HLA DR4w4-binding motifs illustrate the biochemical basis of degeneracy and specificity in peptide-DR interactions.

In the present study, we describe the definition of a DRB1*0401 (DR4w4)-specific motif. The strategy used entailed a three-step process: 1) screening a large set of unrelated peptide ligands to identify good MHC binders; 2) truncation analysis of several DR4w4 binding peptides of high affinity to identify the crucial core-binding regions; 3) the use of single amino acid substitutions of the DR4w4-binding peptide hemagglutinin (HA) 307-319 to elucidate the specific residues crucial for binding. The DR4w4 motif is characterized by the presence of a hydrophobic or aromatic (F, W, Y, L, I, V, M) anchor residue in position 1, and a second hydroxyl (S, T) or aliphatic (L, I, V, or M) anchor residue in position 6. Furthermore, positive charges (R, K) are not allowed in positions 4, 7, and 9, and negative charges (D, E) are not allowed in position 9. This motif was present in 92% of good (IC50 < or = 100 nM) DR4w4-binding peptides, but less than 25% of the negative (IC50 > 45 microM) binders, indicating that the presence of the motif is necessary, but not sufficient for good DR4w4 binding capacity. The results of the present study are discussed in relation to previous work defining binding motifs and rules for other DR alleles, illustrating how different DR alleles bind variations on a similar structural theme. Finally, using two different peptide ligands [tetanus toxoid 830-843 and HA 307-319] as model systems, it is demonstrated how the fine allelic specificity of the DR binders can be predictably modulated by introducing subtle changes in the primary amino acid sequence.

Peptide stability in drug development. II. Effect of single amino acid substitution and glycosylation on peptide reactivity in human serum.

The determination of peptide stability in human serum (HS) or plasma constitutes a powerful screening assay for eliminating unstable peptides from further development. Herein we report on the stability in HS of several major histocompatibility complex (MHC)-binding peptides. Some of these peptides are in development for the novel treatment of selected autoimmune disorders such as rheumatoid arthritis and insulin-dependent diabetes. For most of the l-amino acid peptides studied, the predominant degradation mechanism is exopeptidase-catalyzed cleavage. Peptides that were protected by d-amino acids at both termini were found to be more stable than predicted, based on additivity of single substitutions. In addition, N-acetylglucosamine glycopeptides were significantly stabilized, even when the glycosylation site was several amino acids from the predominant site(s) of cleavage. This indicates that long-range stabilization is possible, and likely due to altered peptide conformation. Finally, the effect of single amino acid substitutions on peptide stability in HS was determined using a model set of poly-Ala peptides which were protected from exopeptidase cleavage, allowing the study of endopeptidase cleavage pathways.

MHC class II molecules bind indiscriminately self and non-self peptide homologs: effect on the immunogenicity of non-self peptides.

Synthetic peptides spanning the entire sequence of both human and mouse beta 2-microglobulin (beta 2M) have been tested for their capacity to bind to three different mouse (I-Ad, I-Ed, and I-Ak) or human (DR1, DR2, and DR5) class II molecules. The results demonstrate that class II molecules do not discriminate between self and non-self peptides. When the immunogenicity of the human beta 2M peptides was measured by their ability to prime H-2d mice for in vitro T cell proliferation, it was found that peptides incapable of binding class II molecules in vitro were also non-immunogenic in vivo. Interestingly, however, several binders, including the human beta 2M peptide 1-16, the best binder in this series to Iad molecules, were found to be non-immunogenic. Since the corresponding mouse beta 2M peptide 1-16 was also capable of binding to Iad molecules, this suggested that lack of responsiveness to the non-self peptide could arise either from central or peripheral tolerance induced by the self homolog. Alternatively, lack of responsiveness could arise from other mechanisms, such as negative selection by other non-homolog sequences or lack of suitable T cell receptor genes. To discriminate between these possibilities, H-2d mice with disrupted beta 2M genes were immunized with the human beta 2M peptide 1-16. This peptide also failed to prime for T cell responsiveness in beta 2M-negative mice, suggesting that a hole in the T cell repertoire for this antigen was not mediated by negative selection or peripheral tolerance induced by self beta 2M peptides.

Functional consequences of engagement of the T cell receptor by low affinity ligands.

The mechanisms involved in TCR antagonism by Ag analog/MHC have been analyzed. A detailed structure-activity relationship study indicated that modification of any of the major T cell contact residues of the peptide molecule can yield a powerful antagonist. It was also shown that as the analog structure increased in similarity to the Ag, the capacity to antagonize Ag-TCR interaction increased up to the point that the analogs themselves became antigenic. These data strongly suggest an affinity-related mechanism whereby a certain affinity is required for signaling through the TCR, and that below this level there can be sufficient affinity to engage the receptor such that triggering does not occur and antagonism can be detected. Taking advantage of this information, antagonist peptides active down to the 10 nM range were engineered. Thus, this approach demonstrates for the first time a rational approach to designing effective, selective low m.w. compounds with high potential in treatment of allergies and autoimmune diseases.

High affinity for class II molecules as a necessary but not sufficient characteristic of encephalitogenic determinants.

A direct binding assay specific for IAs molecules has been developed and its immunological relevance validated by examining, for a panel of nine different synthetic peptides, the correlation between their capacity to bind purified IAs and to inhibit IAs-restricted antigen presentation. The IAs assay thus developed has then been used to study the IAs binding affinity of a set of overlapping peptides spanning the entire myelin basic protein (MBP). It was found that the encephalitogenic MBP region corresponds to peptides with high MHC binding affinities. Other regions of the MBP that have not been described as being pathogenic in the context of IAs molecules have also been found to be high IAs binders, suggesting that variables other than MHC affinity are also involved in determining the pathogenic potential of self-derived determinants.

On the interaction of promiscuous antigenic peptides with different DR alleles. Identification of common structural motifs.

We have investigated the interaction between DR1 molecules and the two antigenic peptides, tetanus toxoid 830-843 and hemagglutinin 307-319, previously known to bind most DR alleles (degenerate binding) and to be recognized by the same T cell clones in the context of different DR alleles (promiscuous T cell recognition). The DR1 affinity of these two peptides was compared with that of two other different T cell epitopes (pertussis toxin 30-42 and ragweed allergen Ra3 51-65). It was found that degeneracy and promiscuity were associated with high affinity interactions, whereas binding and T cell selectivity were associated with weaker interactions. Thus, the selectivity of DR-peptide interactions, as is commonly observed with the antibody molecule, appears to be inversely correlated to affinity. Several singly substituted analogs of the hemagglutinin 307-319 determinant have also been tested for capacity to bind various DR alleles (DR1, DR2, DR5, and DR7). The results obtained suggest that this determinant may bind the different DR alleles in a similar orientation. Similar conclusions were reached when the interaction between the tetanus toxoid 830-843 determinant and three different DR alleles (DR1, DR2, and DR7) was studied following the same experimental approach. When crucial DR-binding residues of the two peptides were compared, it was found that they were very similar in both chemical nature and spacing in the peptide primary structure, suggesting that the two peptides may bind DR in a very similar orientation. Finally, a putative motif has been derived and shown to be present in a majority of the DR binders tested, but only in a minority of the non-DR binding peptides.

Single amino acid changes in DR and antigen define residues critical for peptide-MHC binding and T cell recognition.

Single amino acid substitutions of Ag and MHC were used to analyze the fine structure of the influenza hemagglutinin (HA)-derived epitope (HA 307-319) recognized in the context of DR7 molecules by a T cell clone. Putative T cell (HA 308, 310, 311, 313, and 316) and DR (HA 309, 312, and 317) contact residues of the Ag were identified by the use of single amino acid-substituted analogs that were tested for their T cell-activating and DR-binding capacities. The peptide-DR7-T cell interaction was further characterized by the use of a panel of 13 site-directed DR7 mutant transfectants analyzed for their capacity to present Ag to T cells, and for their purified mutant DR7 molecules to bind HA 307-319 or its single amino acid-substituted analogs. Eight mutants lost their Ag-presenting function, whereas only one had any decrease in peptide binding. Finally, for three of the mutants it was possible to correct the deleterious effects of mutation by using a particular single amino acid-substituted analog of the peptide molecule. The observed pattern of complementation led to a model that predicts that the Ag assumes an extended conformation, with a turn, in the binding groove, such that the following residues are in close proximity: DR 86-HA 309, DR 71-HA 312, DR 30-HA 314, and 315.

Truncation analysis of several DR binding epitopes.

Peptide regions crucial for binding to four different DR alleles (DR1, DR2, DR5, and DR52a) have been localized in five unrelated DR binding peptides (dynorphin 1-13, sperm whale myoglobin 132-153, influenza hemagglutinin 307-319, pigeon cytochrome c 88-104, and tetanus toxoid 830-843) by testing panels of truncated analogs for DR binding. It was found that in most cases, different DR alleles recognize almost identical, albeit distinct, core regions, suggesting that different DR alleles may recognize similar structures on their peptide ligands. Furthermore, it was found that these core regions, notwithstanding their derivation from unrelated sequences, share a common structural pattern. When the sequences of several other unrelated determinants were scrutinized, the structural motif identified was present in some, but absent in other good DR binders, suggesting that good DR binding capacity of peptide molecules may be compatible with more than one single sequence pattern.

Inhibition of experimental autoimmune encephalomyelitis induction in SJL/J mice by using a peptide with high affinity for IAs molecules.

Blocking of the Ag presenting function of MHC by peptides capable of high affinity binding to this molecule has been proposed as a potential immunotherapeutic intervention in MHC-associated diseases. Recent studies have used this strategy to prevent the induction of experimental allergic encephalomyelitis (EAE) in mice. However, because of the close structural relationship between the inhibitor and encephalitogenic peptides, the results of these previous studies have been difficult to interpret with regard to whether MHC blockade was the mechanism by which the inhibitory peptides functioned. In our study, we have determined the capacity of unrelated peptides capable of binding with high affinity to IAs in inhibiting the induction of EAE in SJL/J mice after immunization with the autoantigenic peptide PLP 139-151. Prevention of the disease was accomplished by two methods: 1) when inhibitor was administered together with the encephalitogenic peptide at the time of immunization, as in previous studies, and 2) when inhibitor was administered at a separate site from the autoantigen 1 day before the immunization with that Ag. Inhibition was due to binding of the inhibitor to IAs, as evidenced by the fact that a control peptide incapable of binding to this MHC had no effect on the course of the disease. The finding that inhibitor could also be efficacious when administered at a separate site has implications for potential use of such a strategy to reverse ongoing autoimmune diseases. The inhibitor had to be present during the time of Ag stimulation, and had no long term inhibitory effects, in that a secondary immune response to the encephalitogenic peptide was not inhibited in animals given the inhibitory peptide before the induction of a primary response. This is compatible with the conclusion that MHC blockade was, in fact, the mechanism of the inhibition, rather than as a result of any long term suppressive effects on immunoreactive T cells. Finally, not only did administration of the inhibitory peptide lead to a prevention of the induction of EAE, but it could also be shown to decrease the T cell proliferative response in vitro to the autoantigen.

A novel approach to the generation of high affinity class II-binding peptides.

We have found that if core regions crucial for class II binding are incorporated in multiple copies in the same peptide molecule ("reiterative motifs"), marked enhancement of the binding capacity occurs. Isotype specificity (IAd vs IEd binding capacities) is retained in all three antigenic determinants so far analyzed (lambda rep 12-26, OVA 323-339, and hen egg lysozyme 105-120). The mechanism involved in such an effect is not clear, but experiments involving introduction of a peptide spacer between two repeated core regions do not support the notion that the effect is mediated by cross-linking of more than one MHC molecule, favoring the possibility that conformational effects or distinct subsites of interaction on the MHC molecule may be involved. Based on reiterative structures, a peptide molecule composed of only two different amino acids (Ala and His) has been produced that still retains a very high binding affinity. An 125I-radiolabeled form of this peptide has been used to demonstrate that the high binding detected is mediated by the same binding site involved in the interaction of IAd and OVA 323-339. Inhibition of Ag presentation studies further supports the immunologic relevance of the phenomena observed. Finally, we observed naturally occurring clustered binding sites in proximity of immunodominant protein regions, raising the possibility that the phenomenon might have a physiologic counterpart.

Characterization of the specificity of peptide binding to four DR haplotypes.

This study describes the establishment of a peptide-binding assay for purified, detergent-solubilized DR molecules. For each of the DR specificities and peptides studied, a unique pattern of interaction was observed. Excellent correlation was detected between the DR1-, 2-, 5-, and 52a-binding capacities and the known DR restrictions of a panel of synthetic peptides. This supports the immunologic relevance of the binding assay, and emphasizes the importance of determinant selection in defining the immune response of individuals. We have also examined the capacity of a panel of DR-restricted peptides to compete with one another for binding to DR1. The results obtained are compatible with a single peptide-binding site on DR molecules. The peptide-binding capacity of the four different DR types (DR1, DR2, DR5, and DR52a) has been further examined by testing a collection of 133 different peptides. This collection is unbiased with respect to previously known DR binding and restrictions, and includes peptides of eukaryotic, bacterial, and viral origins. It was found that: 1) approximately 15 to 35% of the peptides tested bound any given DR type; 2) DR-binding capacities appeared to correlate with each other, suggesting that different alleles of the DR isotype may recognize related structures on an Ag molecule; and 3) despite the statistical correlation between binding capacity of different DR types, approximately 50% of the peptides that were positive binders still were specific in that they could bind only one of the four DR molecules tested. Degenerate binding (i.e., binding to most or all the DR molecules tested) was detected in only a minority of the cases analyzed (approximately 25%).

The use of peptide analogs with improved stability and MHC binding capacity to inhibit antigen presentation in vitro and in vivo.

The identification of a core region for OVA 323-339, which is critical in determining binding to IAd, has enabled us to generate a series of analog peptides in which this core region was extended at both the N and C termini with different amino acid residues. When assessed for binding capacity, several peptides were shown to have increased affinity for IAd compared with the parent sequence, and in addition, some peptides had acquired binding specificities for class II MHC haplotypes not present for OVA 323-339. These peptides were next examined for their ability to inhibit T cell responses in vitro and in vivo. The correlation between binding and the ability to inhibit T cell activation in vitro was good. However, when assessed in vivo, it was clear that high Ia binding was not sufficient in itself to define the inhibitory capacity of a given peptide. That this discrepancy was due to differences in degradation of the core-extended peptides was suggested by 1) results from an inhibition of Ag presentation assay, in which the pulse period with Ag and inhibitor was extended to 20 h; and 2) direct analysis of peptide stability by using reverse phase HPLC. Finally, by protecting the peptide from degradation with N- and C-terminal substitutions of D-amino acids, the inhibitory capacity of an unstable core-extended peptide in vitro could be greatly enhanced. These data indicate that the core extension approach may be one method by which antagonists for MHC class II molecules may be generated.

Structural requirements for the interaction between peptide antigens and I-Ed molecules.

We have analyzed the structural characteristics of the interaction between I-Ed molecules and their peptide ligands. It was found that unrelated good I-Ed binders share structurally similar "core" regions that were experimentally demonstrated to be crucial for binding to I-Ed molecules. Single amino acid substitution analogues of one good I-Ed binder, hen egg lysozyme 107-116, were analyzed for their capacity to bind to I-Ed molecules and to activate two different I-Ed-restricted T cell hybridomas. The results illustrate the great permissiveness of I-Ed-peptide interaction and the great specificity of T cell recognition. It was concluded from these analyses that basic residues on the peptide molecule play a crucial role in binding to I-Ed. This contrasts with the structural requirements for binding to the other Iad isotype, I-Ad, the crucial hydrophobic residues. Thus, different class II molecules of the same MHC haplotype may have rather distinct peptide binding specificities, thereby expanding the repertoire of possible immunogenic peptides presented for T cell recognition.

Capacity of intact proteins to bind to MHC class II molecules.

Here we have demonstrated that denatured, but not native protein antigens can interact with Ia molecules. Thus, the failure of native antigens to be recognized as such by T cells appears to be at least in part due to a deficient antigen/Ia interaction. These results also support previous observations that some T cells can recognize denatured antigens without a further processing requirement. Moreover, a striking correlation was observed between the in vitro binding pattern of denatured proteins and the pattern of restriction of T cell responses elicited by immunization with the native antigen, raising the possibility that an unfolding step may actually occur early during in vivo processing and influence the final outcome of Ia-restricted T cell responses.

Effect of conformational propensity of peptide antigens in their interaction with MHC class II molecules. Failure to document the importance of regular secondary structures.

In an attempt to define some of the conformational requirements for binding of the antigenic peptide OVA 323-336 to purified IAd molecules, three distinct experimental approaches were applied. First, the effect of introducing proline or glycine residues within the region of OVA 323-336 crucial for its IAd binding capacity was analyzed. In most instances these substitutions had little or no effect, suggesting that neither alpha-helical nor beta-sheet regular structures may be strictly required for productive interaction with MHC molecules. Some of the same substitutions were also found to have no effect on the capacity of the peptide to stimulate OVA 323-336 specific T cell hybridomas, suggesting that regular structures such as alpha-helices or beta-sheets may not be strictly required for T cell stimulation, either. Second, we introduced, within the OVA 323-336 molecule, structural modifications predicted to alter its dipole characteristics and stabilize helical structures. No improvement of the IAd binding capacity was detected following these structural alterations. Surprisingly, some but not others of these analogs displayed increased antigenicity for OVA 323-336 specific T cell hybridomas. Third, a panel of analogs of OVA 323-336 were synthesized in which the crucial IAd binding core region was linked to non-native sequences of differing conformational propensities. When 22 such analogs were tested for IAd binding, it was found that these non-native sequences could drastically influence the binding capacity, but no correlation was found between their effect and their alpha-helical, beta-sheet, or beta-turn conformational propensity as calculated by the Chou and Fasman algorithm. In summary, all the data presented herein suggest that, at least in the case of OVA 323-336 and IAd, the propensity of the antigen molecule to form secondary structures such as alpha-helices, beta-sheets, or beta-turns does not correlate with its capacity to bind MHC molecules.

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

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 IEd binding was successfully predicted in approximately 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.

Autologous peptides constitutively occupy the antigen binding site on Ia.

Low molecular weight material associated with affinity-purified class II major histocompatibility complex (MHC) molecules of mouse (Ia) had the expected properties of peptides bound to the antigen binding site of Ia. Thus, the low molecular weight material derived from the I-Ad isotype was efficient in inhibiting the binding of 125I-labeled I-Ad-specific peptide to I-Ad, but did not significantly inhibit the binding of an I-Ed-specific peptide to I-Ed; the reciprocal isotype-specific inhibition was demonstrated with low molecular weight material derived from I-Ed. The inhibitory material was predominantly peptide in nature, as shown by its susceptibility to protease digestion. It was heterogeneous as measured by gel filtration (mean molecular weight approximately 3000), and when characterized by high-performance liquid chromatography, it eluted over a wide concentration of solvent. Such self peptide-MHC complexes may have broad significance in the biology of T cell responses, including generation of the T cell repertoire, the specificity of mixed lymphocyte responses, and the immune surveillance of self and nonself antigens in peripheral lymphoid tissues.

The relation between major histocompatibility complex (MHC) restriction and the capacity of Ia to bind immunogenic peptides.

The capacity of purified I-Ad, I-Ed, I-Ak, and I-Ek to bind to protein derived peptides that have been previously reported to be T cell immunogens has been examined. For each of the 12 peptides studied strong binding to the relevant Ia restriction element was observed. All the peptides bound more than one Ia molecule; however, for 11 of 12 peptides, the dominant binding was to the restriction element, whereas in one instance the dominant binding was to a nonrestriction element. When the peptides were used to inhibit the presentation of antigen by prefixed accessory cells to T cells, an excellent correlation was found between the capacity of a peptide to inhibit the binding of an antigen to purified Ia and the capacity of the peptide to inhibit accessory cell presentation of the antigen. Thus, the binding of peptide to purified Ia is immunologically relevant, and Ia seems to be the only saturable molecule on the surface of the accessory cell involved in antigen presentation. Inhibition analysis also indicated that all peptides restricted to a particular Ia molecule competitively inhibited one another, suggesting that each Ia restriction element has a single binding site for antigen. Cross-linking of labeled peptides to Ia followed by electrophoretic analysis and autoradiography suggested that this single binding site is made up of portions of both alpha and beta chains of Ia.

Isolation and characterization of antigen-Ia complexes involved in T cell recognition.

Using equilibrium dialysis, it has been previously demonstrated that immunogenic peptides bind specifically to the Ia molecules serving as restriction elements in the immune response to these antigens. Using gel filtration to study the formation of ovalbumin (OVA) peptide-I-Ad complexes, it is herein demonstrated that the complexes, once formed, are very stable (kd approximately equal to 3 X 10(-6) s-1), but the rate of complex formation is very slow (ka approximately 1 M-1 s-1 explaining the overall low equilibrium constant of approximately 2 X 10(-6) M. Treating the complexes with glutaraldehyde revealed that the ovalbumin peptide was cross-linked solely to the alpha chain of I-Ad. Planar membranes containing I-Ad-OVA complexes stimulated a T cell response with 2 X 10(4) less antigen than required when uncomplexed antigen was used, thus demonstrating the biologic importance of these complexes in antigen recognition.