Prediction of major histocompatibility complex binding regions of protein antigens by sequence pattern analysis.

Abstract

We have previously experimentally analyzed the structural requirements for interaction between peptide antigens and mouse major histocompatibility complex (MHC) molecules of the d haplotype. We describe here two procedures devised to predict specifically the capacity of peptide molecules to interact with these MHC class II molecules (IAd and IEd). The accuracy of these procedures has been tested on a large panel of synthetic peptides of eukaryotic, prokaryotic, and viral origin, and also on a set of overlapping peptides encompassing the entire staphylococcal nuclease molecule. For both sets of peptides, IAd and 1Ed binding was successfully predicted in =75% of the cases. This suggests that definition of such sequence "motifs" could be of general use in predicting potentially immunogenic peptide regions within proteins.There is at present compelling evidence in support of the concept that T cells recognize a complex formed between major histocompatibility complex (MHC) molecules and "processed" antigen. The structural requirements for this interaction have been the object of intense investigation. In a previous study (1), the effect of a large number of single amino acid substitutions of an ovalbumin-derived peptide that binds strongly to ] on its capacity to bind purified IAd molecules was examined. Significant changes were only detected in a six-amino acid core region ] that had also been strongly implicated as crucial for binding by analysis of a set of N-and C-terminal truncated analogs. In a subsequent study (2), it was demonstrated that unrelated peptides that are good lAd binders contained regions ("motifs") that were structurally similar to each other and to . More recently, it was shown (A.S., L. Adorini, S.B., E.A., and H.M.G., unpublished data) that a different amino acid sequence pattern characterizes the interaction between peptide antigens and the other class II MHC molecule, lEd. The motif recognized by IEd molecules appears to be independent of the motif recognized by IAd and is defined by the presence of positively charged amino acids (refs. 3 and 4; A.S., L. Adorini, S.B., E.A., and H.M.G., unpublished data).In the present study, this information was used to derive a procedure that identifies motifs. Using a large panel of peptides, the presence of either an IAd or IEd motif as determined by this procedure was compared to the experimentally determined capacity of a given peptide to bind IAd and IEd. The peptide data base contained 62 peptides and, in addition, a set of peptides encompassing the entire staphylococcal nuclease molecule. In these sets ofpeptides, IAd and IEd binding was successfully predicted in ==75% of the cases, suggesting that similar procedures should be useful in selecting peptides capable of interacting with other MHC specificities, identify possible autoantigens in self molecules or specific blockers of autoimmune diseases, and because ofthe previously demonstrated strong correlation between MHC binding and immunogenicity (5) be useful in selecting potentially immun...

A peptide derived from an autoantibody can stimulate t cells in the (nzb × nzw)f1 mouse model of systemic lupus erythematosus

“…Although Sette and colleagues (16) have shown that a triple-base motif (BXBXB) such as KFKGK, in which 3 positively charged amino acids are separated by 1 or 2 uncharged amino acids, can bind to I-Ed and induce T cell proliferation (f6), we have not studied the sequences within peptide 58-69 that are optimal for stimulation of NZB/W T cells, nor have we proven that they are presented solely by I-Ed. These investigations are in progress.…”

“…Sette et a1 (16) showed that a triple-base motif BXBXB (where B is a basic amino acid and X is an uncharged amino acid) can bind the I-Ed class I1 molecule, which NZB/W mice express. The 12-mer contains such a motif, KFKGK.…”

Induction of cytotoxic T lymphocytes in mice against the principal neutralizing domain of HIV-1 by immunization with an engineered T-cytotoxic-T-helper synthetic peptide construct

“…Based on the observations that the tripalmitoil-S-glycerylcysteinyl-seryl-serine moiety used in the construct of Deres et al ( 13) behaves as a TD,, in another type of construct (18) and the ovalbumin tryptic fragment 229-276, which contains a predicted TDi, according to Sette's algorithm ( 19) and a proven TD, (20) but not the 242-276 (which only contains the TD,) is able to induce CTL in vivo (1 I), we speculated that synthetic colinear peptide constructs of the type TDh-TD, or TD,-TDh might be able to induce CTL in vivo. Ishioka et al (12) have suggested a similar proposition; however, to our knowledge, no experimental evidence has yet been published to validate these hypothesis.…”

Induction of arthritis in HLA–DR4–humanized and HLA–DQ8–humanized mice by human cartilage proteoglycan aggrecan but only in the presence of an appropriate (non‐MHC) genetic background

“…PG was isolated from human cartilage, and the GAG side chains were depleted by digestion with chondroitinase ABC (Seikagaku America, Falmouth, MA) as described previously (38,39). Peptide sequence selection was based on the presence of MHC-binding motifs (40), ␣-helix amphipathic sequence motifs that are common in T cell epitopes (41), and the hydrophobic "strip-of-helix" algorithm (42). As a result of this selection, a total of 143 aggrecan core protein-specific peptides (84 located in the G1 domain and 59 in other regions) covering all predicted sites, with a 3-mer offset (43) and ϳ25% nonpredicted regions, were synthesized by the pin method (Chiron Mimotopes, Raleigh, NC).…”

“…Also, we were not surprised to identify a few, originally nonpredicted, T cell epitopes. However, we found it interesting that peptides spanning 2 regions of the PG core protein (p22-51 and p253-285) proved to be highly immunogenic in both DR4-and DQ8-transgenic mice, since these sequences were not predicted to be potential T cell epitopes for either DRB1*0401 (46,47) or DQB1*0302 alleles in studies using multiple prediction methods (40)(41)(42). By contrast, a few peptides in the G1 domain and a peptide sequence ( 1785 GAYYGSGTPSSFP) in the chondroitin sulfateattachment region with predicted high-affinity binding to DRB1*0401, which were selected as strong T cell epitopes (46,47), were found not to be recognized in the humanized mice.…”