A Self-Reactive TCR Drives the Development of Foxp3⁺ Regulatory T Cells that Prevent Autoimmune Disease

Abstract

Although Foxp3⁺ regulatory T (T_reg) cells are thought to express autoreactive TCRs, it is not clear how individual TCRs influence T_reg cell development, phenotype, and function in vivo. We have generated TCR transgenic mice (termed SFZ70 mice) using Tcra and Tcrb genes cloned from an autoreactive CD4⁺ T cell isolated from a T_reg cell-deficient scurfy mouse. The SFZ70 TCR recognizes a cutaneous autoantigen and drives development of both conventional CD4⁺ Foxp3⁻ T cells (T_conv) and Foxp3⁺ T_reg cells. SFZ70 T_reg cells display an activated phenotype evidenced by robust proliferation and expression of skin-homing molecules such as CD103 and P-selectin ligand. Analysis of Foxp3-deficient SFZ70 mice demonstrates T_reg cells inhibit T_conv cell expression of tissue-homing receptors and their production of pro-inflammatory cytokines. Additionally, T_reg cell suppression of SFZ70 T_conv cells can is overcome by non-specific activation of antigen-presenting cells. These results provide new insights into the differentiation and function of tissue-specific T_reg cells in vivo and provide a tractable system for analyzing the molecular requirements of T_reg cell-mediated tolerance toward a cutaneous autoantigen.

Introduction

Thymocytes expressing TCRs with high affinity for self-antigen are subject to a variety of fates. Many are eliminated via negative selection while agonistic selection diverts others into alternative lineages such as CD4⁺Foxp3⁺ regulatory T (T_reg) cells and CD8αα intestinal epithelial (IEL) T cells (1). However, a fraction of autoreactive thymocytes are neither deleted nor redirected, but instead emerge from the thymus as conventional T (T_conv) cells with the potential to cause autoimmunity. As a result, T cell intrinsic and extrinsic peripheral tolerance mechanisms have evolved to control this population (2).

At barrier tissues such as the skin and intestines, the immune system must mount robust responses against invading pathogens while simultaneously maintaining self-tolerance. Indeed, several lines of evidence indicate that T_reg cells are essential for maintaining proper immune homeostasis in the skin. T_reg cell-deficient scurfy mice develop severe CD4⁺ T cell-mediated cutaneous inflammation and have autoantibodies directed against skin-specific antigens (3, 4). Additionally, human and mouse skin contain a high frequency of T_reg cells, and the majority of human peripheral blood T_reg cells express skin-homing receptors (5, 6). Finally, we have found that selectively blocking T_reg cell migration to the skin impairs their ability to maintain cutaneous tolerance (5, 7). Conversely, in mice expressing the model antigen ovalbumin (OVA) under a skin-specific promoter, T_reg cells do not play a critical role in the regulation of OVA-specific T cell migration to the skin (8). Thus, the extent to which, and mechanisms whereby T_reg cells regulate skin-specific autoreactive T cells merits further investigation.

TCR-transgenic (Tg) mice recognizing endogenously expressed self-antigens have been valuable tools for studying T cell activation and tolerance in autoimmune disease (9–11). To further examine the development and behavior of autoreactive T cells in the skin, we generated a new line of TCR Tg mice, termed SFZ70 mice, using TCR chains cloned from a scurfy mouse–derived CD4⁺ T cell reactive towards a cutaneous self-antigen. These animals have allowed us to explore the behavior of skin antigen-specific T_conv and T_reg cells in vivo and to further define the function of T_reg cells in maintaining cutaneous tolerance.

Materials and Methods

Hybridoma generation and T cell activation assay

Generation and maintenance of β-galactosidase (lacZ)-inducible T cell hybridomas has been previously described (12). Briefly, CD4⁺ T cells isolated from the skin-draining LN of scurfy mice were fused with the lacZ-inducible fusion partner BWZ.36 to generate a lacZ-inducible T cell hybridoma panel. CD11c⁺ APCs were isolated using magnetic bead enrichment (Miltenyi) and cocultured overnight with LacZ-inducible T cell hybridomas. In some experiments the APCs were incubated with various amounts of anti-I-A^b mAb (Y3P) during co-culture as indicated. To assess T cell hybridoma activation, lysis buffer containing the lacZ substrate chlorophenolred-β-D-galactopyranoside (CPRG) was added to overnight cultures (0.15mM CPRG, .125% NP-40, and 9 mM MgCl₂ in PBS). Results show lacZ activity as the absorbance of the CPRG cleavage product at 570 nm (with 630 nm as the reference wavelength).

Mice

C57BL/6J, Rag1^−/− (B6.129S7-Rag1tm1Mom/J), OT-II (C57BL/6-Tg(TcraTcrb)425Cbn/J), and scurfy mice (B6.Cg-Foxp3sf/J) were obtained from The Jackson Laboratory. SFZ70 TCR Tg mice were generated by cloning cDNA encoding the rearranged TCR α and β variable regions (using the TRAV13D-1*01 and TRBV13-1*01 variable gene segments) into cassette vectors containing endogenous TCR regulatory elements (13). Linearized plasmids were then injected into the pronuclei of single-cell mouse embryos. The following PCR primers were used for identifying mice carrying the SFZ70 transgene: TCRα: (forward) GAGACCCGGGCAGCAACCACACAAGCACCA, (reverse) GAGAGCGGCCGCCCCAGCCAAAGCTGGGAATTT, expected product size: 800 bp; TCRβ: (forward) GAGACTCGAGTGGTCGCGAGATGGGCTCCA, (reverse) GAGAGACCGCGGAAATGCTCCCTCCCCTTCATT, expected product size: 600 bp. All animals were housed and bred under specific pathogen-free conditions in the Benaroya Research Institute animal facility, and the Benaroya Research Institute Institutional Animal Care and Use Committee approved all experiments.

Flow cytometry

For surface staining, cells were incubated at 4°C for 30 minutes in staining buffer (HBSS, 3% FBS, 0.09% NaN₃) with the following monoclonal antibodies from eBioscience, Biolegend and BD Biosciences: anti-CD4 (GK1.5), anti-CD8 (53–6.7), anti-CD25 (PC61.5), anti-CD103 (2E7) and anti-CD44 (IM7). For P-selectin ligand staining, cells were incubated with P-selectin–human IgM fusion proteins, followed by biotinylated goat anti–human IgM (Jackson Immunoresearch) and streptavidin-phycoerythrin (eBioscience). For activated caspase staining, cells were incubated with FITC-tagged z- VAD-fmk (Promega). For intracellular staining of IFN-γ, IL-17, TNF-α, IL-4, and Foxp3, cells were surface stained as described above, permeabilized with Fix/Perm buffer (eBioscience), and stained for 30 minutes with the following antibodies purchased from eBioscience or BioLegend: anti-IFN-γ (XMG1), anti-IL-17 (TC11-18H10.1), anti-TNF-α (MP6-XT22), anti-IL-4 (11B11) and/or anti-Foxp3 (FKJ-16s) in BD Perm/Wash. Data were acquired on a LSR II (BD Biosciences) and analyzed using FlowJo software (TreeStar).

In vitro T cell stimulation

Prior to intracellular cytokine staining, cells from the skin-draining LN or spleen were depleted of red blood cells and stimulated in complete medium (RPMI plus 10% FCS, 4 mM L-glutamine, 1 mM sodium pyruvate, 50 units/mL penicillin and 50 μg/mL streptomycin) for 5 hours at 37°C with PMA (50 ng/mL) and ionomycin (1μg/mL) in the presence of monensin (10 μg/mL).

BrdU incorporation assay

BrdU (5-bromodeoxyuridine) was administered to mice in drinking water at a concentration of 0.8 mg/mL for 72 hours prior to analysis. Lymphocytes from treated mice were surface stained as described above, fixed for 20 minutes using BD Cytofix/Cytoperm, and then treated for 1 hour at 37°C with DNase I (Sigma #D4513, 300μg/mL). For analysis of BrdU incorporation, cells were stained intracellularly using antibodies directed against BrdU (MoBU-1) and Foxp3 in BD Perm/Wash 30 min.

Anti-CD40 treatment

Mice were injected intraperitoneally on days 0, 2, and 4 with 25 μg of anti-CD40 (FGK4.5; UCSF Monoclonal Laboratory Core) or 25 μg rat IgG (Sigma) in PBS. Mice were euthanized for analysis on day 6.

Tissue histology

Tissues were fixed using 10% neutral buffered formalin, paraffin embedded, and cut into 5 μm sections which were stained with hematoxylin and eosin.

Statistical analysis

Statistical significance was determined by one-way repeated-measures analysis of variance or two-tailed unpaired Student's t-test where appropriate as indicated in the figure legends.

Results

The SFZ70 TCR recognizes a cutaneous self-antigen

Foxp3-deficient scurfy mice suffer from severe CD4⁺ T cell–mediated skin inflammation, and therefore we reasoned they would provide an abundant source of CD4⁺ T cells specific for cutaneous antigens. We fused CD4⁺ T cells from the spleens and skin-draining lymph nodes (LN) of scurfy mice with the BWZ.36 partner to generate a panel of CD4⁺ T cell hybridomas expressing β-galactosidase under the control of NFAT regulatory elements (12). To identify clones reactive toward cutaneous self-antigens, each hybridoma was co-cultured with CD11c⁺ dendritic cells (DCs) isolated from either the skin-draining LNs or spleens of WT mice. T cell activation was then assessed by measuring cleavage of CPRG, a chromogenic β-galactosidase substrate. We identified several hybridoma clones that selectively recognized antigens presented by DCs from skin draining LN but not from the spleen. Of these, we selected one, termed scurfy-derived LacZ inducible hybridoma 70 (SFZ70) for further study based on its particularly robust, MHC class II-restricted response to skin-draining LN DCs (Fig. S1A). The SFZ70 hybridoma reacted strongly to DCs isolated from either the skin-draining LNs or directly from the skin, but minimally to DCs isolated from the spleen, mesenteric LN, thymus, intestine, liver, and lung (Fig. 1, top panels and data not shown). By contrast, DCs from all tissues were able to activate an ovalbumin-specific hybridoma (OTIIZ) when pulsed with OVA peptide 323–339, confirming their ability to present antigen. When analyzed separately, DCs from the inguinal, axillary, brachial, and superficial cervical skin-draining LNs activated the SFZ70 hybridoma to a similar extent (Fig. S1B).

Figure 1. Skin-specific activation of the SFZ70 hybridoma.

CD11c⁺ APCs were isolated from the indicated tissues of WT mice (top panels), WT and Ccr7^−/− (middle panels) or conventional (CNV) and germ-free (GF) mice (bottom panels). APCs were then co-cultured with 1×10⁵ SFZ70 hybridoma cells (left panels) or pulsed with OVA peptide and co-cultured with 1×10⁵ OTIIZ hybridoma cells (right panels). After 16 hours, cells were lysed and the hybridoma response was assessed by measuring the absorbance (A570-A630) of the chromogenic product released following CPRG cleavage by β-gal. Results are representative of at least two independent experiments with APCs pooled from 2–8 mice per group.

The skin-draining LNs contain multiple subsets of DCs including LN resident DCs and several populations of migratory DCs. Migratory DCs transport cutaneous antigens to the draining LN via efferent lymphatics in a process dependent on the chemokine receptor CCR7 (14). DCs from the skin-draining LNs of CCR7-deficient mice failed to activate the SFZ70 hybridoma but were still capable of presenting antigen to the OTIIZ cells (Fig. 1, middle panels). Additionally, DCs from both germ-free and conventional mice were equally capable of activating the SFZ70 hybridoma, demonstrating that the SFZ70 TCR does not recognize an antigen derived from the cutaneous microflora (Fig. 1, bottom panels). Instead, these results demonstrate that the SFZ70 hybridoma recognizes a cutaneous self-antigen that is transported to the skin-draining LNs in a CCR7-dependent manner.

T cell development in SFZ70.Rag1^−/− TCR transgenic mice

To study the development and behavior of T cells specific for an endogenously expressed cutaneous antigen in vivo, we generated SFZ70 TCR Tg mice which were subsequently backcrossed onto the C57BL/6 background. To prevent the confounding effects of endogenous TCR rearrangement, we further crossed the SFZ70 mice to Rag1^−/−mice (SFZ70.Rag1^−/−) and performed all subsequent experiments using these animals. The majority of SFZ70.Rag1^−/− mice did not display any overt signs of autoimmune or inflammatory disease and appeared externally and histopathologically normal. However, < 2% developed caudal dermatitis reminiscent of that seen in Foxp3-deficient scurfy mice (Fig. S2).

Total thymocyte numbers, as well as the number of CD4/CD8 double positive (DP) and double negative (DN) cells in SFZ70.Rag1^−/− mice were similar to those found in WT mice. However, we observed a 10-fold decrease in the number of CD4⁺CD8⁻thymocytes (Fig. 2A). Additionally, SFZ70.Rag1^−/− mice contained a CD8SP T cell population not found in OVA-specific OT-II.Rag1^−/− TCR Tg mice. CD4SP thymocytes expressing TCRs with higher affinity for self-antigen are thought to be preferentially diverted into the T_reg cell lineage, and T_reg cell development can only be recapitulated in TCR Tg mice that also express the TCR’s cognate antigen. In accordance with this model, we found a high frequency of Foxp3⁺ thymocytes within the SFZ70.Rag1^−/−CD4SP subset, which on average was slightly larger than that found in WT mice, although this difference was not statistically significant (Fig. 2B). Moreover, although the frequency of the CD4SP Foxp3⁺ cells among CD4SP thymocytes varied substantially between mice, we consistently found ~100,000 CD4SP Foxp3⁺ thymocytes in SFZ70.Rag1^−/− animals.

Figure 2. T cell development in SFZ70.Rag1^−/− mice.

(A) Expression of CD4 and CD8 by thymocytes from SFZ70.Rag1^−/−, OT-II.Rag1^−/− or WT mice. Graphs show corresponding absolute numbers of DN, DP, and CD4SP thymocytes. (B) Expression of Foxp3 and CD25 by thymocytes isolated from SFZ70.Rag1^−/−, OT-II.Rag1^−/− or WT mice. Graphs show corresponding percentage and absolute number of CD4SP thymocytes expressing Foxp3. Data are representative of three independent experiments with 1–5 mice per group.

SFZ70.Rag1^−/− mice contain activated peripheral T_reg cells

CD4⁺ T cells were present in the skin-draining LNs, mesenteric LNs, and spleen SFZ70.Rag1^−/− mice (Fig. 3A). In addition, the level of surface TCRβ expressed by these cells was equivalent to that observed in CD4⁺ T cells from WT mice (data not shown), indicating that tolerance in these animals is not the simply the result of TCR downregulation. However, the peripheral lymphoid organs of SFZ70.Rag^−/− mice were highly lymphopenic compared to those of OT-II.Rag1^−/− or WT animals, containing ~10–100-fold fewer CD4⁺ T cells. Moreover, the majority of TCRβ⁺ SFZ70.Rag1^−/− T cells in the periphery did not express CD4 or CD8, whereas TCRβ⁺ CD4/CD8 DN cells were rare in WT and OT-II.Rag1^−/− mice. CD8αα⁺ IEL T cells are thought to originate from T cells expressing self-reactive TCRs, and the development of IEL and T_reg cells has been genetically linked (15, 16). Consistent with this link, the intestinal epithelium of SFZ70.Rag1^−/− mice contained a large population of CD8αα T cells (Fig. S3).

Figure 3. SFZ70.Rag1^−/− mice contain peripheral CD4⁺ and CD8⁺ T cells and Helios⁺ T_reg cells.

(A) Expression of CD4 and CD8 by TCRβ⁺ lymphocytes isolated from the skin-draining lymph nodes (LN), mesenteric lymph node (MLN) and spleens of SFZ70.Rag1^−/−, OT-II.Rag1^−/− or WT mice. Graphs show corresponding absolute numbers of CD4⁺ lymphocytes. (B) Expression of Foxp3 and CD25 (top) or Foxp3 and Helios (bottom) by CD4⁺ lymphocytes isolated from SFZ70.Rag1^−/−, OT-II.Rag1^−/− or WT mice. Statistical significance was determined using a one-way repeated-measures analysis of variance. A Tukey post-test was used to obtain the p-value for the bracketed pairwise comparisons. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Data are representative of two experiments with 3–9 mice per group.

Upon examination of the peripheral SFZ70.Rag1^−/− T_reg cell compartment, we found Foxp3⁺ expression by up to 50% of the CD4⁺ T cells within the spleen, mesenteric LN, and skin-draining LNs (Fig. 3B, top and data not shown). However, no Foxp3⁺ cells were present in either the DN or CD8⁺ T cell populations (data not shown). The disparity in the size of the T_reg cell population in the periphery vs. the thymus of SFZ70.Rag1^−/− mice suggests that peripheral T_reg cells are either derived from T_conv cells that have upregulated Foxp3 or that the T_reg cells possess a peripheral homeostatic proliferative/survival advantage over their T_conv cell counterparts. We examined expression of the Ikaros family transcription factor Helios, which is expressed by essentially all thymically-derived T_reg cells, but not by T_reg cells induced either in vivo in response to antigen feeding, or in vitro by activation in the presence of TGF-β (17). Interestingly, nearly all of the peripheral T_reg cells in SFZ70.Rag1^−/− mice were Helios⁺ indicating that they originated in the thymus, and not through peripheral conversion (Fig. 3B, bottom).

To determine if enhanced proliferation might account for the increased frequency of peripheral T_reg cells and to directly compare T_conv and T_reg cell proliferation in SFZ70.Rag1^−/− mice, we measured incorporation of the thymidine analogue bromodeoxyuridine (BrdU) by T cells following a 72-hour treatment. As a control, we also examined BrdU incorporation in OT-II.Rag1^−/− animals. Indeed, the percentage of BrdU⁺ T_conv cells was significantly greater in the autoreactive SFZ70.Rag1^−/− mice than in OT-II.Rag1^−/− animals (Fig. 4A). Furthermore, consistent with studies documenting the robust proliferation of T_reg cells in vivo (18), BrdU labeling was observed in a larger proportion of the SFZ70.Rag1^−/− T_reg cells than their T_conv cell counterparts.

Figure 4. SFZ70.Rag1^−/− T cells display an activated phenotype.

(A) Incorporation of BrdU by CD4⁺ lymphocytes from the skin-draining LN of SFZ70.Rag1^−/−, OT-II.Rag1^−/−or WT mice. Graph indicates the percent BrdU⁺ cells amongst Foxp3⁺ or Foxp3⁻ cells from each group. (B) Expression of CD44, CD103, and P-selectin ligand (P-lig) by CD4⁺ lymphocytes isolated from the skin draining LN of SFZ70.Rag1^−/−, OT-II.Rag1^−/− or WT mice. Statistical significance was determined using a one-way repeated-measures analysis of variance; a Tukey post-test was used to obtain the p-value for the bracketed pairwise comparison. **, P < 0.01; ***, P < 0.001. Data are representative of three independent experiments with 3–5 mice per group.

Despite their robust proliferative response, SFZ70.Rag1^−/− T cells do not accumulate, suggesting they may undergo increased apoptosis. To address this possibility, we stained T cells from the skin-draining LN of SFZ70.Rag1^−/−, OT-II.Rag1^−/−, and WT mice with FITC conjugated z-VAD-fmk, which specifically binds activated caspases and labels cells undergoing apoptosis (19). Indeed, CD4⁺ T cells from SFZ70.Rag1^−/− mice show high levels of caspase activation compared to those from OT-II.Rag1^−/− or WT animals, indicating that activation-induced apoptosis may help prevent the accumulation of autoreactive T cells and subsequent autoimmune disease in SFZ70.Rag1^−/− mice (Fig. S4).

Both T_conv and T_reg cells in SFZ70.Rag1^−/− mice uniformly displayed a CD44^hi effector/memory cell phenotype (Fig. 4B, top panels). Moreover, SFZ70.Rag1^−/− T_reg cells expressed skin homing receptors such as CD103 and P-selectin ligand (P-lig) (Fig. 4B, middle and bottom panels). Thus, in accordance with their specificity for a cutaneous autoantigen, activated T_reg cells with a skin-homing phenotype comprise a large percentage of the peripheral CD4⁺ T cell compartment in SFZ70.Rag1^−/− mice.

T_reg cells are critical for maintenance of peripheral tolerance in SFZ70.Rag1^−/− mice

Although the SFZ70 TCR recognizes a self-antigen, most SFZ70.Rag1^−/− mice do not develop overt autoimmunity. To examine the role of T_reg cells in preventing autoimmune disease in these animals, we bred SFZ70.Rag1^−/− and Rag1^−/−Foxp3^sf animals to generate T_reg cell-deficient SFZ70 (SFZ70.Rag1^−/−Foxp3^sf) mice. By 24 days of age, SFZ70.Rag1^−/−Foxp3^sf mice displayed a moribund phenotype, and weighed significantly less than their SFZ70.Rag1^−/−Foxp3^WT littermates (Fig. 5A). By contrast, OT-II.Rag1^−/− mice that are unable to express functional Foxp3 do not develop any inflammatory disease and are indistinguishable from WT littermates (20). While there was no visible cutaneous inflammation in the SFZ70.Rag1^−/−Foxp3^sf mice, histological analysis revealed substantial lymphocytic infiltration of the skin and liver that was not observed in the SFZ70.Rag1^−/−Foxp3^WT littermate controls (Fig. 5B). Furthermore, despite their reduced weight and runted appearance, there was no evidence of inflammation in other tissues normally impacted in Foxp3-deficient mice such as the lungs and intestines (data not shown).

Figure 5. T_reg cells are required for maintenance of peripheral tolerance in SFZ70.Rag1^−/− mice.

(A) Weight of SFZ70.Rag1^−/−Foxp3^sf and SFZ70.Rag1^−/−Foxp3^WT mice at 24 days of age. (B) Photomicrographs of hematoxylin and eosin–stained sections of the skin (20×), and liver (20×) from SFZ70.Rag1^−/−Foxp3^sf and SFZ70.Rag1^−/−Foxp3^WT mice. (C) Expression of Ki-67, P-selectin ligand (P-lig), and CXCR3 by CD4⁺ T cells from the skin-draining LN of SFZ70.Rag1^−/−Foxp3^sf (solid black line) or SFZ70.Rag1^−/− Foxp3^WT (gray filled histogram) mice. (D) Production of IFN-γ, TNF-α, IL-4, and IL-17 by SFZ70.Rag1^−/−Foxp3^sf and SFZ70.Rag1^−/−Foxp3^WT CD4⁺Foxp3⁻CD44^hi T cells from the skin-draining LN following in vitro PMA and ionomycin stimulation. ***, P < 0.001. Statistical significance was assessed using a two-tailed, unpaired Student’s t-test.

T_reg cells can suppress autoimmunity by inhibiting the priming and activation of autoreactive T cells. However, in SFZ70.Rag1^−/− mice the majority of T_conv cells were CD44^hi and underwent robust proliferation (Fig. 4). Additionally, expression of the proliferation-associated protein Ki-67 was similar in cells from SFZ70.Rag1^−/−Foxp3^WT and SFZ70.Rag1^−/−Foxp3^sf mice (Fig. 5C, top panels). However, expression of P-lig and CXCR3, two molecules associated with T cell homing to inflamed tissues, were substantially increased in T_conv cells from the skin-draining LN and spleen of SFZ70.Rag1^−/−Foxp3^sf mice (Fig. 5C, middle and bottom panels). Furthermore, following ex vivo restimulation with PMA/ionomycin an increased percentage of SFZ70.Rag1^−/− Foxp3^sf CD4⁺ T cells produced pro-inflammatory cytokines such as IFN-γ, TNF-α, and IL-4 compared to those from SFZ70.Rag1^−/−Foxp3^WT littermate controls, whereas cytokine production by CD8⁺ and DN T cells was unchanged between the two groups (Fig. 5D, and data not shown). Thus, consistent with previous reports (8, 21, 22), T_reg cells in SFZ70.Rag1^−/− mice do not appear to inhibit the initial activation of T_conv cells, but instead function primarily by modulating T_conv cell migration and cytokine production.

Systemic DC maturation overcomes T_reg cell-mediated suppression in SFZ70.Rag1^−/− mice

The activation status of the antigen presenting DCs is a key factor in determining the outcome of T cell priming. Whereas immature DCs drive a tolerogenic T cell response, mature DCs express increased levels of costimulatory molecules and produce pro-inflammatory cytokines, which enable full T cell activation and effector differentiation (23). To examine the relationship between DC activation and T_reg cell function, we treated SFZ70.Rag1^−/− mice over a six-day period with either an agonistic anti-CD40 mAb that potently activates DCs, or control rat IgG. During the last 72 hours of anti-CD40 treatment, BrdU was administered to measure T cell proliferation. Indicative of robust DC activation, we found that proliferation of both T_conv and T_reg cells was increased in the skin-draining LN and spleen of anti-CD40-treated SFZ70.Rag1^−/− mice compared to controls (Fig. 6A). Furthermore, in response to ex vivo restimulation, there was a substantial increase in the percentage of T_conv cells producing the pro-inflammatory cytokines IFN-γ and IL-17 in the anti-CD40 treated mice (Fig. 6B). By contrast, anti-CD40-treatment did not augment proliferation or cytokine production by T_conv cells in non-autoreactive OT-II.Rag1^−/− animals (data not shown). Thus, DC activation can overcome T_reg cell-mediated suppression of T_conv cells, resulting in their increased proliferation and effector cytokine production.

Figure 6. T_reg cell-mediated suppression in SFZ70.Rag1^−/− mice is overcome by systemic DC activation.

On days 0, 2, and 4 SFZ70.Rag1^−/− mice were injected intraperitoneally with 25 μg of anti-mouse CD40 or 25 μg rat IgG (control) in PBS. Mice were euthanized on day 6 and CD4⁺ T cells from the skin-draining LN and spleen were examined for (A) incorporation of BrdU and (B) production of IFN-γ and IL-17 following in vitro PMA and ionomycin stimulation.

Discussion

To examine the behavior of autoreactive T cells specific for an endogenously expressed cutaneous antigen, we generated SFZ70 TCR Tg mice using an autoreactive TCR isolated from a scurfy mouse. Notably, this TCR is compatible with the development of both CD4⁺Foxp3⁻ T_conv cells and Foxp3⁺ T_reg cells. Our analyses of these animals provide new insights into control of skin reactive T cells, and describe a new platform for elucidating the molecular and cellular mechanisms required for generation and maintenance of T_reg cell-mediated immunological tolerance.

In mice containing a polyclonal TCR repertoire, T_reg cell-mediated suppression of autoreactive T_conv cells is required throughout the animal’s lifespan for prevention of autoimmune disease (24). However, the specificity and behavior of the pathogenic T cells that cause disease in the absence of T_reg cells has been poorly defined. Although long inferred, our results, together with a recent report demonstrating that T_reg cell ablation in germ-free mice results in cutaneous inflammatory disease (25), provide direct evidence demonstrating that dysregulated CD4⁺ T cell responses in scurfy mice are aimed in part toward cutaneous autoantigens.

Despite the fact that it was recovered from a CD4⁺ T cell in the periphery of a scurfy mouse, the ability of the SFZ70 TCR to drive the development of CD4SP cells is greatly limited. Instead, a large fraction of TCR⁺ cells present in the thymus and periphery of SFZ70 mice express neither CD4 nor CD8. The development of these mature DN T cells may be due to the autoreactive nature of the SFZ70 TCR. Indeed, T cell coreceptor downregulation has been shown to lower overall TCR avidity for peptide:MHC complexes and has been previously described as a mechanism of tolerance (26). Alternatively, the large DN T cell population may reflect the altered timing of TCR expression during thymic development of SFZ70 T cells. Early expression of Tg TCR α and β chains under the control of endogenous transcriptional regulatory elements results in accelerated T cell development and the emergence of a large population of DN T cells in the thymus and periphery that phenotypically and functionally resemble γδ T cells (27). Similarly, development of CD8αα cells in TCR Tg mice has also been linked to premature TCRα chain expression during thymic development. However a detailed comparison of SFZ70 mice with other systems is complicated by differences in antigen expression and prevalence, MHC I vs. MHC II restriction, and TCR affinity.

In SFZ70 mice, the large proportion of DP thymocytes and relative absence of CDSP cells suggests deletion occurs at the DP to CD4SP transition, and this most likely reflects extensive negative selection of the autoreactive SFZ70 thymocytes. The timing and extent of negative selection in TCR Tg mice is dependent on factors including TCR affinity for peptide:MHC and the location and cell type in which antigen is presented in the thymus (28). Although thymic DCs can induce negative selection, these cells did not potently activate the SFZ70 hybridoma. Thus, thymic DCs may present the SFZ70 antigen at levels sufficient to induce negative selection in vivo, but below the detection limit of our hybridoma assay. For instance, low-level DC migration from the skin to thymus has been reported to contribute to negative selection and may play a role in shaping the SFZ70 thymocyte population (29). Additionally, presentation of the SFZ70 cognate antigen may be mediated in large part by medullary thymic epithelial cells (mTECs) capable of expressing and presenting tissue-specific antigens. Indeed, antigen presentation by mTECs can drive both negative selection and development of Foxp3⁺ T_reg cells (30, 31). Crossing SFZ70 mice to animals deficient in various pathways of cellular migration and antigen presentation will help determine the relative contributions of DCs and mTECs for driving both selection and induction of Foxp3. Additionally, poor positive selection may also contribute to the paucity of SFZ70 CD4SP thymocytes, and we cannot exclude the possibility that the peptide:MHC complex mediating the thymic selection of SFZ70 T cells is distinct from the cutaneous antigen they respond to in the periphery.

In TCR Tg mice generated using TCRs isolated from T_reg cells, only a very small percentage of CD4SP thymocytes were found to express Foxp3 (32, 33). However, mixed bone marrow chimeras revealed that these TCRs drove the development of T_reg cells when their clonal frequency was limited to a small fraction of a polyclonal repertoire. These experiments demonstrated that a specific niche exists for a TCR of any given specificity that can only accommodate a limited number of T_reg cells. In contrast to these mice, T_reg cells comprise a substantial percentage of the thymic CD4SP population in SFZ70 mice at frequencies higher or comparable to that observed in WT mice. However, in agreement with a limited T_reg cell clonal niche size, the absolute number of SFZ70 thymic T_reg cells is consistently around 100,000 cells. Given the low absolute number of CD4SP thymocytes, intraclonal competition for T_reg cell-inducing peptide:MHCII may be modest in SFZ70 mice, thereby accounting for the relatively large percentage of T_reg cells among CD4SP thymocytes observed in these animals.

In the periphery, T_reg cells undergo robust homeostatic proliferation that is thought to be dependent on their recognition of self-antigens presented by DCs on MHCII (34, 35). Additionally T_reg cell homeostasis is controlled in part by IL-2 produced by CD44^hi effector/memory T_conv cells (36). Therefore, the robust proliferation and peripheral expansion of T_reg cells in SFZ70 mice may be driven by a combination of TCR autoantigen recognition and IL-2 produced by the relatively large population of CD44^hi T_conv cells in these animals. Indeed, even when comparing cells of identical TCR specificity, T_reg cells have a substantial proliferative/survival advantage over their T_conv counterparts. Moreover, the large percentage of Foxp3⁺ T cells in SFZ70 mice suggests that the scurfy T cell from which we isolated the SFZ70 TCR may have been a T_reg cell unable to express functional Foxp3. Activated T cells from Foxp3-deficient mice express similar TCRs to those found in the T_reg cells from WT mice (37). These ‘would-be’ T_reg cells share many traits with bona fide T_reg cells, but comprise a distinct peripheral T cell population that lacks functional suppressive activity and is substantially expanded in Foxp3-deficient animals (38, 39).

T_reg cells are not phenotypically and functionally homogenous, but instead can be divided into distinct subsets based activation marker and homing receptor expression. CD44^lo T_reg cells share characteristics with naïve T_conv cells including a low level of homeostatic proliferation and a CD62L^highCD103⁻ phenotype that facilitates their entry into lymphoid tissues. By contrast, CD44^hi T_reg cells resemble effector/memory T_conv cells and are characterized by low CD62L expression, increased expression of homing receptors such as CD103 and E/P-lig that target cells to non-lymphoid sites, and robust homeostatic proliferation (40–42). Although the molecular basis for this diversity is still poorly understood, the uniform effector/memory phenotype of SFZ70 T_reg cells indicates that TCR specificity strongly influences T_reg cell migration, homeostasis and function.

Effector/memory T_reg cells display a high suppressive capability under inflammatory conditions, but the mechanisms and targets of T_reg cell-mediated suppression remain largely unknown. Our finding that an increased percentage of T_conv cells express peripheral homing receptors and produce inflammatory cytokines in T_reg cell-deficient SFZ70 mice argues that T_reg cells function at least in part by limiting trafficking and cytokine production by activated autoreactive T_conv cells. Given the known ability of DCs to imprint T cells with specific patterns of cytokine and homing receptor expression (43), our data indicate that SFZ70 T_reg cells may control T_conv cells indirectly by acting on DCs. However, strong activating stimuli can overcome T_reg cell suppression of DCs (44), and we found that DC activation with an agonistic anti-CD40 mAb resulted in increased effector cytokine production by SFZ70 T_conv cells. Thus, even antigen-matched T_reg cells are not sufficient to inhibit acquisition of effector function by autoreactive T_conv cells in the during periods of robust innate immune stimulation. This suggests that during the course of an infection with a strong innate stimulatory component, autoreactive T_conv cells may override T_reg cell-mediated suppression, potentially contributing to the development of autoimmune disease. Indeed, dysregulation of DC activity can result in spontaneous autoimmunity even in the presence of T_reg cells (45).

T_reg cells have typically been studied as a bulk population composed of cells differing in TCR specificity, activation status, and origin. Thus, although T_reg cells are thought to be largely self-reactive, how their specificity contributes to their unique phenotype, function and homeostasis has been difficult to examine experimentally. The SFZ70 mouse has allowed us to distinguish differences in proliferation, tissue-homing receptor expression, and effector function between T_reg and T_conv cells expressing the same TCR. Given that the SFZ70 mice allow the study of T_reg cells as a monospecific population, they are uniquely poised elucidate the underlying mechanisms involved in the generation and maintenance of T_reg cell-mediated immune tolerance in the skin.