Immobilized Nitro-avidin and Nitro-streptavidin as Reusable Affinity Matrices for Application in Avidin–Biotin Technology

Immobilized Nitro-avidin and Nitro-streptavidin as Reusable Affinity Matrices for Application in Avidin–Biotin Technology

ANALYTICAL BIOCHEMISTRY ARTICLE NO. 243, 257–263 (1996) 0514 Immobilized Nitro-avidin and Nitro-streptavidin as Reusable Affinity Matrices for Appl...

281KB Sizes 0 Downloads 136 Views

ANALYTICAL BIOCHEMISTRY ARTICLE NO.

243, 257–263 (1996)

0514

Immobilized Nitro-avidin and Nitro-streptavidin as Reusable Affinity Matrices for Application in Avidin–Biotin Technology Ely Morag, Edward A. Bayer, and Meir Wilchek1 Department of Membrane Research and Biophysics, The Weizmann Institute of Science, Rehovot 76100, Israel

Received July 25, 1996

Chemically modified forms of egg-white avidin and bacterial streptavidin (termed nitro-avidin and nitrostreptavidin, respectively), in which the binding-site tyrosine was nitrated, were used for several biotechnological applications. The fundamental difference between nitro-avidin and the native protein is that interaction of the modified protein with biotin can be reversed under relatively mild conditions. Consequently, nitro-avidin affinity columns or immobilizing matrices can be reused. Three examples are given to demonstrate the possible uses of such columns: (a) biotinylated protein A was attached to a nitro-avidin affinity column, and immunoglobulin was purified directly from whole rabbit serum; (b) biotinylated transferrin was attached to a nitro-streptavidin column, and anti-transferrin was isolated directly from rabbit anti-serum; and (c) biotinylated b-glucosidase was immobilized onto a nitro-avidin column and used as an enzyme reactor. In each example, the immobilized biotinylated probe could be released selectively from the column and recovered following its utilization. Reusable nitro-avidin thus provides an easy and attractive reversible form of avidin and thereby serves to expand the versatility of avidin–biotin technology. q 1996 Academic Press, Inc.

Avidin–biotin technology (1) is based on the extremely tight binding complex formed between the vitamin biotin and the egg-white glycoprotein avidin (2) (or its bacterial cognate, streptavidin). In this context, a biologically active molecule can be derivatized with biotin, and the resultant biotinylated probe can be attached to a solid support to which avidin has been coupled (3, 4). Due to this strong interaction, one can 1 To whom correspondence should be addressed. Fax: /972-89468256. E-mail: [email protected].

be assured of almost complete attachment of the probe to the affinity matrix, which is used subsequently for the isolation of compounds which interact with the probe. One of the major drawbacks for some applications of this system is the almost irreversible nature of the avidin–biotin complex (5, 6). In order to disrupt the complex, extreme denaturing conditions are required (7, 8). Such drastic conditions would invariably inactivate the biological activity of the biotinylated component, thus rendering it unsuitable for subsequent use. Moreover, despite the relatively robust constitution of the avidin component, its denaturation often leads to a reduction in its activity. In order to overcome the problem of irreversibility, a monovalent avidin column was developed (9, 10). This approach requires treatment of immobilized avidin with denaturing agents which dissociates the avidin tetramer; covalently bound monomers remain on the column, whereas free denatured monomers are removed from the column. Such treatment can be deleterious to both the carrier and the immobilized monomer; in fact, this resin must be stored in the presence of denaturing agent in order to prevent reassociation of immobilized, covalently bound monomers to form the tetramer (9). Moreover, the approach is unsuitable for streptavidin, thus limiting further its broad application. In a recent communication (11), we described the preparation of chemically modified forms of egg-white avidin and bacterial streptavidin, in which the bindingsite tyrosine was modified at its ortho position. In one example, the binding-site tyrosine was nitrated, and the resultant proteins were termed nitro-avidin or nitro-streptavidin. Unlike native avidin or streptavidin, which essentially bind biotin in an irreversible manner, the resultant derivatized avidins exhibited a reversible interaction with biotin. In the present report, columns 257

0003-2697/96 $18.00 Copyright q 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

AID

AB 9839

/

6m23$$$381

11-11-96 17:57:44

aba

AP: Anal Bio

258

MORAG, BAYER, AND WILCHEK

containing immobilized nitro-avidin and nitro-streptavidin were used for affinity chromatography and enzyme immobilization. Following isolation of the desired component or application of the immobilized enzyme, the biotinylated affinity ligand or enzyme could be removed from the column by high-pH buffers or by competition with free biotin. The regenerated nitro-avidin matrix was available for further usage. MATERIALS AND METHODS

Materials Egg-white avidin was obtained from Belovo Chemicals (Bastogne, Belgium) or from STC Laboratories (Winnipeg, Manitoba, Canada). Streptavidin was prepared from culture filtrates of Streptomyces avidinii as described previously (12). Protein A, human transferrin, bovine serum albumin (BSA),2 b-glucosidase (Catalog No. G6906), and p-nitrophenyl glucopyranoside (PNPG) were products of Sigma Chemical Co. (St. Louis, MO). Anti-transferrin antiserum was produced in rabbits as described previously (13), and the serum served as the source of general rabbit immunoglobulin.

Selective Blocking of Unmodified Biotin-Binding Sites

Buffers pH 4 and pH 6.3 buffers: 50 mM citrate–phosphate buffer at the indicated pH. pH 8 and pH 9 buffers: Tris–HCl buffer at the indicated pH and ionic strength. pH 10 buffer: 50 mM sodium carbonate–HCl buffer (pH 10). Biotin-containing buffers consisted of 0.6 mM biotin dissolved in one of these buffers. Methods Nitro-avidin and nitro-streptavidin were prepared in pH 9 buffer using tetranitromethane according to the recently published procedure. A ratio of 0.4 ml/mg protein was used for avidin and 2.4 ml/mg protein was used for streptavidin. Nitrated or unmodified proteins were immobilized on Sepharose 4B-CL (Pharmacia, Uppsala, Sweden) by the CNBr procedure (14, 15), using 0.5 to 2 mg of protein per gram of resin. Immobilized avidin and streptavidin were also nitrated directly on the resin. For this purpose, a sample of 4 ml of avidin–Sepharose or streptavidin–Sepharose (1.4 mg/ml Sepharose) was washed using 50 mM pH 8 buffer and treated with 6 ml of tetranitromethane for 50 min at 237C. The nitro-modified resin was 2 Abbreviations used: BSA, bovine serum albumin; PNPG, p-nitrophenyl glucopyranoside; PBS, phosphate-buffered saline; BNHS, biotinyl N-hydroxysuccinimide ester.

AID

AB 9839

washed extensively, first with 1 M NaCl, then with double-distilled water, and finally with phosphate-buffered saline, pH 7.2 (PBS). Proteins were biotinylated using biotinyl N-hydroxysuccinimide ester (BNHS) according to a previously published protocol (16). For protein A, a 10-mg sample was dissolved in 5 ml of 0.1 M sodium carbonate buffer, pH 8.5, and an aliquot (125 ml) of BNHS solution (2 mg/ml of dimethylformamide) was added. For transferrin, an identical quantity of protein was dissolved in 2 ml of the same buffer, and 71 ml of a BNHS stock (6 mg/ml of dimethylformamide) was introduced. For biotinylation of b-glucosidase, 25 mg in 5 ml of the above buffer was treated with 0.2 ml of BNHS stock (7.6 mg/ml of dimethylformamide). In each case, the reaction was allowed to proceed for 3 h, after which the biotinylated preparation was dialyzed exhaustively against PBS. SDS–PAGE was performed on boiled samples using 10% gels (17). The gels were stained with Coomassie brilliant blue R-250. Protein in solution was estimated according to Bradford (18).

/

6m23$$$382

11-11-96 17:57:44

Unmodified biotin-binding sites of immobilized nitroavidin were blocked selectively with biotin by the following procedure: A solution of biotin-containing buffer (0.6 mM biotin in pH 4 buffer) was applied to the nitroavidin matrix, either by percolation through a column or by batchwise addition to the resin and washing by centrifugation with pH 4 buffer. Biotin molecules which occupy the modified (nitro-tyrosine) binding sites were released using pH 10 buffer; under these conditions, biotin molecules are retained in unmodified binding sites. RESULTS

Repeatability of Nitro-avidin Column The reversibility of the nitro-avidin–resin was examined by repeated application and elution of a biotinylated protein. For this purpose, identical samples of biotinylated BSA (300 mg/ml of pH 4 buffer) were applied to a 0.75-ml column of nitro-avidin–Sepharose. The column was washed with pH 4 buffer and eluted using pH 10 buffer. The procedure was repeated three additional times, and fractions were monitored for protein (Fig. 1A). A small portion of the biotinylated BSA preparation failed to bind to the nitro-avidin column; this fraction may represent either a nonbiotinylated subpopulation of the preparation and/or minor overloading of the column. In any case, the amount of protein in this fraction remained fairly constant for each cycle. As shown in Fig. 1B, pooling of the eluted fractions

aba

AP: Anal Bio

NITRO-AVIDIN AFFINITY COLUMNS IMMOBILIZATION

259

FIG. 1. Repeated application and elution of biotinylated BSA from a nitro-avidin–Sepharose column. Identical samples (300 mg) of the biotinylated protein were applied successively to (Ap) and eluted from (El) the column (0.75 ml), using pH 4 and pH 10 buffers, respectively. Fractions of 1 ml were collected and the protein content was determined by the Bradford method (A). The amount of accumulated protein eluted from the column was very similar (B).

showed essentially identical levels of biotinylated protein bound to and released from the nitro-avidin column per cycle. Example 1: Immunoaffinity purification of IgG. The performance of a nitro-avidin column as a universal affinity resin was examined. In this context, biotinyl protein A was applied to a nitro-avidin–Sepharose matrix and the resultant immunoaffinity column was used to purify immunoglobulin directly from whole rabbit serum (see scheme in Fig. 2A). The immunoglobulin was eluted from the column under standard acidic conditions, and the biotinylated protein A was subsequently released from the immobilized nitro-avidin using a basic (pH 10) buffer or a biotin-containing buffer. The solubilized ligand could be stored and used in ensuing purifications. A sample (2 ml) of nitro-avidin–Sepharose resin (0.5 mg protein/ml of resin) was suspended in 2 ml of pH 4 buffer, and 1.8 mg of biotinylated protein A was added. The resin was washed with 50 mM pH 8 buffer, whole rabbit serum (0.5 ml, diluted fourfold in the same buffer) was applied, and the column was washed with 10 mM pH 8 buffer. The bound immunoglobulin was released from the column by pH 4 buffer. The affinity ligand (biotinylated protein A) was removed subsequently by pH 10 buffer. The results of the chromatography are shown in Fig. 2B. SDS–PAGE of the various peak fractions (Fig. 2C) indicated that the purified immunoglobulin appeared to be as pure as a commercially available sample of an equivalent fraction (compare lanes 3 and 5). The biotinyl protein A fraction (lane 4), which was eluted from the column by alkaline treatment (pH 10 buffer), was found to be similarly pure, compared to commercially obtained protein A (lane 6). Example 2: Immunoaffinity purification of antitransferrin antibodies. In a second example, a specific polyclonal antibody fraction was isolated directly from

AID

AB 9839

/

6m23$$$382

11-11-96 17:57:44

rabbit antiserum, using a biotinylated antigen and a nitro-streptavidin matrix (see scheme in Fig. 3A). Biotinyl human transferrin—the antigen—was immobilized onto a nitro-streptavidin–Sepharose matrix, and the resultant affinity column was used to purify antihuman transferrin antibodies directly from whole rabbit antiserum. A sample (4 ml) of nitro-streptavidin–Sepharose resin (0.3 mg protein/ml of resin) was suspended in 2.5 ml of pH 4 buffer, and biotinylated human transferin (2.5 mg in 0.5 ml of PBS) was added. The column was washed with 50 mM pH 8 buffer, and human transferrin-immunized whole rabbit antiserum (1.2 ml, undiluted) was applied. The column was washed with the same buffer at 10 mM concentration. The bound antibodies were released from the column by pH 4 buffer. Subsequent treatment of the column with pH 10 buffer resulted in a biphasic peak (Fig. 3B). The peak released by pH 4 buffer represented a highly enriched fraction of transferrin antibodies (Fig. 3C, lane 2). A minute amount of rabbit serum albumin contaminated the latter fraction. Following pH 10 treatment, the shoulder of the resultant biphasic peak (lane 3) also contained mostly antibody. The major pH 10 peak (lane 4) showed an enriched fraction of biotinylated human transferrin with residual amounts of antibody. The crude polyclonal antiserum would presumably contain various antibody subpopulations. Indeed, the major portion, comprising lower affinity species, was released from the column by added pH 4 buffer. Subsequently, a smaller fraction of antibody, apparently comprising higher affinity species, was released together with the antigen using the pH 10 buffer. Example 3: Reversible immobilization of biotinylated enzyme. In some cases, particularly for industrial usage, it would be advantageous to remove damaged or

aba

AP: Anal Bio

260

MORAG, BAYER, AND WILCHEK

FIG. 2. Performance of nitro-avidin column in affinity chromatography: Purification of rabbit immunoglobulin using biotinylated protein A as a removable affinity ligand. (A) Schematic description of the nitro-avidin matrix, the affinity ligand (biotinylated protein A), and the target molecule (rabbit IgG). (B) Elution profile (protein versus fraction number) of the affinity column. Rabbit serum was applied to the column (Ap); a large peak, representing unbound protein, immediately passed through the column. The column was washed extensively with pH 8 buffer, and the bound immunoglobulin was released with pH 4 buffer. The column was washed again, and the biotinylated affinity ligand was released upon addition of pH 10 buffer. (C) SDS–PAGE profile of fraction eluates and standards. Lane 1, applied material (whole rabbit serum). Lane 2, column effluent (unbound material); note disappearance of bands representing the immunoglobulin heavy and light chains. Lane 3, peak fraction following addition of pH 4 buffer; major bands conform with heavy and light immunoglobulin chains. Lane 4, peak fraction following addition of pH 10 buffer; major band corresponds to mobility of protein A (which migrates at a position similar to that of the heavy chain); no light chain can be detected. Lane 5, immunoglobulin standard. Lane 6, protein A standard.

inactivated material (i.e., a biotinylated ligand) from an avidin column, thus reconstituting the column for attachment of a new sample of the biotinylated ligand. This is demonstrated by an additional application of the nitro-avidin resin, which involved successive immobilization and desorption of a biotinylated enzyme. For this purpose, a biotinylated preparation of b-glucosidase from a highly thermophilic anaerobic bacterium (Cauldocellum saccharolyticum) was used. A sample (2.5 ml) of nitro-avidin–Sepharose resin (0.5 mg protein/ml of resin) was suspended in 2.5 ml of pH 4 buffer, and 1.5 mg of biotinylated b-glucosidase was added. The resin and biotinylated enzyme were allowed to interact for 40 min at room temperature, after which the column was equilibrated and washed for 40 min at 607C with pH 6.3 buffer at a flow rate of 1 ml/min. The enzymatic reaction was initiated by

AID

AB 9839

/

6m23$$$382

11-11-96 17:57:44

introduction of the substrate (1 mM PNPG in the same buffer at the same temperature), and the flow rate was reduced to 0.28 ml/min. Fractions of 2.8 ml were collected and enzyme activity was determined spectroscopically (A405). After a desired time interval, pH 10 buffer (15 ml) was added, and the biotinylated enzyme was eluted from the column. The column was reequilibrated and washed with pH 6.3 buffer (15 ml). In order to determine whether the release of biotinylated enzyme was complete, substrate was reintroduced, and the column was monitored for residual enzyme activity. A fresh sample of enzyme was then applied to the column under the original conditions, and the process was repeated. The procedure of immobilization and release of biotinylated b-glucosidase was performed three times (Fig. 4). During the first trial, enzymatic activity re-

aba

AP: Anal Bio

NITRO-AVIDIN AFFINITY COLUMNS IMMOBILIZATION

261

FIG. 3. Performance of nitro-streptavidin column in affinity chromatography: Purification of specific rabbit antibodies using biotinylated human transferrin as a removable affinity ligand. (A) Schematic description of the nitro-streptavidin matrix, the affinity ligand (biotinylated transferrin), and the target molecule (rabbit anti-transferrin antibody). (B) Elution profile (protein versus fraction number) of the affinity column. Rabbit antiserum was applied to the column (Ap); a large peak, representing unbound protein, immediately passed through the column. The column was washed extensively, and a distinct peak was released with pH 4 buffer. The column was washed again, and a biphasic peak was released using pH 10 buffer. (C) SDS–PAGE profile of fraction eluates and standards. Lane 1, applied material (whole rabbit serum). Lane 2, peak fraction following addition of pH 4 buffer; major bands conform with heavy and light antibody chains, with very low, but detectable amounts of rabbit serum albumin. Lane 3, shoulder of pH 10 peak fraction; majority of material corresponds to heavy and light antibody chains; residual amounts of albumin and transferrin discernible. Lane 4, major peak fraction following addition of pH 10 buffer; major band corresponds to human transferrin; relatively minor amounts of antibody chains are also visible. Lane 5, transferrin standard.

mained rather steady for the entire period (90 min). Following release and reapplication of a second allotment of enzyme, the activity associated with the column was initially about 70% of the original activity, but dropped to about 50%. After the third application of enzyme, the observed activity appeared to stabilize at about half of the initial enzyme activity. DISCUSSION

The increased use of the avidin–biotin system in affinity chromatography and protein immobilization is due largely to the very strong interaction and the four biotin-binding sites of avidin (3, 4). One of the main limitations in some of these applications is the lack of reversibility of the exceptionally stable complex under mild conditions. In order to dislodge biotin from the

AID

AB 9839

/

6m23$$$382

11-11-96 17:57:44

binding site of avidin, the protein has customarily been denatured by extreme conditions, e.g., 3 M guanidine thiocyanate, 6 M guanidine hydrochloride at pH 1.5, or by heating (ú707C) in the presence of detergents (7, 8, 19). When dealing with the dissociation of a biotinylated probe, such as a protein, from a complex with avidin, most probes would be severely damaged or completely inactivated by such treatment. In a recent publication (11), we described the preparation and characterization of a chemically modified form of avidin, in which the biotin-binding property can be controlled. The modification involved the conversion of the single binding-site tyrosine in avidin (or the conserved tyrosine in streptavidin) to an ortho-substituted derivative, by nitration or iodination. The resultant nitro-avidin (or iodo-avidin) bound biotinylated ligands tightly at low and neutral pH values, but failed

aba

AP: Anal Bio

262

MORAG, BAYER, AND WILCHEK

FIG. 4. Use of nitro-avidin resin as a reversible enzyme-immobilizing matrix. Biotinylated b-glucosidase (from the highly thermophilic bacterium, Cauldocellum saccharolyticum) was applied to the resin, and enzyme activity was monitored by including a chromogenic substrate (PNPG) in the equilibration buffer (pH 6.3 buffer). At the indicated time period (T1), the bound enzyme was released using pH 10 buffer. The column was reequilibrated with pH 6.3 buffer, PNPG was added, and residual activity was determined. At T2 , a fresh sample of biotinylated enzyme was applied, and the column was monitored with PNPG as before. The biotinylated enzyme was again released from the column at T3 , and another sample was applied at T4 .

to do so under basic conditions (pH 10). Biotinylated ligands could also be released from their complex with avidin by exchange using free biotin. In the examples presented in the present communication, we demonstrate that both nitro-avidin and nitro-streptavidin can be successfully used in immobilization of biotinylated ligands. The nitration procedure can be performed either in solution or directly on the protein coupled to the column. In both cases, the nitration reaction of the binding-site tyrosine fails to go to completion, and a given preparation exhibits a mixture of both modified and unmodified binding sites. It is thus imperative to block the latter sites with biotin, in order to provide an efficient reversible biotin-binding preparation. The column is thus primed successively with free biotin and pH 10 buffer, before applying a biotinylated ligand. Biotin binds to all sites (modified and unmodified), and the pH 10 buffer selectively releases the biotin molecules from the modified sites. The latter are then available for interaction with biotinylated ligands. The binding of biotinylated ligands to nitro-avidin is very strong. This is particularly evident from the stable binding of the biotinylated thermophilic enzyme, even upon extended incubation at 607C. This is further demonstrated by comparing the performance of the nitroavidin column to that of conventional avidin-based affinity chromatography (11, 13, 20, 21). Moreover, SDS– PAGE gels of eluent fractions exhibited little or no avidin band, again indicating the integrity of the immobilized nitro-avidin tetramer.

AID

AB 9839

/

6m23$$$382

11-11-96 17:57:44

The strong binding of biotinylated ligands displayed by nitro-avidin columns is superior to that observed for columns containing avidin monomers (10) which also show reversible binding of biotin. Such monomers can only be formed by treatment of covalently immobilized avidin with potent denaturing agents, since avidin monomers are unstable in solution and tend to undergo strong reassociation to form the tetramer (9). The tendency of avidin monomers to reassociate is so strong that such columns must be stored in the presence of denaturant. In fact, columns containing monomeric streptavidin cannot be prepared at all, since its tetramer is even more stable than that of avidin (19). The properties of the monomeric avidin column are impaired significantly, compared with those of the avidin tetramer. The biotin-binding capacity, the binding constant, and the stability of the binding site are all markedly reduced. The structure of the biotin-binding site of avidin implies that these deficiencies can be traced to the absence in the monomer of a bindingsite tryptophan (Trp-110), which, in the tetramer, is contributed from a neighboring monomer (21–23). This may also explain recent site-directed mutagenesis studies in which replacement of this tryptophan resulted in concomitant loss in binding affinity (24, 25). The tetrameric nitro-avidin resin thus exhibits clear advantages over monomeric avidin columns. Finally, in this article we have emphasized the use of nitro-avidin resins for affinity chromatography and enzyme immobilization. This study provides a prototype for the use of reversible nitro-avidin and nitrostreptavidin derivatives for additional future applications, e.g., for the separation and isolation of selected subpopulations of cells, for the production of reusable avidin-based biosensors, and for the generation of phage-display libraries (26). ACKNOWLEDGMENTS This research was supported by grants from the United States– Israel Binational Science Foundation (BSF), Jerusalem, Israel, and from the Baxter Healthcare Corporation (Chicago, IL).

REFERENCES 1. Wilchek, M., and Bayer, E. A. (Eds.) (1990) in Methods in Enzymology, Vol. 184, Academic Press, San Diego. 2. Green, N. M. (1975) Adv. Protein Chem. 29, 85–133. 3. Bayer, E. A., and Wilchek, M. (1980) Methods Biochem. Anal. 26, 1–45. 4. Wilchek, M., and Bayer, E. A. (1988) Anal. Biochem. 171, 1–32. 5. Bayer, E. A., and Wilchek, M. (1978) Trends Biochem. Sci. 3, N237–N239. 6. Wilchek, M., and Bayer, E. A. (1989) in Protein Recognition of Immobilized Ligands (Hutchens, T. W., Ed.), pp. 83–90, A. R. Liss, New York. 7. Green, N. M. (1963) Biochem. J. 89, 609–620.

aba

AP: Anal Bio

NITRO-AVIDIN AFFINITY COLUMNS IMMOBILIZATION 8. Cuatrecasas, P., and Wilchek, M. (1968) Biochem. Biophys. Res. Commun. 33, 235–239. 9. Green, N. M., and Toms, E. J. (1973) Biochem. J. 133, 687–700. 10. Kohanski, R. A., and Lane, M. D. (1990) Methods Enzymol. 184, 194–200. 11. Morag, E., Bayer, E. A., and Wilchek, M. (1996) Biochem. J. 316, 193–199. 12. Bayer, E. A., Ben-Hur, H., Gitlin, G., and Wilchek, M. (1986) J. Biochem. Biophys. Methods 13, 103–112. 13. Bayer, E. A., and Wilchek, M. (1990) J. Mol. Recognit. 3, 102– 107. 14. Kohn, J., and Wilchek, M. (1984) Appl. Biochem. Biotechnol. 9, 285–305. 15. Wilchek, M., Miron, T., and Kohn, J. (1984) Methods Enzymol. 104, 3–55. 16. Bayer, E. A., and Wilchek, M. (1990) Methods Enzymol. 184, 138–160.

AID

AB 9839

/

6m23$$$383

11-11-96 17:57:44

263

17. Laemmli, U. K. (1970) Nature 227, 680–685. 18. Bradford, M. (1976) Anal. Biochem. 72, 248–254. 19. Bayer, E. A., Ehrlich-Rogozinski, S., and Wilchek, M. (1996) Electrophoresis 17, 1319–1324. 20. Bayer, E. A., and Wilchek, M. (1990) Methods Enzymol. 184, 301–303. 21. Weber, P. C., Ohlendorf, D. H., Wendoloski, J. J., and Salemme, F. R. (1989) Science 243, 85–88. 22. Livnah, O., Bayer, E. A., Wilchek, M., and Sussman, J. L. (1993) Proc. Natl. Acad. Sci. USA 90, 5076–5080. 23. Pugliese, L., Coda, A., Malcovati, M., and Bolognesi, M. (1993) J. Mol. Biol. 231, 698–710. 24. Sano, T., and Cantor, C. R. (1995) Proc. Natl. Acad. Sci. USA 92, 3180–3184. 25. Chilkoti, A., Tan, P. H., and Stayton, P. S. (1995) Proc. Natl. Acad. Sci. USA 92, 1754–1758. 26. Balass, M., Morag, E., Bayer, E. A., Fuchs, S., Wilchek, M., and Katchalski-Katzir, E. (1996) Anal. Biochem. 243, 000–000.

aba

AP: Anal Bio