The Stimulation of Low-Affinity, Nontolerized Clones by Heteroclitic Antigen Analogues Causes the Breaking of Tolerance Established to an Immunodominant T Cell Epitope

Female B10.A mice (The Jackson Laboratory) were used at 8–10 wk of age. Invariant chain–deficient B10.BR mice were provided by Dr. D. Mathis (IGBMC, Strasbourg, France) and used at 6–12 wk of age. AND and AD10 transgenic mice containing pigeon cytochrome c (PCC)–specific TCRs were provided by Dr. S. Hedrick (University of California, San Diego, CA), as was the cytochrome c–specific T cell clone, AD10. The B cell lymphoma CH27, provided by Dr. G. Houghton (University of North Carolina, Chapel Hill, NC), was used as APC in some experiments. For IL-2 determinations, the IL-2–dependent T cell line CTLL-2 was used. BW5147α⁻β⁻ cell line was used as fusion partner in the production of T cell hybridomas.

pMCC (ANERADLIAYLKQATK) and the heteroclitic analogue (pMCC-A; AAAAAAAIAYAKQATK) were synthesized on a Rainin Symphony synthesizer (Peptide Technologies) as described previously 27. Peptides were purified by reverse-phase HPLC to >95% purity. Identity of the peptides was substantiated by mass spectrometry.

Tolerance induction and immunization were performed as described previously 26. In brief, to induce tolerance, B10.A mice were injected intraperitoneally with 300 μg pMCC in IFA (Pierce Chemical Co.). 10 d later, tolerized mice were immunized with 20 μg pMCC or the heteroclitic antigen analogue, pMCC-A, in CFA (Difco) subcutaneously in the base of the tail. As controls, normal mice were immunized with 20 μg pMCC or pMCC-A in CFA. 10 d after immunization, draining lymph nodes were removed and single cell suspensions were made and used for proliferation and cytokine assays or to establish short-term T cell lines and hybridomas.

Cells from draining lymph nodes (paraaortic and inguinal) were plated at 2.5 × 10⁵ cells/well along with 2.5 × 10⁵ irradiated (3,000 rad) syngeneic splenocytes and stimulated with the pMCC or pMCC-A peptide as indicated. After 72 h, the cultures were pulsed for an additional 18 h with 1 μCi of [³H]thymidine and analyzed by beta-plate scintigraphy. In experiments to analyze relative immunogenicity of MCC analogues, proliferation assays were performed using the AD10 clone, as described previously 27.

For cytokine analysis, supernatants from the cultures that had been established to measure the proliferative response were removed after 24 h for IL-2 measurement, after 48 h for IL-4 and IL-10, or after 72 h for IFN-γ. IL-2 was measured by bioassay using the CTLL-2 cell line. IL-4, IL-10, and IFN-γ were assayed by ELISA according to the instructions provided by the manufacturer of the reagents (PharMingen). The sensitivities of the ELISA assays were as follows: IFN-γ, 100 pg/ml; IL-4, 50 pg/ml; and IL-10, 50 pg/ml.

Peptides were analyzed for their ability to bind to purified I-E^k molecules as described previously 28. The binding capacity is reported as the concentration of peptide required to obtain 50% inhibition (IC₅₀) of binding of the radiolabeled ligand.

CD4⁺ T cells were purified from the lymph node and spleen of pMCC-specific TCR transgenic mice (AND) by collecting nonadherent cells from a nylon wool column, followed by complement treatment of the cells that had previously been incubated with antibodies to CD8 (3.155), heat-stable antigen (JIID), class II MHC (M5/114 and CA-4.A12), macrophages (M1/70), and dendritic cells (33D1), and subsequently cross-linked with mouse anti–rat kappa antibody (MAR18.5). The percentage of CD4⁺ cells that were Vα11⁺Vβ3⁺ T cells was determined by flow cytometry. This information was then used to calculate the total number of cells that had to be injected to achieve 2.5 × 10⁶ Vα11⁺Vβ3⁺ T cells/injection. Mice were injected intravenously with this number of cells in a vol of 0.4 ml of HBSS (GIBCO BRL) 21. Recipient mice were unirradiated, invariant chain–deficient B10.Br mice. 3 d after cell transfer, the animals were injected intravenously with 300 μg of pMCC in a vol of 250 μl to induce tolerance. 10 d later, tolerized and nontolerized mice were immunized with 20 μg of pMCC or pMCC-A in CFA subcutaneously. After an additional 10 d, draining lymph nodes were removed and single cell suspensions were used for in vitro stimulation and cytokine analysis.

Draining lymph node cells from immunized or tolerance-broken mice were restimulated in vitro with 0.5 μM pMCC or pMCC-A for 2 d and expanded with IL-2 (20 U/ml) for an additional 2 d. The lymphoblasts thus generated were fused with the BW5147α⁻β⁻ cell line at ∼1:3 ratio, and hybridomas were selected in HAT medium and distributed in 96-well plates at ∼10⁴ cells/well 29. Wells were screened for reactivity to pMCC or pMCC-A by measuring IL-2 production. The positive wells were cloned at a concentration of 0.3 cells/well, and clones that produced IL-2 in response to antigen were expanded for further analysis.

Cells were incubated on ice with anti-CD3–FITC, anti-Vα11–FITC, anti-Vβ3–PE (PharMingen), or anti-CD4–PE (Becton Dickinson) for 1 h and washed several times. Live cells were gated by forward and side scatter and analyzed by flow cytometry on a FACScan™ equipped with CELLQuest™ software (Becton Dickinson).

Total RNA was isolated from T cell hybridomas using an RNeasy Mini Kit (QIAGEN) according to the manufacturer's instructions. Oligo dT–primed cDNA was synthesized using a cDNA Cycle kit (Invitrogen Corp.) following the manufacturer's protocol and then amplified by PCR for sequence analysis. The predicted size fragments from PCR amplification were further purified with a QIAquick Gel Extraction kit (QIAGEN) and sequenced using the Taq Dye Deoxy Terminator Cycle Sequencing kit (Applied Biosystems) on an Applied Biosystems 310 DNA Sequencer. The PCR and sequencing primers used were as follows: for α chain PCR: Cα, AAG TCG GTG AAC AGG CAG AG; for Vα11, TCA GGA ACA AAG GAG AAT GG; for α chain sequencing: Cα, AAC TGG TAC ACA GCA GGT TC; for β chain PCR primers: Cβ, AGC ACA CGA GGG TAG CCT T 30; for β-chain sequencing: GGA GTC ACA TTT CTC AGA TCC A. A set of Vβ primers was used for PCR as reported previously 31.

In a previous study to determine whether cross-reactive antigen analogues could break tolerance induced to the immunodominant T cell epitope from MCC, pMCC, it was found that certain heteroclitic antigens (as defined by their heightened antigenicity for a representative T cell clone) were capable of breaking tolerance 26. Specifically, substitutions at the minor MHC contact residue L98 were effective in this regard. In this study, we have used a pMCC analogue with an L98A substitution that also had residues 88–94, which are outside the MHC peptide binding cleft, substituted with alanines. This poly-A MCC analogue (pMCC-A) had a similar binding capacity to I-E^k as the native peptide, but was able to stimulate MCC-specific clones ∼10-fold more efficiently than the native pMCC peptide (Table). The ability of this analogue to break tolerance to pMCC is shown in Fig. 1. Tolerized animals had a markedly reduced capacity to proliferate in response to stimulation with pMCC compared with the response of immunized controls (Fig. 1 A) and did not synthesize detectable amounts of IFN-γ (Fig. 1 B) or other cytokines such as IL-2, IL-4, or IL-10 (data not shown). In contrast, mice that were tolerized to pMCC and then immunized with the heteroclitic analogue, pMCC-A, mounted vigorous proliferative and cytokine responses, albeit requiring 10–100-fold more antigen than the normal immunized controls to achieve comparable responses.

Because of reports that claim administration of antigen intraperitoneally in IFA leads to sequestration of the immune response in the spleen rather than tolerance induction, studies were carried out to evaluate this possibility. Cytokine analysis of spleen and draining lymph nodes after pMCC administration intraperitoneally in IFA followed by pMCC immunization subcutaneously in CFA indicated that tolerance was evident in both lymph node and spleen, with undetectable quantities of IFN-γ and IL-4 found in culture supernatants of both organs. Enzyme-linked immunospot (ELISPOT) analysis also indicated tolerance induction, although less dramatic than the supernatant analysis. IFN-γ–producing cells predominated in normal immune lymph node and spleen (600 and 500 cells/10⁶), respectively, with a very small IL-4–producing population (<20 cells/10⁶). In pMCC-tolerized and immunized animals, there was a 60 and 75% reduction in IFN-γ–producing cells in both spleen and lymph node, with no change or a slight increase in IL-4–producing cells (24 cells/10⁶ in spleen). Thus, with respect to proliferation and the major cytokine, IFN-γ, tolerance was clearly established by the protocol used, although the small number of IL-4–producing cells generated was apparently not affected by the tolerance-inducing protocol.

The results shown in Fig. 1 and those reported previously 26 suggest that the T cells involved in the breaking of tolerance have a lower avidity for antigen compared with normal immune T cells, as evidenced by the shift in the antigen dose–response profile. To study in greater detail the functional activity of the T cells involved in the breaking of tolerance to pMCC, and in order to analyze the structure of their TCRs, a series of T cell hybridomas was made from animals whose tolerance to pMCC was broken by immunization with pMCC-A. As controls, hybridomas were also derived from normal animals immunized with either pMCC or pMCC-A. The response of representative hybridomas derived from normal or tolerance-broken animals to stimulation with pMCC is shown in Fig. 2. The two hybridomas derived from pMCC or pMCC-A normal immune animals behaved similarly to stimulation with pMCC. In contrast, and similar to the data obtained with the bulk T cell response, the hybridoma derived from the tolerance-broken group required ∼10-fold more pMCC to give an IL-2 response comparable to that of the hybridomas derived from the normal immunized animals. The decreased antigen responsiveness of the hybridomas derived from tolerance-broken animals was not due to a decrease in surface expression of either the TCR or the CD4 coreceptor (Fig. 2B and Fig. C; and data not shown).

Fig. 3 summarizes the antigen response data from all hybridomas that reacted to both pMCC and pMCC-A, as a plot of the quantity of pMCC versus pMCC-A required to generate an IL-2 response in each hybridoma analyzed. Several conclusions can be derived from the data shown in Fig. 3. (a) All but one of the hybridomas derived from the pMCC-immunized group responded to lower concentrations of pMCC-A than pMCC, as evidenced by a location on the graph below the diagonal. This indicates that the heteroclitic nature of pMCC-A, originally described with a single T cell clone, is generalizable to the majority of the pMCC-specific T cell repertoire. (b) The hybridomas in the pMCC tolerance-broken group, in general, required higher concentrations of pMCC to stimulate an IL-2 response than the hybridomas from either the pMCC or pMCC-A normal immune groups. Thus, the hybridomas from the tolerance-broken group required an average of 0.88 μM of pMCC to stimulate an IL-2 response, whereas the pMCC immune group required 0.01 μM, and the pMCC-A immune group required 0.07 μM pMCC to stimulate a response. (c) In contrast to the greater antigen requirement of the tolerance-broken group for pMCC, the response to pMCC-A was similar for all three groups. In addition to the cross-reactive hybridomas shown in Fig. 3, a small number of hybridomas from pMCC or pMCC-A immune animals were found to be non–cross-reactive (1 of 10 in the pMCC immune group failed to respond to pMCC-A, and 3 of 14 of the pMCC-A group failed to respond to pMCC; data not shown).

The antigen dose–response data for pMCC indicated that the hybridomas from the tolerance-broken group had a lower avidity for pMCC than hybridomas from normal immune animals. To determine if this functional difference between T cells derived from normal and tolerance-broken animals was correlated with structural features of the TCR, we undertook an analysis of their Vα and Vβ chains. Since previous studies of the Vα and Vβ utilization of MCC-specific T cells indicated that the response to this antigen was dominated by T cells that express Vα11 and Vβ3 32,33,34,35,36,37,38, initially flow cytometric analysis of the T cell hybridomas for expression of Vα11 and Vβ3 was performed. Almost one half (5 of 12) of the T cell clones derived from pMCC-immunized animals were Vα11⁺/Vβ3⁺, with the rest being either Vα11⁺/Vβ3⁻ (4 of 12) or negative for both Vα11 and Vβ3 (3 of 12). The repertoire of clones derived from pMCC-A immune animals differed somewhat from this distribution, in that only about one fourth of the clones (3 of 13) possessed the canonical Vα11⁺/Vβ3⁺ phenotype, and about one half (7 of 13) were Vα11⁺/Vβ3⁻. The data obtained with the T cell clones derived from the tolerance-broken group were strikingly different, in that all eight clones analyzed were Vα11⁺/Vβ3⁻. These data suggest that the typical Vα11⁺/Vβ3⁺ MCC-reactive clones were irreversibly inactivated by the tolerizing dose of pMCC and did not participate in the termination of tolerance induced by pMCC-A. Rather, a subset of the Vα11⁺/Vβ3⁻ clones responsive to pMCC-A was involved in the breaking of tolerance.

The difference in the repertoire elicited by pMCC and pMCC-A immunization with respect to the relative numbers of Vβ3⁺/Vα11⁺ expressing T cells was substantiated by the analysis of short-term T cell lines (Fig. 4). Whereas the T cell line generated from pMCC-immunized animals contained 33% Vα11⁺/Vβ3⁺ cells and 25% Vα11⁺/Vβ3⁻ cells, the pMCC-A immune line contained less than half the percentage of Vα11⁺/Vβ3⁺ cells (14%) and more than twice the percentage of Vα11⁺/Vβ3⁻ cells (59%). These data are consistent with the data obtained with the T cell hybridomas derived from pMCC-A–immunized animals that also demonstrated a skewing toward a Vα11⁺/Vβ3⁻ phenotype. A second independently derived set of T cell lines gave similar results.

To further analyze the TCR repertoire involved in breaking tolerance and to compare it with the repertoire normally involved in the response to MCC, cDNA sequence analysis of the junctional regions encompassing the CDR3 of the α and β chains was performed, together with the identification of the Vβ families other than Vβ3 that were used (Fig. 5). Previous sequence analyses of the Vα11 genes associated with a cytochrome c–specific response indicated that the CDR3 regions were usually 8 residues long, with a conserved glutamic acid (E) at position 93 and S, A, or G at position 95 34,35,36,37,38. Of the 8 Vα11 CDR3 regions analyzed from T cell hybridomas derived from pMCC-immunized animals, all had E at position 93, 5 had S at position 95, and 6 were 8 residues in length (Fig. 5 A). Thus, this set of pMCC-specific clones was similar in the structure of their Vα CDR3 regions to previously studied TCRs. The Vα11 CDR3 region from the tolerance-broken group was, in general, very similar to that of the pMCC immune group, but with a slightly greater deviation from the canonical structure (Fig. 5 B). Thus, all 8 clones had E at position 93, but only 4 had G or A at position 95, and only 4 were 8 residues in length. Of particular interest was the finding of a 4-asparagine (N) repeat from positions 95–98 that was present in 4 of the 8 sequences analyzed. This feature was observed in only one of the sequences from the pMCC immune group. The Vα11 CDR3 region from the pMCC-A immune group had characteristics intermediate between the pMCC immune group and the tolerance-broken group (Fig. 5 C): all 10 sequences contained E at position 93, 6 of 10 contained G, A, or S at 95, 6 of 10 had CDR3 regions 8 residues in length, and 3 sequences contained the asparagine repeat from residues 95–98.

With respect to the Vβ gene analysis, as described above, the most striking difference between the pMCC immune group and the tolerance-broken group was the complete absence of Vβ3 in the latter group of hybridomas. 4 of 8 Vα11⁺ TCR genes sequenced in the pMCC immune group were Vβ3, with the other 4 β chains being derived from Vβ8, 14, 15, and 16 gene families (Fig. 5 A). In contrast, the tolerance-broken group (Fig. 5 B) expressed chains from only 2 gene families: Vβ16 (3 of 8) and Vβ8 (5 of 8). The motif and length of the Vβ CDR3 that has been previously described to be associated with anti-cytochrome c activity is a length of 9 residues, with N at position 100 and a G or A at position 102 34,35,36,37,38. In the pMCC immune group, 6 of 8 β chains analyzed had CDR3 regions 9 residues in length, 6 had N at position 100, and 7 had G or A at 102. In striking contrast to this, none of the 8 Vβ CDR3 regions analyzed from the tolerance-broken group had a length of 9 residues, and none had N at position 100, although 4 of 8 had a G or A at 102. Again, the pMCC-A immune group had characteristics intermediate between those of the pMCC immune and tolerance-broken groups. 3 of the 10 β chains analyzed were Vβ3, with the remainder being either Vβ8 2 or Vβ16 5. 4 of the 10 CDR3 regions were 9 residues long, 2 had N at position 100, and 6 contained G or A at 102.

These data suggest the following conclusions: (a) MCC recognition, even the low-avidity response associated with the tolerance-broken group, is very strongly associated with the presence of an α chain from the Vα11 family, with a CDR3 region that has an E at position 93, a residue previously implicated as being involved in the interaction with one of the immunodominant TCR contact residues in pMCC, K99 39; (b) although the expression of Vβ3 is not necessary for MCC reactivity, it appears to be associated with the higher avidity interactions of the pMCC-immunized group; and (c) the lower avidity TCRs found in the tolerance-broken group expressed Vβ8 or Vβ16 and did not contain the canonical N at position 100. Vβ16 expression appeared to be particularly associated with decreased avidity for MCC. Two of the three clones from the pMCC-A immune group that failed to cross-react at all with pMCC were Vβ16, as was the clone with the least cross-reactivity with pMCC. Also, in the tolerance-broken group, the two lowest avidity clones for pMCC (requiring >1 μM MCC for stimulation) expressed Vβ16.

The data shown in Fig. 3 and Fig. 5 were obtained with hybridomas derived, in the case of the tolerance-broken group, from a pool of lymphocytes from two animals, and in the case of the normal immune group, from two separate fusions of cells from individual mice. A second fusion of cells from another two tolerance-broken animals was consistent with the data shown in Fig. 3 and Fig. 5, in that they had lower avidity for pMCC than normal immune animals, and none of them expressed Vβ3 (data not shown).

To evaluate the possibility that breaking tolerance by pMCC-A might involve reversal of anergy, an adoptive transfer system was established in which the effect of pMCC-A on the tolerance induced in a single clone of cells could be evaluated 21. For this purpose, T cells from cytochrome c–specific TCR transgenic mice (AND) were used. The AND TCR is Vβ3⁺/Vα11⁺ and recognizes pMCC-A as a heteroclitic antigen. As recipient mice, H-2K⁺ invariant chain–deficient animals were used. These animals had previously been shown to lack a T cell repertoire capable of responding to pMCC 40,41. Thus, if tolerance could be broken in adoptively transferred invariant chain–deficient animals, the T cells involved would have to have been derived from the adoptively transferred TCR transgenic cells. In preliminary experiments, the failure of invariant chain–deficient mice to respond to either pMCC or pMCC-A was confirmed (data not shown). These mice were injected with 2.5 × 10⁶ CD4⁺ Vβ3⁺/Vα11⁺ T cells from AND transgenic mice. 3 d after transfer, one group of mice was tolerized by the intravenous injection of 300 μg of pMCC. 10 d after the induction of tolerance, these animals and groups of nontolerized, adoptively transferred animals were immunized with pMCC or pMCC-A and their immune responses were evaluated by analyzing the in vitro response to pMCC of regional lymph nodes harvested 10 d after immunization. Adoptively transferred and tolerized mice contained 1.61 ± 0.4% Vβ3⁺/Vα11⁺ T cells, compared with 0.72 ± 0.23% Vβ3⁺/Vα11⁺ cells in mice that did not receive an adoptive transfer. This difference was significant (P < 0.01), and therefore it could be concluded that after adoptive transfer and tolerance induction, there was a residual transgenic T cell population derived from the adoptively transferred transgenic T cells (1.61 – 0.72 = 0.89%) that could be studied for reversal of anergy. The immune response of T cells derived from adoptively transferred, tolerized, and subsequently immunized animals is shown in Fig. 6. The data, expressed as the amount of IFN-γ per Vβ3⁺/Vα11⁺ cell put into culture, indicated: (a) tolerance could be established in this adoptive transfer system, since animals that received an intravenous injection of pMCC before immunization with pMCC in CFA were severely impaired in their capacity to produce IFN-γ compared with the nontolerized controls (0.15 vs. 0.95 pg/cell); and (b) immunization with the heteroclitic analogue pMCC-A failed to break this tolerant state.

If the model suggested by our data is correct, namely that breaking tolerance involves the activation of T cells with low affinity for the tolerogen but high affinity for the heteroclitic antigen, then it must be further postulated that these T cells, once stimulated by the heteroclitic antigen, have a lowered threshold for stimulation and become responsive to stimulation by the lower affinity tolerogen. To evaluate this postulate, we studied the response of naive and recently primed TCR transgenic T cells to stimulation with either the cognate antigen used to prime the T cells or a nonantigenic TCR antagonist analogue of the antigen with presumably lower affinity for the TCR. T cells from the cytochrome c–specific TCR transgenic line AD10 were used for this purpose, since this TCR had been extensively analyzed previously for reactivity to a large panel of pMCC analogue peptides. Naive T cells or T cells that had been stimulated with pMCC 8–12 d previously were stimulated with varying doses of pMCC or the TCR antagonist peptide T102G, which has a glycine (G) substituted for one of the major TCR contact residues, T102. The data shown in Fig. 7 demonstrate that the T102G analogue, which was nonantigenic for naive T cells, was an agonist when assayed on recently primed T cells, albeit requiring a high concentration (1–100 μg/ml) to achieve this effect.

This study was undertaken to evaluate two potential mechanisms by which antigen analogues might operate in the termination of T cell tolerance to an immunodominant epitope: stimulation of low-affinity nontolerized clones and reversal of anergy. Three sets of data presented in this study favor the conclusion that the first mechanism is operative.

First, T cell hybridomas derived from tolerance-broken animals required, on average, 88-fold higher concentration of antigen than hybridomas derived from normal immune animals in order to be stimulated to produce IL-2. This difference in antigen dose requirement is most likely due to differences in affinity of the TCRs from these two groups of hybridomas, but other possibilities need to be considered, such as differences in the levels of expression of molecules that contribute to the stimulation of the T cells; these include the TCR itself, the CD4 coreceptor, and adhesion molecules such as LFA-1. Two lines of evidence suggest that differential expression of these molecules does not contribute to the antigen dose requirements observed. First, no correlation between the levels of expression of TCR and CD4 and the antigen dose requirement for stimulation was observed (Fig. 2; and data not shown). Second, when I-E^k–transfected fibroblasts that lacked intercellular adhesion molecule (ICAM), B-7, or other known ligands for T cell adhesion/costimulator receptors were used as APC, the large difference in antigen dose requirements for the tolerance-broken and normal immune animals was still observed (576 vs. 22 nM).

The second set of data that supports the hypothesis of stimulation of low-affinity nontolerized clones as a mechanism for breaking tolerance is the finding of a different repertoire of TCRs on T cells from tolerance-broken animals and normal immune animals. Although there was no absolutely unique feature to the TCRs from the tolerance-broken group of hybridomas, there was pronounced skewing of the TCRs compared with those found in the normal immune group. Vβ3, the major Vβ family found in the normal immune group 33,34,35,36,37,38, was not represented at all in the tolerance-broken group. Instead, Vβ8 and Vβ16 were expressed in the tolerance-broken group—Vβ families that were only occasionally expressed in the MCC-specific T cells from normal immune animals. Similarly, the canonical Vβ CDR3 motif and length associated with the response to pMCC were not observed in any of the Vβ CDR3 regions from the tolerance-broken group of hybridomas. If reversal of anergy were the mechanism for breaking tolerance, there is no apparent reason for it to be restricted to a minor subset of clones with the exclusion of the major Vβ3⁺/Vα11⁺ subset. On the other hand, a distinctly different repertoire of TCRs with different Vβ/Vα composition would be consistent with a mechanism of breaking tolerance that involved the stimulation of low-affinity clones that were not capable of being stimulated after MCC immunization nor were tolerizable when MCC was administered in a tolerogenic form.

The third observation in support of the concept that tolerance was broken by low-affinity nontolerized clones was our failure to terminate the tolerant state of adoptively transferred and tolerized transgenic T cells. When transgenic T cells bearing a typical Vβ3⁺/Vα11⁺ TCR specific for MCC were transferred and subsequently tolerized to MCC, the same heteroclitic peptide that was capable of breaking tolerance in conventional animals failed to break tolerance in these animals with a clonal population of tolerized cells.

Taken together, these three sets of data strongly support the hypothesis that breaking tolerance involves the stimulation of nontolerized, low-affinity clones by the MCC heteroclitic analogue. For reversal of anergy to be operative, it would be necessary to postulate that certain (Vβ3⁺/Vα11⁺) cells are selectively incapable of undergoing reversal of anergy, whereas others (Vβ8⁺/Vα11⁺ and Vβ16⁺/Vα11⁺) are able to undergo anergy reversal. Although this is theoretically possible, we consider it unlikely.

Our data are consistent with previous reports that found that tolerance to certain self-epitopes was incomplete and that immunization with the cognate antigen could generate an immune response 42,43,44. However, the responding T cells that were elicited were of relatively low affinity. For instance, immunization of beef insulin (BI) transgenic mice with BI generated a BI-specific T cell response that required higher concentrations of antigen to be elicited than the response elicited in nontransgenic mice 42. Similarly, low-avidity CD8 T cell responses were obtained after immunization with a p53 self-epitope when p53⁺ mice were used. In contrast, p53-deficient mice generated a higher avidity response 43. These reports suggest that when self-tolerance is incomplete, active immunization with the unaltered self-protein or an epitope derived from it can elicit a response of low-avidity T cells that had escaped tolerance. Our data extend these observations to situations in which tolerance to the epitope in question is complete and cannot be reversed by immunization with the native epitope but can be overcome by immunization with a heteroclitic analogue.

The hypothesis that termination of tolerance to pMCC involves the stimulation of low-affinity clones that are neither tolerized nor stimulated by pMCC but can be stimulated by the heteroclitic analogue, pMCC-A, requires the further postulate that after stimulation by the heteroclitic antigen, the affinity threshold for stimulation is lowered such that restimulation with the tolerogen results in a recall response. To test this postulate, we studied the response of naive and previously primed MCC-specific transgenic T cells to a ligand of presumably lower affinity, a TCR antagonist peptide. It was found that, whereas naive and primed T cells responded with similar vigor to the wild-type agonist peptide, only primed cells proliferated in response to stimulation with the antagonist peptide. This finding further supports the low-affinity T cell model of breaking tolerance. The mechanism by which priming results in a lowering of the affinity threshold for TCR-mediated stimulation is unknown. One possible factor that could influence the sensitivity of the T cell response is the level of expression of the CD4 coreceptor. Several studies have documented that T cell clones with the same TCR have heightened responses to suboptimal concentrations of agonist peptide when the T cell is CD4⁺ compared with CD4⁻ variants or when CD4⁺ cells are pretreated with anti-CD4 antibodies 45,46,47,48. Although we have not been able to document any differences in the level of expression of CD4 on the naive and recently primed T cells that we analyzed (data not shown), it is possible that CD4 function differs in the two cell types; e.g., the activity of the CD4-associated Lck may be greater in primed compared with naive T cells. This and other possible differences between naive and recently primed T cells that may be important in the conversion of TCR antagonists into agonists are being investigated.

We thank Joyce Joseph for help in preparation of the manuscript.

This work was supported in part by National Institutes of Health grants CA71541 and AI18634 to H.M. Grey. This is publication no. 299 of the La Jolla Institute for Allergy and Immunology.