Functional Adaptive CD4 Foxp3 T Cells Develop in MHC Class II-Deficient Mice¹

Increasing interest is focused at the basic, translational, and clinical level of research on the biology of regulatory T (T_R)³ cells and their potential for the treatment of immunopathology (1, 2, 3). The forkhead family transcription factor Foxp3 has emerged as the master regulatory factor in the development and function of T_R cells (4, 5, 6). CD4 Foxp3⁺ T_R cells are the best characterized T_R cell population in the large and heterogeneous group of CD4 or CD8 T cells with suppressive functions. The evidence for Foxp3 as a necessary and sufficient factor for T_R cell generation and function comes from different sources. Spontaneous Foxp3 mutations in mice (scurfy) and humans (immune dysfunction, polyendocrinopathy, enteropathy, X-linked; IPEX), or the knockout of Foxp3 in mice show that the deficiency of this transcription factor leads to T cell hyperreactivity resulting in autoimmune disease. In contrast, transgene-driven overexpression of Foxp3 rescues mice from scurfy disease but strikingly impairs the establishment of T cell immunity. T cells expressing a transduced gene encoding Foxp3 develop a T_R phenotype (7, 8). The knockin of a reporter into the Foxp3-encoding locus shows that expression of this gene coincides with expression of the T_R phenotype (9). There is thus convincing evidence that Foxp3⁺ T cells are a major T_R cell population, although the heterogeneity in development, phenotype, and suppressive effects of Foxp3⁺ T_R cell populations is an unresolved issue.

T_R cells are categorized as natural or adaptive (3, 10). The relative contribution of natural (thymus-derived) and adaptive (peripherally generated) T_R cells to the pool of peripheral Foxp3 T_R cells is uncertain. Natural CD4 CD25^high T_R cells express Foxp3 and are generated in the thymus independent of IL-2/CD25 or TGF-β. Their intrathymic development is independent of commitment to either the CD4 or CD8 lineage but depends on positive selection by medium-to-high affinity TCR-MHC interactions that generates a repertoire skewed toward recognition of self-Ags (11). Positive selection of CD4 Foxp3 T_R cells in the thymus may dependent on MHC class II expression by cortical epithelium (12) and/or thymic stromal lymphopoietin-conditioned plasmacytoid dendritic cells (DC) in the medulla of the thymus (13). Natural T_R cells preferentially home to T cell zones of secondary lymphoid tissues where they control effector T cell development from naive precursors. Studies in conditional MHC-II-deficient mice indicate that peripheral homeostasis of natural Foxp3 T_R cells depends at least partially on MHC-II-dependent TCR stimulation and IL-2 availability (14).

Adaptive T_R cells may or may not express Foxp3. Recent data demonstrated the conversion of peripheral, naive CD4 CD25⁻Foxp3⁻ T cells into adaptive, regulatory CD4 CD25⁺Foxp3⁺ T_R cells through TGF-β-mediated induction of Foxp3 expression in the absence of IL-6 (9, 15, 16, 17, 18, 19, 20, 21). Different DC populations seem to generate and/or regulate T_R cell homeostasis in the periphery. Peripheral immature or mature myeloid DC that constitutively engulf and process self Ags under steady state conditions may maintain the natural T_R cell pool (22). In allotolerance, an exclusive interaction of Foxp3 T_R cells with plasmacytoid DC has been demonstrated (10, 23). Most natural CD4 Foxp3 T_R cells are constitutively CD25^high, but some CD4 (CD25⁺ or CD25⁻) Foxp3 T_R cells express CD103 (the α_E chain in the cadherin-binding integrin α_Eβ₇) and are a potent, possibly adaptive T_R cell subpopulation, particularly in epithelial environments (24, 25). These T_R cells are found in mucosal tissues (26) and efficiently migrate to inflamed sites (27). CD103 seems to mediate the homing of T_R cells and their subsequent retention in the dermis in Leishmania infection (25, 28).

We have reported the generation of effector function and regulatory activity in the MHC class II-independent CD4 αβ T cell population of completely MHC class II-deficient mice (29, 30). Adoptive transfer of CD4 αβ T cells from wild-type (wt) or MHC class II-deficient (Aα^−/− or Aβ^−/−) C57BL/6J (B6) donor mice into congenic, immunodeficient RAG^−/− hosts induced an aggressive transfer colitis (29). CD4 T cells developing in mice deficient in MHC class II comprised a major (80%) single-positive (SP) CD4⁺ CD8⁻ subset and a minor (20%) double-positive (DP) CD4⁺ CD8⁺ subset. Although a single-cell assay for Foxp3 was not available at the time, PCR data suggested that T_R cells were present mainly in the DP T cell population from spleen, mesenteric lymph nodes, and colonic lamina propria of MHC class II-deficient mice. These T_R cells could partially control the proinflammatory potential of SP CD4 T cells in the transfer colitis model (30). Here, we test whether Foxp3⁺ T_R cells develop in the thymus or in the periphery of MHC class II-deficient mice, whether class II-independent Foxp3⁺ T_R cells express CD4 or CD8, or CD103 and/or CD25, and whether they are functional in vitro and in vivo.

Wild-type C57BL/6J mice (H-2^b) (B6) mice and H-2 class II-deficient (Aα^−/−) B6 mice (31) were bred and kept under standard pathogen-free conditions in the animal colony of Ulm University (Ulm, Germany). Female and male mice were used at 8–12 wk of age. All animal experiments were performed according to the guidelines of the local Animal Use and Care Committee and the Provincial Animal Welfare Law.

Single cell suspensions were aseptically prepared from spleen, inguinal, and mesenteric lymph nodes or liver as described (32). T cells were purified using the CD8 T cell (Miltenyi Biotec) or CD4 T cell MACS isolation kit (Miltenyi Biotec). The purity of the isolated CD8⁺ or CD4⁺ T cells was >98% as verified by flow cytometry (FCM).

Cells washed twice (in PBS, 0.3% w/v BSA supplemented with 0.1% w/v sodium azide) were preincubated with mAb 2.4G2 (BD Biosciences) to block nonspecific binding of Abs to FcR. Cells were washed, incubated for 30 min at 4°C with 0.5 μg/10⁶ cells of the relevant mAb, and washed. In most experiments, cells were subsequently incubated for 20 min at 4°C with a second-step reagent. Four-color FCM analyses were performed by FACSCalibur (BD Biosciences). The forward narrow angle light scatter was used as an additional parameter to facilitate exclusion of dead cells and aggregated cell clumps. Data were analyzed using the FCS Express (DeNovo) software. The following reagents from BD Biosciences or eBioscience were used: allophycocyanin-conjugated anti-CD8 mAb 53-6.7 and anti-CD4; biotinylated anti-CD8, anti-CD25, anti-CD44, anti-CD28, anti-CD80, anti-CD86, anti-ICOSL, anti-CD94, anti-B7-H3, and anti-B7-H4; PE-conjugated anti-CD25 mAb PC61, anti-CD103, anti-CD69, anti-PD-1, anti-CTLA-4, anti-BTLA, anti-CD3, anti-PD-L2, anti-PD-L1, anti-ICOS, anti-NK1.1 and anti-CD54; CD1d-IgG1 fusion protein DimerX; and FITC-conjugated anti-Foxp3.

For intracellular staining of BTLA, CTLA, and PD-1, cells were surface stained with allophycocyanin-conjugated anti-CD4 mAb followed by fixation with 2% paraformaldehyde, resuspension in permeabilization buffer (HBSS, 0.5% BSA, 0.5% saponin, and 0.05% sodium azide), incubation with Ab for 30 min at room temperature and two washes in permeabilization buffer. Stained cells were resuspended in PBS, 0.3% w/v BSA supplemented with 0.1% w/v sodium azide. Foxp3 staining was performed using a kit (eBioscience) following the manufacturer’s instruction.

To perform the CD1d-Dimer Assay, we incubated 4 μg of the soluble, divalent mouse CD1d-IgG1 fusion protein DimerX overnight with 100 ng of α-galactosylceramide (αGalCer) at 37°C and neutral pH. αGalCer was a gift from Dr. Yasuhiko Koezuka (Pharmaceutical Research Laboratory, Kirin Brewery, Gunma, Japan). The αGalCer-loaded CD1d-IgG1 dimers were incubated with PE-coupled anti-IgG1 (BD Biosciences) for 60 min at 4°C. Mouse NKT cells were labeled with αGalCer-loaded CD1d-IgG1 dimers at 4°C for 60 min.

CD4 T cells were suppressed in mice by i.p. injections of the anti-CD4 mAb YTS 191.1 (days 4 and 1 prevaccination, at the time of vaccination, and days 4 and 9 postvaccination) of 200 μl of PBS containing 100 μg Ab. FCM analyses of PBMC populations demonstrated that >98% of the CD4 T cells were deleted at the time of vaccination.

We used the hepatitis B surface Ag (HBsAg) to assay the effect of Foxp3 T_R cells on specific CD8 T cell responses because we previously demonstrated that different vaccination strategies efficiently elicit K^b-restricted, spleen (S_190–197) VWLSVIWM peptide (S2)-specific CD8 T cell responses to this Ag (33, 34). HBsAg particles produced in Hansenula polymorpha (strain RB10) were purified from crude yeast extracts by adsorption to silica gel, column chromatography, and isopycnic ultracentrifugation (35). These particles were kindly provided by Dr. K. Melber (Rhein Biotech, Düsseldorf, Germany). The indicated dose of HBsAg particles was mixed in 50 μl of PBS with 20 μg of the oligonucleotide (ODN) TCATTGGAAAACGTTCTTCGGGGCG (MWG-Biotech) and injected i.m.

Spleen cells (1 × 10⁷/ml) from immunized mice were incubated for 4 h in RPMI 1640 with 2.5 μg/ml synthetic K^b-binding S_190–197 VWLSVIWM peptide (S2) of HBsAg (obtained from Jerini BioTools) in the presence of 5 μg/ml brefeldin A (Sigma-Aldrich). Harvested cells were surface stained with allophycocyanin-conjugated anti-CD8 mAb fixed with 2% paraformaldehyde and stained with FITC-conjugated anti-IFN-γ mAb. Alternatively, cells were washed twice in FACS buffer (PBS, 0.3% w/v BSA supplemented with 0.1% w/v sodium azide) and incubated for 30 min at 4°C with allophycocyanin-conjugated anti-CD8 mAb and the PE-conjugated tetramer K^b/S2 (K^b with bound VWLSVIWM) provided by the National Institute of Allergy and Infectious Diseases Tetramer Facility. Cells were washed twice in FACS buffer. Frequencies of IFN-γ⁺ CD8⁺ or tetramer⁺ CD8⁺ T cells/10⁵ CD8 T cells were determined by FCM. Mean numbers of CD8⁺ IFN-γ⁺ or tetramer⁺ CD8⁺ T cells/10⁵ CD8 T cells of five individual mice are shown.

CD4 or CD8 T cells were washed twice with PBS and incubated with 5 μM CFSE (Invitrogen Life Technologies) for 15 min at 37°C. CFSE labeling was stopped with cold FCS, and the cells were washed twice with medium.

We used CD4⁺ CD25^high T cells electronically sorted from the spleen of normal or Aα^−/− B6 mice injected 18 h previously with 50 μg of anti-CD3 mAb 145-2C11 as in vivo preactivated T_R cells. These T_R cells were cocultured at a 1:1 ratio for 2–4 days with MACS-isolated, CFSE-labeled splenic CD25⁻CD4 or CD8 T cells from naive B6 mice in the presence of anti-CD3/CD28 dynabeads (DYNAL) at a bead-to-T cell ratio of 1:2.

IL-2 in supernatants was detected by conventional double-sandwich ELISA using mAb JES6-1A12 and biotinylated mAb JES6-5H4. Extinction was analyzed at 405/490 nm on a Tecan microplate ELISA reader (Tecan) using EasyWin software (Tecan). For detection of IL-10, we used the OptEIA kit from BD Biosciences.

Foxp3 expression was analyzed at the single-cell level in nonstimulated, splenic CD4 or CD8 T cells from wt B6 mice (Fig. 1,A). Foxp3⁺ cells were found almost exclusively in CD4 (but not CD8) CD3 T cell populations, with 8–14% of all CD4 T cells being Foxp3⁺ (Fig. 1, B and C). Of the gated CD4 Foxp3⁺ T_R cells, 50–60% were CD25^high, 15–20% were CD103⁺ and about one-third was CD25⁻CD103⁻ (Fig. 1,D). The analysis of the marker profile of Foxp3⁺ vs Foxp3⁻ CD4 T cells revealed similar expression levels of (intracellular and surface) PD-1 and BTLA and of (surface) CD28, CD54, CD94, and PD-L1 in both CD4 T cell subsets. No (or very low) surface expression of CTLA-4, ICOS, CD80, CD86, ICOSL, PD-L2, B7-H3, B7-H4, or NK1.1 was detectable in both CD4 T cell subsets (Fig. 1,E). Some Foxp3⁺ CD4 T_R cells expressed enhanced levels of CD69 and CD44 on the surface and CTLA-4 within the cell when compared with the corresponding Foxp3⁻ CD4 T cells (Fig. 1 E). Foxp3⁺ CD4 T_R cells did not express the (αGalCer-loaded) CD1d dimer-binding TCR, indicating that Foxp3 is not expressed by NKT cells with an invariant TCR. We did not detect differences in the marker profiles of CD25⁺ vs CD103⁺ Foxp3⁺ CD4 T_R cells (data not shown). Similar to the spleen, mesenteric and inguinal lymph node CD4 T cell populations contained a fraction of 8–14% Foxp3⁺ cells (data not shown). Hence, Foxp3 T_R cells represent a fairly constant fraction of 10–15% of the CD4 T cells in secondary lymphoid tissues.

Only low numbers of CD4 αβ T cells are found in B6 mice completely deficient in MHC class II molecules (31, 36, 37). The origin and function of these MHC class II-independent CD4 T cells are unknown, but some of these T cells are CD1d-restricted NKT cells (38). We asked whether Foxp3-expressing cells can be found in the small CD4 T cell population or the major CD8 T cell population of MHC class II-deficient Aα^−/− B6 mice. The small CD4 T cell population in Aα^−/− B6 mice (5 × 10⁵ cells/spleen in Aα^−/− B6 mice vs 10⁷ cells/spleen in wt B6 mice) represented only 6% of the total splenic T cell population, i.e., was 20 times smaller than the CD8 T cell population. CD8 T cells represented >90% of the total splenic T cell population of Aα^−/− mice (6 × 10⁶ cells/spleen in Aα^−/− B6 mice and 5 × 10⁶ cells/spleen in wt B6 mice). Foxp3⁺ was expressed by 20% of the cells in the CD4 T cell population (10⁵ cells/spleen in Aα^−/− B6 mice vs 10⁶ cells/spleen in wt B6 mice) and 1% of all the cells in the CD8 T cell population (10⁵ cells/spleen in Aα^−/− B6 mice and 10⁴ cells/spleen in wt B6 mice) (Fig. 2). Therefore, almost equal numbers of Foxp3⁺ T cells (10⁵/spleen) were found in the splenic CD4 and CD8 T cell populations of Aα^−/− mice. Foxp3⁺ T_R cells can hence develop in the absence of MHC class II molecules and are present in CD4⁺ as well as CD8⁺ T cell populations in MHC class II-deficient mice.

The surface expression profile of CD4/CD8 coreceptors of splenic or lymph node CD4 T cell populations from Aα^−/− B6 mice was unusual. Whereas 80% of the CD4 T cells displayed a SP (CD4⁺CD8⁻) surface phenotype, 20% of the CD4 T cells were DP (CD4⁺CD8⁺; Fig. 2). Within the SP T cell population, 12–15% of the cells were Foxp3⁺ (similar to the frequency of Foxp3⁺ T_R cells in splenic CD4 T cell populations from wt mice). In contrast, 40% of the DP T cells were Foxp3⁺ T_R cells (data not shown). Foxp3⁺ T_R cells were hence 3-fold enriched in the splenic DP T cell population from Aα^−/− B6 mice, confirming our previous report (30).

The surface phenotype of CD4 Foxp3⁺ T_R cells from Aα^−/− and wt B6 mice differed. Most (80–90%) Foxp3⁺ cells from Aα^−/− B6 mice were CD103⁺, whereas only a minor fraction (20%) was CD25⁺ (Fig. 2). A similar prevalence of CD103 surface expression was found in the CD8 Foxp3⁺ T cell population. Hence, Foxp3⁺ T cells generated under MHC class II-deficient conditions show a different distribution in the CD4 and CD8 T cell subset and a different surface phenotype compared with those found in wt congenic controls. Further surface profiling of CD25⁺ vs CD103⁺ CD4 (Foxp3⁺ vs Foxp3⁻) T cells from wt vs Aα^−/− mice (using the marker panel described in Fig. 1 E) revealed no major differences between these subsets. As in wt B6 mice, a fraction of the CD4 Foxp3⁺ T cells from Aα^−/− mice was activated (CD69⁺) and expressed surface PD-L1 but not (or only low levels of) PD-1, BTLA, or CTLA-4 (data not shown). The main difference between CD4 Foxp3⁺ T cells from wt vs Aα^−/− mice was thus the surface expression of CD25 vs CD103.

We searched for Foxp3⁺ T cells in the thymus of young (5-wk-old) or adult (15-wk-old) Aα^−/− or wt B6 mice. A subset of 10–15% Foxp3⁺ cells was readily detectable in the single-positive CD4⁺ CD8⁻ but not CD4⁻ CD8⁺ thymocyte population of wt B6 mice (Fig. 3A). Most of these CD4 T_R cells were CD25^high but CD103^low (Fig. 3,B). This CD4 Foxp3 T_R cell population was similar in young and adult wt mice (Fig. 3,A). In contrast, no CD4 Foxp3⁺ cells were detected in thymocyte populations of young Aα^−/− B6 mice and only variable but low numbers (<0.5%) were found in thymocyte populations from adult Aα^−/− B6 mice (Fig. 3,A). The numbers of Foxp3⁺ CD4 or CD8 T cells increased in the spleen of Aα^−/− but not wt B6 mice with age (Fig. 3 A). Peripheral Foxp3⁺ T_R cells in Aα^−/− B6 mice are hence unlikely to be thymus derived and expand in the periphery with age.

We tested if CD4 Foxp3⁺ T cells from Aα^−/− B6 mice can be activated in vivo. Fifty micrograms of anti-CD3 Ab were injected i.p. into wt or Aα^−/− B6 mice and the activation of T cells was analyzed 18 h postinjection. The % Foxp3⁺ T cells in the splenic CD4 T cell population increased in response to polyclonal activation but the absolute numbers of Foxp3⁺ CD4 T cells per spleen remained similar (due to a decline in Foxp3⁻ CD4 T cells after polyclonal stimulation). Foxp3⁺ CD4 T cells up-regulated surface expression of CD69, CD44, and CD25 but not CD103 and down-regulated surface expression of BTLA after polyclonal stimulation (Fig. 4 A, data not shown). Enhanced CD25 surface expression by CD4 Foxp3⁺ T cells from wt and Aα^−/− B6 mice in response to polyclonal in vivo activation indicates that these T_R cells are functional.

CD25^low, CD25^intermediate, and CD25^high subsets could be readily distinguished in the activated splenic CD4 T cell population. Only few Foxp3⁺ T cells were found in the CD25^low or CD25^intermediate fractions, whereas 80–90% of the cells in the CD25^high subset were Foxp3⁺ (Fig. 4 B). The CD4 CD25^high subset of the splenic T cell population from treated wt or Aα^−/− mice that can be readily sorted for functional studies is thus highly enriched for Foxp3⁺ T cells.

To assess the regulatory activity of CD4 Foxp3⁺ T cells from wt vs Aα^−/− B6 mice in vivo, we used a vaccine (i.e., HBsAg particles mixed with immunostimulating CpG ODNs) that efficiently primes CD4 T cell help-independent CD8 T cell responses (39). wt or Aα^−/− B6 mice were depleted of CD4 T cells by repeated Ab treatment that reduced the detectable CD4⁺ and Foxp3⁺ T cell numbers in spleen and lymph nodes by >98% during the entire period of the ongoing immune response (Fig. 5,A). The vaccine-induced CD8 T cell response of wt and Aα^−/− B6 mice to the well-defined, K^b-restricted HBsAg epitope (34) was enhanced by depleting CD4 T cells (Fig. 5,B, Table I). This was apparent when different doses of the vaccine were used, and when the response was detected either by staining tetramer⁺ CD8 T cells or by measuring the number of CD8 T cells with specifically inducible IFN-γ expression. No tetramer⁺ CD8 T cells expressing Foxp3 were found in either mouse line. Although this enhancement was only 2-fold, it is in line with previously published data (40) and may well be of biological significance. In wt as well as MHC class II-deficient mice, CD4 T cells can thus down modulate specific CD8 T cell responses.

We confirmed in vitro that MHC class II-independent CD4 CD25 T cells down-modulate CD4 and CD8 T cell responses. Sorted splenic CD25^high CD4 T cells from anti-CD3 Ab-treated, wt or Aα^−/− B6 mice (Fig. 4,B) were cocultured with purified, naive (CFSE-labeled) CD25⁻ CD4 or CD8 T cells (at a 1:1 ratio) from wt B6 mice in the presence of microbead-coupled anti-CD3 and anti-CD28 Abs. Proliferation and IL-2 release of activated CD4 as well as CD8 T cells were blocked in the presence of CD25^high CD4 T_R cells from wt as well as Aα^−/− B6 mice (Fig. 6). Anti-CD3/CD28-stimulated T_R cells from wt but not Aα^−/− B6 mice produced large amounts of IL-10. CD25^high CD4 T cells from Aα^−/− B6 mice thus qualify as T_R cells, similar to their counterparts from MHC class II-expressing wt B6 mice.

We demonstrate in this study that CD4 Foxp3 T_R cells are found in secondary lymphoid tissues and the periphery but not the thymus of severely MHC class II-deficient mice. These T_R cells constitutively express CD103 but up-regulate CD25 surface expression in response to TCR-mediated activation. MHC class II-independent T_R cells down-modulate specific CD8 T cell responses in vivo and IL-2 release and proliferation of CD4 or CD8 T cells primed in vitro. Functional CD4 Foxp3⁺ T_R cells thus develop in the complete absence of MHC class II molecules presumably in peripheral tissues.

We found CD4 Foxp3⁺ T_R cells in spleen, lymph nodes, and liver of Aα^−/− B6 mice but not in the thymus. The surface phenotype of Foxp3 T_R cells from Aα^−/− B6 mice was heterogeneous, but most of these T_R cells were CD103^high and CD25^low. CD4 Foxp3 T_R cells from Aα^−/− B6 mice up-regulated surface expression of CD25, CD69, and CD44 in response to TCR-mediated activation. It has been hypothesized that natural Foxp3 T_R cells constitutively express high surface levels of CD25 because they are primed in a TGF-β-independent manner by self Ags in the thymus and are continuously challenged by self Ags in peripheral tissues, thereby exerting tonic suppression, which contributes to self tolerance (3, 9, 41). In contrast, adaptive Foxp3 T_R cells are primed in peripheral tissues by (CD103-inducing) TGF-β from naive Foxp3⁻ precursors by specific challenge of foreign Ag and can be repeatedly challenged by this Ag in, e.g., chronic infection (28). If correct, MHC class II-deficient Aα^−/− B6 mice would harbor adaptive but not natural Foxp3 T_R cells. This would imply that establishment of natural T_R cell control in the thymus is strictly MHC class II dependent whereas establishment of adaptive T_R cell control in the periphery is more flexible in terms of restriction element usage.

We found no CD4⁺ Foxp3⁺ cells in thymi of young Aα^−/− B6 mice, whereas substantial numbers of Foxp3⁺ cells were readily detected in SP CD4⁺CD8⁻ thymocyte populations of sex- and age-matched, congenic wt B6 mice. The minor and variable subset of CD4 Foxp3⁺ T cells emerging in thymi of adult Aα^−/− B6 mice may result from reentry of activated T blasts into the thymus (42). Only few CD8 Foxp3⁺ T cells were found in thymi of Aα^−/− B6 mice, although the splenic CD8 T cell population of these animals contained a Foxp3⁺ subset that (although small) was 4- to 10-fold expanded when compared with that in wt B6 mice. The CD4 and CD8 Foxp3 T_R cell populations expanded in the spleen of Aα^−/− B6 mice with age. Although we did not detect Ag-specific (tetramer⁺) CD8 Foxp3⁺ T cells in vaccinated Aα^−/− B6 mice, these T_R cells may be expanded by repeated Ag challenge.

Although NKT cells are frequent in the CD4 αβ T cell compartment of Aα^−/− B6 mice, splenic CD4 Foxp3⁺ T_R cells from Aα^−/− B6 mice were not conventional NKT cells. These T cells did not express an invariant TCR that binds glycolipid-loaded CD1d dimers. CD1d^−/− B6 mice do not lack CD4 Foxp3⁺ CD103^highCD25^low T_R cells (data not shown). We thus found no evidence that point to NKT cells as part of the Foxp3 T_R cell population, although suppressive NKT cells have been identified (43, 44, 45, 46) and Foxp3⁺ T_R cells and NKT cells have been shown to cooperate in the control of autoimmune disease (47). NKT-like cells with a variant TCR (that are ill defined) may contribute to the Foxp3⁺ T_R cell pool.

Recognition and restriction specificity of CD4 Foxp3⁺ T_R cells were resolved when TCR transgenic systems were used (20) but not under natural conditions. The restriction element in the specific recognition of Foxp3⁺ CD4 T_R cells from Aα^−/− B6 mice is unknown. There are few leads that suggest candidates. It is intriguing that Foxp3⁺ T_R cells preferentially express CD4 even in the absence of conventional MHC class II molecules. MHC class II (-like) molecules present in Aα^−/− B6 mice (DM and DO molecules) do not present peptides to T cells. Alternatively, MHC class I-like molecules (such as Qa-1, TL, and H2M3) may present epitopes to these T_R cells that would make this subset NKT-like. A coreceptor function for CD103 is not known, although enhancement of CD3/TCR-induced activation of intestinal intraepithelial T cells by stimulation with an CD103-binding Ab has been described (48). Aα^−/− mice seem to be a novel model for the study of adaptive Foxp3⁺ T_R cells (isolated from the natural Foxp3⁺ T_R cell compartment).

We greatly appreciate the expert technical assistance of Ellen Allmendinger. We gratefully acknowledge the gift of the MHC class II-deficient Aα^−/− B6 mice from Dr. H. Bluethmann (Roche, Basel, Switzerland) and the HBsAg/K^b tetramers from the National Institute of Allergy and Infectious Diseases Tetramer Facility (Emory University Vaccine Center, Atlanta, GA). We thank Dr. G. Hämmerling (German Cancer Research Center, Heidelberg, Germany) for helpful discussions.

The authors have no financial conflict of interest.