Diversity of TCRs on Natural Foxp3⁺ T Cells in Mice Lacking Aire Expression Free

The majority of regulatory CD4⁺ lymphocytes (regulatory T cells [Tregs]) originate from the thymus, where they undergo positive and negative selection and turn on lineage-specific transcription factor Foxp3 (reviewed in Refs. 1, 2). Currently, the role/specificity of self-Ags presented in the cortex or medulla that positively select Tregs and may influence the lineage commitment of their precursors remains controversial. It has been shown that Tregs develop normally in mice expressing MHC class II localized to cortical epithelium, and in transgenic mice bearing a TCR specific for influenza hemagglutinin expression restricted to cortical thymic epithelial cells, positive selection of Foxp3⁺ thymocytes is enhanced (3–5). Though these data suggest that Treg lineage commitment occurs in the cortex, immature Foxp3⁺ thymocytes are rare, and it was reported that thymocytes upregulate Foxp3 late while trafficking through the medulla (6, 7). Further, several reports hypothesize that Treg lineage commitment is regulated by TCR recognition of tissue-specific Ags (TSAs) on medullary epithelium or bone marrow-derived APCs (8–10). In support of these hypotheses, it has been shown that medullary thymic epithelial cells (mTECs) and Tregs preferably colocalize to the same medullary microenvironments, and, in TCR-transgenic mice in which thymocytes recognized a specific agonist self-Ag expressed on mTECs, Foxp3⁺ thymocytes were efficiently selected (9, 11). In this model, the pattern of intrathymic expression of agonist self-Ag was controlled by the Aire gene promoter, mimicking the natural expression of TSAs. In addition, patients suffering from autoimmune polyendocrinopathy candidiasis ectodermal dystrophy, who lack Aire-controlled TSA expression on mTECs due to mutations in the Aire gene, have reportedly fewer, less effective Tregs, and the diversity of their TCR repertoire is compromised, contributing to this autoimmune manifestation (12). Collectively, these findings suggest that although recruitment of thymocytes to the Treg lineage can start in the cortex, many thymocytes become committed to this lineage in the medulla upon contact with TSAs, and, specifically, that the absence of Aire-controlled TSA expression has an impact on Treg development, function, and TCR repertoire diversity (reviewed in Ref. 13).

Aire expression is essential during the perinatal period for the prevention of multiorgan autoimmunity, but not required thereafter (14). Interestingly, Aire-deficient and day 3 thymectomized mice manifest similar disease phenotypes, differing from mice totally deficient in Tregs (i.e., Scurfy mice) (15). Coincidentally, in neonatal wild-type mice thymii, TdT mRNA is not detected until day 4 postbirth, and it functions in the addition of nucleotides to the 3′ end of each coding gene at the V(D)J junction, contributing to 90% of TCRαβ diversity. In TdT-deficient mice, the TCRαβ repertoire is estimated to be only 10⁵ unique TCRs, which is 5–10% of the wild-type repertoire (16). In lieu of this, the autoimmune manifestation seen in Aire-deficient and day 3 thymectomized mice may be the result of a lack of TCR diversity, mainly affecting the Treg subpopulation, considering that it is naturally more diverse than the naive subpopulation (17, 18). On the contrary, it was reported that Aire-deficient and wild-type mice have a similar number and proportion of Tregs, and they appear to be functional, suggesting that in the absence of Aire-controlled TSA expression, the Treg subpopulation is unaffected (19). However, this study did not examine the effect of Aire deficiency on the diversity of individual Treg TCRs in the thymus or nonlymphoid organs.

It has also been postulated that Tregs recognize self-Ags with high affinity as a result of positive selection by agonist self-Ags (4, 9, 20). Perhaps this inherent self-reactivity allows Tregs to become activated in the absence of foreign Ags, helping them to suppress autoreactive T cells. However, ubiquitously expressed self-Ags are rarely recognized as agonists by TCRs expressed on Tregs, but if TSAs expressed on mTECs select thymic Treg precursors, their self-reactivity could remain concealed in assays utilizing conventional APCs (21, 22). In this report, we examine the impact of Aire-controlled TSA expression on Treg diversity and self-reactivity by analyzing individual Treg TCRs expressed on thymic and nonlymphoid organs, utilizing a novel transgenic mouse model (Aire⁻ mice expressing a miniature TCR repertoire [TCR^mini], along with GFP driven by the Foxp3 promoter [Foxp3^GFP]).

TCR^mini mice express a polyclonal but restricted set of TCRs. All T cells express the same transgenic Vβ14-Jβ2.6 chain but different TCRα-chains that form natural rearrangements of the transgenic Vα2-Jα26(Jα2) minilocus. TCR^mini mice lack endogenous TCRα-chains; thus, the diversity of the TCR repertoire is limited to the CDR3α region of the transgenic TCRα-chain, allowing one to consistently estimate the frequencies of specific TCRs (17). All mice were heterozygous for the TCRα minilocus to ensure expression of a single TCR. The targeted disruption in the Aire gene was designed to mimic the most common human autoimmune polyendocrinopathy candidiasis ectodermal dystrophy mutation currently identified (23), and the Foxp3^GFP mice harbor a bacterial artificial chromosome transgene, which encodes the full locus control region of Foxp3 driving GFP expression (24). All mice were housed under specific pathogen-free conditions in the animal care facility at the Medical College of Georgia (Augusta, GA). All experiments were conducted at the Medical College of Georgia under protocols approved by the Animal Care and Use Committee.

Hybridomas (1 × 10⁵) and thymocytes (1 × 10⁴) were stimulated with either plate-bound anti-CD3 mAbs (2 μg/ml) or purified mTECs (1 × 10⁵) from neonatal thymii. The mTECs were purified by Ficoll after 5 d of culture in the presence of dGuo as described (25). After 24 h in coculture, the supernatant from the hybridomas was evaluated for the presence of IL-2 as described (21). Alternatively, the activation of CD4⁺ thymocytes cocultured with mTECs was evaluated by monitoring CD69 expression by flow cytometry.

Single-cell suspensions were prepared from all lymphoid organs by mechanical disruption through nylon mesh; nonlymphoid organs required additional digestion with collagenase IV and DNAse I. The following Abs were used for flow cytometry analysis and sorting: anti-CD4–APC, anti-CD8–PE-Cy7, anti-CD25–FITC, anti-CD25–PE-Cy5, anti-CD62L–PE, anti-Vα2–PE, and anti-Vβ14–FITC. All mAbs used in this study were purchased from BD Biosciences (San Jose, CA) or eBioscience (San Diego, CA). Intracellular staining for Foxp3 expression was performed using an intracellular staining kit according to the manufacturer’s protocol (eBioscience).

RNA was isolated from sorted populations using Trizol according to the manufacturer’s protocol (Invitrogen, Carlsbad, CA). cDNA synthesis was done with M-MLV reverse transcriptase (Promega, Madison, WI) and Random Hexamers. TCR α-chains were amplified in a standard PCR reaction with Vα2-specific primers. A total of 2 μl PCR product was used as a template for a runoff reaction with a fluorescent primer labeled with Cy3 (synthesized by Integrated DNA Technologies, Coralville, IA). The denatured fluorescent products were then subjected to two-dimensional electrophoresis. The first fluorescent products were separated on a denaturing 8% polyacrylamide gel in capillary electrophoresis. Gel noodles were pooled out of capillaries, trimmed, and loaded on a second dimension of nondenaturing MDE (Cambrex, East Rutherford, NJ) slab gel. Fluorescent images were acquired by scanning the slab gel in a Typhoon 9410 imager and with Image Master 5.0 Platinum software (Amersham Biosciences, Piscataway, NJ).

cDNA was prepared as mentioned above then amplified by nested PCR with Vα2-specific primers. PCR product was ligated into pCR 2.1-TOPO cloning vector and transformed into Mach1 competent bacteria (Invitrogen). Individual colonies were picked, incubated, and amplified with M13 forward and reverse primers. Alternatively, T cells were single cell-sorted onto 96-well plates containing 5 ml RT buffer and 2 U RNAsin (Promega). cDNA synthesis was done with M-MLV reverse transcriptase and Random Hexamers (all from Promega). The entire 10 ml cDNA reaction was used for two rounds of nested PCR via PerfectTaq polymerase (5 Prime, Gaithersburg, MD) according to the protocol. Products from the second PCR for Vα2 were directly sequenced with Va2_287s primer and analyzed as described (21).

To address the impact of Aire-controlled TSA expression on TCR diversity, we created a mouse model (Aire⁻ TCR^miniFoxp3^GFP) in which we crossed previously described Aire-deficient mice (23) lacking Aire-dependent TSA expression with our existing TCR^mini mice, harboring a restricted TCR repertoire, allowing the TCRα-CDR3 regions of Vα2.9-Jα26(Jα2) to be used as a specific marker of individual T cell clones. Then we crossed them with Foxp3^GFP transgenic mice, in which GFP is driven by the Foxp3 promoter, labeling all Tregs (Supplemental Fig. 1) (24). Previous reports characterizing TCR^mini mice showed bias in the selection of CD4⁺ thymocytes due to the inherent specificity of the original TCR to A^b; therefore, the proportion of CD4⁺ thymocytes is higher than in wild-type thymii (17). FACS analysis and MoFlo sorting of Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP thymii revealed similar proportions of CD4⁺ and CD8⁺ single-positive, CD4⁺CD8⁺ double-positive, and CD4⁻CD8⁻ double-negative as well as gated CD4⁺CD25⁺ and CD4⁺Foxp3⁺ thymocytes (Fig. 1A), and the total number of cells recovered from Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP thymii following sorting of CD4⁺Foxp3⁺ thymocytes was similar (data not shown), which agrees with a previously published report demonstrating that Aire deficiency in mice does not affect the total number of Tregs (19).

To determine the TCR diversity of CD4⁺Foxp3⁻ and CD4⁺ Foxp3⁺ thymocytes in Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP mice thymii, we compared their TCR repertoires using two-dimensional fluorescent single-stranded conformational polymorphism (2D-F-SSCP) analysis. In this method, the sequences of TCRα-CDR3 regions (referred to as CDR3α regions) derived from different subpopulations of T cells are amplified with fluorescently labeled primers and resolved in two-dimensional gel according to the CDR3α’s length (first dimension) and then conformation (second dimension). This method is capable of separating >200 spots representing CDR3α regions from the most frequently used TCRs and provides a snapshot of the core diversity of TCRs analyzed (17). The data acquired can then be compared between different samples to determine their similarity. As shown in Fig. 1B, CD4⁺Foxp3⁻ thymocytes found in Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP mice had TCR repertoires overlapping at 91%. Likewise, the TCRs expressed by CD4⁺ Foxp3⁺ thymocytes derived from Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP mice were similarly diverse, but these repertoires overlapped at 71.5% (Fig. 1C). Because the TCR repertoire of Tregs is inherently more diverse than the repertoire of CD4⁺Foxp3⁻ T cells, this finding is consistent with previous reports showing similar overlaps of ~70% and 90%, respectively, for thymic and peripheral lymphoid TCR repertoires in Aire⁺ TCR^mini mice (17). To support this finding, we performed single-cell RT-PCR analysis in which we sorted and then compared the most frequent individual CD4⁺Foxp3⁻ and Treg CDR3α regions from Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP mice thymocytes. As a cutoff value for the determination of frequent TCR clones, we chose clones found five or more times. In Fig. 1D, we show 29 different Treg CDR3α regions found with a frequency above four in Aire⁺ TCR^miniFoxp3^GFP mice and then compared their frequencies within the Aire⁻ TCR^miniFoxp3^GFP Treg CDR3α regions. This comparison found 18 Treg CDR3α regions to be shared, equaling a 62% overlap. Also, out of the 25 different CDR3α regions from CD4⁺Foxp3⁻ T cells found with a frequency above four in Aire⁺ TCR^miniFoxp3^GFP mice, 24 of them were shared with Aire⁻TCR^miniFoxp3^GFP, CD4⁺Foxp3⁻, CDR3α regions, equaling a 96% overlap (Fig. 1E). Lastly, we performed statistical analysis of all 2225 CD4⁺Foxp3⁻ and CD4⁺ Foxp3⁺ thymocyte TCR sequences obtained from Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP mice using the Morisita-Horn index to compare the overlap between these repertoires. As shown in Fig. 1F, the repertoires of TCRs found on naive and regulatory thymocytes from Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP mice had the highest degrees of overlap, 93% and 49%, respectively. Thus, both the 2D-F-SSCP analysis of CD4⁺Foxp3⁺ thymocytes and sequences matched from single-sorted CD4⁺Foxp3⁺ thymocytes that developed in Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP mice indicated that the selection of dominant TCRs on Tregs is Aire independent.

It has been reported that mice with a polyclonal but restricted repertoire have many TCRs dominantly expressed on Tregs that are rare or not found on naive T cells and that this is established in the thymus (17, 26). However, the mechanism(s) responsible for the dissimilarity of the TCR repertoires of CD4⁺Foxp3⁻ T cells and Tregs is currently unclear. It was hypothesized that it may be a consequence of the asymmetric distribution of TCRs, selection of Tregs on a different pool of self-Ags, or positive selection of Treg precursors by high-affinity TCRs (4, 9, 20, 26). To determine if this trend is conserved in the absence of Aire-controlled TSA expression in the medulla, we compared CD4⁺Foxp3⁻ and CD4⁺Foxp3⁺ CDR3α regions in Aire⁻ TCR^miniFoxp3^GFP mice thymii. A similar asymmetric pattern was observed for Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP mice (Fig. 2A, 2B). Secondly, when we looked at the CDR3α region lengths for CD4⁺Foxp3⁻ T cells and Tregs, the same CDR3α regions were abundant in the Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP mice (Fig. 2C). Together, these data revealed that the lack of Aire-controlled TSA expression does not affect positive selection of a highly diverse Treg TCR repertoire in the thymus, and its diversity continues to exceed the diversity of TCRs found on CD4⁺Foxp3⁻ T cells. Thus, if Aire-controlled TSA expression does play a role in Treg selection, these cells may account for only a nominal fraction of this subpopulation and therefore be hidden among the analysis of central lymphoid organs. Perhaps their existence may be visible in nonlymphoid organs, especially if their tropism depends upon their specificity to the TSAs they were selected upon.

Although Aire⁺ TCR^miniFoxp3^GFP mice remain healthy during their lifespan, Aire⁻ TCR^miniFoxp3^GFP mice develop organ-specific autoimmunity similar to that of wild-type Aire-deficient mice, marked by circulating autoantibodies and T cell infiltrates (D.L. Daniely, A. Cebula, A.K. Ignatowicz, J.X. She, L. Ignatowicz, manuscript in preparation). It has been shown that wild-type Aire-deficient mice develop mild/moderate autoimmunity, mainly affecting the eye, stomach, salivary gland, lacrimal gland, adrenal gland, and gonads, with little/no autoimmune manifestation in the lung, kidney, or heart (27). In addition, of the 30 genes mostly downregulated by the lack of Aire-controlled TSA expression on mTECs, salivary protein 1 and salivary protein 2 were found to be specific to one tissue, whereas only one gene was downregulated for the lung, cytochrome P450, which had shared specificities for liver and intestine, indicating that Aire plays a major role in tolerance induction toward the salivary gland (27). Therefore, we chose to examine TCRs in the salivary gland, an organ with mild/moderate infiltration, and compare them to TCRs found in the lung, an organ with little/no infiltration in Aire⁻ TCR^miniFoxp3^GFP mice (Fig. 3) to determine the impact of Aire-controlled TSA expression on the Treg repertoire in nonlymphoid organs. We sequenced Treg TCRs from the lung and salivary gland of Aire⁻ TCR^miniFoxp3^GFP mice, as well as the nondraining lymph nodes, and cross-compared them to the TCRs expressed on thymocytes of Aire⁺ TCR^miniFoxp3^GFP mice. One would expect that if Aire-controlled TSAs are involved in the selection of tissue-specific Tregs in Aire⁺ mice, then those Tregs would be absent in Aire⁻ mice. FACS analysis showed that Tregs were more frequent among CD4⁺ T cells in the salivary gland than in the lung, suggesting that in the salivary gland, tissue-specific Tregs (not selected by Aire-controlled TSAs) can infiltrate and expand (Fig. 4A). Our cross-comparison of CDR3α regions revealed that ~40–50% of total Treg TCRs were shared between Aire⁻ TCR^miniFoxp3^GFPmice organs and Aire⁺ TCR^miniFoxp3^GFP mice thymus, including rare sequences that were found only once in the Aire⁺ TCR^miniFoxp3^GFP thymus. Of the unique/different Treg TCRs analyzed, 35% were found to be shared for the lung and 30% for the salivary gland, indicating that the diversity of Treg TCRs shared for the lung and salivary gland were similar, though the salivary gland is more infiltrated due to the lack of TSAs than the lung (Fig. 4B). We found some Treg TCRs to be common between both lung and salivary gland, whereas others were found only in one organ, presumably indicating that the first group of TCRs may recognize ubiquitously expressed Ags, and the second may include Treg clones expressing tissue-specific TCRs. Tregs represent ~5–10% of the peripheral CD4⁺ population, and though their numbers are higher in mouse models of disease than in healthy mice, the total numbers found in nonlymphoid organs remain low. Therefore, we show a sampling from the two nonlymphoid organs we analyzed and validate our sampling size by showing that the percentage of unique Treg TCRs out of the total Treg TCRs analyzed for each organ was comparable (40–60%) irrespective of the number of sequences retrieved (Fig. 4B, table). In conclusion, we found that in Aire⁻ TCR^miniFoxp3^GFP mice, diseased organs are infiltrated by Tregs expressing a diverse set of TCRs, and many of these TCRs can also be found on Tregs selected in Aire⁺ TCR^miniFoxp3^GFP mice thymus, meaning that the selection of these Tregs is likely Aire controlled, TSA independent.

It has also been proposed that the Treg repertoire is enriched in self-reactive TCRs as a consequence of either being selected by agonist self-Ags in the context of self-MHC class II complexes, and therefore, they are capable of recognizing self-Ags with high affinity or by being diverted from naive self-reactive thymocytes upon self-Ag recognition (4, 9, 21). To determine if Aire-controlled TSAs are frequently recognized with high affinity as agonists by TCRs expressed on Tregs, we examined the specificities of Treg-derived hybridomas cocultured with autologous mTECs. Considering that established mTEC cell lines poorly express Aire, we used dGuo-treated neonatal thymic cultures to isolate fresh EpCAM⁺A^b+ and Y3P⁺ (MHC class II) mTECs (Fig. 5A). In the past, T cell hybridomas have been successfully used to detect low abundant endogenous and exogenous peptides and to study Treg TCR specificity (21, 28, 29). Therefore, we established hybridomas from Tregs isolated from wild-type (C57BL/6), TCR^mini, or A^bEp single-peptide mice (30) and from CD4⁺ T cells from Scurfy mice. As shown in Fig. 5B, all hybridomas responded to anti-CD3 stimulation. However, only hybridomas derived from A^bEp (these TCRs were not tolerant to naturally processed self-Ags) and Scurfy mice (established from naturally arising self-reactive T cells) responded to self-Ags presented by mTECs, suggesting that TCRs expressed on Tregs rarely recognize self-derived peptides presented by mTECs as agonists. However, Treg-derived hybridomas from TCR^mini mice frequently responded when challenged with non–self-Ags and recognized Ags derived from commensal bacteria following adoptive transfer into lymphopenic hosts (21). Therefore, though Tregs do not recognize Aire-controlled self-Ags with high affinity, they can strongly respond to a variety of foreign Ags.

It has been proposed that during thymic ontogeny, CD25 expression can precede Foxp3⁺ expression on CD4⁺ thymocytes and binding of the TCR to an agonist peptide in the medulla may turn on Foxp3 expression, leading to the final lineage commitment of CD25⁺Foxp3⁻ thymocytes to Treg lineage (31). Hence, in the second experiment, we compared CD69 upregulation as a marker of TCR signaling/activation between CD4⁺CD25⁻ and CD4⁺CD25⁺ thymocytes from Aire⁺ and Aire⁻ mice that were cocultured with fresh mTECs isolated from Aire⁺ mice. As shown in Fig. 5C, when CD4⁺ thymocytes were in the presence of agonist-mimicking activators (immobilized anti-CD3 mAb or Aire⁺ mTECs for A^bEp-derived thymocytes), a majority of them upregulated CD69 expression. In contrast, when Aire⁺ and Aire⁻ thymocytes were cocultured with Aire⁺ mTECs, little to no upregulation of CD69 was observed, respectively. These findings suggest that similar to ubiquitously expressed self-Ags, TSAs present on mTECs are rarely recognized as cognate Ags by TCRs naturally selected on Tregs.

To address the role of Aire-controlled TSA recognition on the diversion of Tregs from self-reactive thymocytes, we hypothesized that Aire-controlled TSA-dependent Tregs found in Aire⁺ TCR^miniFoxp3^GFP mice would express identical TCRs also found on self-reactive T cells that escaped negative selection in Aire⁻ TCR^miniFoxp3^GFP mice. By FACS analysis, the proportion of potentially self-reactive T cells or those activated CD4⁺ T cells that have downregulated CD62L and migrated to the nonlymphoid organs of Aire⁻ TCR^miniFoxp3^GFP mice was higher in the salivary gland than in lung (Fig. 6A). However, our sampling of dominant CDR3α regions found on activated T cells in the lung and salivary gland of Aire⁻ TCR^miniFoxp3^GFP mice, extrapolated back to the sequences found on Tregs in Aire⁺ TCR^miniFoxp3^GFP thymii, showed only seven unique sequences for the lung and one unique sequence for the salivary gland (Fig. 6B, marked with asterisks in Fig. 6C). However, a majority of those CDR3α regions found to be shared with Aire⁺ TCR^miniFoxp3^GFP Tregs were also found in the naive repertoire of Aire⁺ TCR^miniFoxp3^GFP mice, indicating that their selection was not dependent upon Aire-controlled TSAs.

In this report, we examined the impact of Aire-controlled TSA expression on the diversity and self-reactivity of TCRs expressed on natural CD4⁺Foxp3⁺ T cells in Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP mice. The restricted TCR repertoire allowed us to intricately compare TCRs at the level of a single T cell clone. One may argue that the restriction of the TCR repertoire in our mouse model could unintentionally compromise its diversity, regardless of Aire-controlled TSA expression. Nevertheless, when TCR^mini mice were crossed with autoimmune-prone strains, such as NOD (R. Pacholczyk, personal communication), Aire-deficient, or Scurfy (P. Kraj, personal communication), it resulted in multiorgan autoimmunity or death, respectively. In contrast, when TCR^mini mice were crossed with nonautoimmune-prone wild-type C57BL/6 or 129 mouse strains, these mice remained healthy throughout their life span. Therefore, not only does the restricted TCR repertoire house sufficient autoreactive TCRs capable of driving the development of autoimmune disease, it also houses a sufficient amount of Tregs capable of maintaining the health of nonautoimmune-prone mice. In addition, our results show that the CD4⁺Foxp3⁺ TCR repertoires in Aire⁺ and Aire⁻ TCR^miniFoxp3^GFP mice were similar and more diverse than their repertoires on CD4⁺Foxp3⁻ thymocytes, which were almost identical. Many CD4⁺Foxp3⁺ thymocyte TCRs were not found on CD4⁺ Foxp3⁻ thymocyte TCRs, whereas the majority of CD4⁺ Foxp3⁻ thymocyte TCRs could be found in the regulatory population. Thus, as previously reported, the asymmetric distribution of TCRs on CD4⁺Foxp3⁻ and CD4⁺Foxp3⁺ thymocytes is preserved in Aire-deficient mice, and the diversity of their thymic TCR repertoires is not compromised in the absence of Aire-controlled TSAs.

It has been reported that the selection of transgenic immature thymocytes by high affinity, self-reactive interactions with MHC/peptide ligands in the medulla can induce Foxp3 expression and direct CD4⁺CD25⁺Foxp3⁻ Treg precursors toward the regulatory lineage. Yet, when we tested TCRs derived from Tregs toward self-Ags presented by thymic medullary epithelial cells, we found no evidence of abundant, self-reactive TCRs specific for Aire-controlled TSAs. Additionally, our experiments also show that CD4⁺CD25⁺ thymocytes that contain both natural CD4⁺Foxp3⁺ Tregs and their immediate precursors do not express TCRs capable of recognizing auto-Ags with high affinity on thymic medullary epithelium. In TCR^mini thymii, the relative proportion of CD25⁺Foxp3⁻ thymocytes is higher than in wild-type thymii. We hypothesize that this subpopulation may be enriched in thymocytes that are currently undergoing recruitment, whereas the CD25⁺Foxp3⁺ population of thymocytes may represent those that are already committed (Fig. 1A) (24, 31).

It has been shown that self-Ags are involved in the development of tissue-specific Tregs (32). Therefore, our data do not exclude the existence of infrequent Aire-controlled TSA-dependent Treg clones, which may contribute to organ-specific tolerance. However, transplantation of Aire⁺ fetal thymii into Aire⁻ hosts by another group did not rescue the host from developing organ-specific autoimmunity, which questions the significance of those Tregs (19). In addition, we show that nonlymphoid organs subject to autoimmune manifestation in Aire-deficient mice harbor many Tregs identical to those found in wild-type mice, suggesting that selection and TSA-associated tropism of those Tregs is Aire independent. Also, our study does not account for the occurrence of inducible Tregs that may arise in an autoimmune-prone environment (33). However, the TCRs expressed on the most abundant naive and Treg clones in the salivary gland of Aire⁻ TCR^miniFoxp3^GFP mice were nonoverlapping, suggesting that there is no major recruitment of dominant CD4⁺Foxp3⁻ cells to the regulatory lineage.

Our data, along with a previously published report, suggest that perhaps the thymic precursors of Tregs use Aire-independent self-Ags for their selection and lineage commitment, which, in the absence of Aire-controlled TSAs, similar to self-reactive thymocytes, populate nonlymphoid organs (34). As a consequence, those Tregs and self-reactive thymocytes would express dissimilar TCR repertoires if the former repertoire were chosen independently of Aire and selected within a specific niche, whereas the latter repertoire escaped negative selection as a result of the lack of Aire (7). To determine the similarity between the TCRs of Tregs and those of activated T cells in Aire⁻ TCR^miniFoxp3^GFP mice, we examined the dominant TCRs on Tregs and activated T cells in the lung and salivary gland and observed little overlap in the lung and none in the salivary gland (Fig. 6C). These results implied that in Aire⁻ TCR^miniFoxp3^GFP thymii, selection generates Tregs that can specifically accumulate in diseased organs and express TCRs unlike TCRs expressed on activated T cells. However, these Tregs are not sufficient to fully control the development of autoimmunity. Therefore, it appears that the majority of Tregs are not selected by Aire-controlled TSAs. Thus, although the most anticipated effect of Aire deficiency one would expect is not a quantitative but rather a qualitative compromise of Treg diversity and antigenic specificity, based on our analysis, that appears to be unseen.

We thank Drs. Jin-Xiong She and Piotr Kraj for providing the Aire-deficient and Foxp3^GFP mouse strains, respectively, Dr. Rafal Pacholczyk for personal communication (NOD-TCR^mini), Dr. Grzegorz Rempala and Michal Seweryn for biostatistical analysis, M. Kuczma for editing the manuscript, and P. Merai and H. Ignatowicz for performing technical assistance.

Disclosures The authors have no financial conflicts of interest.