Systematic Exploration of the Antigen Binding Activity of Synthetic Peptides Isolated from the Variable Regions of Immunoglobulins

Abstract

Sets of short (12 residues) cellulose-bound synthetic overlapping peptides derived from the sequences of the variable regions of the heavy and light chains of three different antibodies (an anti-thyroglobulin antibody, the HyHEL-5 anti-lysozyme antibody, and an anti-angiotensin II antibody) were used to systematically assess the antigen binding capacity of peptides from the antibody paratope outside their natural molecular context. Peptides enclosing one or several of the complementarity determining region (CDR) residues had antigen binding activity, although the most active peptides were not necessarily those bearing the greatest number of CDR residues. Several residues from the framework region, preceding or following the CDR, were found to play a role in binding. Affinity constants from 4.1 × 10−7 to 6.7 × 10−8m−1 for the soluble form of 9 lysozyme-binding dodecapeptides were measured by BIAcore analysis. Alanine scanning of lysozyme-binding hexapeptides from the HyHEL-5 sequence identified 38 residues important for binding, of which 22 corresponded to residues that had been shown by x-ray crystallography to be at the interface between HyHEL-5 and lysozyme. Our results could be of interest for the rational identification of biologically active peptides derived from antibody sequences and in providing an experimental basis for mutagenesis of the antibody paratope. Sets of short (12 residues) cellulose-bound synthetic overlapping peptides derived from the sequences of the variable regions of the heavy and light chains of three different antibodies (an anti-thyroglobulin antibody, the HyHEL-5 anti-lysozyme antibody, and an anti-angiotensin II antibody) were used to systematically assess the antigen binding capacity of peptides from the antibody paratope outside their natural molecular context. Peptides enclosing one or several of the complementarity determining region (CDR) residues had antigen binding activity, although the most active peptides were not necessarily those bearing the greatest number of CDR residues. Several residues from the framework region, preceding or following the CDR, were found to play a role in binding. Affinity constants from 4.1 × 10−7 to 6.7 × 10−8m−1 for the soluble form of 9 lysozyme-binding dodecapeptides were measured by BIAcore analysis. Alanine scanning of lysozyme-binding hexapeptides from the HyHEL-5 sequence identified 38 residues important for binding, of which 22 corresponded to residues that had been shown by x-ray crystallography to be at the interface between HyHEL-5 and lysozyme. Our results could be of interest for the rational identification of biologically active peptides derived from antibody sequences and in providing an experimental basis for mutagenesis of the antibody paratope. Antibody molecules bind antigens with high affinity and specificity by synergistically using multiple noncovalent forces. The combining site (paratope), whose shape is complementary to the epitope on the antigen, is made up of the hypervariable regions, also called complementarity determining regions (CDRs) 1The abbreviations used are: CDR, complementarity determining region; VH, variable region of the heavy chain; VL, variable region of the light chain; Fmoc,N-(9-fluorenyl)methoxycarbonyl; HPLC, high pressure liquid chromatography. (1Wu T. Kabat E. J. Exp. Med. 1970; 132: 211-250Crossref PubMed Scopus (935) Google Scholar). It is commonly accepted that there are three CDRs in the light chain (L1, L2, and L3) and in the heavy chain (H1, H2, and H3). These CDRs fold into turn structures that are stabilized by the β-sheet framework of the variable domains. The interface between antibodies and antigens has been precisely described by x-ray crystallographic studies, and several complexes between Fab fragments of monoclonal antibodies and peptide or protein antigens have been recently described (for reviews see Refs.2Davies D. Sheriff S. Padlan E. J. Biol. Chem. 1988; 263: 10541-10544Abstract Full Text PDF PubMed Google Scholar, 3Davies D. Padlan E. Annu. Rev. Biochem. 1990; 50: 439-473Crossref Scopus (693) Google Scholar, 4Wilson I. Ghiara J. Stanfield R. Res. Immunol. 1994; 145: 73-78Crossref PubMed Scopus (26) Google Scholar). The structures of antibody-antigen complexes indicate that at least four of the CDRs, and in some cases all six CDRs, contribute to antigen binding (5Wilson I. Stanfield R. Curr. Opin. Struct. Biol. 1993; 3: 113-118Crossref Scopus (264) Google Scholar). Residues in the framework have rarely been reported to participate in this interaction (6Sheriff S. Silverton E. Padlan E. Cohen G. Smith-Gill S. Finzel B. Davies D. Proc. Natl. Acad. Sci. U. S. A. 1987; 84: 8075-8079Crossref PubMed Scopus (599) Google Scholar, 7Padlan E. Silverton E. Sheriff S. Cohen G. Smith-Gill S. Davies D. Proc. Natl. Acad. Sci. U. S. A. 1989; 86: 5938-5942Crossref PubMed Scopus (468) Google Scholar). Antibody-peptide or antibody-protein complexes are excellent model systems to study the physicochemical requirements for molecular recognition. Unfortunately, it is a difficult task to obtain crystals suitable for the structural elucidation of antibody fragments in complex with proteins or peptides. Therefore, other approaches to obtain information about the key residues involved in the interaction would be very useful, in particular for paratope mutagenesis. Some workers have demonstrated that synthetic peptides derived from the amino acid sequences of CDRs bind antigens with specificities similar to those of the original antibody molecules (8Kang C.-Y. Brunck T. Kieber-Emmons T. Blalock J. Kohler H. Science. 1988; 240: 1034-1036Crossref PubMed Scopus (83) Google Scholar, 9Taub R. Gould R. Garsky V. Ciccarone T. Hoxie J. Friedman P. Shattil S. J. Biol. Chem. 1989; 264: 259-265Abstract Full Text PDF PubMed Google Scholar, 10Williams W. Moss D. Kieber-Emmons T. Cohen J. Myers J. Weiner D. Greene M. Proc. Natl. Acad. Sci. U. S. A. 1989; 86: 5537-5541Crossref PubMed Scopus (91) Google Scholar, 11Williams W. Kieber-Emmons T. VonFeldt J. Greene M. Weiner D. J. Biol. Chem. 1991; 266: 5182-5190Abstract Full Text PDF PubMed Google Scholar, 12Welling G. Van Gorkum J. Damhof R. Driijfhout J. J. Chromatogr. 1991; 548: 235-242Crossref PubMed Scopus (32) Google Scholar, 13Igarashi K. Asai K. Kaneda M. Umeda M. Inoue K. J. Biochem. (Tokyo). 1995; 117: 452-457Crossref PubMed Scopus (20) Google Scholar, 14Zhang X. Piatier-Tonneau D. Auffray C. Murali R. Mahapatra A. Zhang F. Maier C. Saragovi H. Greene M. Nat. Biotechnol. 1996; 14: 472-475Crossref PubMed Scopus (42) Google Scholar, 15Zhang X. Gaubin M. Briant L. Srikantan V. Murali R. Saragovi U. Weiner D. Devaux C. Autiero M. Piatier-Tonneau D. Greene M. Nat. Biotechnol. 1997; 15: 150-154Crossref PubMed Scopus (38) Google Scholar). Such peptides have very often been chosen in the CDR3 of the VH sequence, which is considered to play a prominent role in defining antibody reactivity. However, the capacity of synthetic peptides derived from the variable regions of a given antibody to bind an antigen has never been probed in a systematic manner, i.e. in assessing the antigen binding capacity of every overlapping peptide from the VH and VL sequences. In this study, we present the results obtained by measuring the ability of an antigen to bind to sets of immobilized overlapping peptides of uniform size covering the amino acid sequences of the VHand VL domains of three different antibodies. The peptides were prepared by the Spot method (16Frank R. Tetrahedron. 1992; 48: 9217-9232Crossref Scopus (925) Google Scholar), which has previously been successfully used to identify peptide epitopes recognized by anti-protein antibodies and further developed to map protein-protein interaction sites (17Reineke U. Sabat R. Kramer A. Stigler R.D. Seifert M. Michel T. Volk H.D. Schneider-Mergener J. Mol. Diversity. 1996; 1: 141-148Crossref PubMed Scopus (48) Google Scholar). Our results indicate that numerous peptides show antigen binding capacity, most of them exhibiting measurable affinities in BIAcore, and that paratope residues important for antigen recognition can be identified by peptide analysis. Hen egg white lysozyme was from Sigma, and human thyroglobulin was from UCB-Pharma (Nanterre, France). Synthetic angiotensin II and an N-terminally biotinylated derivative were prepared as described below. 2 mg of the antigen in 2 ml of bicarbonate buffer (pH 8.6) were biotinylated by using a commercial reagent (Amersham RPN2202) according to the manufacturer's protocol. Biotinylated antigens were incubated with 0.1 m glycine (1 h, 37 °C) and then stored in phosphate-buffered saline at −20 °C. The numbering of the amino acid sequences of the variable regions was that of Wu and Kabat (1Wu T. Kabat E. J. Exp. Med. 1970; 132: 211-250Crossref PubMed Scopus (935) Google Scholar). The amino acid sequence of the anti-angiotensin II antibody 4D8 was established by sequencing the cDNAs corresponding to the heavy and light chains after reverse transcription of the mRNA from the 4D8 hybridoma (18Simon D. Romestand B. Huang H. Badouaille G. Fehrentz J.A. Pau B. Marchand J. Corvol P. Clin. Chem. 1992; 38: 1963-1967Crossref PubMed Scopus (35) Google Scholar). The amino acid sequences of the anti-lysozyme antibody HyHEL-5 and the thyroglobulin antibody were from the E. Mol. Immunol. 1994; PubMed Scopus Google Scholar, D. T. I. M. F. Pau B. M. M. J. Immunol. 1996; PubMed Scopus Google Scholar). The has been described previously F. D. C. Pau B. C. Res. 1996; Google Scholar). were obtained from amino were from was used for the peptides were at their the peptide sequences had been the were by acid (16Frank R. Tetrahedron. 1992; 48: 9217-9232Crossref Scopus (925) Google Scholar). The was the as for epitope (16Frank R. Tetrahedron. 1992; 48: 9217-9232Crossref Scopus (925) Google for the of a biotinylated antigen at 37 the a of an was incubated for at was by of a a on those the of was obtained with the after scanning the as described F. D. C. Pau B. C. Res. 1996; Google Scholar). were with to a as the of a the of the it was by m and acid in as to the and molecules to peptides. The of antigen was in or three soluble peptides were on an by for angiotensin a sequence was at the of every peptide by a Peptides were and from the by acid in the of Peptides were and their was by peptides were to The BIAcore was from BIAcore were at °C. N-terminally biotinylated peptides in saline were immobilized on a The was at a of The for immobilized peptides was found to be about after of the which to The binding of lysozyme to immobilized peptides was by lysozyme in saline buffer at a of was in buffer at a of The of the binding were using R. H. L. B. 1994; Scopus Google Scholar). The was from a of the at the a was from a of The was from the a The of the complex Sheriff S. Davies D. 1996; PubMed Scopus Google were used with the II to identify CDR residues in with amino from the framework that were found to be important by Spot peptide analysis. The capacity of short peptides derived from the variable regions of three different antibodies to bind the antigen was in a systematic The VH and VL amino acid sequences were as sets of overlapping dodecapeptides according to the Spot method (16Frank R. Tetrahedron. 1992; 48: 9217-9232Crossref Scopus (925) Google Scholar, F. D. C. Pau B. C. Res. 1996; Google Scholar). In this the peptides to the used for their and their is probed by the with a the an the results obtained with peptides derived from the variable regions of an anti-lysozyme antibody (6Sheriff S. Silverton E. Padlan E. Cohen G. Smith-Gill S. Finzel B. Davies D. Proc. Natl. Acad. Sci. U. S. A. 1987; 84: 8075-8079Crossref PubMed Scopus (599) Google Scholar). Biotinylated lysozyme (1 × to several peptides derived from the VH and VL sequences of HyHEL-5 of this interaction is between the of Peptides and CDR in the The binding was not by the biotinylated lysozyme in a of human i.e. in the of a high of proteins to lysozyme However, lysozyme was to the binding binding was with the complex on peptides that have the sequence recognized by J. L. P. Science. 1990; PubMed Scopus Google Scholar). The binding of lysozyme to immobilized peptides is In the model biotinylated thyroglobulin (1 × was also found to bind several peptides from the VH and VL domains of the anti-thyroglobulin antibody M. T. Pau B. PubMed Scopus Google Scholar). The was in six regions of the that corresponded to peptides CDR residues not The was in the of an of In of the VH and of the high affinity anti-angiotensin II antibody 4D8 × were by overlapping Biotinylated angiotensin II (1 × to peptides corresponding to the three CDRs of VL and to peptides corresponding to and biotinylated angiotensin II (1 was incubated with angiotensin the was with of the three antigens on peptides derived from antibodies. of results that peptides derived from the amino acid sequence of the VH and of three different antibodies are of recognized by their the results of the of the binding of biotinylated lysozyme to cellulose-bound peptides derived from the amino acid sequence of The of peptides framework residues binding and However, several peptides and that also framework residues were The molecular basis for this is of Residues to and with Residues by of the between the amino acid sequence and the binding that binding capacity was CDR residues were present in the peptide the sequence from had activity, the following sequence the to binding. residues from CDRs were identified as from at peptide from at peptide and from at peptide Some residues not to CDRs also to binding at peptide at peptide at peptide at peptide at peptide at peptide and at peptide a in the binding capacity of peptides residues from the CDRs from the peptide sequence, the in the of peptides and could be to the of in peptide The following CDR residues were from in binding to peptide from from and from residues not to the CDR were from the amino acid sequence of a binding the antigen binding capacity was of in the sequence of peptide of in peptide of in peptide of in peptide and of in peptide Therefore, it that the binding of antigen to cellulose-bound peptides is on the in their sequence of residues from the CDRs and in several of framework residues the of the dodecapeptides derived from the VH and VL of antibody and their with biotinylated residues are in and CDR residues are in to peptides binding. in a residues are in and CDR residues are in to peptides binding. Alanine scanning of hexapeptides derived from previously identified binding sequence from the VH and VL domains of HyHEL-5 was to identify the residues to antigen binding. The study of the CDR region is given as a see of for the amino acid numbering of the VH and VL hexapeptides and and of their six were by the Spot method and for with biotinylated lysozyme. of peptide by an to a in the antigen binding capacity, of to of the binding. of the other amino of the peptide not the binding 2 peptide the three amino and to be important Alanine scanning of peptide that is important 2 In we that the of a given amino acid to binding was not the the sequence to be a in peptides and not a in peptide the of in sequence in other sequences. In several hexapeptides had to be to the residues to the of residues found to contribute to the binding of lysozyme to peptides with residues involved in the interface in the HyHEL-5 lysozyme complex Sheriff S. Davies D. 1996; PubMed Scopus Google residues are in Residues to residues in with lysozyme in the was considered to be a the binding was by at least it was by an The amino acid sequences of the VH and VL of HyHEL-5 are given below. CDR residues are and The for the region was to with all four residues to the the were for CDR the region, the was identified as one to the CDR, and four residues were from the framework sequence preceding the region, the was Residues and were of the CDR, in the the region, the four residues from the CDR and four residues from the the region, the a framework and six residues in the of the CDR the region, the had amino from the CDR and four residues from the framework and on the and of the The of residues for binding the antigen a for the of framework for peptides an sequence that is of the CDR peptides and could be of the of the sequence very similar to the of The of peptide could be to a sequence with peptides from the region as with a some of the in framework similar for the of peptides and was The residues identified as important for the binding of lysozyme to peptides derived from the sequence of the VH and VL domains of HyHEL-5 were then with the residues involved in antigen binding in the of HyHEL-5 (6Sheriff S. Silverton E. Padlan E. Cohen G. Smith-Gill S. Finzel B. Davies D. Proc. Natl. Acad. Sci. U. S. A. 1987; 84: 8075-8079Crossref PubMed Scopus (599) Google Scholar, Sheriff S. Davies D. 1996; PubMed Scopus Google the region, three amino identified by peptide are in the crystallographic was found to be a in the region, one CDR amino acid was by crystallography in the Alanine scanning of peptides from this region the role of and of the of four residues preceding the region, the residues of the complex were identified by peptide the of an the region, four CDR residues and one framework were found by crystallography to be in antigen by scanning of the amino were found to be However, three other amino outside the CDR to be in the interaction of the peptide with the the region, which is a CDR, amino from the CDR and one in the framework play a role in the interaction as by x-ray using the Spot the of three amino of the CDR and and of the of the framework was the crystallographic results the and the of the CDR, the of the CDR to play an important role peptides were the region, the found by crystallography is similar to the found by the scanning of However, amino which were by peptide as were not in the interaction as in the four of them to the framework to the CDR, and to the The between residues important for lysozyme binding to and peptides and residues involved in interaction to identify of the residues in crystallographic with the antigen and 22 of the residues that are in or at the interface Sheriff S. Davies D. 1996; PubMed Scopus Google residues that were found to be important in the binding of biotinylated lysozyme to peptides not to amino at the The of them (12 of were framework residues. that all from the VL framework and six of important residues from the VH framework to the of framework residues from the as residues that CDR and the to antigen J. G. J. Mol. Biol. 1992; PubMed Scopus Google Scholar). using the of the it was found for and in the all the other framework residues identified as to antigen binding by peptide in the with at least residues from the CDR Therefore, this of residues could contribute by the peptide an for antigen of framework residues found to be important for antigen binding by peptide from the to the from the CDR as The of is not J. G. J. Mol. Biol. 1992; PubMed Scopus Google as The of is not in a on the sequences of the peptides used in the Spot a of peptides derived from the variable regions of HyHEL-5 was by and used in BIAcore interaction A. L. J. Immunol. 1991; 145: PubMed Scopus Google Scholar). a of the interaction of an immobilized biotinylated peptide with the of the experimental and with the that a interaction model described the binding. The were made for all other measurable peptide the results of the BIAcore study in which the a were soluble peptides corresponding to sequences in the Spot at least one peptide from of the six hypervariable regions of the antibody was found to measurable binding with in the × of other soluble synthetic peptides and measurable binding in BIAcore analysis. and corresponding to peptides from the framework of HyHEL-5 exhibiting binding in an affinity that was similar to that measured for the other peptides. peptides or peptides derived from sequences of or binding in the BIAcore of These results indicate that several short (12 peptides identified on the basis of their in the Spot to interaction of with lysozyme to be by interaction analysis. However, there were some peptides in the Spot that not show measurable it difficult to BIAcore from the results of the Spot of the binding of the interaction between lysozyme and biotinylated peptides derived from the VH and VL sequences of of the of the were measured after of the with biotinylated at 0.1 not antibody m not The of the were measured after of the with biotinylated at 0.1 not not in a binding molecular with several of the CDRs of the VH and of the antibody F. Proc. Natl. Acad. Sci. U. S. A. PubMed Scopus Google Scholar, A. R. S. R. Scopus Google Scholar). systematic of the capacity of every peptide from the VH and VL regions to bind the antigen was in this study to assess the of the sequence information the molecular of the paratope. It was found that in the of three different antigens egg white and human in their numerous immobilized peptides from the VH and VL regions of the antibody bind the antigen in a we have results that in the of other anti-protein antibodies the could be in of the of the for anti-protein antibodies. Our present that peptides with antigen binding have one or several residues from the CDR in their amino acid Synthetic peptides enclosing CDRs binding peptides CDRs amino from sequences the CDRs were also to the binding of some residues outside the CDR It is not residues contribute by the antigen or by the to the that residues often to the of framework residues that could the of the CDR E. Mol. Immunol. 1994; PubMed Scopus Google J. G. J. Mol. Biol. 1992; PubMed Scopus Google Scholar). It is that the of peptides immobilized on the is very to the of the sequence in the paratope. this is the of a peptide for would antigen binding in the that of the corresponding framework would binding by CDR It was that some peptides framework residues bind to the this is in cases to in the framework sequence of with the to antigen binding. It has been shown that a sequence of three residues in a peptide is to to E. Van Mol. Immunol. 1991; PubMed Scopus Google Scholar). However, in several there is that the or structural basis for this to be of of synthetic peptides corresponding to those active in the Spot was by BIAcore interaction analysis. The constants of the interaction between soluble peptides and lysozyme in the are 2 or of the of the between the antibody and lysozyme D. D. Smith-Gill S. G. 1992; PubMed Scopus Google Scholar). The of binding were by to a number of between the peptide and the the binding capacity of peptides derived from antibody it is that could be in an active X. Piatier-Tonneau D. Auffray C. Murali R. Mahapatra A. Zhang F. Maier C. Saragovi H. Greene M. Nat. Biotechnol. 1996; 14: 472-475Crossref PubMed Scopus (42) Google Scholar, Greene M. 1992; 1: Scopus Google Scholar). peptides that were active in the Spot not measurable affinities in the BIAcore that of antigen binding could be by the high of peptides at the of the R. Tetrahedron. 1992; 48: 9217-9232Crossref Scopus (925) Google Scholar), which is not the in the BIAcore the peptide is as as The identification of short amino acid sequences recognized by antigens have important for the of peptides antibody sequences W. Kieber-Emmons T. VonFeldt J. Greene M. Weiner D. J. Biol. Chem. 1991; 266: 5182-5190Abstract Full Text PDF PubMed Google in the to the M. M. U. D. H. H. J. B. Proc. Natl. Acad. Sci. U. S. A. 1993; PubMed Scopus Google with the of S. Nat. Struct. Biol. 1996; 3: PubMed Scopus Google Scholar). Our systematic a to on an experimental peptide sequences with antigen binding The results indicate that sequences not to peptides CDR in residues from the framework were found to play a role in the of K. Asai K. Kaneda M. Umeda M. Inoue K. J. Biochem. (Tokyo). 1995; 117: 452-457Crossref PubMed Scopus (20) Google Scholar). Several residues that were found to contribute to interaction were not residues at the interface by x-ray were that this of a particular role in the of the interaction The role of or residues in the binding of peptides on a sequence of the has been X. Piatier-Tonneau D. Auffray C. Murali R. Mahapatra A. Zhang F. Maier C. Saragovi H. Greene M. Nat. Biotechnol. 1996; 14: 472-475Crossref PubMed Scopus (42) Google Scholar, 15Zhang X. Gaubin M. Briant L. Srikantan V. Murali R. Saragovi U. Weiner D. Devaux C. Autiero M. Piatier-Tonneau D. Greene M. Nat. Biotechnol. 1997; 15: 150-154Crossref PubMed Scopus (38) Google Scholar). The identification of residues to the binding of the antigen to peptides was in the HyHEL-5 antibody model for which information on the interface between the antibody paratope and the protein is (6Sheriff S. Silverton E. Padlan E. Cohen G. Smith-Gill S. Finzel B. Davies D. Proc. Natl. Acad. Sci. U. S. A. 1987; 84: 8075-8079Crossref PubMed Scopus (599) Google Scholar, Sheriff S. Davies D. 1996; PubMed Scopus Google Scholar). Our identification was made in a and further with the results of x-ray 22 of the residues as important in peptides for binding biotinylated lysozyme to the of residues in with the antigen or in the interface (6Sheriff S. Silverton E. Padlan E. Cohen G. Smith-Gill S. Finzel B. Davies D. Proc. Natl. Acad. Sci. U. S. A. 1987; 84: 8075-8079Crossref PubMed Scopus (599) Google Scholar, Sheriff S. Davies D. 1996; PubMed Scopus Google Scholar), that the binding that was with short peptides was with the binding of the antigen to the The capacity of the method we used to residues involved in the interface could be for the of at in this not residues identified as important play a structural or role would not be that peptide with binding site R. A. J. J. Mol. Biol. 1996; PubMed Scopus Google Scholar, 1991; PubMed Scopus Google could be a for antibody The of in the is A. for in molecular