|Home | About | Journals | Submit | Contact Us | Français|
Foxp3 plays an indispensable role in establishing stable transcriptional and functional programs of regulatory T (Treg) cells. Loss of Foxp3 expression in mature Treg cells results in a failure of suppressor function, yet the molecular mechanisms ensuring steady heritable Foxp3 expression in the Treg cell lineage remain unknown. Using Treg cell-specific gene targeting we found that Runx-CBFβ complexes were required for maintenance of Foxp3 mRNA and protein expression in Treg cells. Consequently, mice lacking CBFβb exclusively in the Treg cell lineage exhibited a moderate lymphproliferative syndrome. Thus, Runx-CBFβ complexes maintain stable expression of high amounts of Foxp3 and serve as an essential determinant of Treg cell lineage stability.
The regulatory T (Treg) cell lineage is indispensable for suppressing autoimmunity and preventing over-exuberant responses to pathogens1. The X-chromosome encoded forkhead winged-helix transcription factor Foxp3 (http://www.signaling-gateway.org/molecule/query?afcsid=A002750) is a lineage specification factor required for Treg cell differentiation and function2–4. Furthermore, continuous expression of Foxp3 in mature Treg cells is required for the maintenance of suppressor function and of the Foxp3-dependent transcriptional program5. Genome-wide analysis of Foxp3 binding sites coupled to gene expression profiling suggested that Foxp3 can act both as a transcriptional activator and repressor6–8. For example, Foxp3 binding to the promoters and 5′ regulatory regions of Ctla4 and Il2ra genes results in their activation, whereas Foxp3 binding to Il2 and Il7ra promoters facilitates repression of these genes. The dual activity of Foxp3 defines characteristic features of Treg cells, which express high amounts of CTLA-4 and CD25 but low amounts of IL-7R, and lack of IL-2 production6,7.
Although the identity of Foxp3 target genes and understanding of their roles in Treg cell biology have began to emerge, the mechanisms regulating expression of the Foxp3 gene itself remain poorly understood. A number of sequence-specific transcription factors including STAT5, nuclear factor of activated T cells (NFAT), CREB, Smad3, and NF-KB were implicated in transcriptional regulation of Foxp3 expression, yet questions remain about their redundancy and direct or indirect effect on Foxp3 transcriptional activation versus survival of Treg cells or their precursors9–12. For example, STAT5 was proposed to be essential for Foxp3 induction yet forced expression of Bcl2 in Treg precursors rescues Foxp3 expression in the absence of STAT59,10 (Stephen Malin and Meinrad Busslinger, personal communication).
Furthermore, the molecular mechanisms by which Foxp3 controls gene expression in Treg cells appear to be complex. Foxp3 interactions with other sequence-specific transcription factors were proposed to be essential for establishment of the Foxp3-dependent transcriptional program. In this regard, Foxp3 can interact with NFAT; these two proteins cooperatively activate Ctla4 and Il2ra and antagonize NFAT-AP1-dependent Il2 expression13. More recently, the interaction between Foxp3 and the RUNT domain containing transcription factor Runx1 (http://www.signaling-gateway.org/molecule/query?afcsid=A000523) was suggested to be critical for Treg cell function. Mutations in Foxp3 that abrogate its interaction with Runx1 result in a loss of Foxp3 function. Unlike wild-type Foxp3 protein, the mutant form of Foxp3 is unable to repress Il2 production or activate Il2ra or Ctla4 when ectopically expressed in CD4+CD25−T cells14. Using a proteomics approach, we found that Runx1 and its co-factor CBFβ are prominently represented among binding partners of Foxp3 (Rudra et al., in preparation).
Transcriptional activity of Runx proteins is dependent upon heterodimerization with CBFβ (http://www.signaling-gateway.org/molecule/query?afcsid=A000524)15. CBFβ stabilizes Runx protein-DNA interactions and prevents ubiquitin-mediated Runx degradation, thus serving as an indispensable component of Runx transcriptional complexes16,17. The three known Runx factors (Runx1–3) are broadly expressed in hematopoietic cells including T cells. where they play important roles at various stages of differentiation18. Expression of T cell receptor (TCR) genes requires Runx1, which also acts as a transcriptional repressor of the Cd4 locus in CD4−CD8− thymocytes19–23. Runx3 is expressed in high amounts and represses Cd4 expression in CD8+ T cells23. Runx3 has also been suggested to interact with the T helper type 1 (TH1) lineage specification factor T-bet and to facilitate Ifng activation and Il4 repression in TH1 cells24.
The documented importance of Runx in cell fate decisions during T cell differentiation and Runx interactions with Foxp3 prompted us to investigate the role of Runx proteins in Treg cell biology in vivo using a genetic approach. We took advantage of the fact that transcriptional activity of all the three Runx proteins is dependent upon CBFβ and the ablation of the Cbfb gene results, therefore, in a complete loss of Runx function. To address the role for Runx protein function specifically in the Treg cell lineage we induced deletion of a floxed Cbfb allele in Treg cells by crossing Cbfbfl and Foxp3YFP-Cre mice25,26. Unexpectedly, we found that Runx factors were largely dispensable for Foxp3 target gene expression and Treg cell suppressor function. However, Runx protein function was critical for the maintenance of expression the Foxp3 gene itself. Thus, Runx proteins play an essential role in maintaining high amounts of Foxp3 expression ensuring Treg cell lineage identity.
To determine the functional consequence of ablating Runx activity in Treg cells, we crossed mice harboring a floxed Cbfb allele (Cbfbfl) with mice expressing a YFP-Cre recombinase fusion protein under the control of the Foxp3 regulatory elements (Foxp3YFP-Cre)25,26. Efficient Cre-mediated deletion of the Cbfbfl allele was confirmed by PCR analysis of genomic DNA isolated from sorted YFP-Foxp3+ Treg cells from Cbfbfl/+Foxp3YFP-Cre and Cbfbfl/fl Foxp3YFP-Cre mice and subsequent reduction in CBFβ mRNA was shown by real-time PCR analysis of RNA isolated from these cells (Supplementary Fig. 1a–d). Furthermore, immunoblot analysis confirmed the absence of the CBFβ protein only in sorted Treg cells derived from Cbfbfl/fl Foxp3YFP-Cre mice (Supplementary Fig. 1e). DNA, RNA, and protein expression analyses indicated that the deletion of the Cbfbfl allele was limited to CD4+YFP-Foxp3+ Treg cells and was not observed in the CD4+YFP− “non-Treg” cells.
Cbfbfl/flFoxp3YFP-Cre mice were born at the expected Mendelian ratio and showed no clinical signs of autoimmunity until 12–14 weeks of age. Nevertheless, examination of the secondary lymphoid organs in 5–8 week old mutant mice revealed lymphadenopathy and splenomegaly (Fig. 1a, b). Importantly, lymphoproliferative syndrome in these mice was relatively minor in comparison to the one observed in Foxp3-deficient mice, or Foxp3DTR mice subjected to Treg ablation upon diphtheria toxin treatment2,27. Examination of tissue pathology revealed moderate to marked lymphohistiocytic, and occasionally plasmacytic, inflammation in the affected tissues of the diseased mice (Fig. 2). Notably, the lung had multifocal perivascular and peribronchiolar lymphohistiocytic pneumonitis and arteritis with marked proliferative arteriopathy and bronchiolar goblet cell hyperplasia (Fig. 2).
Next, we examined relative sizes of thymocyte subsets as well as lymphoid and myeloid cell subsets, including CD4 and CD8 T cells, B cells, natural killer (NK) cells, dendritic cells (DCs) and macrophages in the spleen and lymph nodes (LNs), and failed to find marked differences between mutant and wild-type mice (Fig. 1c and Supplementary Fig. 2). To assess the extent of CD4+Foxp3− T cell activation in these mice, we analyzed the expression of a panel of known activation markers including CD25, GITR, ICOS, CTLA-4, and CD62L and observed a modest increase in the proportion of activated T cells in Cbfbfl/flFoxp3YFP-Cre mice (Fig. 1d). Likewise, the numbers of proliferating cells assessed based on Ki-67 expression were only slightly increased (Fig. 1d), whereas changes in TH1 and TH2 cytokine production were insignificant in mice harboring CBFβ-deficient Treg cells (data not shown). Thus, the mild dysregulation of T cell responses in the presence of CBFβ-deficient Treg cells contrasted sharply with the highly aggressive and severe immune tissue lesions and T cell activation and population expansion in mice lacking Treg cells27. Thus, these observations in Cbfbfl/flFoxp3YFP-Cre mice implied a partial impairment of Treg cell suppressor function in the absence of CBFβ.
To directly assess the functional potential of Treg cells in Cbfbfl/flFoxp3YFP-Cre mice we first examined Foxp3 expression in these cells, as high amounts of Foxp3 are required for Treg cell suppressor function7,28,29. Unexpectedly, intracellular Foxp3 staining combined with flow cytometric analysis revealed a significant reduction in the amount of Foxp3 protein in peripheral Treg cells lacking CBFβ (Fig. 3a). A similar reduction in the amount of YFP-Cre fusion protein was observed (Supplementary Fig. 3), suggesting that down-regulation of Foxp3 expression occurred at a transcriptional level. Indeed, real-time PCR analysis of RNA isolated from sorted CD4+YFP+ Treg cells showed a marked reduction in Foxp3 mRNA in the absence of CBFβ (Fig. 3b). In contrast to peripheral Treg cells, amounts of Foxp3 protein were comparable in Foxp3+ thymocytes in Cbfbfl/flFoxp3YFP-Cre and Cbfbfl/+Foxp3YFP-Cre mice likely due to the carryover of CBFβ protein from Foxp3− precursor thymocytes. In agreement with this idea, in Cbfbfl/flCD4-Cre+ mice Foxp3 expression in CD4 SP thymocytes was diminished although to a lesser extent than in the periphery (Supplementary Fig. 4a). A similar cell-intrinsic effect of CBFβ deficiency, characterized by marked reduction in Foxp3 expression primarily in the periphery, was observed upon co-injection of CD45.2+ Cbfbfl/flCD4-Cre+ and wild-type C57BL/6 CD45.1+ bone marrow transfer into RAG-deficient recipients (Supplementary Fig. 4b). Although it was formally possible that selective outgrowth or survival of Foxp3low cells in the absence of CBFβ accounted for the observed decrease in Foxp3 expression in the CBFβ-deficient Treg cell population, we found that Foxp3high cells lacking CBFβ divide more than their Foxp3low counterparts (Supplementary Fig. 5), in agreement with an earlier finding that high amounts of Foxp3 expression confer proliferative potential to Treg cells7. Furthermore, CBFβ-deficient Foxp3high Treg cells did not exhibit increased apoptosis as compared to Foxp3low cells (data not shown). Together, these results indicate that Runx-CBFβ complexes control Foxp3 expression in Treg cells.
To conclusively demonstrate that the reduced amounts of Foxp3 in Cbfbfl/flFoxp3YFP-Cre mice were due to the inactivation of the Runx-CBFβ complexes and not a hitherto unknown Runx independent function of CBFβ, we generated Runx1fl/flFoxp3YFP-Cre mice in which Runx1, which was more prominently expressed than Runx3 in Treg cells, was ablated upon Cre-mediated deletion of a conditional Runx1fl allele (Supplementary Fig. 6a). Similarly to the Cbfbfl/flFoxp3YFP-Cre mice, the Runx1fl/flFoxp3YFP-Cre mice exhibited a significant reduction in amounts of Foxp3 in peripheral Treg cells (Supplementary Fig. 6b). Additionally, Runx1-deficient Foxp3+ thymocytes also showed a modest reduction in Foxp3 amounts on a per cell basis, albeit to a lesser extent than the peripheral Treg cells (Supplementary Fig. 6b). The difference in Foxp3 amounts present in Runx1- and CBFβ-deficient Foxp3+ thymocytes in Cbfbfl/flFoxp3YFP-Cre and Runx1fl/flFoxp3YFP-Cre mice may be attributed to a shorter half-life and lesser abundance of Runx1 in comparison to CBFβ.
CBFβ-deficient Treg cell populations in Cbfbfl/flFoxp3YFP-Cre mice, but not in healthy littermate Cbfbfl/+Foxp3YFP-Cre mice, contained increased proportions of GITRhi, ICOShi, as well as Ki-67+ cells, indicative of heightened activation and proliferation likely driven by proliferation and activation of non-Treg cells (Fig. 3c; data not shown). CBFβ-deficient Treg cells suppressed proliferative responses of CD4+ Foxp3− T cells in vitro in a manner comparable to their CBFβ-sufficient counterparts (Fig. 3d). Thus, in agreement with the slowly progressing immune lesions and moderate lymphoproliferative syndrome in Cbfbfl/flFoxp3YFP-Cre mice, the CBFβ-deficient Treg cell population as a whole does maintain suppressor capacity, likely due to the presence of Foxp3hi Treg cells recently emigrated from the thymus.
To assess competitive fitness of CBFβ-deficient Treg cells in the absence of inflammation we examined relative sizes of CBFβ-sufficient and CBFβ-deficient Treg subsets in healthy heterozygous Foxp3YFP-Cre/+Cbfbfl/fl and Foxp3YFP-Cre/+Cbfbfl/+ females. The proportion of peripheral YFP-Cre+ Treg cells in Foxp3YFP-Cre/+Cbfbfl/fl mice was markedly lower than in Foxp3YFP-Cre/+Cbfbfl/+ mice (Fig. 3e, f). In these disease-free mice, CBFβ-deficient YFP-Cre+ Treg cells also expressed markedly reduced amounts of Foxp3 on a per cell basis in comparison to CBFβ-sufficient counterparts, in agreement with findings in diseased Foxp3YFP-CreCbfbfl/fl mice. Thus, CBFβ-deficiency in Treg cells results in diminished amounts of Foxp3, likely leading to impaired suppressive capacity and diminished competitive fitness in the presence of CBFβ-sufficient Treg cells.
Although the analysis of Treg cells in Runx1fl/flFoxp3YFP-Cre and Cbfbfl/flFoxp3YFP-Cre mice made it evident that the Runx-CBFβ deficiency impairs Foxp3 expression, the extent to which Foxp3 expression was affected in these mice was difficult to assess due to aforementioned continuous thymic output. To directly assess the potential loss of Foxp3 expression in a cohort of peripheral Treg cells lacking CBFβ we adoptively transferred sorted CD45.2+CD4+YFPhi cells from Cbfbfl/flFoxp3YFP-Cre mice or Cbfbfl/+Foxp3YFP-Cre mice mixed with CD4+ effector T cells from CD45.1+Foxp3− mice into Rag2−/− recipients. Flow cytometric analysis of transferred cells revealed that in addition to diminished Foxp3 amounts, the CBFβ-deficient Treg population exhibited a more pronounced loss of Foxp3with considerably faster kinetics than the CBFβ-sufficient Treg cell population (Fig. 4a–d). Thus, Runx-CBFβ regulates both the amount and stability of Foxp3 expression in Treg cells.
Apart from a conserved promoter region, the Foxp3 locus contains two recently described proximal conserved non-coding sequence (CNS1, CNS2) elements downstream of the transcription start site11,12. Both CNS1, containing a Smad-NFAT response element, and CNS2, containing CREB-ATF and STAT5 binding sites, were proposed to act as enhancers important for Foxp3 induction in the thymus and in the periphery 11, 12. To test whether Runx-CBFβ complexes bind to the Foxp3 promoter and CNS elements in vivo, we performed CBFβ chromatin immunoprecipitation (ChIP) using nuclear lysates isolated from wild-type Treg cells. We found a marked enrichment of CBFβ bound to CNS2. In addition, we reproducibly observed an enrichment of CBFβ at the Foxp3 promoter despite a very high background in the control IgG ChIP (Fig. 5). The Tcrb enhancer, known to bind Runx1-CBFβ, was used as a positive control in these experiments20.
To test whether Runx-CBFβ binding to the Foxp3 promoter and CNS2 directly facilitates Foxp3 transcription, we performed luciferase reporter assays using primary Treg cells and the EL-4 T cell line. We assessed the role of putative Runx binding sites in the promoter and CNS2 in the direct enhancement of Foxp3 transcription mediated by these elements. In agreement with previous reports, the minimal Foxp3 promoter did not exhibit measurable activity in the reporter assay (Supplementary Fig. 7a,b)11. Although the non-chromatinized DNA constructs containing the minimal Foxp3 promoter and CNS1 (construct B) or CNS2 (construct C) showed an increased luciferase signal in the presence of PMA and ionomycin with and without transforming growth factor-β (TGF-β), mutations of the predicted Runx binding sites had no measurable effect on CNS-mediated enhancement of Foxp3 promoter activity in EL-4 cells or purified CBFβ-sufficient Treg cells (Supplementary Figure 7c–e). Similarly, the CNS2-containing reporter drove comparable luciferase expression in CBFβ-deficient and CBFβ-sufficient Treg cells (Supplementary Fig. 7e). These results suggested that Runx-CBFβ complexes might regulate Foxp3 expression through epigenetic changes at the Foxp3 locus; this type of regulation cannot be accounted for in a reporter assay. In support of this idea, we detected a marked decrease in permissive histone H3K4me3 modifications at the Foxp3 promoter and a concomitant widespread increase in inhibitory histone H3K9me3 modifications at the 5′ end of the Foxp3 locus (Fig. 6) in CBFβ-deficient Treg cells. In contrast to a pronounced increase in H3K9me3 inhibitory marks, the distribution of another prototypic non-permissive histone modification, H3K27me3, across the Foxp3 locus was unchanged in CBFβ-deficient Treg cells (Fig. 6). Thus, it seems likely that Runx-CBFβ complexes impart permissive epigenetic modifications and oppose certain inhibitory modifications on the Foxp3 locus.
Reduced and unstable Foxp3 expression may result in attenuated in vivo Treg cell suppressor function, which might contribute to the lymphoproliferative syndrome in Cbfbfl/flFoxp3YFP-Cre mice. However, it was also possible that Runx-CBFβ–Foxp3 complexes contribute to Treg functionality by cooperatively regulating the Treg cell gene expression signature. To directly distinguish between these two possibilities, we examined expression of the characteristic Treg cell surface molecules CTLA-4, CD25, and GITR upon “forced’ Foxp3 expression in CBFβ-sufficient or CBFβ-deficient non-Treg CD4+ cells (Fig. 7). Previous studies showed that retroviral transduction of CD25−CD4+ T cells with a Foxp3-expressing retrovirus conferred a characteristic Treg cell surface phenotype and suppressor function upon non-Treg cells2,3. It is noteworthy that the amount of Foxp3 expression in retrovirally transduced cells in these experiments is comparable to, or lower than that observed in Treg cells2,3. We employed the previously described retroviral bi-cistronic vector to express both Foxp3 and GFP, or GFP alone, in activated CD25−CD4+ T cells from Cbfbfl/flCD4-Cre+ or Cbfb+/+CD4-Cre+ control mice and examined the aforementioned surface markers and in vitro suppressor activity of FACS purified Foxp3+GFP+ and control GFP+ T cells. As the retroviral vector expressing Foxp3 cDNA was devoid of the endogenous Foxp3 regulatory elements that serve as targets of Runx-CBFβ transcriptional complexes, comparable Foxp3 expression was found in transduced CBFβ-deficient and CBFβ-sufficient T cells (Fig. 7a). Furthermore, GITR, CTLA-4, CD25 were present in similar amounts in CBFβ-deficient and CBFβ-sufficient Foxp3-expressing T cells (Fig. 7a) and these cells exhibited comparable suppressor activity in vitro (Fig. 7b). In contrast, the GFP vector transduced CBFβ-deficient and CBFβ-sufficient T cells demonstrated no suppressive capacity. Instead, they showed enhanced proliferation in comparison to freshly isolated responder T cells, presumably due to their prior activation during the retroviral transduction (Fig. 7c).
To assess in vivo suppressor capacity of Treg cells lacking CBFβ but retrovirally transduced with a Foxp3-encoding retrovirus, we isolated Foxp3+ Treg cells from Cbfbfl/flFoxp3YFP-cre or control Cbfbfl/+Foxp3YFP-cre mice and transduced them with another bi-cistronic retroviral vector containing the Foxp3 coding sequence followed by an IRES-driven tailless human CD2 reporter. As a control, cells were transduced with the retroviral vector containing the IRES-driven tailless human CD2 reporter but lacking the Foxp3 insert. FACS sorted transduced Treg cells were co-transferred with effector T cells isolated from Foxp3− mice into Tcrb−/− Tcrd−/− recipient mice. Transfer of effector Foxp3− T cells alone led to a characteristic lymphoproliferative systemic immune-mediated syndrome associated with the severe weight loss, lymphadenopathy, splenomegaly, and increase in production of TH1 and TH2 cytokines (Fig. 7d and Supplementary Fig. 8). Co-transfer of Foxp3- or control vector-transduced control CBFβ-sufficient Treg cells or Foxp3-transduced CBFβ-deficient Treg cells led to an essentially complete rescue of the lymphoproliferative syndrome in the recipient mice. In agreement with relatively mild late-onset autoimmune lesions in unmanipulated Cbfbfl/flflFox3pYFP-cre mice, CBFβ-deficient Treg cells transduced with the control vector exhibited only a mild impairment in suppressor function, reflected in partial rescue of recipient mice from the weight loss (Fig. 7d). It is noteworthy that both Foxp3 and control vector-transduced CBFβ-deficient Treg cells appeared somewhat less efficient at limiting IL-4 production by effector T cells compared to control CBFβ-sufficient Treg cells (Supplementary Fig. 8e). In agreement with the weight loss data, small inflammation foci in the liver, skin and lung found in recipient mice which were co-transferred with effector T cells and control vector-transduced CBFβ-deficient Treg cells were not observed in the presence of Foxp3 transduced CBFβ-deficient or -sufficient Treg cells or control vector transduced CBFβ-sufficient Treg cells (Supplementary Fig. 8f). Thus, Foxp3+ Treg cells in the absence of CBFβ exhibit only modestly diminished in vivo suppressor function and ectopic Foxp3 expression largely restores this function. The latter results are in agreement with unimpeded in vitro suppression capacity of Foxp3-transduced CBFβ-deficient CD4+ T cells (Fig. 7b). These results suggest that Foxp3 is able to confer Treg cell-specific gene expression and suppressive capacity in the absence of functional Runx-CBFβ complexes and that the moderate decline in suppressor function of Treg cells lacking CBFβ is largely due to progressively diminishing Foxp3 expression.
Our studies suggest that in Treg cells Runx-CBFβ activity is required for maintaining the expression of Foxp3. The lack of clinical signs of overt autoimmune phenotype in Cbfbfl/flFoxp3YFP-Cre mice up to 8–10 months of age was in sharp contrast to the very short 3–5 week lifespan of mice lacking Foxp3. Consistent with the late occurrence of gross clinical manifestations, the lymphadenopathy, splenomegaly, and CD4+Foxp3−effector T cell activation in Cbfbfl/flFoxp3YFP-Cre mice were relatively mild. It is of interest, however, that the only pronounced tissue pathology noticeable in mice harboring CBFβ-deficient Treg cells was a marked pulmonary arteriopathy similar to that in two recent studies, which proposed a role for Treg cells in the pathogenesis of this disease30,31. Notably, human lupus patients, who, like aged Cbfbfl/flFoxp3YFP-Cre mice, exhibit a diminished size of the Treg cell subset, often develop severe angioproliferative pulmonary hypertension with analogous histopathological lesions32. Except for the lung arteriopathy, immune mediated inflammation in other tissues was far less severe and its progression was much slower than that in mice with a complete loss of Treg function or lack of Treg cells2,27. Of note, ablation of CBFβ in CD4+ T cells in Cbfbfl/flCD4-Cre+-mice results in asthma-related symptoms due to enhanced TH2 response presumably due to impaired silencing of Il4 locus in TH1 cells25.
Consistent with these results, Runx-CBFβ activity was unlikely a prerequisite for Treg function since ectopic expression of Foxp3 in CBFβ-deficient CD4+CD25 T cells derived from Cbfbfl/flCD4-Cre+mice resulted in a prototypical pattern of Foxp3 target gene expression and conferred suppressive capacity upon these cells. Therefore, it appears that in Treg cells the main function of Runx-CBFβ is to regulate the amount and stability of Foxp3 expression.
Runx-CBFβ nuclear factors may influence Foxp3 expression in three different yet not mutually exclusive ways. They can bind to Foxp3 and facilitate its transcription or regulate the locus by modifying the chromatin structure and keeping it in an open configuration. In addition, the effect of Runx-CBFβ on Foxp3 expression in Treg cells can be indirect, if a known or unknown regulator of the Foxp3 gene is under the control of Runx proteins. Our observation that in Treg cells CBFβ occupies the Foxp3 promoter and CNS2 supports the first two possibilities. However, the presence or absence of Runx-CBFβ complexes in primary Treg cells had little, if any, effect on expression of a luciferase reporter construct containing Foxp3 promoter and CNS2 sequences, suggesting the possibility that Runx-CBFβ complexes maintain Foxp3 expression in Treg cells via epigenetic modifications of the Foxp3 locus. Indeed, ablation of CBFβ in Treg cells was associated with a pronounced decrease in permissive H3K4me3 modifications immediately downstream of and at the Foxp3 promoter, and a selective increase in non-permissive H3K9me3, but not in H3K27me3 modifications, at the Foxp3 locus. Thus, these results are consistent with the idea that Runx-CBFβ complexes promote the active state of the Foxp3 locus via an epigenetic mechanism. Further evidence in favor of this possibility comes from our recent finding that mice lacking CNS2 exhibit a progressive loss of Foxp3 expression in the progeny of dividing Foxp3+ cells similar to that in Cbfbfl/flFoxp3YFP-Cre mice (Y. Zheng et al., in preparation).
In contrast to peripheral Foxp3+ T cells, Cre mediated deletion of the Cbfbfl allele in Foxp3-expressing cells did not result in a decrease in the amount of Foxp3 expressed in the thymus. This finding could suggest that Runx-CBFβ mediated regulation of Foxp3 expression is dispensable in the thymus. However, another possible explanation is highlighted by the previously documented carryover of CBFβ protein 25, which lingers in thymocytes long after Cbfb allele deletion. In this regard, a recent study showed that even after CD4-Cre mediated deletion of a conditional Cbfbfl allele in CD4+CD8+ thymocytes, CBFβ protein was readily detectable by immunoblotting in CD4+CD8− thymocytes, but not in peripheral CD4+CD8− T cells25. Indeed, in our analysis of Cbfbfl/flCD4Cre mice we observed that the expression of Foxp3 in CD4 SP thymocytes was diminished in comparison to Cbfbfl/flFoxp3YFP-Cre mice, presumably due to earlier deletion of the Cbfbfl allele in the former mice. Further support for a role for CBFβ-Runx complexes in regulation of Foxp3 expression in the thymus came from comparison of Treg cell-specific ablation of Runx1 or CBFβ Although Treg cell-specific ablation of Runx1 or CBFβ results in a similar reduction in Foxp3 in the peripheral cells, Runx1-deficient Foxp3+ thymocytes exhibited markedly decreased amounts of Foxp3 likely due to faster turnover or lower amounts of Runx1. Thus, our studies suggest that Runx-CBFβ complexes play an important non-redundant role in the maintenance of Foxp3 expression both in the thymus and in the periphery.
We thank K. Forbush, T. Chu and L. Karpik for managing the mouse colony; I. Taniuchi (RIKEN Research Center for Allergy and Immunology), Y. Tone and M. Tone (University of Pennsylvania) for advice. This work was supported by NIH grants (A.Y.R.). D.R is supported by Arthritis Foundation postdoctoral fellowship. T.E is a Leukemia and Lymphoma Society Fellow. M M.W. Chong is a recipient of a Helen and Martin Kimmel Stem Cell Fellowship. A.Y.R. and D.R.L are Howard Hughes Medical Institute investigators.