|Home | About | Journals | Submit | Contact Us | Français|
Whereas antigen recognition mediated by the T cell receptor (TCR) influences many facets of Foxp3+ regulatory T (Treg) cell biology, including development and function, the cell types that present antigen to Treg cells in vivo remain largely undefined. By tracking a clonal population of Aire-dependent, prostate-specific Treg cells in mice, we demonstrated an essential role for dendritic cells (DCs) in regulating organ-specific Treg cell biology. We have shown that the thymic development of prostate-specific Treg cells required antigen presentation by DCs. Moreover, Batf3-dependent CD8α+ DCs were dispensable for the development of this clonotype and had negligible impact on the polyclonal Treg cell repertoire. In the periphery, CCR7-dependent migratory DCs coordinated the activation of organ-specific Treg cells in the prostate-draining lymph nodes. Our results demonstrate that the development and peripheral regulation of organ-specific Treg cells are dependent on antigen presentation by DCs, implicating DCs as key mediators of organ-specific immune tolerance.
CD4+Foxp3+ regulatory T (Treg) cells are required throughout life for the maintenance of immune homeostasis and the prevention of autoimmunity (Kim et al., 2007; Sakaguchi et al., 2008). A growing body of evidence suggests that the Treg cell repertoire contains organ-specific Treg cells reactive to tissue-restricted self antigens, and that these cells may be critical for the protection of organs from autoimmune attack. T cell receptor (TCR) profiling reveals an asymmetric distribution of Treg cell specificities in lymph nodes throughout the body, suggesting that the anatomical distribution of Treg cells is shaped by the presentation of regional organ-specific antigens (Lathrop et al., 2008). In transfer experiments, Treg cells isolated from organ-draining lymph nodes are more efficient than Treg cells from non-draining lymph nodes at suppressing organ-specific autoimmunity at multiple target sites (Samy et al., 2005; Setiady et al., 2006; Wheeler et al., 2009). Recently, we identified a clonal population of prostate-specific Treg cells, named “MJ23” Treg cells, that are enriched in the prostate-draining lymph nodes and prostate tumor lesions of male mice (Malchow et al., 2013), providing direct evidence of the existence of organ-specific Treg cells.
For some time, it has remained unclear whether organ-specific Treg cells originate in the thymus or develop extrathymically upon encounter with tissue-derived antigens. Recently, we and others demonstrated that the thymic development of some naturally occurring Treg cell specificities, including prostate-specific MJ23 Treg cells, is dependent on the expression of Autoimmune Regulator (Aire) (Malchow et al., 2013; Perry et al., 2014). Aire is a transcription factor expressed by medullary thymic epithelial cells (mTECs) that promotes the promiscuous expression of a variety of genes, many of which encode peripheral tissue-restricted antigens (TRAs) (Anderson et al., 2002; Derbinski et al., 2005). These findings provide mechanistic clarity, demonstrating that Aire-dependent mirroring of the “peripheral self” (Anderson et al., 2002) plays a critical role in promoting the thymic development of Treg cells reactive to peripheral organ-specific antigens.
Substantial evidence indicates that in vivo, the development, differentiation, and function of Treg cells are dependent on TCR-mediated antigen recognition. In the thymus, Treg cell development is dependent on MHC-II-restricted antigen recognition, and occurs via a TCR-instructive process in which distinct TCR specificities facilitate efficient differentiation into the Treg cell lineage (Bautista et al., 2009). In the periphery, a substantial proportion of Treg cells perceive TCR signals and actively divide at steady state (Fisson et al., 2003; Smigiel et al., 2014). Moreover, conditional ablation of the TCR on Foxp3-expressing cells disrupts Treg cell function, resulting in systemic autoimmunity (Levine et al., 2014; Vahl et al., 2014), demonstrating a crucial role for TCR-dependent signals in Treg cell-mediated immune tolerance. Thus, antigen recognition is critical for many aspects of the life cycle and lifestyle of Treg cells. Despite the importance of TCR engagement, little is known about the cell types that present antigen for recognition by Treg cells. These antigen presenting cells (APCs) are likely to orchestrate many aspects of Treg cell biology, providing antigenic and accessory signals that dictate Treg cell activation, anatomical distribution, spatial positioning within lymphoid and non-lymphoid organs (Gerner et al., 2012), and access to distinct environmental cues. Thus, identifying the cellular interaction partners that interface with Treg cells in the thymus and periphery is critical for gaining a complete understanding of Treg cell biology and the mechanisms by which immune tolerance is established and maintained.
Antigen presentation by both mTECs and dendritic cells (DCs) is required to establish a replete Treg cell pool, suggesting that these presentation pathways collaborate to form a complete Treg cell repertoire (Klein et al., 2009; Perry et al., 2014; Proietto et al., 2008). For the development of Aire-dependent Treg cells in the thymus, two models have been proposed to describe the presentation of antigens encoded by Aire-dependent transcripts (Klein et al., 2009). In the first model, Aire-expressing mTECs function in a cell-autonomous fashion to directly present Aire-dependent antigens to thymocytes. In the second model, antigens expressed by mTECs are transferred to neighboring DCs, which then present acquired antigen to coordinate Treg cell differentiation. In this regard, the presence of multiple DC subsets in the thymic medulla, including plasmacytoid DCs, CD8α+ classical DCs (cDCs) and Sirpα+ cDCs (Guerri et al., 2013; Stritesky et al., 2013), raises the possibility that a distinct DC subset may specialize in presenting Aire-dependent antigens to developing T cells.
Perhaps even less is known about the APCs that present antigen to Treg cells outside of the thymus, and the impact of these APCs on the peripheral homeostasis and function of Treg cells. In one study, it was demonstrated that following the depletion of CD11c-expressing cells or expansion of Flt3L-dependent DCs, the frequency of splenic Treg cells decreases or increases, respectively, suggesting that the homeostasis of some splenic Treg cells is orchestrated by CD11c+ cells (Darrasse-Jeze et al., 2009). In a separate study, it was shown that expansion of Flt3L-dependent DCs or administration of anti-ICOSL antibody induces the selective expansion or reduction, respectively, of CD44highCD62Llow splenic Treg cells (Smigiel et al., 2014). Since the spleen contains multiple subsets of CD11c-expressing cells (Mildner and Jung, 2014), it is unknown whether a distinct DC subset supports Treg cell homeostasis at this site. Furthermore, it is unclear whether the same dependency on CD11c+ cells is operative in the lymph nodes, which contain various APC populations that are not found in the spleen (Mildner and Jung, 2014). For the presentation of organ-derived antigens to organ-specific Treg cells in the draining lymph nodes, at least two possible scenarios can be envisioned (Itano et al., 2003). In the first scenario, lymph node resident DCs or macrophages capture organ-derived antigen draining to the lymph nodes via the afferent lymphatics, and present this antigen to organ-specific Treg cells. In an alternate but not mutually exclusive scenario, DCs acquire antigen within the organ, migrate to the draining lymph nodes, and process and present antigen for recognition by organ-specific Treg cells (Scheinecker et al., 2002).
Here, we demonstrate that DCs play a critical role in the thymic development and peripheral homeostasis of prostate-specific Treg cells. Our findings identify the cellular interaction partners of an archetypal population of Aire-dependent, organ-specific Treg cells, providing a conceptual framework for the broader understanding of organ-specific Treg cells of diverse specificities, and the APC types that choreograph the establishment and enforcement of organ-specific immune tolerance.
In this study, we used two approaches to characterize the APCs required for the development and homeostasis of MJ23 Treg cells, a clonal population of Aire-dependent, prostate-specific Treg cells that has been identified previously (Malchow et al., 2013). In the first approach, we generated bone marrow chimeric mice (BMCs) in which irradiated CD45.2+ hosts were reconstituted with donors cells of defined genotype, including a low frequency of bone marrow cells from CD45.1+ MJ23 TCR transgenic (MJ23Tg) mice on the Rag1−/− background. In the second approach, we utilized intrathymic injection to introduce CD45.1+ MJ23Tg Rag1−/− thymocytes, which are devoid of Foxp3+ Treg cells (Malchow et al., 2013), into CD45.2+ recipients. Both approaches enabled us to track the development, anatomical distribution, and activation of MJ23 T cells at low clonal frequencies post-transfer or engraftment.
Treg cell development in the thymus requires both MHC-II-dependent TCR stimulation and CD80 or CD86-dependent co-stimulation via the CD28 receptor (Hsieh et al., 2012). In addition, Aire is required for the thymic development of prostate-specific MJ23 Treg cells, implying that Aire-expressing mTECs are the most likely source of MJ23 antigen in the thymus (Malchow et al., 2013). Using BMCs, the role of bone marrow-derived APCs and radioresistant host APCs (thymic epithelial cells, TECs) in coordinating Treg cell development could be assessed. We generated lethally irradiated BMCs in which wild-type, Aire−/−, or Cd80−/− Cd86−/− host mice were reconstituted with bone marrow from wild-type mice, Cd80−/−Cd86−/− mice, or H2dlAb-Ea homozygous mice deficient in all MHC-II genes (hereafter referred to as “MHC-II deficient” mice), and MJ23 Treg cell development was assessed. Given that positive selection of CD4+ T cells is substantially impaired in MHC-II null hosts (Markowitz et al., 1993) due to a lack of MHC-II on cortical TECs, we did not analyze MJ23 Treg cell development in MHC-II-deficient hosts. Instead, we assessed the role for TECs in Treg cell development using Cd80−/−Cd86−/− hosts lacking both CD80 and CD86, which are required for Treg cell development (Hsieh et al., 2012; Salomon et al., 2000) but do not impact positive selection at earlier stages of development. We found that MJ23 Treg cells did not develop in BMCs reconstituted with MHC-II deficient or Cd80−/−Cd86−/− bone marrow, but developed efficiently in BMCs generated in Cd80−/−Cd86−/− hosts (Figures 1A, 1B and S1A). As expected, Aire expression by radioresistant host cells was crucial for MJ23 Treg cell development (Figure 1B), confirming previous findings (Malchow et al., 2013). These results suggest that MJ23 antigen is transferred from Aire-expressing mTECs to bone marrow derived APCs, which coordinate MJ23 Treg cell development by the provision of MHC-II-restricted antigen and co-stimulatory signals through CD28. Analysis of non-MJ23Tg polyclonal Treg cells revealed a similar pattern, in which BMCs reconstituted with MHC-II deficient or Cd80−/−Cd86−/− bone marrow exhibited a partial reduction of Treg cells (Figures 1A, 1C and S1B), suggesting that these requirements may apply to a substantial fraction of the polyclonal Treg cell repertoire. Additionally, deficiency of CD80 and CD86 on radioresistant TECs led to a marked increase in the total number of polyclonal Treg cells in the thymus (Figure S1B), suggesting a potential role for CD80 and/or CD86 expression by TECs in the negative selection of some specificities.
In order to identify the bone marrow-derived APCs that coordinate MJ23 Treg cell development, we utilized models of inducible or constitutive APC deficiency. We assessed the requirement for antigen presentation by CD11c-expressing DCs, using both intrathymic injection and BMC approaches. First, we performed intrathymic injection of CD45.1+ MJ23Tg Rag1−/− thymocytes into Itgax-Cre+ (also known as CD11c-Cre) Ab1flox/−− mice, which exhibit conditional deletion of the I-Ab MHC-II molecule on CD11c-expressing cells (Figure 2A). MJ23 Treg cell development failed to occur in Itgax-Cre+ Ab1flox/− hosts, but did occur in littermate controls lacking the Itgax-Cre transgene (Figures 2B, 2C and S2A). MJ23 Treg cell development was lost in Itgax-Cre+ Ab1flox/− mice despite detectable amounts of MHC-II remaining on the surface of CD11c+ cells (Figure 2A), which may reflect the acquisition of MHC-II molecules from mTECs (Koble and Kyewski, 2009). Second, we generated BMCs in which lethally irradiated wild-type hosts were reconstituted with a mixture of Itgax-DTR+ (also known as CD11c-DTR) and MHC-II deficient bone marrow, together with a low frequency of CD45.1+ MJ23Tg Rag1−/− bone marrow. Itgax-DTR+ mice express the human diphtheria toxin receptor (DTR) under the control of the Itgax promoter (Jung et al., 2002), permitting the inducible ablation of CD11c-expressing cells following administration of diphtheria toxin (DT). In these chimeras, DT administration induced the depletion of >95% of MHC-II-expressing DCs (Figure S2C), allowing us to assess the role of antigen presentation by CD11c-expressing cells. Our data revealed that MJ23 Treg cell development was abrogated in DT-treated chimeras (Figures S2D and S2E), demonstrating that MHC-II expression by CD11c+ DCs is required for MJ23 Treg cell development. In control experiments, injection of DT alone into MJ23Tg BMCs did not impact the number of MJ23Tg or polyclonal Treg cells in the thymus or periphery (Figures S2G-S2L). Loss of MHC-II on CD11c-expressing cells in either context (genetically or by DT administration) resulted in a reduction in the frequency of polyclonal thymic Treg cells (Figures 2B, 2D, S2D, and S2F), suggesting that the development of a substantial fraction of Treg cells is dependent on antigen presentation by DCs. Finally, while recent reports demonstrate that a subpopulation of thymic B cells exhibits Aire expression (Yamano et al., 2015) and may impact T cell selection (Perera et al., 2013; Yamano et al., 2015), MJ23 Treg cell development and polyclonal Treg cell frequencies were not impacted by B cell deficiency in Ighm−/− mice (Figures 2C, 2D, S3B, and S3C). Taken together, our results demonstrate that MHC-II-restricted antigen presentation by CD11c+ DCs is required to support MJ23 Treg cell development in the thymus.
The thymus contains three major classes of DCs, including plasmacytoid DCs (pDCs), Sirpα+ cDCs, and CD8α+ cDCs (Klein et al., 2014; Mildner and Jung, 2014). To assess the role of pDCs in Treg cell development, we utilized CLEC4C-DTR+ (also known as BDCA2-DTR) mice, which express DTR exclusively in pDCs (Swiecki et al., 2010). In experiments in which MJ23Tg thymocytes were injected intrathymically into CLEC4C-DTR+ mice (Figures 3A-3D, S3D, and S3E), and in BMCs in which MJ23Tg bone marrow was introduced into CLEC4C-DTR+ hosts (Figures S3F and S3G), DT-mediated ablation of pDCs did not impact the development of MJ23 or polyclonal Treg cells, indicating that pDCs are dispensable for the development of these cells.
Batf3−/− mice exhibit a deficiency of CD8α+ cDCs in the thymus (Figures 3E, S3H, and S3I) and secondary lymphoid organs (Hildner et al., 2008), which allowed us to test the requirement for CD8α+ cDCs in orchestrating Treg cell development. In mice injected intrathymically with MJ23Tg thymocytes and in MJ23Tg BMCs, MJ23 Treg cells developed efficiently in Batf3−/− hosts (Figures 3F, 3G, S3J, and S3M), indicating that CD8α+ cDCs are dispensable for MJ23 Treg cell development. Thus, our data showing a requirement for antigen presentation by DCs (Figures 2 and and3),3), coupled with our data demonstrating that pDCs and CD8α+ cDCs are dispensable for this process (Figure 3), suggest either functional redundancy of multiple DC subsets, or a critical role for Batf3-independent Sirpα+ cDCs in mediating MJ23 Treg cell development. It is currently not possible to test a specific requirement for thymic Sirpα+ cDCs in vivo due to a lack of available models. These findings are inconsistent with the conclusions of Perry and colleagues (Perry et al., 2014), who suggest that CD8α+ cDCs play a specialized role in coordinating the thymic development of Aire-dependent Treg cells. To determine whether our observed results are applicable to other Aire-dependent Treg cell specificities, we evaluated the thymic development of a second Aire-dependent Treg cell clone named RT83 (Malchow et al., 2013), which does not exhibit reactivity to a prostate-associated antigen. Using a combination of intrathymic injection and BMC approaches, we found that the thymic development of RT83Tg Treg cells was abrogated in MHC-II deficient or Aire−/− hosts, but occurred efficiently in hosts lacking B cells, pDCs, or CD8α+ cDCs (Figure S4).
To broaden our analysis, we assessed the impact of Batf3 deficiency on the polyclonal Treg cell repertoire using a deep TCR sequencing approach (Malchow et al., 2013). Briefly, we isolated Treg cells from the thymus of male Batf3+/+ or Batf3−/− mice expressing a fixed transgenic TCRβ chain and a Foxp3GFP reporter, and performed sequencing of the complete TCRα repertoire, regardless of variable-region usage (see Table S1). Presence or loss of CD8α+ cDCs in the spleen was confirmed for all mice analyzed (data not shown). On average, we isolated approximately 1.0 × 105 Treg cells from each thymus sample, yielding 1.0 × 106 TCR sequence reads per sample with a complexity of 9 × 103 unique TCRs, allowing us to assess the broad impact of Batf3 on the selection of thousands of Treg cell specificities.
Comparison of these Treg cell TCR catalogs revealed that Batf3 deficiency had negligible impact on the Treg cell repertoire in the thymus (Figure 4). Of the 8,423 unique TCR clonotypes recurrently expressed by Treg cells in the thymus of either Batf3+/+ or Batf3−/− mice, differential analysis of TCR frequency revealed a lack of Batf3-dependent specificities that were significantly underrepresented or absent in Batf3−/− thymi (Figure 4A, right arm of volcano plot). This finding was further illustrated by plotting the frequencies of the most prevalent thymic Treg cell specificities (Figure 4B), which demonstrated that the frequency of these clonotypes was not significantly altered by Batf3 deficiency (when corrected for multiple comparisons). Analysis using the Morisita-Horn (MH) similarity index (Magurran, 1988) also revealed a high degree of similarity between the thymic Treg cell TCR catalogs from Batf3+/+ and Batf3−/− mice (Figure 4C). Finally, unlike Aire−/− mice that display diminished Treg cell frequencies in the thymus (Figure 2D and (Anderson et al., 2005)), Batf3−/− mice exhibited elevated numbers and frequencies of polyclonal Treg cells in the thymus (Figures 3H S3K, and S3N), supporting the idea that Batf3-deficiency is not associated with a paucity of Treg cells. Our data suggest that Batf3-dependent CD8α+ cDCs do not play a specialized role in orchestrating Treg cell development.
To further examine the functional role of distinct thymic DC subsets in coordinating MJ23 Treg cell development, we performed in vitro assays in which fluorescence activated cell sorting (FACS)-purified thymic CD8α+ or Sirpα+ cDCs were cultured with MJ23Tg thymocytes. Both thymic cDC subsets induced proliferation, CD69 expression, and Foxp3 induction by a fraction of MJ23Tg thymocytes (Figures S5A-S5C). This induction was blocked by addition of anti-MHC-II antibody, and was not conferred by co-culture with splenic CD11c+ cells (Figures S5A-S5C). Furthermore, when cultured with exogenous prostate-tissue extract, both CD8α+ and Sirpα+ cDCs were able to process and present antigen, inducing robust stimulation and Foxp3 upregulation (Figures S5D-S5F). Thus, our cumulative data suggest that both cDC subsets are capable of acquiring antigen from mTECs and promoting MJ23 Treg cell development in vitro, suggesting possible functional redundancy of these DC subsets in vivo.
Having identified a role for DCs in coordinating the thymic development of MJ23 Treg cells, we next aimed to elucidate the role of antigen recognition in dictating the peripheral homeostasis and anatomical distribution of MJ23 Treg cells. Previously, we demonstrated that MJ23 Treg cells are selectively enriched in the prostate-draining periaortic lymph nodes (pLNs) of male mice (Malchow et al., 2013). Phenotypic analysis of MJ23Tg BMC mice revealed that MJ23 Treg cells in the pLNs of male mice expressed markers indicative of recent T cell activation, including CD69 and Egr2 (Figures 5A and 5B). In contrast, MJ23 Treg cells in female hosts or in the skin-draining lymph nodes and spleen of male hosts did not express these markers (Figures 5A and 5B).
To test the hypothesis that MJ23 Treg cell enrichment in the pLNs is dependent on continuous antigen recognition, we performed studies in which MJ23Tg BMCs were subjected to castration. We reasoned that castration, which leads to rapid involution of the prostate (Figure 5C and (Kyprianou and Isaacs, 1988)), would reduce prostate antigen density, thereby impacting MJ23 Treg cell enrichment. Thus, we assessed the frequency, anatomical distribution, and activation status of MJ23Tg Treg cells in castrated mice, sham castrated mice, or mice subjected to castration followed by testosterone supplementation, a procedure that leads to complete regeneration of the prostate (Isaacs, 1987). While MJ23 Treg cells in sham castrated mice expressed high amounts of CD69 and exhibited the expected enrichment in the prostate-draining pLNs, enrichment and activation of MJ23 Treg cells was lost upon castration and involution of the prostate (Figures 5D-5F). Finally, the enrichment and activation of MJ23 Treg cells in the pLNs was restored in castrated mice subjected to testosterone supplementation (Figures 5D-5F). In sum, although antigen-independent hormonal effects of castration can not be excluded, these findings suggest that the enrichment of prostate-specific MJ23 Treg cells in the pLNs requires the continued presence of antigen, and that antigen-driven enrichment in the organ-draining lymph nodes is reversible.
To determine the role of DCs in coordinating antigen-driven MJ23 Treg cell enrichment in the pLNs, we analyzed MJ23 Treg cell distribution in the periphery of BMCs reconstituted with a mixture of MJ23Tg, Itgax-DTR+, and MHC-II deficient bone marrow, described above. In these mice, DT-mediated ablation of MHC-II+ CD11c-expressing cells (Figure 6A) led to a significant reduction of MJ23 Treg cell frequency and total number in the pLNs (Figures 6B and S6A), indicating that MHC-II expression by CD11c+ DCs is required for optimal enrichment of MJ23 Treg cells in the pLNs. At the polyclonal level, DC depletion also induced a significant reduction of Treg cell frequency in the lymph nodes and spleen (Figure 6C), suggesting that a fraction of polyclonal Treg cells at these sites are dependent on antigen presentation by DCs. In other experiments, we determined the impact of systemic DC expansion (Figure 6D) induced by subcutaneous challenge with B16 melanoma cells producing Flt3L (Curran and Allison, 2009). B16-Flt3L tumor challenge induced significant increases in polyclonal Treg cell frequency in the spleen and inguinal & brachial lymph nodes, but did not induce further enrichment of MJ23 Treg cells in the pLNs (Figures 6E, 6F and S6D) or thymus (Figure S6H). Furthermore, as previously described for the spleen (Darrasse-Jeze et al., 2009), we found that polyclonal Treg cell frequencies were positively correlated with DC abundance in the lymph nodes following DC depletion or expansion (Figures S6C and S6F). These findings suggest that MJ23 enrichment in the pLNs is limited by antigen availability, and is not further augmented by the systemic or local expansion of Flt3L-responsive DC populations.
Given the emerging appreciation that DCs exhibit considerable functional and phenotypic heterogeneity (Haniffa et al., 2013; Hashimoto et al., 2011), we set out to determine whether a distinct DC subset is responsible for antigen presentation to MJ23 Treg cells. Lymph node mDCs, which are characterized by an MHC-IIhiCD11cint phenotype (Figure 7A), require CCR7-dependent signals to traffic from the tissues into the draining lymph nodes (Randolph et al., 2008). Analysis of DCs in mouse prostate tissue revealed that both CD103+ and CD11b+ cDCs are present at steady state, and that these populations do not express CCR7 over background (Figures S7A and S7B). In contrast, mDCs in the pLNs exhibit a CCR7+ phenotype (Figures S7C and S7D), suggesting that mDCs upregulate CCR7 prior to or during transit from the tissues to the draining lymph nodes. To determine the extent to which mDCs are required for MJ23 Treg cell accumulation in the pLNs, we evaluated the anatomical distribution and phenotype of MJ23 Treg cells following intrathymic injection of CD45.1+ MJ23Tg Rag1−/− thymocytes into Ccr7+/+ or Ccr7−/− mice. In Ccr7+/+ hosts, MJ23 Treg cells developed efficiently (Figures S7E-S7G), preferentially accumulated in the pLNs, and adopted a CD69+ phenotype indicative of TCR signaling (Figures 7B-7D). In contrast, in Ccr7−/− mice, MJ23 Treg cells developed efficiently in the thymus (Figures S7E-S7G), but failed to accumulate in the pLNs and did not exhibit hallmarks of TCR signaling (Figures 7B-7D). Notably, this effect was specific to the pLNs, as MJ23Tg Treg cell frequency was not reduced in the non-draining lymph nodes of Ccr7−/− mice. In parallel experiments using RT83Tg T cells, which are not reactive to a prostate-specific antigen, the anatomical distribution of RT83 Treg cells did not differ between Ccr7+/+ and Ccr7−/− hosts (Figure 7E). These findings suggest that CCR7-dependent mDCs are required for the enrichment of MJ23 Treg cells in the pLNs of male mice, implying that this DC subset may play a critical role in coordinating the peripheral homeostasis of prostate-specific Treg cells, thereby promoting immune tolerance to this anatomical site.
In this study, we identified the APCs that coordinate the thymic development and peripheral activation of a naturally occurring specificity of organ-specific Treg cells. Our results support a model in which DCs in the thymic medulla acquire promiscuously expressed prostate antigen from Aire+ mTECs, and promote the differentiation of MJ23 precursors into the Treg cell lineage. In the periphery, our results are consistent with a model in which DCs acquire antigen within the prostate, migrate to the draining lymph nodes via a CCR7-dependent process, and present antigen for recognition by MJ23 Treg cells, inducing TCR signaling and enrichment in the prostate-draining lymph nodes. Our data implicate DCs as key mediators of immune tolerance, choreographing the thymic development and peripheral homeostasis of organ-specific Treg cells. Moving forward, it will be important to determine the extent to which these principles apply to organ-specific Treg cells of diverse specificities, and the molecular and cellular properties that endow DCs with the capacity to coordinate Treg cell biology.
Our findings are consistent with work involving the study of “disease-specific” polyclonal Treg cells by Tung and colleagues (Samy et al., 2005; Setiady et al., 2006; Wheeler et al., 2009). In a representative study, it has been shown that the local enrichment of Treg cells that possess the capacity to suppress experimental autoimmune prostatitis is eliminated by neonatal orchiectomy and restored by subsequent 5α-dihydrotestosterone treatment, suggesting the existence of prostate-specific Treg cells that are dependent on antigen availability (Setiady et al., 2006). Our current study extends these findings by providing direct evidence that TCR signaling and anatomical enrichment of a naturally occurring organ-specific Treg cell population requires sustained antigen presentation and is reversible.
DCs exhibit considerable phenotypic, developmental, and functional diversity, suggesting that distinct DC subsets may play unique roles in regulating tolerance, immunity, and inflammation. Based on our finding that CCR7-dependent mDCs are required for the enrichment of MJ23 Treg cells in the pLNs, we hypothesize that mDCs bearing organ-derived antigen are specialized to interface with organ-specific Treg cells in lymph nodes throughout the body. Idoyaga and colleagues have demonstrated that the experimental targeting of antigen to mDCs, delivered by anti-DEC-205 or anti-Langerin antibodies, induces the peripheral differentiation of antigen-specific T cells into Foxp3+ cells (Idoyaga et al., 2013). While the extrathymic Treg cell induction observed by Idoyaga et al. is inherently different from the enrichment of thymic-derived organ-specific Treg cells demonstrated in our study, together these findings suggest that mDCs are endowed with the capacity to support the activation, enrichment, and differentiation of Treg cells within the lymph nodes. Given the heterogeneity of mDCs in the lymph nodes (Randolph et al., 2008), it will be important to determine whether antigen presentation to organ-specific Treg cells is a general property of all mDC subsets, or whether a distinct mDC subset is specialized to interface with these Treg cells. For example, the findings of Gerner et al. suggest that different mDC subsets are localized to distinct regions within the lymph node microenvironment (Gerner et al., 2012). mDC localization may be critical for the proper positioning of organ-specific Treg cells in order to place them in direct proximity to conventional T cells or other environmental niches.
The thymus contains a diverse array of MHC-II-expressing APCs, including TECs, B cells, macrophages, and multiple DC subsets (Klein et al., 2014; Mildner and Jung, 2014). Here, we have demonstrated that MHC-II-restricted antigen presentation by DCs was required for the thymic development of prostate-specific MJ23 Treg cells. Furthermore, we have demonstrated that pDCs and Batf3-dependent CD8α+ cDCs are dispensable for the development of two distinct Aire-dependent Treg cell specificities, MJ23 and RT83, and that CD8α+ cDCs are dispensable for the generation of a fully diverse polyclonal Treg cell repertoire. Cumulatively, our results are consistent with the idea that there is considerable functional redundancy amongst thymic DC subsets in the capacity to promote the thymic development of Treg cells reactive to Aire-dependent antigens. These findings are in agreement with previous work demonstrating that both Sirpα+ and CD8α+ cDCs are capable of acquiring antigen from mTECs in vivo (Koble and Kyewski, 2009; Perry et al., 2014), and in vitro studies showing that both Sirpα+ and CD8α+ cDC subsets are capable of supporting Treg cell development of non-Treg cell TCR transgenics upon provision of exogenous model antigens (Guerri et al., 2013; Proietto et al., 2008; Wirnsberger et al., 2009).
Our findings that Batf3-dependent CD8α+ cDCs are largely dispensable for Aire-dependent Treg cell development are incongruous with the conclusions of Perry and colleagues, who suggest that CD8α+ cDCs play a specialized role in coordinating the development of Aire-dependent Treg cells (Perry et al., 2014). In that study, the authors demonstrated that the thymic development of four Aire-dependent Treg cell clones is abolished in Batf3−/− hosts. However, these four clones differ by a single amino acid substitution, suggesting that they may be reactive to a single antigen. In contrast, we present data demonstrating that Batf3-dependent CD8α+ cDCs are dispensable for the development of two well-characterized Aire-dependent Treg cell clones. In addition, to bring further clarity to this discrepancy, we performed deep TCR sequence analysis of polyclonal Treg cells in the thymus of Batf3+/+ versus Batf3−/− mice. We found that Batf3 deficiency had negligible impact on the >8,000 recurrent Treg cell TCR clonotypes identified in the analysis, indicating that CD8α+ cDCs are dispensable for the formation of a complete Treg cell repertoire.
While the results discussed here suggest that there may be functional redundancy in the capacity of thymic DC subsets to coordinate the development of Aire-dependent Treg cell specificities, it remains formally possible that Sirpα+ cDCs play a unique role in this process. Moving forward, it will be critical to develop approaches that enable the inducible or constitutive ablation of Sirpα+ cDCs, in an effort to elucidate the role of these cells in T cell selection. Such studies may help to elucidate the mechanisms and pathways that are ultimately responsible for dictating whether an autoreactive thymocyte undergoes clonal deletion or differentiation into the Treg cell lineage.
All mice were on the C57BL/6J (B6) background and maintained in accordance with the animal care and use regulations of the University of Chicago. The following mouse strains were purchased from the Jackson Laboratory and maintained in our facility: C57BL/6J (B6), CD45.1+ B6.SJL-Ptprca Pepcb/BoyJ, Aire−/− B6.129S2-Airetm1.1Doi/J, Foxp3GFP B6.Cg-Foxp3tm2Tch/J, Rag1−/− B6.129S7-Rag1tm1Mom/J, Itgax-DTR B6.FVB-Tg(Itgax-DTR/EGFP) 57Lan/J, BDCA2-DTR C57BL/6-Tg(Clec4b1-HBEGF)956Cln/J, MHC-II deficient B6.129S2-H2dlAb1-Ea/J, Cd80−/−Cd86−/− B6.129S4-Cd80tm1ShrCd86tm2Shr/J, Ccr7−/− B6.129P2(C)-Ccr7tm1Rfor/J, Itgax-Cre B6.Cg-Tg(Itgax-cre)1-1Reiz/J, Ab1flox/flox B6.129X1-H2-Ab1tm1Koni/J, Ab1−/− B6.129-H2-Ab1tm1Gru/Tac, Batf3−/− B6.129S(C)-Batf3tm1Kmm/J, and Ighm−/− B6.129S2-Ighmtm1Cgn/J. MJ23Tg, RT83Tg, and “TCRβTg” mice were generated as described previously (Malchow et al., 2013).
5 × 106 T cell-depleted (via CD90.2 MACS beads (Miltenyi)) bone marrow cells were retro-orbitally injected into sublethally (500 rads) or lethally (1100 rads) irradiated host mice. Mixed donor cells typically consisted of a low-frequency (5%) of TCRTg bone marrow and polyclonal “filler” cells, all of which taken from female donor mice. Unless otherwise noted, all mice were analyzed at 6 weeks post-engraftment.
Thymocytes were harvested from 4-6 week old female TCRTg donor mice and depleted of dendritic cells using CD11c MACS beads (Miltenyi). 4 × 106 cells were then intra-thymically injected into 4-6 week old hosts and analyzed at either seven days, to evaluate thymic development, or three weeks, to evaluate the anatomical distribution and activation of TCRTg cells.
Sublethally irradiated MJ23Tg bone marrow chimeras were generated as described above. 6 weeks post-engraftment, mice were anesthetized with ketamine & xylazine and either surgically castrated via removal of the testes or sham castrated. Four weeks post-procedure, testosterone was restored in half of the castrated mice via subcutaneous implantation of a testosterone slow-release capsule. All mice were analyzed four weeks post-implantation.
For most analyses involving APC subsets, lymphoid organs were injected and digested with Liberase TL (400 μg/mL, Roche) and DNase (800 μg/mL, Roche) in RPMI for 30 min at 37°C. When isolating APCs from the thymus, EDTA (10 mM) was added to digests and enriched by layering digested thymocytes on top of a discontinuous Percoll gradient (GE Healthcare) at 1.115 g/mL in PBS, followed by centrifugation at 1350 g for 30 min and isolation of cells settling at the Percoll interface (Aschenbrenner et al., 2007).
All antibodies were purchased from eBioscience, Biolegend, BD Biosciences or Miltenyi Biotec. Flow cytometry was performed on either an LSR-II or an LSRFortessa flow cytometer (BD Biosciences), using FlowJo data analysis software (Tree Star). Fluorescence-activated cell sorting was performed using a FACSAria (BD Biosciences).
Data were analyzed using Prism software (GraphPad). For the comparison of two groups, the Student's t-test (two-tailed) or the nonparametric Mann-Whitney test were used, depending on whether data were normally distributed. For comparison of multiple groups, one-way ANOVA coupled with Tukey's multiple comparison test was employed.
Thymic CD4+Foxp3+ Treg cells were flow cytometry-purified from 9-week-old TCRβ transgenic Foxp3GFP males on a Batf3+/+ or Batf3−/− background. RNA from flow cytometry-sorted T cell subsets was subjected to TCRα sequencing using the Amp2Seq service from iRepertoire, a platform based on semi-quantitative multiplex PCR coupled with Illumina sequencing that allows analysis of the complete TCRα repertoire, regardless of variable-region usage. TCRs were analyzed solely based on the predicted CDR3 sequence, regardless of V-region usage, using edgeR from the RvBioconductor package. We first filtered for TCRs with CDR3 segments between 7-17 amino acids in length and in order to focus on recurrent TCRs we removed those TCRs with counts that were less than 10 in Ns-1 samples were Ns was the size of the smallest group. Samples were then normalized using the standard edgeR normalization routine (calcNormFactors) and p-values were computed using the edgeR standard test. P-values were corrected using the FDR method with a standard threshold of 0.05. To enable fold-change calculations, a pseudocount approach was used. To analyze TCR repertoire similarity, we also analyzed CDR3 elements using the Morisita-Horn similarity index (Magurran, 1988).
We thank Bana Jabri for critical reading of the manuscript. We thank Hannah Brechka, Jacob Kach, Yi Cai, and Zheng Zhong for technical assistance. We thank Dengping Yin and the University of Chicago Microsurgery Core Facility for their assistance with intrathymic injections. This work was funded by the following sources (to P.A.S.): R01 (R01CA160371), R01 (R01AI110507), a Cancer Research Institute Investigator Award, and the University of Chicago Comprehensive Cancer Center. J.K. was supported by R01 (R01CA166770). D.S.L. was supported by an NIH/NCI F31 predoctoral fellowship (CA183357). N.D.S. was supported by MSKCC Comprehensive Cancer Center (Support Grant #P30 CA008748).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
D.S.L. and P.A.S. conceived the project, designed experiments, and performed data analysis. D.S.L., S.M., D.C.G., J.M.B., S.N., and V.L. performed the experiments. D.E.K., J.K., D.J.V., and H.H. assisted with experimental design and data interpretation. N.D.S. and P.A.S. performed computational and statistical analysis of TCR sequence data. D.S.L. and P.A.S. wrote the manuscript.