Peptide-MHC Class II Complex Stability Governs CD4 T Cell Clonal Selection

The random rearrangement of TCR genes in the thymus enables the adaptive immune system to recognize a variety of pathogen-derived molecules (1). Ag-specific T cells are selected from this vast pool of diverse, naive cells based on the ability of their surface TCR to recognize peptide-MHC complexes. Several studies of infection in mice (2) and in monkeys (3, 4) have suggested that the nature of the responding T cell repertoire can be important for effective pathogen control. There is also evidence that the progression of autoimmune disease in the murine experimental autoimmune encephalomyelitis is linked to the presence of high-avidity public clonotypes (5, 6). Although these studies suggest that the nature of the TCR repertoire selected may be crucial for pathogen control and for the development of autoimmunity, the mechanisms governing the clonal composition of the T cell response remain poorly understood.

The recognition of peptide-MHC class II (pMHCII) complexes by the TCR lies at the center of the CD4 T cell response. The pMHCII complexes are stabilized by the interaction of peptide anchor residues with polymorphic pockets and depressions in the MHC class II (MHCII) binding groove (7). Previous studies have shown that the stability of pMHCII complexes impacts CD4 T cell response, and more particularly peptide immunogenicity (8), immunodominance (9), and CD4 T cell differentiation (10, 11). However, the importance of pMHCII stability for CD4 T cell clonal selection remains unclear.

The I-E^k-restricted murine response to pigeon cytochrome c (PCC) (12, 13) provides an ideal experimental model to determine the mechanisms of clonal selection in vivo. Immunization of B10.BR mice with PCC protein induces Vα11Vβ3-expressing CD4 T cells with restricted CDR3 regions that confer specificity to one dominant epitope (PCC_88–104) (13). In this model, clonal dominance is established during the first week of the primary T cell response (13) and is based on threshold levels of TCR-pMHCII affinity (14). Adjuvants have been shown to alter this TCR-based selection (15), but the factors setting the affinity threshold during an immune response are poorly understood.

In the current studies, immunization with cytochrome c peptides mutated at different critical MHCII anchor residues revealed the importance of pMHCII stability for CD4 T cell clonal selection. Although both low- and high-stability peptides induced extensive clonal expansion of Ag-specific CD4 T cells, the diversity of the Ag-specific TCR repertoire correlated directly with pMHCII stability. Peptides lacking key anchor residues for I-E^k skewed the CD4 response toward high-affinity clonotypes with limited TCR repertoire diversity. Increasing peptide stability broadened TCR repertoire diversity by recruiting lower affinity clonotypes without eliminating high-affinity clonotypes. However, this increased TCR repertoire diversity was regulated by Ag dose, with low dose of high-stability peptides favoring higher affinity clonotypes. Thus, pMHCII stability controls the clonal composition of the Ag-specific CD4 T cell response in a dose-dependent manner.

B10.BR, B10.BR-Thy1.1 congenic, and ANDαβ transgenic mice were maintained under pathogen-free conditions at The Medical College of Wisconsin. The Medical College of Wisconsin and the Institutional Animal Care and Use Committee reviewed and approved all experiments.

Peptides were synthesized by standard solid-phase methods, purified by HPLC, and confirmed by mass spectrometry. Prior to labeling with NHS-LC-Biotin (Pierce, Rockford, IL) an aminohexanoic acid spacer (EMD Biosciences, Gibbstown, NJ) was added to the amino terminus of the peptide resin. The labeled peptides were then deprotected and purified by HPLC.

I-E^k molecules were expressed as described previously (14) and loaded with CLIP peptide (Anaspec, San Jose, CA) to stabilize the molecule. For dissociation assays, soluble I-E^k (8 μM final concentration) was loaded with biotinylated peptide (160 μM final concentration) in 50 mM NaH₂PO₄, 50 mM sodium citrate (pH 5.3), and protease inhibitor for 72 h at 37°C. I-E^k/peptide complexes were purified by buffer exchange to PBS using a Centricon-30 spin column (Millipore, Bedford, MA). For dissociation, reaction complexes (400 nM final concentration) were incubated with a 100-fold molar excess of unlabeled competitor peptide, moth cytochrome c (MCC_88–103), to prevent rebinding of dissociated peptide. At indicated time points, aliquots were removed and immediately put on ice. Remaining complexes were quantified with an europium-labeled streptavidin-based solid-phase immunoassay. Maxisorb microtiter plates (Nunc, Rochester, NY) were coated with 14.4.4 (anti-I-E^k) Ab. Complexes were captured for 4 h at 4°C and detected with streptavidin-europium. Plates were measured in a Wallac VICTOR counter (PerkinElmer Wallac, Waltham, MA). Data were normalized and expressed as the percentage of biotinylated peptide/I-E^k complex remaining relative to the complex at t = 0. Graphpad Prism 5.0 (GraphPad Software, San Diego, CA) was used to fit the data to an exponential decay model for the determination of complex half-life.

Soluble I-E^k (40 nM final concentration) was incubated with biotinylated MCC_88–103 peptide (1000 nM final concentration) and various concentrations of unlabeled competitor peptide (ranging from 0–320 μM) in 50 mM NaH₂PO₄ and 50 mM sodium citrate (pH 5.3), and protease inhibitor for 72 h at 37°C. Complexes were quantified as described previously but were captured for 1 h at 37°C. Data were normalized and fit to a three-parameter sigmoid function. IC₅₀ values were calculated using SigmaPlot 2000 software (Systat Software, San Jose, CA).

Mice were immunized s.c. at the base of the tail with 60 μg or the indicated dose of peptide in combination with monophosphoryl lipid A-based adjuvant [laboratory formulation based on procedures in ref (16)]. For adoptive transfer, 10⁵ total splenocytes from ANDαβ transgenic mice containing 10⁴ naive PCC-specific CD4 T cells were transferred i.v. into B10.BR-Thy1.1 congenic mice at the time of immunization.

Cell suspensions from lymphoid tissues in PBS with 5% FCS were labeled for 45 min at 4°C at a density of 2.0 × 10⁸ cells/ml with predetermined optimal concentrations of the following fluorophore-labeled mAbs: FITC-conjugated anti-Vα11 (RR8.1), PE-conjugated anti-Vβ3 (KJ25; all produced in the laboratory); Cy7-APC -conjugated anti-CD44 (IM7), APC-conjugated anti-CD90.2 (30-H12), Cy5-PE-conjugated anti-B220 (6B2), anti-CD8α (53-6.7) and anti-CD11b (M1/70; all from Biolegend, San Diego, CA); APC-Alexa Fluor 750 conjugated anti-CD44 (Pgp-1; eBioscience, San Diego, CA); APC conjugated anti-CD62L (MEL-14; BD Biosciences, San Jose, CA). PE-MCC/I-E^k tetramers (pMHCII tetramer), prepared as previously described (14), were incubated for 2 h at room temperature at a final concentration of 300 nM. After staining, cells were suspended in 1.5 μg/ml DAPI (Invitrogen, Carlsbad, CA) (for dead cells exclusion) for analysis. For Annexin V staining, cells that had been incubated with antisurface marker Abs were washed and incubated with Annexin V-PE (BD Biosciences) and DAPI for 15 min at room temperature. As positive control, splenocytes were incubated with 5 μM staurosporine for 3 h. Data were collected with FACS Diva software (BD Biosciences) and were analyzed with FlowJo software (Tree Star, Ashland, OR). Profiles are presented as 5% probability contours with outliers.

Single cells with the appropriate surface phenotype were sorted for repertoire analysis on a FACS Aria and CloneCyt software (BD Biosciences). The synthesis of cDNA and amplification of TCR Vα11 and Vβ3 regions were carried out as previously described (13). Direct sequencing of the purified PCR products was carried out with a Vα11-specific or Vβ3-specific primer by a commercial vendor (Integrated DNA Technologies, Coralville, IA). The frequency of obtaining a sequenceable PCR product from single cells varies between TCR-α and TCR-β with minimal variation across different peptides.

Cell suspensions from draining lymph nodes (LNs) in RPMI 1640 supplemented with 10% FCS, 50 nM 2-ME, were plated in 24-well plates (3 × 10⁶ cells/ml) and restimulated with titrated dilutions of MCC_88–104 or PCC_103K peptide for 4 d. IFN-γ levels in supernatants were measured by sandwich ELISA using mAb from Biolegend. Plates were read at 405 nm by a VERSAmax microplate reader and data were analyzed with SoftMax Pro software (Molecular Devices, Sunnyvale, CA). Detection limit was 15 pg/ml.

Statistical differences between experimental groups were determined by the Student t test. Correlations between parameters were assessed by the Spearman correlation analysis. Statistical analysis was conducted with Prism software (GraphPad Software, La Jolla, CA). p < 0.05 was considered statistically significant.

To examine the importance of pMHCII stability for CD4 T cell clonal selection, we mutated two well-characterized cytochrome c peptides (PCC_88–104 and MCC_88–103) at I-E^k anchor residues to produce peptides with altered binding strengths. PCC_88–104 is known to raise a heteroclitic response to MCC_88–103, which means that all T cell clones raised by immunization with PCC_88–104 respond better to MCC_88–103 (17). An important difference between MCC_88–103 and PCC_88–104 is that PCC_88–104 has an alanine at position 103, which partially fills the P9 pocket (18) and reduces its binding affinity and stability to I-E^k (19). To produce a PCC_88–104 variant with increased stability, an Ala→Lys substitution at P9 pocket residue (PCC_103K) was synthesized. Most peptides that bind to I-E^k have a large hydrophobic residue filling the P1 pocket of I-E^k (Ile or Leu). To produce a MCC_88–103 peptide variant with decreased stability, an Ile→Ala substitution at P1 (MCC_95A) was synthesized. Purified I-E^k molecules were loaded with biotinylated peptide in vitro at endosomal pH and 37°C for 3 d and pMHCII half-lives were determined. PCC_103K displayed the longest half-life (t_1/2 = 229 h) and was three times more stable than MCC_88–103 (t_1/2 = 73 h) (Fig. 1A, 1C). In contrast, peptides lacking one key anchor residue displayed reduced pMHCII stability. PCC_88–104 displayed a half-life of only 5 h, ∼50-fold less than its higher stability variant PCC_103K. MCC_95A displayed the shortest half-life of all peptides tested, (t_1/2 = 0.6 h). The same hierarchy of binding was found when peptides were assayed by competition (Fig. 1B, 1C). Collectively, our measurements of relative stability established the following hierarchy of peptides: MCC_95A<PCC_88–104<MCC_88–103<PCC_103K.

The s.c. immunization of B10.BR mice with 400 μg of whole PCC protein triggers a PCC-specific CD4 T cell response that reaches a maximum population size in draining LNs 7 d after immunization (13, 15) (Fig. 2A). When we compared the day 7 responses induced by an equivalent molar amount of peptides, we found that all four peptides elicited similar numbers of Ag-specific CD4 T cells (Vα11⁺Vβ3⁺CD44^hiCD62L^lo) regardless of their differences in pMHCII stability (Fig. 2B, 2C). Ag-specific CD4 T cells accumulated earlier in draining LNs of mice immunized with high-stability peptides PCC_103K (Fig. 2D), but the magnitude of the CD4 T cell response between PCC_88–104 and PCC_103K was similar at later time points. There was also an increase of Vα11⁺Vβ3⁺CD44^hiCD62L^hi cells for all peptides at day 7, but their total number paralleled what was found in the CD44^hiCD62L^lo compartment (Supplemental Fig. 1). Therefore, the level of accumulation of Ag-specific CD4 T cells in draining LNs appears minimally impacted by pMHCII stability.

To determine the clonal composition of the responding T cells induced by peptides with varying binding half-lives with MHCII molecules, we sorted single Ag-specific CD4 T cells (Vα11⁺Vβ3⁺CD44^hiCD62L^hi cells) from draining LNs 7 d after peptide immunization and sequenced their CDR3α and CDR3β regions using a single-cell RT-PCR approach. Previous studies have shown that the CD4 T cell response to whole PCC protein is dominated by clonotypes expressing TCR with eight preferred CDR3 features (13). Four preferred features have been defined in the TCRα-chain: 1) glutamic acid at positions 93, 2) serine at positions 95, 3) a CDR3 length of 8aa, and 4) preference for Jα16, 22, 34, or 17 gene segment. Similarly, four CDR3β features are predominantly found in PCC-specific CD4 T cells: 1) asparagine at position 100, 2) alanine or glycine at position 102, 3) a CDR3 length of 9aa, and 4) preference for Jβ1.2 or Jβ2.5 gene segment. All four peptides favored Ag-specific CD4 T cells expressing six or more of these CDR3 features in their TCR (>70% of responders express six or more of the eight preferred features for all peptides, Fig. 3A). However, there was a significant inverse relationship between the average number of preferred TCR features expressed by the responding CD4 T cells and the pMHCII half-life (Spearman: r = −0.77 and p = 0.003; Fig. 3B). This inverse correlation was evident for both the TCRα (Spearman: r = −0.61 and p = 0.03; Fig. 3C) and TCRβ-chain (Spearman: r = −0.74 and p = 0.006; Fig. 3D). These data suggest that pMHCII stability determines the clonotypic diversity of the responding CD4 T cell compartment.

We have previously shown that, within the dominant clonotypes (clones expressing six or more of the eight preferred features), there were considerable variations in TCR binding properties that broadly correlated with J region usage. PCC-specific TCRβ transgenic cells expressing Jβ2.5 displayed overall lower affinity for pMHCII than cells expressing Jβ1.2 (14, 15). Similarly, different Jα genes pairing with the same TCRβ also conferred different binding kinetics (14). Hence, we next examined the prevalence of individual CDR3 features selected by low- and high-stability peptides. At the level of the CDR3α region, there were some variations predominantly based on Jα region usage (Supplemental Fig. 2). All peptides selected similar frequencies of the canonical Jα but low-stability peptides favored clonotypes expressing Jα16, whereas higher stability peptides favored dominant clonotypes expressing Jα22 (Supplemental Fig. 2A). All other CDR3α features were highly restricted across all peptides, except for the highest stability peptide PCC_103K that selected a higher number of clonotypes with longer CDR3α (Supplemental Fig. 2B) and more diverse amino acids at positions α93 and α95 (Supplemental Fig. 2C, 2D, respectively). However, the most consistent and systematic change between low- and high-stability peptides was observed at the level of the CDR3β region (Supplemental Fig. 3), particularly at the level of Jβ region usage (Fig. 4A, Supplemental Fig. 3A). Similar to what was observed with the whole PCC protein, PCC_88–104 peptide induced three times more Jβ1.2- than Jβ2.5-expressing clonotypes (Fig. 4A) (15). The lowest stability peptide MCC_95A further exaggerated the frequency of Jβ1.2-expressing clonotypes (ratio Jβ1.2/Jβ2.5 = 13 ± 4, Fig. 4A). In contrast, higher stability peptides did not display this bias and recruited similar numbers of Jβ1.2- and Jβ2.5-expressing clonotypes (Fig. 4A). The CDR3β feature of G or A at β102 correlated with the Jβ region differences (Supplemental Fig. 3B), whereas the CDR3β length of 9aa (Supplemental Fig. 3C) and the selection for N at β100 (Supplemental Fig. 3D) were similarly restricted across all peptides. Hence, pMHCII stability skews J region gene usage of the dominant Ag-specific CD4 T cell compartment.

We have previously shown that the CD4 T cell response to PCC whole protein was dominated by Jβ1.2 clonotypes expressing one specific public (present in most of the mice) CDR3 rearrangement (SLNNANSDY or 5C.C7β-chain) (15) that conferred high pMHCII binding affinity (14, 15). Next, we analyzed the CDR3β amino acid sequences selected by high- and low-stability peptides and divided the sequences based on their relative prevalence distinguishing public, recurrent (present in more than one mouse), and private (specific to individual mice) TCRβ-chains (Fig. 4B, Supplemental Fig. 4). Interestingly, the frequency of public TCRβ sequences in the Ag-specific CD4 T cell repertoire correlated inversely with the pMHCII stability (Spearman: r = −0.9 and p < 0.0001; Fig. 4C). Responses to low- and intermediate-stability peptides MCC_95A, PCC_88–104, and MCC_88–103 were dominated by clones expressing public TCRβ-chains, whereas response to the high-stability peptide PCC_103K was dominated by clones displaying private TCRβ-chains (Fig. 4B, Supplemental Fig. 4). When we compared the public TCRβ-chains selected by individual peptides, we observed that low-stability peptides predominantly selected the 5C.C7β-chain (SLNNANSDY) and the related C.F6β-chain (SLNSANSDY) (20) (Fig. 4B). Both 5C.C7β- and C.F6β-expressing clonotypes were found in MCC_88–103 immunized mice but this response was dominated by a third public CDR3β-chain (SLNRGQDTQ) (Fig. 4B) that was also selected by low-affinity PCC-specific CD4 T cells in mice immunized with PCC protein and IFA or Alum (15). Overall, our results demonstrate that pMHCII stability controls the clonotypic diversity of the CD4 T cell compartment and suggest that decreasing pMHCII stability focuses Ag-specific CD4 T cells toward public clonotypes expressing high-affinity TCR.

The pMHCII tetramer staining provides a distinct method to detect PCC-specific CD4 T cells in vivo (15, 21). Importantly, the measurement of pMHCII tetramer mean fluorescence intensity (MFI) at optimal concentrations also provides a reliable indicator of differential TCR-binding affinity (14, 15, 22–25). For three of four peptides, the number of Ag-specific CD4 T cells detected by pMHCII tetramer staining was similar to the one estimated by V region-based staining (Fig. 5A, 5B). However, for the highest stability peptide PCC_103K, significantly fewer Ag-specific CD4 T cells were recognized by pMHCII tetramer staining (Fig. 5A, 5B), suggesting the presence of lower affinity clonotypes that fall below the level of detection of pMHCII tetramer reagents (15). Consistent with the TCR repertoire studies, we observed an inverse relationship between the pMHCII tetramer staining MFI of the responding CD4 T cells and the pMHCII stability (Spearman: r = −0.6 and p = 0.0019; Fig. 5C). The highest stability peptide PCC_103K induced the Ag-specific T cell population with the lowest MFI intensity compared with all other peptides (Fig. 5C). For lower stability peptides (PCC_88–104 and MCC_95A), there was a significant increase in MFI corresponding to the increased presence of higher affinity cells (Fig. 5A, 5C). This difference in pMHCII tetramer staining was not caused by a difference in TCR expression levels, as the Vα11 staining MFI by Vα11⁺pMHCIItet⁺CD44^hi cells were similar across all peptides (Fig. 5D).

Because it is possible that differences in pMHCII tetramer staining intensities may not reflect differences in TCR affinity but differences in activation status (26), we also assessed the functional avidities of CD4 T cells responding to low- and high-stability peptides. The functional avidity of the responding populations was measured by assessing their ability to produce IFN-γ ex vivo in response to decreasing concentration of the high-stability peptide MCC_88–103. For all immunization conditions, we found that the maximal response was detected with 10⁻⁵ M peptide and that replacing MCC_88–103 by PCC_103K did not affect the magnitude of this response (Fig. 5E). When restimulated with optimal MCC_88–103 peptide concentration, Ag-specific CD4 T cells induced by high-stability peptides (PCC_103K, MCC_88–103) produced more IFN-γ than those elicited by low-stability peptides (PCC_88–104 and MCC_95A) (Fig. 5E), but with lower peptide concentrations in culture, the magnitude of the IFN-γ response induced by high-stability peptides decreased rapidly, showing a half-maximal response (EC₅₀) between 0.6 × 10⁻⁶ M and 1 × 10⁻⁶ M (MCC_88–103 and PCC_103K, respectively; Fig. 5F). In contrast, Ag-specific CD4 T cells induced by low-stability peptides displayed a superior EC₅₀ between 0.8 × 10⁻⁷ M and 1 × 10⁻⁷ M (PCC_88–104 and MCC_95A, respectively; Fig. 5F). Consistent with the pMHCII tetramer binding assays, there was a direct relationship between the EC₅₀ and the pMHCII stability (Spearman: r = 0.5 and p = 0.01; Fig. 5G). Together, these data suggest that pMHCII stability alters the prevalence of high-affinity cells within the Ag-specific CD4 T cell population.

The low prevalence of T cells bearing high-affinity TCR in response to high-stability peptides could result from their elimination by apoptosis (5). To investigate this possibility, we used high-affinity AND PCC-specific TCRαβ-transgenic CD4 T cells in adoptive-transfer experiments. AND CD4 T cells express the 5C.C7β-chain paired with a Jα16-expressing TCRα-chain (CDR3α: EASSGQKL) conferring high affinity for pMHCII binding (14) (Supplemental Fig. 5). We transferred CD90.2-expressing transgenic cells into CD90.1 recipient mice and immunized recipients with either low-stability peptide PCC_88–104 or high-stability peptide PCC_103K. Seven days after immunization, comparable numbers of high-affinity AND CD4 T cells were recovered from draining LNs of recipients immunized with high- or low-stability peptides (Fig. 6A, 6B). Furthermore, there was no evidence for increased apoptosis of AND cells in mice immunized with the highest stability peptide PCC_103K (Fig. 6C). Hence, immunization with high-stability peptides does not cause the elimination of high-affinity clonotypes.

Using whole PCC protein Ag, we had previously demonstrated that TCR-based selection was independent of Ag dose (14, 15). Because of the low-stability of the PCC immunodominant peptide, it was important to readdress the importance of the Ag dose in the context of a higher stability peptide such as PCC_103K. Decreasing 100-fold the dose of low-stability peptide PCC_88–104 significantly altered Ag-specific CD4 T cell clonal expansion (Fig. 7A, 7B). In contrast, a 10,000-fold reduction in the dose of high-stability peptide PCC_103K was necessary to observe a significant decrease in the number of Ag-specific CD4 T cells (Fig. 7A, 7B). To determine the clonal composition of the responding T cells induced by these lower doses of PCC_103K, we isolated single Ag-specific CD4 T cells from mice immunized with 0.06 μg peptide for repertoire analysis. At this dose, there was a significant increase in the number of clonotypes expressing eight preferred features (Fig. 7C, Supplemental Fig. 6) and an increase in the prevalence of clonotypes expressing Jβ1.2 instead of Jβ2.5 (Fig. 7D). At the TCRβ-chain peptidic level, we also observed an increased prevalence of high-affinity public clonotypes (SLNNANSDY and SLNSANSDY) (Fig. 7E). Consistent with this TCR repertoire change, Ag-specific CD4 T cells induced at a low Ag dose had a greater pMHCII tetramer staining intensity than Ag-specific CD4 T cells induced at a higher dose (Fig. 7F, 7G). Thus, pMHCII stability regulates the clonotypic diversity of the Ag-specific CD4 T cell response in a dose-dependent manner.

Our studies revealed the remarkable adaptation of the Ag-specific CD4 T cell response and its TCR repertoire to pMHCII stability. By mutating key anchor residues, we generated four peptides with marked differences in their capacity to form stable pMHCII complexes. Despite these differences in pMHCII stability, all four peptides induced comparable clonal accumulation of Ag-specific CD4 T cells in vivo. There was, however, a direct correlation between pMHCII stability and the clonotypic diversity of the CD4 T cell response as measured by the prevalence of preferred CDR3 features or by the frequency of public TCRβ-chains in the Ag-specific CD4 T cell repertoire. Lower stability peptides (MCC_95A, PCC_88–104) selected clonotypes expressing more restricted TCR with public TCRβ-chains and displaying high avidity for their pMHCII ligand. In contrast, higher stability peptides (MCC_88–103, PCC_103K) selected clonotypes expressing less restricted TCR, more private TCRβ-chains and displaying lower avidity for their pMHCII ligand. The Ag dose altered this clonal selection, and at a low-dose, high-stability peptides recruited more restricted TCR repertoire. Hence, our studies established that the clonotypic diversity of the Ag-specific CD4 T cell compartment that regulates the adaptive immune response is governed by pMHCII stability and the Ag dose in vivo.

Anchor residues in peptides determine the stability of binding to MHCII molecules. Kersh et al. (27) have, however, shown that a single amino acid substitution at the P6 I-E^k anchor position of the hemoglobin peptide, that did not alter the peptide stability, had sufficient effect on the solvent-exposed face of the pMHCII complex to modify CD4 T cell recognition. In our studies, substituting the P9 anchor residue of PCC_88–104 or the P1 anchor residue of MCC_88–103 leads to the predicted enhancement or attenuation of binding to I-E^k. Although we cannot exclude that these substitutions also produce some changes in the solvent-exposed face of the pMHCII complex offered to the T cells, the remarkable correlations between the peptide binding half-lives with I-E^k and the complexity and avidity of the responding CD4 TCR repertoires strongly argue that pMHCII stability is the major determinant of the clonotypic diversity of the Ag-specific CD4 T cell compartment in our study. One of the surprising results derived from the peptide-binding stability studies was that PCC_103K displayed significantly longer pMHCII half-life than MC_88–103, despite sharing the same core amino acid sequence and all MHC anchor residues, indicating that positions outside the “formal” peptide-binding groove could play a role in the overall binding energy of the peptide/MHC complex. It is possible that the lysine located outside of the binding groove in PCC_103K at position P10 participates in the peptide anchoring to I-E^k. The P10 position has been found to play a role in complex stability in a number of studies (28, 29) and a structural correlate has been determined in that the side chain can lie on a shallow external shelf (28).

We have proposed that the expansion of Ag-specific CD4 T cells in vivo is limited by a TCR affinity threshold (14). One critical aspect of this model is that, above a threshold affinity, all clonotypes undergo similar clonal expansion regardless of further differences in their TCR-binding properties. In the current studies, we identify pMHCII stability as a critical regulator of this affinity-based selection. Our results suggest that high-stability peptides lower the CD4 T cell affinity-based selection threshold allowing the expansion of a large number of lower affinity clonotypes. Our results do not support the idea that peptides with long half-lives drive the apoptosis of high-affinity CD4 T cells (5). Instead, consistent with our threshold selection model, we interpret the low prevalence of high-affinity clonotypes in response to high-stability peptides by the lack of proliferative advantage of CD4 T cells with high-affinity TCR when the selection threshold is decreased. The discrepancy between the two models could come from the fact that Anderton et al. (6) were studying the fate of autoreactive CD4 T cells that mediate experimental autoimmune encephalomyelitis. In this model, high-affinity autoreactive CD4 T cells ultimately disappear from the immune repertoire as the animal recovers (6). In contrast, high-affinity Ag-specific CD4 T cells generated by immunization with foreign Ag in Ribi adjuvant preferentially develop into effector follicular Th cells (25), persist as memory follicular Th cells in draining LNs (24), and dominate the response to secondary Ag challenge (12, 13). Hence, the elimination of high-affinity autoreactive CD4 T cells in the periphery may represent a form of peripheral tolerance and not a general mechanism of negative selection of high-affinity clonotypes by strong peptide agonists.

How the Ag dose impacts the clonotypic composition of the CD4 T cell compartment is not well understood. Our previous studies using whole PCC protein immunization reported that Ag dose did not significantly impact CD4 T cell clonal selection (14, 15). Rees et al. (30) using a peptide immunization model found that low Ag dose promoted the selection of high-affinity CD4 T cells but this selection was only apparent by tetramer staining at later stages of the primary CD4 T cell response. Our current study shows that Ag dose does indeed impact the clonal composition of the effector CD4 T cell compartment and this impact can be seen at the peak of the primary immune response. This dose effect may not be as pronounced with lower stability peptides because the Ag-specific TCR repertoire obtained at high Ag dose is already very restricted and low-stability peptides do not promote a CD4 T cell response at a very low dose (14, 15). The capacity of high-stability epitopes to differentially expand high-affinity CD4 T cells at low and high Ag doses may have important implications in defining the Ag-specific CD4 T cell repertoire to complex pathogens. The effective Ag dose presented to the Ag-specific CD4 T cells postinfection is, in large part, determined by the capacity of the innate immune response to prevent the pathogen dissemination. A robust innate immune response would lower the Ag dose presented to CD4 T cells and thus favor the development of high-affinity CD4 T cells. Consistent with this, we have previously reported that mice resistant to Leishmania major infection, that effectively prevent the parasite dissemination (31) very rapidly focus their parasite-specific CD4 T cell response toward high-affinity clonotypes, whereas susceptible BALB/c mice that fail to prevent the parasite dissemination (31) recruit lower affinity parasite-specific CD4 T cells (32).

The emergence of high-affinity clonotypes with similar TCR repertoire usage in situations of low Ag dose and low pMHCII stability suggests that pMHCII stability and Ag dose regulate one unique parameter that determines the affinity selection threshold. Henrickson et al. (33) have shown that pMHC class I (pMHCI) stability and Ag dose ultimately determine the number of pMHCI complexes displayed by Ag-presenting dendritic cells in draining LNs. The pMHCI complex density on the surface of dendritic cells appears to set a threshold for CD8 T cell activation (33). The pMHCII density on the surface of the Ag-presenting cells may set a similar threshold for CD4 T cell activation by regulating the magnitude of TCR engagement (34–36). Because CD4 T cells required pMHCII contacts throughout their expansion phase (37, 38), decreasing pMHCII density on the surface of the Ag-presenting cells at the initiation of the CD4 T cell response (immunization with low Ag dose) or during CD4 T cell clonal expansion (immunization with low-stability peptide) may set a higher activation threshold that favors the expansion of high-affinity clonotypes.

In conclusion, we have demonstrated that pMHCII stability and Ag dose are important determinants of the clonotypic diversity of Ag-specific CD4 T cell responses. Our findings provide new insights into the molecular mechanism of CD4 T cell clonal selection during an immune response. Together with our previous studies (15), an important implication of our observations is that several parameters (Ag dose, nature of CD4 T cell epitope, adjuvant) in a protein subunit vaccine or immunotherapy have the potential to change the affinity and clonotypic diversity of Ag-specific CD4 T cell responses and may thereby affect the quality of the protective immune response.

We thank Dr. Bonnie Dittel for helpful discussion, Dr. Aniko Szabo for help with statistical analysis, and Trudy Holyst for technical assistance with the reagents.

Disclosures The authors have no financial conflicts of interest.