276

Peptides Mimicking Sialyl-Lewis A Isolated from a Random Peptide Library and Peptide Array

INSUG O, THOMAS KIEBER-EMMONS,a LASZLO OTVOS, JR., AND MAGDALENA BLASZCZYK-THURINb

The Wistar Institute and aUniversity of Pennsylvania, Philadelphia, Pennslvania 19104, USA

Cell surface carbohydrate structures are an important class of tumor antigens.SA-Lea and its structural isomer SA-LeX have been identified as carbohydratedeterminants expressed on many carcinomas,1 and both have been shown to repre-sent functional ligands of selectins. The crucial role of selectin-dependent neutrophiladhesion in their recruitment process and metastasis was confirmed by in vivo block-age of E-selectin–dependent interaction.2 These data imply that analogous molecu-lar mechanisms do indeed underlie inflammation and metastasis and that similartherapeutic approaches can be used to intervene with both processes. Highly diversepeptide libraries offer many distinct advantages over difficult chemical or enzymaticsynthesis of complex carbohydrates, providing notably inexpensive and rapid iden-tification and optimization of novel ligands. The peptides described here provide ex-cellent leads for the development of potent antagonists of carbohydrate-proteininteraction and in particular antimetastatic and antiinflammatory therapeutic agents.

MATERIALS AND METHODS

Monoclonal antibody (mAb) NS19-9 was generated at the Wistar Institute.3 Pep-tides were synthesized at Research Genetics, Inc. (Huntsville, Alabama). SA-Lea

was obtained from Glycotech, Inc. (Rockville, Maryland). The 12-mer peptide li-brary for these studies was obtained from Invitrogen Inc. (Carlsbad, California).4

The ability of peptides to block mAb recognition of SA-Lea carbohydrate was deter-mined in competition ELISA, using various peptide concentrations. IC50 valueswere calculated by nonlinear least-squares regression of a four-parameter logisticequation. An array of 163 synthetic 12-mer peptides was synthesized by standard F-moc chemistry on polyethylene glycol modified cellulose membrane at the WistarInstitute and tested for binding of mAb NS19-9.5

bAddress for correspondence: Dr. M. Blaszczyk-Thurin, The Wistar Institute, 3601 SpruceStreet, Philadelphia, PA 19104. Phone, 215/898-3829; fax, 215/898-3868.

e-mail, mthurin@wistar.upenn.edu

277O et al.: PEPTIDES MIMICKING SIALYL-LEWIS A

RESULTS

Random Peptide Library Screening. Several bacterial clones that bind the SA-Lea–specific NS19-9 mAb were isolated and sequenced. Clones isolated with thecarbohydrate-specific mAb in the final selection cycle were tested for protein ex-pression using SDS-PAGE and identified after probing by Western blot with themAb. Peptide library screening yielded families of peptides with unique consensussequences. Two distinct consensus sequences, GXWXXVLEG and VVGXP, wereidentified in families of peptides isolated with mAb NS19-9 (TABLE 1). This may in-dicate that peptides based on two different motifs isolated with the same mAb canmimic different structural topographies of SA-Lea carbohydrate, and these subsetsof peptides may very likely represent nonoverlapping surfaces of cognate antigen.

Identification of Sequences Critical for mAb Binding Using Peptide Array. Toidentify the residues critical for mAb binding and to analyze amino acid substitu-tions that might improve peptide-mAb interaction, we generated an array of 163 pep-tides based on the sequence of peptide #4 isolated from family II and probed withmAb NS19-9 (FIG. 1). Results suggest that most substitutions at the very N-terminus(residues 1 and 2) were well tolerated and did not influence mAb binding, whereascritical residues were clearly identified at positions 3 to 5. The most important singleamino acid for mAb binding was Trp at position 3, because most of the amino acidsubstitutions at this position blocked mAb recognition. Replacement of Trp3 with H,Y, A, D, and S completely abolish mAb binding, whereas M significantly decreasedit. Similarly, substitutions of W at position 5 with H, A, R, and K abolished mAbbinding. Several of the substitutions within the identified consensus sequence, inparticular VVGK, were not tolerated by the mAb, whereas others allowed for mAbbinding. No preference for amino acids was evident from the substituted peptideswithin positions 10 to 12 of the C-terminus. By contrast, several amino acid replace-ments resulted in increased binding of the peptides for mAb. The most favorableamino acid with respect to mAb binding was substitution of Trp at position 5 withPhe resulting in the sequence DLWDFVVGKPAG (A#44) that displayed increasedbinding affinity as compared to the original peptide. Amino acid substitutions withinthe consensus sequence, at positions 6 to 10 mainly with K, R, and E, also resultedin higher binding of mAb as compared with the original peptide. These data demon-strate that peptides with higher binding properties for the anticarbohydrate mAbwere selected using the peptide array approach. These results further confirm that ar-

TABLE 1. Peptide sequence families mimicking SA-Lea carbohydrate structure

I #2: VGIWSVVSEGSR II #1: RCSVGVPFTMES

#3 QDGVWEHVLEGG #4: DLWDWVVGKPAG

#15: VELSGRGGLCTW #12: VIGAASHDEDVD

#18: TIEPVLAEMFMG #14: DKETFELGLFDR

#15: FSGVRGVYESRT

#19: PDDAPMHSTRVE

278 ANNALS NEW YORK ACADEMY OF SCIENCES

omatic amino acids may play an important role in mimicking carbohydratestructures.

Synthetic Peptides Mimic the Carbohydrate Antigen Detected with mAb NS 19-9.To test whether peptide #4 and peptide A#44 are true mimics of SA-Lea and to de-termine the concentration of peptides required for blocking of 50% of mAb bindingto native carbohydrate antigen, competition ELISA was carried out. Both peptidesblocked the binding of NS19-9 to a fixed amount of carbohydrate antigen in a dose-dependent manner. The IC50 for peptide #4 blocking of mAb–SA-Lea binding was700 µM. Peptide A#44 showed more prominent inhibition of the mAb-SA-Lea bind-ing as compared to peptide #4, as demonstrated with the IC50 value of 70 µM. Thissuggests that substitution of Phe within the original amino acid sequence generateda peptide with higher affinity for the mAb better mimicking topography for SA-Lea.Overall, our data demonstrate that the peptides sterically interfere with mAb bindingto carbohydrate antigen, implying that the sequences DLWDWVVGKPAG and DL-WDFVVGKPAG represent solvent-accessible epitopes and that the peptides repre-sent cognate determinants for the antibody.

Secondary Structure of Peptides Mimicking Carbohydrate. Both peptides #4 andA#44 highlight the functional role played by the aromatic-X-aromatic motif withinthe peptide. Secondary structure analysis indicates some propensity for extendedand helix structures centered on the W/FVVG region using a neutral net analysis. Itis possible that these structure types might be realized within the antibody-combin-ing site. This is consistent with modeling and crystal structure analysis of this motiftype which suggested their ability to adopt type I and type II turns within the anti-body-combining site. The increased binding of peptide with substitution of Phe forTrp would suggest that the hydrophobic stacking interactions are important for in-creased antibody binding and consequently antigenic mimicry.

FIGURE 1. Reactivity of mAb NS19-9 with a series of solid-phase single amino acidsubstituted dodecapeptides based on peptide #4. (Lower panel) Background binding to themembrane, using BSA instead of primary antibody (upper panel).

279O et al.: PEPTIDES MIMICKING SIALYL-LEWIS A

COMMENTS

Functional equivalence of chemically dissimilar molecules such as carbohydratesand proteins sharing common surface topology has been identified previously as anaturally occurring phenomenon. Combinatorial technologies available in recentyears have provided an avenue to dissect the molecular basis for such mimicry. Usinga combinatorial library screening approach we isolated families of mimics of tumor-associated antigen and E-selectin ligand. We chose one sequence as a lead peptide todelineate the specific residues that may contribute to the mimicry of carbohydratestructures by re-screening of a peptide array. The mAb could tolerate a variety ofamino acid substitutions within the lead peptide sequence, still retaining functionalspecificity. Furthermore, cross-reactive peptides of higher affinity were identified.This suggests that different amino acids can improve structural mimicry within iden-tified peptide. The peptides with various consensus sequences that were identifiedwith the same mAb suggest that indeed different residues act as structural mimics.Alternatively, distinct consensus sequences mimic nonoverlapping topographies of acarbohydrate epitope recognized by the antibody. The prospect of finding mimickingpeptides of carbohydrate tumor antigens that competitively inhibit carbohydrate-specific receptor interaction will allow for the design of antagonists of E-selectin andother endolectins with enhanced therapeutic potential to prevent metastasis.

REFERENCES

1. BECHTEL, B., A.J. WAND, K. WROBLEWSKI, H. KOPROWSKI & J. THURIN. 1990. Con-formational analysis of the tumor-associated carbohydrate antigen 19-9 and its Leablood group antigen component as related to the specificity of monoclonal antibodyCO19-9. J. Biol. Chem. 265: 2028–2037.

2. BRODT, P., L. FALLAVOLLITA, R.S. BRESALIER, S. METERISSIAN, C.R. NORTON & B.A.WOLITZKY. 1997. Liver endothelial E-selectin mediates carcinoma cell adhesion andpromotes liver metastasis. Int. J. Cancer 71: 612–619.

3. MAGNANI, J.L., M. BROCKHAUS, D.F. SMITH, V. GINSBURG, M. BLASZCZYK, K.F.MITCHELL, Z. STEPLEWSKI & H. KOPROWSKI. 1981. A monosialoganglioside is amonoclonal antibody defined antigen of colon carcinoma. Science 212: 55–56.

4. LAVALLIE, E.R., E.A. DIBLASIO, S. KOVACIC, K.L. GRANT, P.F. SCHENDEL & J.M.MCCOY. 1993. A thioredoxin gene fusion expression system that circumvents inclu-sion body formation in the E. coli cytoplasm. Bio/Technology 11: 187–193.

5. RUDIGER, S., L. GERMEROTH, J. SCHNEIDER-MERGENER & B. BUKAU. 1997. The sub-strate specificity of the DnaK chaperone determined by screening cellulose-boundpeptide libraries. EMBO J. 16: 1501–1507.