Molecular Biology Reports 16: 207-212, 1992. 9 1992 Kluwer Academic Publishers. Printed in Belgium.

207

Use of synthetic peptides for the detection and quantification of autoantibodies Keith B. Elkon The Hospital for Special Surgery-Cornell University Medical College, New York, N Y 10021, USA

Key words: autoantibody, autoantigen, synthetic peptide Abbreviations: r: recombinant, SLE: systemic lupus erythematosus

The characteristics of the humoral immune response to protein autoantigens have been reviewed recently [ 1,2]. Most of the autoantibodies demonstrate properties similar to those described for a secondary immune response against foreign protein antigens. Briefly, the autoantibodies are polyclonal, directed against multiple epitopes, react predominantly with conformational epitopes, are composed of the IgG isotype, IgG1 and 3 subclasses (although exceptions are noted [3 ]), and frequently bind to multiple components within the same RNP particle. The signal difference between autoantibodies and antibodies to foreign proteins is the selection of epitopes - autoantibodies usually bind to highly conserved epitopes on self proteins [1,2] whereas antibodies to foreign proteins select epitopes most different from self [4]. X-ray diffraction studies of monoclonal antibody-antigen complexes indicate that the antibodies contact 15-22 amino acids on several loops on the surface of the antigen [reviewed in 5]. Despite this fact, some antisera to foreign or self antigens contain populations of antibodies that also recognize short peptides ('anti-peptide antibodies'). Peptide antigens are therefore thought to correspond either to part of a conformational epitope or to the less ordered structure of the termini of the protein. The distinction between a conformational (discontinuous/ assembled) or linear (continuous) epitope is made

operationally by epitope mapping (see this volume). Antibodies that bind to a short peptide define a linear epitope whereas, when antibodies fail to bind to fragments of the protein antigen, these antibodies are presumed to be directed against amino acids that are located on different parts of the protein but are apposed during folding of the molecule (conformational/discontinuous epitope). This distinction is of course not entirely accurate since even small peptides may assume a more complex tertiary structure under appropriate experimental conditions [6]. The peptides may also adopt different conformations in solution, of which only a small number are recognized by the antibody [7]. Once a linear epitope has been defined, a synthetic peptide antigen can be synthesized. Since, for the reasons mentioned above, the antibody is likely to bind to the synthetic peptide with low affinity [8], it is important to test the antibody for binding to a recombinant peptide or peptide obtained by partial proteolysis under denaturing conditions (eg. by Western blotting or by ELISA). If the antibody retains binding to the denatured peptide, it is reasonable to consider synthesizing a peptide. When epitope mapping has been performed with overlapping short synthetic peptides, only the boundaries and length of the peptide require consideration. Greater specificity is obtained with longer ( > 2 0 amino acids) synthetic peptides.

208

Advantages and potential disadvantages of synthetic peptides as antigens Solid-phase immunoassays (RIA, ELISA) utilizing synthetic peptides as antigens are amongst the most sensitive and specific assays currently available for autoantibody detection. Most importantly, these assays are quantitative. In addition, the peptides are synthesized in milligram amounts. Since these reagents are essentially pure, they provide an ideal source of antigen for diagnostic and experimental use. Although recombinant (r) proteins also offer unlimited supplies of antigen, "purified" preparations are frequently contaminated with E. coli proteins which may produce false positive results in immunoassays. Furthermore, most r autoantigens have been synthesized as hybrid proteins fused to an E. coli protein (eg. B-galactosidase, TrpE). Disadvantages of synthetic peptide antigens are the initial expense of peptide synthesis and possibilities for false positive results. False positive results may occur because the peptide shares amino acid sequence homology with other foreign or self antigens (cross-reactivity can occur with homologous tripeptide sequences [9]) or because chemical conjugation alters the antigenicity of the peptide. A third potential disadvantage of synthetic peptide antigens is the absence of posttranslational modifications, although an absolute requirement for autoantibody binding to posttranslationally modified protein antigen has not yet been reported.

Peptide synthesis Methods for chemical synthesis of peptides are beyond the scope of this review and are discussed elsewhere [10]. Some practical considerations should, however, be kept in mind. Chemical synthesis of peptides is an expensive and slow process. For this reason, it is advantageous to map the location of the epitope as precisely as possible prior to synthesis. When the smallest fragment containing the epitope has been identified, it may be possible to further narrow down the location

of the epitope by using computer algorithms. Programs that utilize the primary amino acid sequence to calculate the hydrophobicity, flexibility or predicted secondary structure of the protein have been claimed to accurately predict the location ofepitopes. However, as discussed elsewhere [ 11,12], these programs have, at best, a 60~o success rate. These limitations are largely due to the relatively poor predictive power of the algorithms for secondary structure derived from the primary sequence of the protein [13]. Insufficient information is available on the accuracy of computer prediction of epitopes on autoantigens. However, since autoantibodies may have additional constraints induced by tolerance, it is unlikely that these algorithms will have a higher success rate with autoantigens. Correlations between the predicted and actual epitopes have been accurate for the systemic lupus erythematosus antigen, P2 but not for La (SSB) or Sm (see below). Once synthesized the peptides should, at minimum, be tested for homogeneity by HPLC and for amino acid composition. The peptides can either be applied directly or, following conjugation to non-antigenic proteins (BSA, OVA), to plastic microtiter plates. We favor the latter approach since binding to the plate does not depend on the solubility of the peptide and less peptide is used per assay. It is very important to verify that the peptide adsorbs to the microtiter plate (direct assay) or is conjugated to the carrier protein in the expected molar ratio. The remainder of this review will focus on the utilization of synthetic peptides for the detection and quantification of autoantibodies in human autoimmune diseases. Application of synthetic peptide antigens in murine lupus [ 14-16] and the use of synthetic peptides for epitope mapping (this volume) are discussed elsewhere.

Organ specific autoimmune diseases Myasthenia gravis Anti-acetylcholine receptor (AChR) antibodies in myasthenia gravis preferentially bind to conformational epitopes on the c~subunit of the AChR

209 [17]. Although the N-terminal 80 residues has been proposed to contain the "main immunogenic region" (MIR) of the ct subunit [ 18], at least 3 linear epitopes outside of this region (peptides 195-212, 257-269 and 310-327) were found to be reactive with ,-~ 50-80~o of patient sera [ 19]. The diagnostic utility of these synthetic peptides remain to be determined. Ohta et. al. found that 52~o of patient sera bound to the synthetic peptide 125-148 but that the correlation between total anti-AChR and anti-peptide 125-148 levels was poor [20]. Graves" disease

Anti-thyrotropin (TSH) antibodies bind to, and stimulate, the TSH receptor. Although a deletion of 8 amino acids (A 38-45) abrogated antibody binding [21], reactivity of anti-TSH antibodies with a synthetic peptide corresponding to this region is controversial [22]. In contrast, using a fluid-phase binding assay with an 125I-labeled synthetic peptide, Mori et al. have identified binding to an epitope located between residues 333343 [23]. Other disease controls were not tested in this assay. Primary Biliary Cirrhosis (PBC)

PBC sera bind to several proteins located within the mitochondria. One synthetic peptide, corresponding to residues 81-100 of the antigen dihydrolipoamide acetyltransferase, was able to absorb anti-dihydrolipoamide acetyltransferase activity in most patient sera tested although only at high serum dilutions (1/80,000) [24]. The clinical value of this peptide has not yet been reported.

Multisystem autoimmune diseases S L E and mixed connective tissue disease (MCTD) Anti-ribosomal P The first autoantigen mapped at the amino acid sequence level was the P2 protein [25]. Approx-

imately 10-20 ~o of SLE patients have antibodies directed against the shared C-terminus of the three P proteins located on the large ribosomal subunit [26, 27]. The remarkable uniformity in epitope recognition under denaturing conditions [25, 26] suggested the feasibility of using a synthetic peptide as an antigen [25]. All SLE sera designated anti-P by immunoblotting bound to the synthetic peptide by RIA or ELISA (28). The ability to detect anti-P antibodies in a quantitative assay, demonstrated that anti-P antibody levels fluctuate as much as 10-fold during clinically active disease [28]. This was initially documented in patients with neuropsychiatric lupus [28] and confirmed in other studies [29, 30] although specificity for lupus psychosis is not absolute [31]. Prospective studies will be required to evaluate the predictive value of elevations in anti-P peptide levels for lupus psychosis. Anti-Sm and anti-RNP The B/B' and D proteins are the immunodominant proteins targeted by SLE anti-Sm sera as determined by immunoblotting [ 32-34]. Epitope mapping of SmB/B' revealed that almost all antiSm sera bound to the carboxyl (C)-terminus of SmB/B' [35, 36] which contains a proline-rich motif [G/A (P)n G/M M/G R - where n = 3 or 4 ]. This proline rich motif is repeated in SmB/ B' as well as in other snRNP proteins [37]. Hines et al. synthesized a 27 amino acid peptide corresponding to the C-terminus of Hela SmB and evaluated 45 SLE sera [22 anti-Sm negative and 23 anti-Sm positive as determined by counterimmunoelectrophoresis (CIE)] for binding in an ELISA [38]. Greater than 9 0 ~ of anti-Sm sera but only 14~o of the SLE anti-Sm negative sera were positive in this assay [38]. Some of the CIE negative, ELISA positive, sera were explained by increased sensitivity of the synthetic peptide ELISA. Of patients with other autoimmune disorders, only 3/42 (7~o) were positive in the ELISA. Two of these sera were from patients with MCTD who had anti-RNP antibodies. In view of the shared proline-rich motif present in U 1 RNP proteins [37], it seems likely that these positive results were due to antibodies cross-

210 reactive with SmB. There was an excellent correlation between anti-Sm antibody levels in ELISAs utilizing either the synthetic peptide or rSmB as antigens [38]. Barakat et al. tested SLE and control sera for binding to 7 overlapping synthetic peptides of SmD [39]. Of sera positive for SmD by immunoblotting, 89~o bound to peptide 44-67 and 67~o bound to peptide 1-20 [39]. Interestingly, almost 60~o of apparently unselected SLE sera were positive for binding to peptide 1-20 compared to the reported frequency of anti-Sin of 20-30~/o using less sensitive tests. The low frequency of binding of control sera (4-6 ~o) to these peptides suggest that they may be useful reagents for diagnosis. Longitudinal studies of a small number of SLE patients failed to reveal significant increases in anti-peptide titers during disease activity [40]. Sera from patients with MCTD bind preferentially to proteins located in the U 1 snRNP viz. the 68Kd, A and C proteins. 94~o of MCTD sera were positive for binding to a synthetic peptide corresponding to residues 35-58 of the A protein whereas only 19~o of SLE and none of the control sera bound to this peptide [41]. Unexpectedly, the epitopes on the A protein identified using synthetic peptides were different from the immunodominant sites localized with r A protein [37].

Anti-Ro (SSA) and anti-La (SSB) A synthetic peptide based on the N-terminal sequence of a protein thought to be Ro was sequenced and found to be antigenic for a proportion of SLE sera [42]. This protein was subsequently identified as calreticulin [43 ]. Although SLE sera contain antibodies to this protein, its identity as a Ro antigen remains controversial [44]. A subset of SLE and Sjogren's sera contain antibodies that bind to short peptides close to the C-terminus of the 60 Kd Ro antigen [45]. The exact frequency and clinical utility of these peptides remain to be determined. Epitope mapping of the La protein localized a dominant antigenic site within the C-terminal 72 amino acids of the protein [46]. Although a hydrophilicity plot predicted an epitope within the

C-terminal 26 residues, anti-La sera did not bind to this peptide (unpublished observations). Failure to bind to this peptide is most likely explained by the striking preference of anti-La sera for conformational epitopes [46].

Anti-histone Of 151 SLE sera tested for reactivity to natural and synthetic peptides of the four core histones, H2A, H2B, H3 and H4, the highest frequency (76~o) bound to peptide 1-21 of H3 [47]. Since only 47~o of the same SLE sera were positive for binding to the intact H3 protein [47], the antibodies either preferentially recognize the denatured form of the N-terminus of H3 or some degree of non-specific binding occurs in this assay.

Adult (RA) and juvenile (JRA) rheumatoid arthritis Antihistone antibodies have been reported in a variable proportion of patients with RA but are consistently found in approximately 50% of patients with JRA [48]. Using a similar panel of core histone synthetic peptides tested in SLE (see above), Tuaillon et al. found that 51 and 44~o of RA and JRA sera respectively were positive for binding to the synthetic peptide 1-21 of H3 [49]. As in the case of the SLE sera, a higher frequency of positive results was observed in the peptide ELISA compared with the ELISA using isolated H3. In contrast, 68~o of sera positive for binding to H3 on immunoblots were positive in the peptide ELISA. These findings emphasize that epitope display using synthetic peptides and immunoblotting are frequently similar.

Summary and conclusions The rapid progress made over the last 10 years in the identification of individual autoantigens and in the localization of the epitopes involved, has resulted in a parallel reduction in the complexity of the antigen required for the detection of au-

211 toantibodies. The ability to use synthetic peptides as antigens is a remarkable culmination of this process considering that many antigenic particles contain multiple proteins (eg. Sm consist of 8 or more individual proteins). Despite the fact that patients with SLE have a polyclonal hypergammaglobulinemia, excellent correlations between ELISAs utilizing the P2 or SmB/B' synthetic peptides, ELISAs utilizing r proteins and immunoblotting were obtained [28, 38, 50]. However, false positive/non-specific binding to a P2-BSA-glutaraldehyde conjugate has been observed with serum from old MRL/lpr mice (unpublished observations). In addition, some of the results obtained in human autoinamune diseases suggest that non-specific binding may be problematic in some instances. It is difficult, at present, to know whether the higher frequencies of detection of autoantibodies to certain synthetic peptide antigens reflect increased sensitivity or decreased specificity. Synthetic peptide antigens have beeen used to detect autoantibodies in both organ specific and multisystem autoimmune diseases. In only a small number of cases have these reagents been rigorously tested for sensitivity and specificity. Despite this, synthetic peptides have been shown to be valuable for detection and quantification of autoantibodies in certain clinical situations. Undoubtedly, further progress in epitope mapping of autoantigens coupled with technological advances in protein synthesis and improved prediction of protein structure will lead to a large number of synthetic peptide antigens for research and clinical applications. It is unlikely that short synthetic peptides will substitute for native proteins in all instances since some autoantibodies show a striking preference for conformational epitopes.

Acknowledgements Supported by Grant AR - 38915 and a Research Career Development Award from the National Institutes of Health, U.S.A. Collaboration with Drs. N. Brot, H. Weissbach and W. Danho,

Roche Institute of Molecular Biology is gratefully acknowledged

References 1. Gharavi AE, Chu JL & Elkon KB (1988) Arthritis Rheum. 31:1337-1345 2. Tan EM, Chan EKL, Sullivan KF & Rubin RL (1988) Clin. Immunol. Immunopathol. 47:121-141 3. Bonfa E, Llovet R & Elkon KB (1988) J. Immunol. 140: 2231-2236 4. Benjamin DC, Berzofsky JA, East IJ, Gurd FRN, Hannum C, et al. (1984) Annu. Rev. Immunol. 2:67-101 5. Laver WG, Air GM, Webster RG & Smith-Gill SJ (1990) Cell 6I: 553-556 6. Leach SJ (1983) Biopolymers 22:425-431 7. Niman HL, Houghten RA, Walker LE, Reisfeld RA, Wilson IA, Hogle JM & Lerner RA (1983) Proc. Natl. Acad. Sci. USA 80:4949-4953 8. Jemmerson R (1987) Proc. Natl. Acad. Sci. USA 84: 9180-9184 9. Trifilieff E, Dubs MC, van Regenmortel MHV (1991) Molec. Immunol. 28:889-896 10. Kent SBH (1988) Annu. Rev. Biochem. 57:957-989 11. Huang NC, Sayer JM & Kang AK (1990) J. Immunol. Methods 129:77-88 12. Pelleguer JL, Westhof E & van Regenmortel MHV (1991) Methods Enzymol. 203:176-201 13. Thornton JM (1988) Nature 335:10-I1 14. Bonfa E, Marshak-Rothstein A, Weissbach H, Brot N & Elkon KB (1988) J. Immunol. 140:3434-3437 15. Portanova JP, Cheronis JC, Blodgett JK & Kotzin BL (1990) J. Immunol. 144:4633-4639 16. St. Clair EW, Kenan D, Burch JA, Keene JD & Pisetsky DS (1991) J. Immunol. 146:1885-1892 17. Lindstrom J (1985) Annu. Rev. Immunol. 3:109-131 18. Tzartos SJ, Kokla A, Walgrave SL & Conti-Tronconi BM (1988) Proc. Natl. Acad. Sci. USA 85:2899-2903 19. Brocke S, Brautbar C, Steinman L, Abramsky O, Rothbard J, Neumann D, Fuchs S & Mozes E (1988) J. Clin. Invest. 82:1894-1900 20. Ohta M, Ohta K, Itoh N, Nishitani H, Hara H & Hayashi K (1990) Neurology 40:1776-1778 21. Wadsworth HL, Chazenbalk GD, Nagayama Y, Russo D & Rapoport B (1990) Science 249:1423-1425 22. Murakami M & Mori M (1990) Biochem. Biophys. Res. Comm. 171:512-518 23. Mori T, Sugawa H, Piraphatdist T, Inove D, Enomoto T & Imura H (1991) Biochem. Biophys. Res. Comm. 178: 165-172 24. Van de Water J, Gershwin ME, Leung P, Ansari A & Coppel RL (1988) J. Exp. Med. 167:1791-1799 25. Elkon KB, Skelly S, Parnassa A, Moller W, Weissbach H & Brot N (1986) Proc. Natl. Acad. Sci. USA 83: 7419-7423

212 26. Elkon K, Parnassa AP & Foster CL (1985) J. Exp. Med. 162:459-471 27. Francoeur AM, Peebles CL, Heckman KJ, Lee JC & Tan EM (1985) J. Immunol. 135:2378-2384 28. Bonfa E, Golombek SJ, Kaufman L, Skelly S, Weissbach H, Brot N & Elkon KB (1987) N. Engl. J. Med. 317: 265-271 29. Schneebaum AB, Singleton JD, West SG, Blodgett JK, Allen LG, Cheronis JC & Kotzin BL (1991) Am. J. Med. 90:54-62 30. van Dam AP, Derksen RHWM, Gmelig Meyling F et al. (1990) Ann. Rheum. Dis. 49:779-782 31. van Dam AP, Nossent H, de Jong J, Meilof J, Ter Borg E-J, Swaak T & Smeenk R (1991) J. Rheumatol. 18: 1026-1034 32. Billings PB & Hoch SO (1983) J. Immunol. 131:347-351 33. Habets WJ, de Rooij DJ, Hoet MH, van de Putte LBA & van Venrooij WJ (1985) Clin. Exp. Immunol. 59: 457466 34. Elkon KB & Jankowski PW (1985) J. Immunol. 134: 3819-3824 35. Elkon KB, Hines JJ, Chu JL, & Parnassa AP (1990) J. Immunol. 145:636-643 36. Rokeach LA, Jannatipour M & Hoch SA (1990) J. Immunol. 144:1015-1022 37. Habets WJ, Sillekens PTG, Hoet MH, McAllister G, Lerner MR & van Venrooij WJ (1989) Proc. Natl. Acad. Sci. USA 86:4674-4678 38. Hines JJ, Danho W & Elkon KB (1991) Arthritis Rheum. 34:572-579

39. Barakat S, Briand J-P, Weber J-C, van Regenmortel MHV & Muller S (1990) Clin. Exp. Immunol. 81: 256262 40. Muller S, Barakat S, Watts R, Joubaud P & Isenberg DL (1990) Clin. Exp. Rheumatol. 8:445-453 41. Barakat S, Briand J-P, Abauf N, van Regenmortel MHV & Muller S (1991) Clin. Exp. Immunol. 86:71-78 42. Lieu T-S, Newkirk MM, Capra JD & Sontheimer RD (1988) J. Clin. Invest. 82:96-101 43. McCauliffe DP, Zappi E, Lieu T-S, Michalak M, Sontheimer RD & Capra JD (1990)J. Clin. Invest. 86:332-335 44. Rokeach LA, Haselby JA, Meilof JF, Smeenk RJT, Unnasch TR, Greene BM & Hoch SO (1991) J. Immunol. 147:3031-3039 45. Scofield RH, Dickey WD, Jackson KW, James JA & Harley JB (1991) J. Clin. Immunol. 11:378-388 46. Bini P, Chu JL, Okolo C, & Elkon KB (1990) J. Clin. Invest. 85:325-333 47. Muller S, Bonnier D, Thiry M & van Regenmortel MHV (1989) Int. Arch. Allergy Appl. Immunol. (1989) 89: 288296 48. Bernstein RM, Hobbs RN & Lea DJ (1985) Arthritis Rheum. 28:285-293 49. Tuaillon N, Muller S, Pasquali J-L, Bordigoni P, Youinou P & van Regenmortel MHV (1990) Int. Arch. Allergy Appl Immunol. 91:297-305 50. Magsaam J, Gharavi AE, Parnassa AP, Weissbach H, Brot N & Elkon KB (1989) Clin. Exp. Immunol. 76: 165-171