Escherichia coli of a 13kDa sequence attached to the target protein, combined with affinity purification of the fusion protein on monomeric streptavidin, thus enabling mild elution conditions (5 mm biotin) to be employed. Recently, a Lac-repressor-fusion-based peptide library was developed where a 13-amino-acid sequence was found to be sufficient to mimic efficiently the much larger natural domain normally recognized by the biotinylating enzyme 1°. Both N- and C-terminal fusions to the tail have been demonstrated to be functional. The use of this much smaller tag would result in a higher product: tail ratio. The FlagTM system is based on the Ca2+-dependent binding ofa monoclonal antibody (mAb) to an eightamino-acid peptide containing an enterokinase recognition site fused N-terminally to the target protein. The use of a low-Ca 2+ buffer as the eluant makes this system suitable for the recovery of sensitive target proteins. However, for large-scale applications, the cost of the affinity resin must be taken into consideration. Fusions to maltose-binding protein (MBP) are also eluted using gentle conditions (sugar), and can be produced both intracellularly or, alternatively, as secreted proteins. Polyhistidine tails are widely used as affinity tags and offer the possibility of purifying the recombinant protein by immobilized-metal-ion affinity chromatography (IMAC) under denaturing conditions, such as 6M guanidine hydrochloride or 8M urea 11,12. An intracellular expression system based on fusion to glutathione-S-transferase (GST), relies on the affinity of GST for glutathione .3. Numerous examples have shown that high-level cytoplasmic production of heterologous proteins is frequently associated with precipitation of the product into insoluble inclusion bodies. However, if the target protein is fused to a highly soluble partner such as the GST moiety, up to 45% of the total cell protein can be produced intracellularly in E. coli in a soluble form TM. In general, there appears to be a trend to design or select short affinity tails, particularly as the affinity tag can then be genetically fused to the target gene by polymerase chain reaction (PCR) techniques (Fig. 1). TIBTECHMAY1994(VOL12) 186 reviews Handle primer Target gene Cleavage site for processing Iral [[ Primer PoR Target gene [ Gene expression Figure 1 Fusion of a target protein to an affinity tag to enable simplifiedrecovery using polymerasechain reaction (PCR).The tag sequenceis introduced via a non-complementaryhandle sequence in one of the PCR primers used for amplification. This PCR-based strategy for subcloning the gene fragment into an expressionvector might simplify conventionalengineeringof the gene product for adaptationto downstream bioprocessing. native N-terminal after cleavage, and has been used successfully for site-specific cleavage of a number of fusion proteins ls,19,25. H64A subtilisin has not yet been commercialized, but other available enzymes, such as enterokinase, thrombin and factor Xa, which all leave native N-terminals have, in several cases, also been shown to give high cleavage yields (Table 2). However, these proteases, which all cleave after a basic residue, sometimes display nonspecffic cleavages. Thus, the yield of native protein can be dramatically decreased. However, the IgA-protease from Neisseria 2onorrhoeae 26, which requires a proline in the P2' (see footnote*) position, has been shown to be very efficient2°. This enzyme has the additional advantage of being produced as a recombinant protein in E. coli, thus simplifying its production and purification. For certain applications, it might be convenient to perform the site-specific cleavage with the fusion protein still immobilized on the affinity column. Using this strategy, the DNA-binding domain of the glucocorticoid receptor fused to protein A (Ref. 27), and mutants of the bovine pancreatic trypsin inhibitor (BPTI) (R.ef. 28) were produced without the need to elute the protein prior to cleavage. The latter case is an elegant example of the integration of unit operations, since the BPTI mutants were affinity purified on a chymotrypsin column, which also processed a chymotrypsin recognition site in the fusion protein. Integration of unit operations This trend might be further enhanced by novel phagedisplay and other powerful selection techniques, which have the potential to provide a rich source of short tails that are suitable for use in purifying and detecting recombinant proteins. For industrial production, the economic aspects of systems based on proteinaceous ligands must be considered, for reasons such as the need for 'cleaning-in-place' (CIP), in order to decontaminate columns for repeated use. Site-specific removal of affinity handles When a native gene product is desired, site-specific cleavage of the expressed fusion protein has to be performed. A vast number of cleavage methods, both chemical and enzymic, have been investigated for this purpose 2,3,15. Table 2 describes some of the more widely used cleavage methods that have been shown to be relatively efficient and specific. To illustrate the practical conditions for such cleavages, a representative exdhaple for each method is presented. Chemical-cleavage methods have a rather low specificity and the relatively harsh cleavage conditions can cause chemical modification of the released products 1s'22. Such methods are, therefore, primarily used in the preparation of peptides and smaller proteins. However, for certain products, chemical-cleavage methods have been used successfully, and they offer the attraction of being cost-effective and relatively easy to scale up 23. Among the enzymic methods, H64A subtilisin24 holds great promise, since it is highly specific, leaves a IBTECH MAY 1994 (VOL 12) An important part of modern biotechnology is to develop simplified schemes for the production and downstream processing of recombinant proteins by the integration of unit operations 29. Gene-product design can be used in several ways to implement such schemes, i.e. by influencing the yield and localization of the product, as well as adapting the gene product by fusion technology to specific unit operations suitable for large-scale downstream processing. The isoelectric point (pI) of a protein can be altered by the addition of charged amino acid residues, enabling the use of easily decontaminated ionexchange-chromatography media in the separation process 3°. Recently, an expanded-bed adsorption procedure was used for efficient recovery of a secreted recombinant fusion protein directly from a crude fermenter broth, without prior cell removal. The fusion protein was designed to have a relatively low pI to allow anionic exchange adsorption at pH 5.5, at which most host E. coli proteins are not adsorbed. This strategy enabled integration of the cell-separation step with ion-exchange adsorption of the gene product with simultaneous volume reduction (Fig. 2A). An overall yield of > 90% was obtained, including a subsequent affinity-chromatography polishing step 31. An alternative strategy to simplify downstream processing was investigated by Kthler and co-workers 32, * T h e cleavage w i n d o w for a protease is defined as: P 3 - P 2 - P I , ~ P I ' P2'-P3', w h e r e cleavage occurs at the b o n d b e t w e e n the P l and P I ' residues. 187 reviews Table 2. Examples of methods for site-specific cleavage of fusion proteins a Cleavage agent Chemical agents Hydroxylamine CNBr Fusion example Cleavage sequence b Cleavage conditions Ratio c (enzyme : substrate) ZZ-IGF-I ZZ-IGF-II Asn ~ Gly Met ~ 6h, 45°C, pH9.2 12h, 25°C, O. 1 M HCI 2M 50 : 1 (molar excess) FlagTm-lL-2 ZZ-PTH MS2-CD8 MBPparamyosin ZZ-PTH (Asp) 4 Lys ~ X Phe Ala His Tyr ~ X Pro Pro ~Thr Pro lie Glu Gly Arg ~ X 16h, 37°C, pH 8.0 4h, 37°C, pH 8.6 lh, 37°C, pH 7.4 lh, 25°C, pH 7.2 1 : 20 [w/w] 1 : 20 1:100 1 : 100 Phe Phe Pro Arg ~ X 6h, 20°C, pH3 6.5 1 : 100 Site-specific cleavage (%) Ref. 80 80 16 17 90 100 100 100 18 19 20 21 70 19 Enzymes Enterokinase H64A Subtilisin IgA protease Factor Xa Thrombin aAbbreviations: CNBr, cyanogen bromide; IGF-I, insulin-like growth factor I; IL-2, interleukin 2; PTH, parathyroid hormone; MBP, maltose-binding protein. bThe cleavage sequence shown here relates to the specific example shown. For some of the enzymes, alternative sequences are also reported to be functional. CThew/w ratio only relates to the enzymic cleavages. For the CNBr cleavage, the ratio is defined as mole CNBr:mole Met residues. in which recombinant technology was applied to alter the partitioning properties of a gene-fusion product, enabling it to be recovered efficiently by aqueous twophase extraction as a primary purification step. An attractive approach to simplify the recovery of products that may be produced with modified primary sequences, e.g. industrial enzymes, would be to alter the partitioning properties by substituting tryptophan residues for one or more surface amino acid residues that are not involved in biological activity 32. This type of protein engineering will probably be used in the future to reduce production costs for bulk proteins. A third example of how unit operations can be integrated by careful design of the gene product and the recovery process was recently demonstrated for the production of native human insulin in E. coli [R Jonasson, j. Nilsson, E. Samuelsson and M. Uhl6n, unpublished]. Human proinsulin was fused genetically to an IgG-binding affinity tail (ZZ) (Fig. 2B), and recovered from inclusion bodies by solubilization and refolding of the intact fusion protein. Native insulin, composed ofcysteine-bridged A and B chains, could be released by a single-step trypsin treatment that cleaved off the ZZ-tail at the same time as processing the C-peptide. a I Fermentation Fermentation Cell separation Concentration Chromatography I Chromatography II Affinity Chromatography I11 chromatography Polishing Polishing I b z z IgG-binding Human proinsulin Enhanced solubility for in vitro refolding The recovery of biologically active or native proteins by in vitro refolding from insoluble inclusion bodies is often hampered by the aggregation of the prodttct during the procedure, leading to low overall yields. To make the procedure more efficient, several improved protocols have been developed, including the addition of'folding enhancers', such as L-arginine 33-35, chaperones 36, polyethylene glycol (PEG) (Ref. 37), or even monoclonal antibodies 3s. Alternatively, the protein itself can be engineered to facilitate the refolding. The presence of hydrophilic peptide extensions, during the refolding can dramatically improve the folding yield, probably by conferring a higher overall solubility Figure 2 Examples of integration of unit operations enabled by genetic design of the product. (a) Several unit operations can be integrated by engineering the product to enable it to be recovered by an expanded-bedadsorption procedure. The left-hand panel lists the number of unit operations that are normally included in the downstream processing of a gene product. Several chromatographic separation steps based on ionexchange, size exclusion, hydrophobic interaction etc., might be utilized. The righthand panel illustrates that unit operations such as cell separation, concentration and chromatographic recovery can be combined into a single procedure. (b) The fusion protein from which native insulin can be recovered by a single trypsin treatment that simultaneously releases the ZZ-tail and processes the C-chain. To obtain completely native insulin, the C-terminal arginines of the B-chain are processed using carboxypeptidase. The arrows indicate processing sites for trypsin. TIBTECHMAY1994(VOL12) 188 reviews on the protein 34. Samuelsson and co-workers 39 showed that the reshuffling of misfolded disulfides in recombinant insulin-like growth factor I (IGF-I) was greatly facilitated by the highly soluble Z Z fusion partner. Compared with unfused IGF-I, the fusion could be successfully refolded at >100-fold higher concentrations (1-2 mg ml<), without the formation of precipitates 4°. Second generation therapeutic products Protein engineering offers the possibility of producing 'second-generation' therapeutics (compounds modified compared with the native molecule) to enhance therapeutic values and/or to facilitate production and purification processes. For example, tissue plasminogen activator (tPA) naturally comprises several discrete domains with different functions. However, a non-glycosylated deletion variant produced in E. coli by in vitro refolding showed improved thrombolytic and pharmacokinetic characteristics compared with the parent molecule34,4E Genetic fusion to one or more moieties can confer additional characteristics on a protein. For example, fusion to fragments of mAbs enables in vivo targeting of toxins42, while fusion to stable protein domains can increase serum hail-life43,44. By using such protein-engineering techniques, the need for chemical linking to achieve the combined activities is circumvented. For vaccine development, coherent antigens can be produced where carrier and/or immunostimulating properties are added to the immunogen by fusion to suitable moieties 45. Conclusions During the past decade, an increasing number of applications involving genetically engineered proteins have emerged. Integrated bioprocesses based on rational design have the advantage that the production and recovery of recombinant proteins, such as pharmaceuticals or industrial enzymes, can be achieved with high yields and at relatively low costs. Charge distribution, solubility, pI and other biophysical parameters can be altered and employed during downstream processing. This 'biotech' approach to protein engineering can also be used to adapt the enzyme for directed immobilization 46,47 in, for example, enzyme reactors and biosensors. Acknowledgements We thank Rainer Rudolph and Peter Lind for valuable information and stimulating discussions. References 1 Uhl6n, M., Nilsson, B., Guss, B., Lindberg, M., Gatenbeck, S. and Philipson, L. (1983) Gene 23, 369 378 2 Uhl~n, M., Forsberg, G., Moks, T., Hartmanis, M. and Nilsson, B. (1992) Curr. Opin. Bioteehnol. 3, 363-369 3 Flaschel, E. and Friehs, K. (1993) in Biotechnology Advances (MooYoung, M. and Glick, B. R., eds), pp. 31-78, Pergamon Press 4 Hanunarberg, B. et al. (1990)3". Biotechnol. 14, 423-438 5 Nilsson, B. and Abrahms6n, L. (1990) Methods Enzymol. 185, 144-161 6 Sterling, D. A., Petrie, A., Pulford, D. J., Paterson, D. T. W. and Stark, M . J . R . (1992) Mol. Mierobiol. 6, 703-713 BTECH MAY 1994 (VOL 12) 7 Zueco,J. and Boyd, A. (1992) Anal. Biochem. 207, 348-349 8 Andersons, D., Engstr6m, A.,Josephson, S., Hansson, L. and Steiner, H. (1991) Biockem.J. 280, 219-224 9 Schmidt, T. G. M. and Skerra, A. (1993) Protein Eng. 6, 109-122 10 Schatz, P.J. (1993) Bio/Teehnology 11, 1138-1143 11 Hochuli, E., Bannwarth, W., D6beli, H., Gentz, R. and Stiiber, D. (1988) Bio/Technology 6, 1321-1325 12 Ljungquist, C., Breitholtz, A., Brink-Nilsson, H., Moks, T., Uhl6n, M. and Nilsson, B. (1989) Eur.J. Biochem. 186, 563-569 13 Smith, D. B. and Johnson, K. S. (1988) Gene 67, 31-40 14 R.ay, M. V. L. et al. (1993) No~Technology 11, 64-70 15 Carter, P. (1990) in Fundamentals and Practice of Large-Scale Protei~ Purification (Am. Chem. Soc. Syrup. Ser. No. 427) (Ladisch, M. R., Willson, R. C., Palnton, C. C. and Builder, S. E., eds), pp. 181-193, American Chemical Society 16 Nilsson, B., Forsberg, G. and Hartmanis, M. (1991) Methods Enzymol. 198, 3-16 17 Forsberg, G., Samuelsson, E., Wadensten, H., Moks, T. and Hartmams, M. (1992) in Techniques in Protein Chemistry III (Angletti, R. H., ed.), pp. 329-336, Academic Press 18 Hopp, T. P. et al. (1988) Bio/Technology 6, 1204-1210 19 Forsberg, G. et al. (1991)J. Prot. Chem. 10, 517-526 20 Pohlner, J., Klauser, T., Kuttler, E. and Halter, R. (1992) Bio/Teehnology 10, 799-804 21 Malna, C. V. et al. (1988) Gene 74, 365-373 22 Lepage, P., Heckel, C., Humbert, S., St~hl, S. and Rautmann, G. (1993) Anal. Biochem. 213, 40-48 23 Moks, T. et al. (1987) Bio/Technology 5, 379-382 24 Carter, P., Nilsson, B., Bumier, J. P., Bnrdick, D. and Wells, J. A. (1991) Biochemistry 30, 6142-6148 25 Forsberg, G., Baastmp, B., Rondahl, H., Pohl, G., Hartmanis, M. and Lake, M. (1992)J. Prot. Chem. 11,201-211 26 Ambrosius, D., Dony, C. and R,udolph, R.. (1992) European Patent Application No. EP 0495398A I 27 Dahlman, K. et al. (1989) J. Biol. Chem. 264, 804-809 28 Altman, J. D., Henner, D., Nilsson, B., Anderson, S. and Kuntz, i. D. (1991) Protein Eng. 4, 593 600 29 Datar, R`. V., Cartwright, T. and Rosen, C-G. (1993) Bio/Technology 11,349-357 30 Dalboge, H., Dahl, H-H. M., Pedersen, J., Hansen, J. W. and Christensen, T. (1987) Bio/Technology 5, 161-164 31 Hansson, M., St~hl, S., Hjorth, R., Uhl~n, M. and Moks, T. No~Technology 12, 285-288 32 K6hler, K., Ljungquist, C., Kondo, A., Veide, A. and Nilsson, B. (1991) Bio/Technology 9, 642-646 33 Buchner, J. and Rudolph, P,. (1991) Bio/Technology 9, 157-162 34 Rudolph, R,. (1994) in Principles and Practice of Protein Engineering (Cleland, J. L. and Craik, C. S., eds), Hanser (in press) 35 Buchner, J., Pastan, 1. and Brinkmann, U. (1992) Anal. Biochem. 205, 263-270 36 Buchner, J., Brinkmann, U. and Pastan, I. (1992) Bio/Technology 10, 682-685 37 Cleland, J. L., Builder, S. E., Swartz, J. R., Winkler, M., Chang, J. Y. and Wang, D. L C. (1992) Bio/Technology 10, 1013-1019 38 Carlson, J. D. and Yarmush, M. L. (1992) No~Technology 10, 86-91 39 Samuelsson, E., Wadensten, H., Hamnanis, M., Moks, T. and Uhl6n, M. (1991) Bio/Technology 9, 363-366 40 Samuelsson, E., Moks, T., Nilsson, B. and Uhl6n, M. Biochemistry (in press) 41 Martin, U. et aI. (1992) Arch. Pharmacol. 346, 108-113 42 Chaudhary, V. K., Batra, J. K., Gallo, M. G., Willingham, M. C., Fitzgerald, D. J. and Pastan, I. (1990) Proc. Natl Acad. &i. USA 87, 1066-1070 43 Capon, D.J. et al. (1989) Nature 337, 525-531 44 Nygren, P-~,., Flodby, P., Andersson, R.., Wigzell, H. and Uhltn, M. (1991) in Vaccines 91 (Chanock, R`. M., Ginsberg, H. S., Brown, F. and Lerner, R,. A., eds), pp. 363-368, Cold Spring Harbor Laboratory Press 45 SjOlander, A., St~thl, S. and Perhnann, P. (1993) lmmunomethods 2, 79-92 46 Katchalski-Katzir, E. (1993) Trends Biotechnol. 11,471-478 47 Carter, P. and Wells, J. A. (1987) Science 237, 394-399