Confirmation of Cross-Reactivity between LymeAntibody H9724 and Human Heat Shock Protein 60by a Combinatorial Approach

Zhiguang Yu, Jian Tu, and Yen-Ho Chu*

Department of Chemistry, The Ohio State University, Columbus, Ohio 43210

A library of Ac-XXXXXPAPRM decapeptides was preparedon a TentaGel solid support using the approach of splitsynthesis and the strategy of partial chain termination.Epitope screening of the library (175 ∼ 1.4 × 106

decapeptides) with a Lyme monoclonal antibody (H9724)and subsequent MALDI-MS analysis of candidate peptidesfrom colored beads revealed a consensus structure of Xi-DLSXj (Xi ) V, L, Y; Xj ) G, R). These identifiedsequences presented no homology to the natural epitopefrom Borrelia burgdorferi flagellin. However, they werefound to resemble a fragment at the N-terminus of humanheat shock protein (Hsp60). Our results confirmed thatH9724 cross-reacts between bacterial and human pro-teins and provided support for an autoimmunity mecha-nism of Lyme disease.

Peptides are useful in serodiagnosis and vaccine developmentfor infectious diseases. Lyme disease, an inflammatory disorder,was responsible for more than 90% of the arthropod-associatedillnesses reported in the United States during 1992.1 This diseaseis caused by Borrelia burgdorferi spirochete, which is transmittedto humans by ticks carried by deer.2 H9724 is a mouse mono-clonal antibody generated against B. burgdorferi flagellin, the mostprevalent Lyme antigen.3 The epitope from B. burgdorferi flagellinfor H9724 was previously identified as a linear dodecapeptide,EGVQQEGAQQPA (residues 213-224).4,5 This 12-mer epitope,without regions cross-reacting with other pathogens, has recentlyshown promise in increasing the specificity of Lymeserodiagnosis.6-8

Combinatorial libraries are widely used to identify novel leadcompounds, mostly for drug discovery.9-12 We are interested in

exploiting the molecular diversity of combinatorial libraries tosearch ligands as diagnostic agents for disease markers and,ultimately, employing affinity capillary electrophoresis13 as thebinding method for the proposed multiple probes-one diseasediagnosis in solution. Small molecules are excellent diagnosticagents because of their long shelf life and ease of manipulationin electrophoresis. Once identified, the combinatorially selectedligands may have applications in the development of highly specificserodiagnosis and vaccine development for the prevention of Lymedisease. This paper reports our initial effort in developingcombinatorial methods for the search of ligands.

EXPERIMENTAL SECTIONMaterials. TentaGel S NH2 resin was purchased from

Advanced ChemTech. PAL resin was obtained from PerSeptiveBiosystems. Fmoc-protected amino acids, coupling reagentsHOBT and HBTU, and the ninhydrin test kit were from AnaSpec.Ethyldiisopropylamine (DIEA), 1,2-dithioethane, trifluoroaceticacid (TFA), thioanisole, and guanidine hydrochloride were theproducts of Fluka Chemicals. N,N-Dimethylformamide (DMF),acetic anhydride, and acetonitrile of HPLC grade were purchasedfrom Fisher Scientific. Ethyl ether was from Mallinckrodt.Acetylglycine, mercaptoacetic acid, phenol, gelatin, goat anti-mouse whole IgG-alkaline phosphatase conjugate, tablets contain-ing 5-bromo-4-chloro-3′-indolyl phosphate p-toluidine and nitro-blue tatrazolium chloride or p-nitrophenyl phosphate, and cyanogenbromide were obtained from Sigma. Mouse monoclonal antibodyH9724 supernatant was kindly provided by Drs. Stanley Stein andLeonard H. Sigal at the University of Medicine and Dentistry ofNew Jersey. Mouse ascites fluid containing H9724 was purchasedfrom the Health Science Center of the University of Texas at SanAntonio. H9724 antibody was purified and isolated from a proteinA column following the manufacturer’s protocol. Experimentsusing matrix-assisted laser desorption/ionization mass spectrom-etry (MALDI-MS) were carried out using a Kratos KompactMALDI-III instrument.

Construction of the Peptide Library. A linear peptide libraryof Ac-XXXXXPAPRM was prepared on TentaGel S NH2 resin (0.26mmol/g loading) using the standard Fmoc chemistry and the

* Address reprint requests and correspondence to Dr. Yen-Ho Chu, Depart-ment of Chemistry, The Ohio State University, 100 West 18th Ave., Columbus,OH 43210. Tel.: (614) 688-4079. Fax: (614) 292-1685. E-mail: chu@chemistry.ohio-state.edu.(1) Centers for Disease Control and Prevention. Morbid. Mortal. Weekly Rep.

1993, 42, 345-348.(2) Barbour, A. G.; Fish, D. Science 1993, 260, 1610-1616.(3) Barbour, A. G.; Hayes, S. F.; Heiland, R. A.; Schrumpf, M. E.; Tessier, S. L.

Infect. Immun. 1986, 52, 549-554.(4) Fikrig, E.; Berland, R.; Chen, M.; Williams, S.; Sigal, L. H.; Flavell, R. A.

Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 183-187.(5) Schneider, T.; Lange, R.; Ronspeck, W.; Weigelt, W.; Kolmel, H. W. Infect.

Immun. 1992, 60, 316-319.(6) Cretella, S.; Gordon, S.; Flavell, R. A.; Fikrig, E. Eur. J. Clin. Microbiol.

Infect. Dis. 1995, 14, 609-613.(7) Robinson, J. M.; Pilot-Matias, T. J.; Pratt, S. D.; Patel, C. B.; Bevirt, T. S.;

Hunt, J. C. J. Clin. Microbiol. 1993, 31, 629-635.(8) Yu, Z.; Carter, J. M.; Huang, S. Y.; Lackland, H.; Sigal, L. H.; Stein, S.

Bioconjugate Chem. 1996, 7, 338-342.

(9) Lam, K. S.; Lebl, M.; Krchnak, V. Chem. Rev. 1997, 97, 411-448.(10) Balkenhohl, F.; von dem Bussche-Hunnefeld, C.; Lansky, A.; Zechel, C.

Angew. Chem., Int. Ed. Engl. 1996, 35, 2288-2337.(11) Terrett, N. K.; Gardner, M.; Gordon, D. W.; Kobylecki, R. J.; Steele, J.

Tetrahedron 1995, 51, 8135-8173.(12) Gordon, E. M.; Gallop, M. A.; Patel, D. V. Acc. Chem. Res. 1996, 29, 144-

154.(13) Chu, Y.-H.; Cheng, C. C. Cell. Mol. Sci., submitted for publiation (invited

review). Chu, Y.-H.; Avila, L. Z.; Gao, J.; Whitesides, G. M. Acc. Chem. Res.1995, 28, 461-468

Anal. Chem. 1997, 69, 4515-4518

S0003-2700(97)00840-8 CCC: $14.00 © 1997 American Chemical Society Analytical Chemistry, Vol. 69, No. 22, November 15, 1997 4515

method of split synthesis. X represents positions randomized withvarious amino acids. In total, 3.5 g of resin was used to produceapproximately three copies of each peptide in the library. Foreach coupling step in the solid phase synthesis, amino acids (5molar equivalents) were activated with 0.25 M HOBt/HBTU inDMF containing 10% DIEA. Double coupling was employed ineach step. For randomized positions, 10% acetylglycine wasincluded to partially terminate the peptide chain on the resin. TheN-termini of peptides in the library were acetylated using aceticanhydride. The library was deprotected using a solution of 82%TFA, 4% water, 4% thioanisole, 6% mercaptoacetic acid, and 4%phenol at room temperature for 4 h, and the resulting peptidelibrary was washed with ethyl ether to remove scavengers.

Screening of the Peptide Library. Before use for libraryscreening, each peptide sublibrary was washed with doublydistilled H2O and 1× PBST buffer (8 mM Na2HPO4, 1.5 mM KH2-PO4, 137 mM NaCl, 2.7 mM KCl, and 0.1% Tween-20, pH 7.2)three times. The beads in the libraries were blocked with 0.1%gelatin in 2× PBS buffer, pH 7.2, for 1 h. Each library was firstincubated with a 1:2000 diluted goat anti-mouse whole IgG-alkalinephosphatase conjugate for 1 h. After washing with 1× PBST threetimes, 2× PBS twice, and 1× TBS (2.5 mM Tris, 13.7 mM NaCl,and 0.27 mM KCl, pH 8.0) once, a standard alkaline phosphatasesubstrate, 5-bromo-4-chloro-3′-indolyl phosphate p-toluidine saltand nitro-blue tatrazolium chloride (BCIP/NBT), was added, andthe staining reaction was allowed to proceed for 30 min. Severalcolored beads were picked up and set aside. The libraries wereregenerated by washing with 6 M guanidine hydrochloride (pH1.0). Then, supernatant containing mouse monoclonal antibodyH9724 was diluted 1:100 with the blocking buffer and incubatedwith the libraries for 1 h at room temperature. All following stepswere the same as described above in the control experiment.

MALDI-MS Analysis of Peptide Ligands. During libraryscreening, purple colored beads were identified under microscope(60×), individually picked out, and treated with 6 M guanidinehydrochloride (pH 1.0) for 2 h to remove bound antibodies.Peptides were released from each single bead in individual tubesby incubating with a solution of CNBr (30 µg/µL) in 70% formicacid for 16-24 h in the dark. This reaction mixture waslyophilized to remove excessive cyanogen bromide. The residuein each tube was dissolved in 0.1% TFA (20 µL). This peptidesolution (1.0 µL) was mixed with the saturated R-cyano-4-hydroxycinnamic acid in acetonitrile/0.1% TFA (1:1 v/v) (1.0 µL)on the sample plate, air-dried, and analyzed by MALDI-MS.

Synthesis and Purification of Individual Peptides. Epitopepeptides selected and identified from the libraries and four othercontrol peptides were synthesized on PAL resin by standard Fmocchemistry. Each coupling step was monitored by the ninhydrintest. The resin was treated with Reagent K (82.5% TFA, 2.5% EDT,5.0% water, 5% thioanisole, and 5% phenol) for 4 h to release thepeptides. Each peptide was either precipitated from a TFAsolution using cold ethyl ether or extracted with a saturatedammonium bicarbonate solution. Crude peptides were furtherpurified by HPLC on a C18 column (Supelcosil LC-318, 25 cm ×4.6 mm, 5 µm), using a linear gradient (1% min-1) running from100% buffer A (0.1% TFA in water) to 30% buffer B (0.1% TFA inacetonitrile), at a flow rate of 1.0 mL min-1. Quality of peptideswas ascertained by MALDI-MS.

Determination of the IC50 Values by Capture ELISA. TheTentaGel resin-bound peptide Ac-VQQEGABBRM (B ) â-alanine)

was suspended in a blocking buffer (0.1% gelatin in 2× PBScontaining 0.05% Tween 20, pH 7.2). Aliquots of resin suspensionwere evenly distributed into microtiter wells (0.3 mL; 0.25 mgresin/well) and allowed to incubate for 1 h. The synthetic peptide,used for competition, was weighed as its powder form and addedto a solution of the blocking buffer containing 1:50 diluted H9724supernatant. This peptide-antibody mixture was then systemati-cally diluted using the 1:50 diluted H9724 supernatant to preparepeptide concentrations ranging from 10 nM to 10 mM, whilekeeping the antibody concentration constant. These mixtureswere equilibrated at 4 °C for 12 h. At each peptide concentration,the peptide-antibody mixture was added to the resin-containingmicrotiter wells in triplicate and allowed to incubate for 45 min.The solution in each well was removed using a microtippedsyringe. After being washed with 1× PBST twice, the beads wereincubated for an additional hour with a 1:1000 diluted solution ofgoat anti-mouse IgG alkaline phosphatase (100 mL). Followingthe final wash with 1× PBST twice and 2× PBS once, the solublesubstrate p-nitrophenyl phosphate (pNPP) was added (0.3 mL/well), and color development was allowed to occur for 30 min.The reaction of pNPP hydrolysis was stopped with 3 M NaOH(0.2 mL), and the solution was subsequently diluted to 1.0 mLwith water. Absorbance readings were recorded at 405 nm.

RESULTS AND DISCUSSIONLibrary Construction and Screening. Peptide libraries were

prepared on TentaGel resin using the standard Fmoc chemistryand the strategy of partial chain termination.14 A divide/couple/recombine protocol was employed to synthesize an Ac-XXXXX-PAPRM decapeptide library, in which five positions toward theN-terminus were randomized with 17 amino acids (A, D, E, F, G,H, K, L, N, P, Q, R, S, T, V, W, and Y). During the synthesis ofthe library, acetyl glycine (Ac-G) as the capping reagent was usedto acquire partial termination at each coupling step, so that eachof the resin beads carried not only the full length product butalso small amounts of sequence-specific termination products. Acommon linker, PAPRM, was flanked at the C-terminus, wherethe methionine residue was engineered to mediate the cleavageof peptides from the resin for sequence determination. Thearginine was included in peptides to ensure greater efficiency ofionization; therefore, peptides cleaved from a single bead couldbe readily detected by MALDI-MS.14 PAP served as a spacer topresent the randomized peptides.

Using ELISA as the library screening method (H9724 as theprimary antibody, goat anti-mouse IgG conjugated with alkalinephosphatase as the secondary antibody, and BCIP/NBT as thesubstrate), colored beads containing candidate peptides weremanually picked up under microscope (Figure 1) and treated with6 M guanidine hydrochloride (pH 1.0) to remove bound antibod-ies. These beads were separated individually, and then peptidesfrom each bead were released using cyanogen bromide in formicacid and analyzed by MALDI-MS without prior chromatographicseparation or purification. Peptide sequences were simply deter-mined from the mass differences between adjacent members ofthe termination series. These differences uniquely define mostamino acid residues. Figure 2 shows a representative MALDI-MS spectrum of a candidate peptide (Ac-VDLSG) from a singlebead. Sequence determination of the peptides from the colored

(14) Youngquist, R. S.; Fuentes, G. R.; Lacey, M. P.; Keough, T. J. Am. Chem.Soc. 1995, 117, 3900-3906.

4516 Analytical Chemistry, Vol. 69, No. 22, November 15, 1997

beads revealed a new, unexpected consensus structure of XiDLSXj

(Xi ) V, L, Y; Xj ) G, R).Evaluation of the New Ligands. Decapeptides of Ac-XiDLSXj

PAPRM (Xi ) V, L, Y; Xj ) G, R), Ac-VNNSGPAPRM and Ac-VNLSGPAPRM (slightly altered sequences), and Ac-VQQEGPA-PRM (part of the natural epitope) were individually synthesizedon the TentaGel resin. Upon incubation with the antibody H9724,only Ac-XiDLSXj PAPRM (Xi ) V, L, Y; Xj ) G, R) peptides showedpositive ELISA reactions, demonstrating that the peptide-antibodyrecognition was sequence specific.

As PAPRM linker may also contribute to the antibody binding,the linker in the decapeptide Ac-VDLSGPAPRM was truncatedfrom the C-terminus one-by-one. The ELISA results showed thatthe minimum binding sequence was Ac-VDLSGP (data notshown). Therefore, one proline from the linker participated inantibody binding.

These newly identified sequences had no similarity to thenatural epitope in B. burgdorferi flagellin. To uncover the reason,the previously reported 12-mer epitope for H9724 was systemati-cally shortened and studied on a TentaGel support using ELISA.14

Results showed that the length of the peptide greatly influencedits antibody recognition: a minimum of 6-mer peptide (VQQEGA)was required for binding if it was properly presented on the solidsupport.15 Since our library was constructed and randomized atonly five positions, it became evident that the possibility of findinga consensus sequence similar to the natural epitope from thislibrary was eliminated.16

Although the new peptide sequences show no homology tothe native bacterial epitope, they resemble a fragment at theN-terminus of human heat shock protein 60 (Hsp60).17 Heatshock proteins are a family of evolutionarily conserved proteinsthat are known to facilitate the correct folding of misfolded orunfolded proteins. Human Hsp60 protein was previously reportedto cross-react with H9724.18,19 A fragment of human Hsp60,

MLQGVDLLADAVA (residues 14-26), was suggested to beresponsible for this cross-reactivity.20 We studied the interactionof H9724 with the commercially available, recombinant humanHsp60 and found that this human protein competes with resin-bound Ac-VDLSGPAPRM peptide for antigen binding sites inH9724. Under identical ELISA condition, human serum albuminused as a control protein showed no inhibition to the bindinginteraction of H9724 to the resin-bound peptide. In molecularmimicry, it has been documented that a high degree of similarityin primary sequences may not be necessary for cross-reactivity.21

In some cases, two amino acid residues are sufficient to facilitatethe antibody binding.22 As our combinatorially identified se-quences share the homology with a fragment in human Hsp60(i.e., Ac-VDLSGPAPRM and MLQGVDLLADAVA), the previouslyreported cross-reactivity between human Hsp60 and H9724 is,therefore, confirmed.23

Determination of IC50 Values of New Ligands. To verifythat the newly identified peptide ligands bind to H9724 in solution,six hexapeptides Ac-XiDLSXj P (Xi ) V, L, Y; Xj ) G, R), both a12-mer peptide Ac-EGVQQEGAQQPA (the natural epitope) anda hexapeptide Ac-VQQEGA (the minimum binding element) fromB. burgdorferi flagellin, a hexapeptide Ac-MLQGVD (the minimumbinding fragment) from human Hsp60,20 and a negative controlpeptide Ac-FLTDGT were synthesized on the PAL resin. Thesepeptides were cleaved from the resin using the reagent K andpurified by HPLC. Each peptide gave its expected molecularweight in MALDI-MS analysis.

A capture ELISA assay24,25 was employed to measure the IC50

values of these peptides in the inhibition of H9724 binding to theepitope Ac-VQQEGABBRM on beads. Results in Table 1 showthat the newly identified peptide ligands give IC50 values rangingfrom 60 to 300 µM, which correlated well with the literature valuesoften observed for the binding systems of small peptide-antibodyinteractions.26 The 12-mer natural epitope has its IC50 value aslow as 105 nM. The 6-mer peptides from the B. burgdorferiflagellin (Ac-VQQEGA) and human Hsp60 (Ac-MLQGVD) giveIC50 values similar to those of the newly identified ligands (Table1). A 6-mer control peptide, Ac-FLDTGT, does not bind H9724up to 6 mM. It is, therefore, concluded that new ligands identifiedfrom combinatorial libraries bind specifically to H9724 in solution.

In general, the size of a chemical library is limited to sixrandomized positions (∼108 peptides). Ligands with weak tomedium affinities are likely selected and identified from the first-generation libraries. These identified lead compounds can then

(15) Yu, Z.; Chu, Y.-H. Bioorg. Med. Chem. Lett. 1997, 7, 95-98.(16) Our preliminary results from the library screening of Ac-XXXXXXBBRM

decapeptides suggested that both the bacterial flagellin epitope and humanhsp60 epitope could be identified when a different linker was employed topresent the peptides (Tu, J.; Yu, Z.; Chu, Y.-H., unpublished results).

(17) Jindal, S.; Dudani, A. K.; Singh, B.; Harley, C. B.; Gupta, R. S. Mol. CellBiol. 1989, 9, 2279-2283.

(18) Sigal, L. H. J. Infect. Dis. 1993, 167, 1372-1378.

(19) Dai, Z.; Lackland, H.; Stein, S.; Li, Q.; Radziewicz, R.; Williams, S.; Sigal, L.H. Biochim. Biophys. Acta 1993, 1181, 97-100.

(20) Dai, Z.; Stein, S.; Williams, S.; Sigal, L. H. In Stress Proteins in Medicine;van Eden, W., Young, D. B., Eds.; Marcel Dekker: New York, 1996; pp147-162.

(21) For example, PESDQPDL in the acetylcholine receptor is highly similar tothe fragment PESDQDQL in polyoma virus middle T antigen, but no cross-reacting antibodies could be induced. On the contrary, PNATQPEL inherpes simplex virus glycoprotein D induces strong immunological cross-reactivity to the acetylcholine receptor (Colman, P. M.; Laver, W. G.;Varghese, J. N.; Baker, A. T.; Tulloch, P. A.; Tir, G. M.; Webster, R. G.Nature 1987, 326, 358-363).

(22) Anda, P.; Backenson, P. B.; Coleman, J. L.; Benach, J. L. Infect. Immun.1994, 62, 1070-1078.

(23) Sigal, L. H.; Curran, A. S. Annu. Rev. Publ. Health 1991, 12, 85-109.(24) Friguet, B.; Chaffotte A. F.; Djavadi-Ohaniance, L.; Goldberg M. E. J.

Immunol. Methods 1985, 77, 305-319.(25) Kim, B. B.; Dikova, E. B.; Sheller, U.; Dikov, M. M.; Gavrilova, E. M.;

Egorov, A. M. J. Immunol. Methods 1990, 131, 213-222.(26) Houghten, R. A.; Pinilla, C.; Blondelle, S. E.; Appel, J. R.; Dooley, C. T.;

Cuervo, J. H. Nature 1991, 354, 84-86.

Figure 1. ELISA assay of a library of Ac-XXXXXPAPRM decapep-tides (1.4 × 106) against the Lyme antibody H9724. Only a portion ofthe beads in the library is shown [60×]. The bead in the center wasidentified as a hit.

Analytical Chemistry, Vol. 69, No. 22, November 15, 1997 4517

provide the basis for the design of tight-binding ligands. Forexample, employing suitable linkers, two or more copies of thelead sequence may be engineered as a single molecule, so thenewly designed compound may bind tightly to the target receptorin multivalent fashions.27,28 A second generation of peptide librarycan also be generated by incorporating the sequence from thefirst-generation leads onto the C- or N-terminus (such as VDLSGPin our case), while other positions toward the N- or C-terminusare further randomized. Tight-binding ligands may be discoveredafter two or three cycles of optimization.

CONCLUSIONSIn combination of the strategy of partial chain termination and

the sensitivity of MALDI-MS, the one-bead-one-compound com-

binatorial library is a powerful procedure to search ligands forantibodies and other protein receptors. It has been reported thatthe design of libraries, including the length of peptides and thelinkers used on the solid support, could dramatically affect theoutcome of the ligand search.15 In our experiments, a library ofAc-XXXXXPAPRM decapeptides was constructed on the TentaGelresin. Using ELISA, screening of the library with a Lyme antibodyH9724 and subsequent MALDI-MS analysis of candidate peptidesfrom colored beads revealed a consensus structure. Though theidentified new sequences are not homologous to the minimumbinding epitope sequence in the bacterial flagellin, they resemblea part of human heat shock protein 60, which has been proposedto cross-react with the monoclonal antibody H9724 and otherpolyclonal antibodies in Lyme patients’ sera. The competitionELISA experiments verified that these peptide ligands bindspecifically to the antibody in solution. Our combinatorial ap-proach confirmed that H9724 cross-reacts with human Hsp60. Thiscombinatorial method should be an efficient tool in identifyingboth the natural and cross-reactive epitopes and may be extendedfor the study of autoimmune diseases.

ACKNOWLEDGMENTWe are grateful to Ya-Huei Tu (Whitehead Institute for

Biomedical Research, Cambridge, MA) for her assistance withFigure 1 and the Society for Analytical Chemists of Pittsburgh(SACP Starter Grant) and the American Association for ClinicalChemistry (AACC Van Slyke Research Grant) for their supportof this work.

Received for review August 5, 1997. Accepted September29, 1997.X

AC9708405(27) Appel, J. R.; Muller, S.; Benkirane, N.; Houghten, R. A.; Pinilla, C. Pept.

Res. 1996, 9, 174-182.(28) Uma, N. S.; Griffin, J. H.; Nicas, J. Am. Chem. Soc. 1996, 118, 13107-

13108. X Abstract published in Advance ACS Abstracts, November 1, 1997.

Table 1. Relative Affinities of Peptides Binding to theH9724 Antibody

peptide IC50a

Ac-VDLSRP 63 µMAc-LDLSRP (2)b 62 µMAc-YDLSRP 105 µMAc-VDLSGP (2) 130 µMAc-LDLSGP (2) 103 µMAc-YDLSGP 301 µMAc-VQQEGA 118 µMAc-MLQGVD 25 µMAc-EGVQQEGAQQPA 105 nMAc-FLTDGT >6 mM

a IC50 values were calculated for inhibition of antibody binding tothe TentaGel-bound Ac-VQQEGABBRM peptide. Estimated RSD, 20%.b Shown in parentheses is the number of copies of ligands identifiedfrom the libraries.

Figure 2. MALDI mass spectrum of peptides (M*, homoserine lactone) cleaved from a single bead. This bead was recognized by H9724 andcontained a peptide sequence of Ac-VDLSGPAPRM. Note that only 5% of the peptides from a single bead was used for sequence decoding.

4518 Analytical Chemistry, Vol. 69, No. 22, November 15, 1997