We have used TCR transgenic mice directed to different MHC class II-restricted determinants from the influenza virus hemagglutinin (HA) to analyze how specificity for self-peptides can shape CD4+CD25+ regulatory T (Treg) cell formation. We show that substantial increases in the number of CD4+CD25+ Treg cells can occur when an autoreactive TCR directed to a major I-Ed-restricted determinant from HA develops in mice expressing HA as a self-Ag, and that the efficiency of this process is largely unaffected by the ability to coexpress additional TCR α-chains. This increased formation of CD4+CD25+ Treg cells in the presence of the self-peptide argues against models that postulate selective survival rather than induced formation as mechanisms of CD4+CD25+ Treg cell formation. In contrast, T cells bearing a TCR directed to a major I-Ad-restricted determinant from HA underwent little or no selection to become CD4+CD25+ Treg cells in mice expressing HA as a self-Ag, correlating with inefficient processing and presentation of the peptide from the neo-self-HA polypeptide. These findings show that interactions with a self-peptide can induce thymocytes to differentiate along a pathway to become CD4+CD25+ Treg cells, and that peptide editing by DM molecules may help bias the CD4+CD25+ Treg cell repertoire away from self-peptides that associate weakly with MHC class II molecules.
The CD25+ regulatory T (Treg)4 cells comprise a distinct CD4+ T cell subset, displaying unique gene transcription profiles and phenotypic characteristics that confer the capacity to suppress immune responses (1). The deleterious autoimmune responses that develop in mice and humans lacking key molecules associated with CD4+CD25+ Treg cell function (e.g., Foxp3) show that these cells play a crucial role in regulating immune responses to self-Ags (2). CD4+CD25+ T cells have been found to display evidence of specificity for self-Ags, because they could exhibit tissue specificity in preventing autoimmune disease (3, 4). Similarly, repertoire analysis of CD4+CD25+ vs CD4+CD25− T cells found that peripheral CD25+ Treg cells are enriched for TCRs with higher affinities for self-Ags (5, 6, 7); still, how TCR specificity directs the formation and activity of CD4+CD25+ Treg cells is not fully understood. Additionally, how these processes might be affected by the diversity of peptides (possessing varying rates of anabolism and catabolism in APCs) that can potentially be generated from individual self-Ags has not been addressed.
Several studies aimed at determining how specificity for self-peptides shape CD4+CD25+ Treg cell formation have used TCR transgenic mice, which offer the advantage of simplifying the otherwise highly diverse CD4+ T cell repertoire (8, 9, 10, 11, 12, 13). Early studies using TCR transgenic mice suggested a role for dual TCR expression in CD25+ Treg cell formation (11, 14), as has also been suggested in recent studies of human CD25+ Treg cells (15). However, other studies using TCR transgenic mice indicated that dual TCR α-chain expression is not required for CD25+ Treg cell formation (12, 16). How TCR specificity directs CD4+CD25+ Treg cell formation became clearer when mice expressing transgenic TCRs directed to nominal Ags that are not normally expressed in the mouse were compared with mice expressing the TCR along with its cognate Ag as a transgenic neo-self-peptide (8, 9, 10). CD4+ T cells expressing the transgenic TCRs were enriched for CD25+ Treg cells when they developed in the presence of their cognate peptide, suggesting that specificity for self-peptides can direct TCRs to undergo CD25+ Treg cell selection during their development in the thymus. However, whether this represents self-peptide-induced formation of CD25+ Treg cells or selective survival of CD25+ Treg cells in the context of overall thymocyte deletion has been questioned (13). As a result, the issue of whether self-peptides can act as ligands that induce CD25+ Treg cell formation remains controversial. In some cases, studies using mice coexpressing TCRs and their cognate peptides have been performed in RAG−/− backgrounds to establish that it is the specificity of the TCR for a self-peptide that directs it to undergo CD25+ Treg cell selection (8, 9, 10). These studies have also been questioned on grounds that the gross perturbations of thymic architecture and lymphocyte compartments that are a consequence of the RAG−/− mutation may cause selection events to occur that would not take place in mice with less perturbed immune systems (13).
In the studies here, we have used TCR transgenic mice that react with different determinants from the influenza virus hemagglutinin (HA) to further analyze how peptide specificity can contribute to CD4+CD25+ Treg cell formation. We show that one of these CD4+ T cell determinants can induce the formation of substantially higher numbers of CD4+CD25+ T cells when the HA is expressed as a neo-self-Ag in HA transgenic mice. However, the other CD4+ T cell determinant induces little or no CD4+CD25+ Treg cell formation, reflecting inefficient processing and presentation from the neo-self-HA polypeptide and correlating with enhanced sensitivity to DM-induced peptide editing. The findings show that individual self-peptides generated from a self-Ag can differ substantially in their ability to induce CD4+CD25+ Treg cell formation.
Materials and Methods
TS1 and TS1 × HA28 mice have been previously described (9, 17). TS2 mice have also been previously described (18); these mice are also designated HNT-TCR mice (19) and encode a transgenic TCR specific for the major I-Ad-restricted determinant of the influenza virus A/PR/8/34 (PR8) HA (S2) (20). TS1, TS2, HA28, and TS1 × HA28 mice had been backcrossed at least 10 generations with BALB/c mice before use. TS1.Cα−/− and TS1 × HA28.Cα−/− mice were generated through mating of either TS1 or TS1 × HA28 mice to TCR Cα−/− mice (21) that had been bred to homozygosity for H-2d haplotype and then backcrossed four generations with BALB/c mice. Mice were maintained in sterile microisolators at The Wistar Institute facility. All experiments involving animals were performed under protocols approved by The Wistar Institute Institutional Animal Care and Use Committee.
Viruses and peptides
PR8 was propagated in the allantoic cavity of 11-day-old embryonated hens’ eggs, and virus titers were determined as hemagglutinating units (HAU) as described (22). Synthetic peptides corresponding to S1 (SFERFEIFPKE) and S2 (HNTNGVTAACSHE) were synthesized and purified by HPLC at The Wistar Institute Peptide Synthesis Facility.
Single-cell suspensions of thymus, spleen, or pooled axillary, inguinal, brachial, and superficial cervical lymph nodes (LN) were stained for flow cytometric analysis with the following mAbs: anti-CD4-Alexa 700 (RM4-5), anti-CD4-FITC (GK1.5), anti-Vβ 8.3-FITC (1B3.3), anti-CD25-FITC (7D4), anti-CD25-PE (PC61), anti-CD8-PerCP-Cy5.5 (53-6.7) (all obtained from BD Pharmingen), and anti-clonotypic TS1 TCR Ab 6.5-biotin (23). Streptavidin-allophycocyanin (BD Pharmingen) or streptavidin-RED670 (Invitrogen: Life Technologies) was used to detect biotinylated Abs. TCR Vβ usage was determined by flow cytometry using a panel of 15 FITC-conjugated TCR Vβ-specific Abs (BD Pharmingen). Intracellular Foxp3 staining was performed using the reagents provided with an anti-Foxp3-PE Ab (FJK-16s) according to the manufacturer’s directions (eBioscience). Live events (100,000–400,000) were collected on either a FACSCalibur (BD Pharmingen) or DakoCytomation CyAn and analyzed using either FlowJo (Tree Star) or Summit (DakoCytomation) software.
CFSE labeling and T cell proliferation assays
LN cells from TS1 or TS2 mice were labeled with CFSE (Molecular Probes) as previously described (24). Briefly, LN cells were prepared as single-cell suspensions in serum-free supplemented IMDM (SF medium) at 37°C, incubated with 5 mM of CFSE at 1 × 107 cells/ml for 4 min, and then washed in SF medium. For in vitro proliferation analyses, CFSE-labeled cells (1 × 106/ml) were incubated either with allantoic fluid containing influenza PR8 virus or with S1 or S2 peptides and irradiated splenocytes (5 × 106/ml) in supplemented IMDM containing 10% FBS. After 3 days cells were analyzed by flow cytometry. For in vivo proliferation analyses, 0.5–1.5 × 107 CFSE-labeled cells were injected into the tail veins of BALB/c, HA28, and HA104 recipients, and 5 days later recipient LNs were harvested and analyzed by flow cytometry.
Bone marrow (BM) chimeras
Donor BM was harvested from TS1.Cα−/− and TS2.Cα−/− mice and depleted of T lymphocytes using MACS columns. BM was stained with biotinylated Abs to CD4, CD8, and I-Ed followed by incubation with streptavidin-conjugated magnetic beads, and depletion was performed according to the manufacturer’s instructions (Miltenyi Biotec). BALB/c or HA28 mice (4–12 wk of age) were exposed to 900 rad of gamma irradiation and reconstituted with 4–5 × 106 T cell-depleted BM cells via tail vein injection 24 h later. Mice received neomycin sulfate (0.2 mg/ml) on the day of irradiation and every 2 days for 2–3 wk after irradiation and were analyzed 2 mo following reconstitution.
Ag presentation assays
Murine fibroblast L cells transfected with MHC genes encoding Ad, Ii, Ed, and in some cases DM (25, 26) were used as APCs to present PR8 virus, S1 peptide, or S2 peptide to S1- or S2-specific T cell hybridomas as previously described (27). Briefly, 5 × 104 T cell hybridomas were incubated with 4 × 104 L cells expressing Ad, Ii, Ed, and either bearing or lacking DM overnight, followed by analysis of culture supernatant. The capacity of the T cell hybridomas to respond to the presented Ags was measured by the production of IL-2, as determined by coculture with the IL-2-dependent cell line CTL.L as previously described (25).
Increased formation of S1-specific CD4+CD25+ Treg cells in TS1 × HA28 and TS1 × HA28.Cα−/− mice
TS1 mice express a transgenic TCR that is specific for the major I-Ed-restricted T cell determinant of HA (termed S1, represented by the synthetic peptide SFERFEIFPKE) and that can be recognized by the anti-clonotypic mAb 6.5 (23). HA28 mice express the influenza virus PR8 HA as a transgenic neo-self-Ag. We previously reported that TS1 × HA28 mice contain increased percentages, relative to TS1 mice, of 6.5+CD4+ T cells and 6.5+CD4+CD8− thymocytes that are CD25+ and can exert regulatory function (9). To analyze the formation of these regulatory T cells in greater detail, we generated TS1 × HA28.Cα−/− mice, which cannot generate endogenous TCR α-chain gene rearrangements. We then examined CD4+CD8− thymocytes and CD4+ T cells from TS1 × HA28.Cα−/−, TS1.Cα−/−, TS1 × HA28, and TS1 mice for their expressions of 6.5, CD25, and Foxp3, which are closely linked to regulatory T cell function (28).
The total numbers of thymocytes and of CD4+CD8− thymocytes were similar between TS1 and TS1 × HA28 mice and between TS1.Cα−/− and TS1 × HA28.Cα−/− mice (Fig. 1,A and data not shown). Additionally, the frequency of CD4+CD8− thymocytes expressing high levels of the 6.5 TCR doubled in the Cα−/− backgrounds (on average 80 vs 39% 6.5+CD4+CD8− thymocytes in TS1.Cα−/− vs TS1 mice, and 67 vs 28% 6.5+CD4+CD8− thymocytes in TS1 × HA28.Cα−/− vs TS1 × HA28 mice), and in both TS1 × HA28 and TS1 [mice] HA28.Cα−/− mice the average number of 6.5+CD4+CD8− thymocytes was reduced by approximately half relative to mice that do not coexpress the HA transgene (Fig. 1,B). Significantly, however, TS1 × HA28 and TS1 − HA28.Cα−/− mice each contained on average at least 10-fold higher numbers of 6.5+CD4+CD8−CD25+ thymocytes than TS1 or TS1.Cα−/− mice (Fig. 1 C).
Because of the increased representation of 6.5+CD4+CD8−CD25+ thymocytes in mice coexpressing the S1 peptide, the CD4+CD8−CD25+ thymocyte populations in TS1 × HA28 and TS1 × HA28.Cα−/− mice were heavily enriched for cells expressing high levels of the 6.5 TCR (Fig. 1,C). In contrast, the small fraction of CD4+CD8− thymocytes from TS1 mice that were CD25+ mostly expressed low levels of the 6.5 TCR, suggesting that they use endogenous TCR gene rearrangements to undergo CD25+ T cell selection in the absence of S1 peptide. Consistent with this result, even fewer CD4+CD8− thymocytes in TS1.Cα−/− mice were CD25+, further suggesting that inhibiting rearrangement of endogenous TCR α-chains can suppress the formation of CD4+CD8−CD25+ thymocytes in mice expressing the 6.5 TCR but lacking the S1 peptide (Fig. 1,C). It was also notable that many of the 6.5+CD4+CD8−CD25+ thymocytes in TS1 × HA28 and TS1 × HA28.Cα−/− mice coexpressed Foxp3, and that each strain also contained sizable populations of 6.5+CD4+CD8−CD25−Foxp3+ thymocytes, resembling studies in nontransgenic mice expressing a GFP reporter gene to identify Foxp3+ cells (Fig. 1 D) (29, 30).
CD4+ LN cells from TS1.Cα−/− and TS1 × HA28.Cα−/− mice were also substantially enriched for expression of 6.5+ cells relative to TS1 and TS1 × HA28 mice. In this case, however, whereas the numbers of 6.5+CD4+ cells did not differ in TS1 vs TS1 × HA28 mice, there were fewer 6.5+CD4+ (and fewer total CD4+) cells in the LNs of TS1 × HA28.Cα−/− mice than in those of TS1.Cα−/− mice (Fig. 2,A and data not shown). Nevertheless, the LNs of TS1 × HA28.Cα−/− and TS1 × HA28 mice each contained significantly higher numbers of 6.5+CD4+CD25+ T cells than were present in LNs from TS1.Cα−/− and TS1 mice. As a result, the CD4+CD25+ LN cells from TS1 × HA28 and TS1 × HA28.Cα−/− mice were in each case enriched for cells expressing high levels of the 6.5 TCR, whereas those in TS1 mice mostly expressed low 6.5 levels, indicating usage of endogenous TCR chains (Fig. 2,B). Notably, the LN cells of TS1.Cα−/− mice also contained a population of CD4+CD25+ T cells (in contrast to CD4+CD8− thymocytes of TS1.Cα−/− mice, of which very few were CD25+) (compare Figs. 1,C and 2,B), indicating that some CD4+CD25+ T cell formation can occur even in TS1 mice that cannot generate endogenous TCR α-chain rearrangements. This finding resembles that from a previous study that showed that CD4+CD25+ T cells were detectable in the periphery of TCR transgenic DO11.10.Cα−/− mice despite being largely absent within the thymus (12). Finally, many of the 6.5+CD4+CD25+ LN cells in each background coexpressed Foxp3, and TS1 × HA28 and TS1 × HA28.Cα−/− mice again contained sizable populations of 6.5+CD4+CD25−Foxp3+ LN cells that were much less abundant in TS1 and TS1.Cα−/− mice (Fig. 2 C).
Endogenous TCR β-chain rearrangements facilitate CD4+CD25+ Treg formation in TS1.Cα−/− mice
Previous studies using several systems have shown that expression of a transgenic TCR in a RAG−/− background can prevent the formation of CD4+CD25+ T cells unless the cognate Ag for the TCR is also expressed (8, 9, 10). It was noteworthy, then, that TS1.Cα−/− mice contained a population of CD4+CD25+ LN cells (including cells expressing high levels of the 6.5 TCR) even though they lack expression of S1 peptide. Since endogenous TCR α-chain rearrangement is prevented in TS1.Cα−/− mice, we examined whether the CD4+CD25+ T cells that develop in this background express endogenous TCR β-chains (in addition to the transgene-encoded Vβ8.2 gene segment) that facilitate their selection in the absence of S1 peptide. Indeed, the frequencies of T cells coexpressing Vβ3, Vβ6, Vβ8.3, and Vβ17 chains were substantially higher in 6.5+CD4+CD25+ T cells from TS1.Cα−/− than from TS1 × HA28.Cα−/− mice (Fig. 3). The increased representation of 6.5+CD4+ T cells using additional TCR β-chains was selective for CD25+ T cells from TS1.Cα−/− mice, because it was not observed in 6.5+CD4+CD25− LN cells from either TS1.Cα−/− or TS1 × HA28.Cα−/− mice.
Individual T cell determinants from HA differ in their abilities to induce CD4+CD25+ Treg formation
The preceding studies showed that the S1 peptide derived from the neo-self-HA induces the formation of 6.5+CD4+CD25+ Treg cells in TS1 × HA28 and TS1 × HA28.Cα−/− mice. We were next interested to examine whether a second MHC class II-restricted CD4+ T cell determinant derived from the neo-self-HA can also induce CD4+CD25+ T cell formation. To address this, we made use of TS2 mice, which express a transgenic TCR that recognizes the major I-Ad-restricted determinant from HA (termed S2, represented by the synthetic peptide HNTNGVTAACSHE). There is no reagent available to detect either the Vα15 chain or the Vα15/Vβ8.3 combination that makes up the TS2 TCR. Accordingly, we generated TS2.Cα−/− mice to prevent endogenous TCR α-chain expression, and we used these mice to produce chimeras in which BM from TS2.Cα−/− mice was introduced into lethally irradiated BALB/c or HA28 mice. Following reconstitution, we used an anti-Vβ8.3 reagent to analyze development of CD4+CD8− thymocytes and CD4+ T cells expressing the TS2 TCR, in comparison with comparable chimeras in which BM from TS1.Cα−/− mice was used to reconstitute irradiated BALB/c and HA28 mice.
Consistent with previous reports that CD4+CD25+ Treg cell formation can be efficiently directed by radio-resistant thymic elements in HA28 mice (9, 31), the 6.5+CD4+CD8− thymocytes and 6.5+CD4+ T cells present in TS1.Cα−/− → HA28 mice contained much higher percentages of CD25+ cells than were present in TS1.Cα−/− → BALB/c mice (Fig. 4). The 6.5+CD4+CD8− thymocytes and 6.5+CD4+ LN cells from TS1.Cα−/− → HA28 mice also exhibited lower levels of the 6.5 TCR than were present in TS1.Cα−/− →BALB/c mice (mean fluorescence intensity of the 6.5 TCR of 267 vs 188 in 6.5+CD4+CD8− thymocytes from TS1.Cα−/− → BALB/c vs TS1.Cα−/− → HA28 mice, and 278 vs 165 in 6.5+CD4+ LN cells from TS1.Cα−/− → BALB/c vs TS1.Cα−/− → HA28 mice), which is typical of cells having undergone CD25+ Treg cell selection (9). There was also evidence that the 6.5 TCR is subjected to modest deletion by the S1 peptide, because the percentages of CD4+CD8− thymocytes and CD4+ LN cells expressing the 6.5 TCR were lower in TS1.Cα−/− → HA28 mice than in TS1.Cα−/− → BALB/c mice. Strikingly, however, there was no difference in the frequency of Vβ8.3+CD4+CD8−CD25+ thymocytes, or of Vβ8.3+CD4+CD25+ LN cells in TS2.Cα−/− → HA28 mice relative to TS2.Cα−/− → BALB/c mice. The overall percentages of Vβ8.3+CD4+CD8− thymocytes and of Vβ8.3+CD4+ LN cells also did not differ in TS2.Cα−/− → HA28 mice relative to TS2.Cα−/− → BALB/c mice, indicating that the TS2 TCR was not induced to undergo CD25+ Treg cell selection (Fig. 4).
We generated additional chimeric mice in which BM cells from TS1.Cα−/− or TS2.Cα−/− mice were used to reconstitute HA104 mice, which express higher levels of HA mRNA than do HA28 mice (32). We have previously shown that the 6.5 TCR is subjected to substantial deletion in TS1 × HA104 mice (33), reflecting these higher levels of HA mRNA expression. Consistent with these previous findings, TS1.Cα−/− → HA104 mice contained significantly fewer 6.5+CD4+CD8− thymocytes and 6.5+CD4+ LN cells than were present in either TS1.Cα−/− → BALB/c or TS1.Cα−/− → HA28 mice (Fig. 4). Additionally, the percentages (but not the absolute numbers) of 6.5+CD4+CD8− thymocytes and 6.5+CD4+ LN cells that were CD25+ were significantly higher in TS1.Cα−/− → HA104 than in TS1.Cα−/− → BALB/c mice (Fig. 4). When we examined TS2.Cα−/− → HA104 mice, we found that the percentage of Vβ8.3+CD4+ LN cells that were CD25+ was significantly higher in the HA104 background. However, the number of Vβ8.3+CD4+CD25+ LN cells was not increased in TS2.Cα−/− → HA104 mice relative to either TS2.Cα−/− → BALB/c or TS2.Cα−/− → HA28 mice, indicating little conversion of CD4+ cells expressing the TS2 TCR to become CD25+ in TS2.Cα−/− → HA104 mice.
The S2 peptide is inefficiently presented from the HA polypeptide
To examine the basis for the differing abilities of the TS1 and TS2 TCRs to undergo selection to become CD25+ Treg cells in response to the HA self-polypeptide, we first examined CD4+ T cells from TS1 and TS2 mice for their sensitivity to stimulation either by synthetic peptides representing the S1 and S2 determinants, or in response to whole virus particles. CFSE-labeled CD4+ T cells from TS1 and TS2 mice underwent similar degrees of division in response to equivalent doses of the S1 and S2 peptides, indicating that the TS1 and TS2 TCRs possess similar sensitivities for their respective determinants (Fig. 5,A). The CD4+ T cells from TS1 and TS2 mice also each underwent multiple rounds of division in response to high concentrations of PR8 virus (Fig. 5,B). However, at lower virus concentrations the TS2 CD4+ T cells divided less extensively than was the case for TS1 CD4+ T cells, indicating that the S2 peptide is less efficiently presented than S1 peptide when processed from an intact HA polypeptide (Fig. 5 B).
We next analyzed CFSE-labeled CD4+ T cells from TS1 and TS2 mice for their ability to undergo division in response to the HA self-polypeptide in vivo. CFSE-labeled LN cells from TS1 and TS2 mice were injected into BALB/c, HA28, or HA104 mice, and after 5 days the extent of proliferation in the different backgrounds was determined by flow cytometry. The 6.5+CD4+ cells from TS1 mice underwent multiple rounds of division in HA28 mice, and more extensive division in HA104 mice (Fig. 5,C). This division was in response to S1 peptide presented from the HA polypeptide, because it did not occur in BALB/c mice. By contrast, there was no division of Vβ8.3+CD4+ cells following transfer of LN cells from TS2 mice into either HA28 or BALB/c mice, and only a modest degree of proliferation of these cells in HA104 mice, indicating that S2 peptide is processed and presented much less efficiently than S1 peptide from the HA self-polypeptide in HA28 and HA104 mice (Fig. 5 C).
To examine factors that might contribute to the differing presentation of S1 and S2 peptides from an intact HA polypeptide, L cells that were transfected with I-Ad, I-Ed, and an invariant chain, and that either did or did not also express DM, were generated for use as APCs. These APCs were then used to stimulate CD4+ T cell hybridomas that had been generated from TS1 or TS2 mice following incubation either with inactivated PR8 virus or with synthetic peptides representing the S1 and S2 determinants. The presence of DM led to an ∼10-fold increase in the amount of virus required to induce half-maximal stimulation of the S2-specific hybridoma, indicating that DM expression antagonized presentation of the S2 peptide by MHC class II molecules (Fig. 5,D). In contrast, ∼100-fold less virus was required to stimulate the S1-specific hybridoma in the presence of DM than in its absence, indicating that DM enhances presentation of the S1 epitope. Notably, these differences were related to requirements for the processing and presentation of the S1 and S2 peptides from the HA polypeptide, because they were not observed when synthetic peptides rather than whole virus were used as Ags (Fig. 5 D).
We have examined how specificity for self-peptides contributes to CD4+CD25+ Treg cell formation on several levels. First, we have shown that interactions with the S1 self-peptide can cause the formation of significantly higher numbers of 6.5+CD4+CD8−CD25+ thymocytes and 6.5+CD4+CD25+ T cells than occurs in mice that do not express S1 peptide. Moreover, this process of peptide-induced 6.5+CD4+CD25+ Treg cell formation can occur with similar efficiency regardless of whether additional TCR α-chains can be expressed along with the 6.5 TCR. Second, when thymocytes and T cells are forced to express the 6.5 TCR without endogenous TCR α-chains and in the absence of S1 peptide (i.e., in TS1.Cα−/− mice), rare precursor cells coexpressing endogenous TCR β-chains can partially restore the peripheral CD4+CD25+ Treg cell pool, implying that a strong pressure exists to fill a CD4+CD25+ Treg cell niche. Third, we found that individual CD4+ T cell determinants from the same self-polypeptide can differ in their ability to induce CD25+ Treg cell formation, reflecting differences in the efficiency of their processing and presentation to CD4+ T cells.
The findings herein extend previous studies showing that interactions with the S1 peptide in TS1 × HA28 mice can induce the formation of 6.5+CD4+CD8+CD25+ thymocytes that can exert regulatory function in vitro (9). We have shown herein that these cells also express the transcription factor Foxp3, and, moreover, that the number of 6.5+CD4+CD8−CD25+ thymocytes is on average ∼10-fold higher in TS1 × HA28 mice than in TS1 mice. These increased numbers of 6.5+CD4+CD8−CD25+ thymocytes contrast a recent report using TAND mice, which coexpress the AND TCR and its agonist peptide (moth cytochrome c) as a self-Ag (13). In TAND mice, CD4+CD8−CD25+ thymocyte formation was found to be associated with increased percentages of CD4+CD25+ thymocytes, but with lower numbers of CD4+ thymocytes expressing the clonotypic TCR. Based on these findings, it was argued that CD4+CD25+ Treg cell formation in response to agonist self-peptides is typically a result of selective survival of CD4+CD25+ thymocytes, reflecting a decreased sensitivity to deletion relative to CD4+CD8−CD25− counterparts and giving a false impression of induced differentiation. However, as we show herein, there is a significant increase in the numbers of 6.5+CD4+CD8−CD25+ thymocytes (and of 6.5+CD4+CD25+ T cells) in TS1 × HA28 mice relative to TS1 mice, arguing against a model in which CD4+CD25+ Treg cell formation reflects a selective survival of cells that had committed to differentiate along this pathway in response to either stochastic processes or alternative nonpeptide cues (13); instead, our data show that S1 can induce more 6.5+CD4+CD8− thymocytes to develop along a pathway to become CD25+ than occur in its absence. In this regard, it is also notable that S1 does not mediate as efficient formation of CD4+CD25+ Treg cells in TS1.Cα−/− → HA104 mice as in TS1.Cα−/− → HA28 mice. Indeed, the findings in TS1.Cα−/− → HA104 mice are more similar to those reported in TAND mice, in which the percentages (but not the absolute numbers) of 6.5+CD4+CD8−CD25+ thymocytes were found to increase relative to mice lacking the HA peptide. It is also clear that the efficiency of CD4+CD25+ Treg cell formation can be sensitive to differences in self-peptide expression in different mouse lineages, because the extent of 6.5+CD4+CD25+ Treg cell formation differed substantially in TS1.Cα−/− → HA28 vs TS1.Cα−/− → HA104 mice. Therefore, it may be that induced formation of clonotype-positive CD4+CD25+ Treg cells was not observed in the TAND system because the self-Ag was expressed in amounts and/or cell types that induce efficient deletion and relatively little CD4+CD25+ Treg cell formation, as occurs in TS1.Cα−/− → HA104 mice. It is also possible that differences in the affinity of each autoreactive TCR for its target peptide and/or the stability of each self-peptide can impact the extent of CD4+CD25+ Treg cell formation in the HA and TAND systems.
The finding that CD4+CD25+ Treg cell formation occurs efficiently in TS1 × HA28.Cα−/− mice excludes a requirement for dual expression of TCR α-chains in Treg cell formation. This finding resembles previous studies showing that dual TCRs are not required for CD4+CD25+ Treg cell formation (12, 16), and it contrasts with a recent report that most human thymic and peripheral blood CD4+CD25+ Treg cells express two TCR α-chains (15). The data here also clarify early findings made using TCR transgenic mice in which it was concluded that endogenous TCR α-chains are required to generate CD4+CD25+ T cells because these cells were greatly diminished when TCR transgenes were expressed in RAG−/− or Cα−/− backgrounds (11, 14). In those studies, it is likely that the TCR transgenic mice lacked expression of a selecting peptide that could induce formation of CD4+CD25+ T cells expressing the clonotypic TCR. Indeed, as we have shown here, the 6.5 TCR undergoes much more efficient selection into the CD4+CD25+ Treg pathway in both TS1 × HA28 and TS1 × HA28.Cα−/− mice (where it encounters its agonist peptide) than in TS1 or TS1.Cα−/− mice (which lack S1 peptide). Moreover, the CD4+CD25+ T cells that are present in TS1 mice mostly expressed low levels of the 6.5 TCR, in contrast to those in TS1 × HA28 mice, which were enriched for cells expressing high levels of 6.5. The 6.5lowCD4+CD25+ T cells in TS1 mice likely coexpressed endogenous TCR chains, which in the absence of the S1 peptide allowed T cells expressing the 6.5 TCR to undergo CD4+CD25+ Treg cell selection on alternative self-peptides. Interestingly, the 6.5+CD4+CD25+ T cells that were present in TS1.Cα−/− mice were enriched for coexpression of endogenous β-chains, which again presumably conferred specificity for selecting self-peptide(s) that can mediate CD4+CD25+ Treg cell formation by 6.5+ T cells in the absence of S1 peptide. The presence of these cells is noteworthy because, in contrast to TCR α-chains, allelic exclusion of TCR β-chains is quite efficient (34, 35), implying that there exists a niche that can promote the formation and/or expansion of CD4+CD25+ Treg cells from rare precursors when other pathways are constrained.
The studies here also provide evidence that the differential presentation of individual MHC class II-restricted peptides from a self-polypeptide can affect the ability to induce auto-reactive CD4+ T cells to undergo CD25+ Treg cell selection. Few or no CD4+CD25+ T cells expressing the TS2 TCR were formed in either TS2.Cα−/− → HA28 or TS2.Cα−/− → HA104 mice, in striking contrast to the TS1 TCR, which underwent abundant CD4+CD25+ T cell selection in TS1.Cα−/− → HA28 mice. It is possible that the presentation of S1 and S2 peptides by I-Ed vs I-Ad molecules, respectively, in some way affects their ability to mediate CD4+CD25+ T cell selection, although the CD4+CD25+ Treg cell development that occurs in genetic backgrounds that lack I-E expression indicates that I-A molecules can support CD4+CD25+ Treg cell formation (36, 37, 38). Significantly, S2 peptide was found to be presented from the self-HA polypeptide less efficiently than S1 peptide, because S2-specific CD4+ T cells underwent less division than S1-specific CD4+ T cells following transfer into either HA28 or HA104 mice. Moreover, presentation of S2 peptide from an intact HA polypeptide was inhibited by the presence of DM molecules in L cell transfectants, whereas DM substantially enhanced presentation of S1 peptide. Because DM catalyzes removal of low-affinity peptides from MHC class II molecules during peptide loading in APCs (39, 40), these data suggest that S1 peptide is more stably associated with MHC molecules when generated from the HA polypeptide than is S2 peptide. Indeed, in vitro dissociation assays with purified class II and synthetic peptide support this view, as the S1 peptide has a dissociation half-time of ∼120 h from class II at pH 5.3 whereas the S2 peptide has a dissociation half-time of ∼25 h ((39) and F. Chaves and A. Sant, unpublished observation). These data are also consistent with previous studies showing that S2 possesses a much shorter half-life in virus-pulsed APCs than does S1 (41) and that S2 can behave as a cryptic self-Ag in virus-immunized HA transgenic mice (32). Collectively, these results suggest that DM editing of the S2 peptide lowers its epitope density on APCs in vivo.
The inefficient presentation of S2 peptide in the presence of DM molecules is noteworthy in light of previous studies indicating that S2 is one of the major CD4+ T cell determinants recognized by HA-specific CD4+ T cells in virus-immunized BALB/c mice (20). Presumably the high amount of S2 peptide generated in the context of an infection compensates for its relative instability and allows S2 to induce robust CD4+ T cell responses. However, in the steady-state conditions of immune repertoire formation, this instability appears to limit the amount of S2 that is presented from the self-HA polypeptide. Thus, while DM-induced peptide editing plays an important role in the development of effective antimicrobial immune responses by promoting the formation of stable MHC-peptide complexes, it may in some cases reduce the expression of a self-peptide to a lower level than is required for CD4+CD25+ Treg cell formation.
We thank Laura Panarey and Malinda Aitken for excellent assistance, and Dr. Heath M. Guay for thoughtful discussions.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Grants AI59166, AI24541, and CA09140 from the National Institutes of Health, by the Lupus Foundation of Southeastern Pennsylvania, and by the Commonwealth Universal Research Enhancement Program, Pennsylvania Department of Health.
Abbreviations used in this paper: Treg, regulatory T; BM, bone marrow; HA, hemagglutinin; LN, lymph node; PR8, influenza virus A/PR/8/34; SF medium, serum-free supplemented IMDM.