A Biological Circuit Involving Mef2c, Mef2d, and Hdac9 Controls the Immunosuppressive Functions of CD4+Foxp3+ T-Regulatory Cells

The Mads/Mef2 (Mef2a/b/c/d) family of transcription factors (TFs) regulates differentiation of muscle cells, neurons and hematopoietic cells. By functioning in physiological feedback loops, Mef2 TFs promote the transcription of their repressor, Hdac9, thereby providing temporal control of Mef2-driven differentiation. Disruption of this feedback is associated with the development of various pathologic states, including cancer. Beside their direct involvement in oncogenesis, Mef2 TFs indirectly control tumor progression by regulating antitumor immunity. We recently reported that in CD4+CD25+Foxp3+ T-regulatory (Treg) cells, Mef2d is required for the acquisition of an effector Treg (eTreg) phenotype and for the activation of an epigenetic program that suppresses the anti-tumor immune responses of conventional T and B cells. We now report that as with Mef2d, the deletion of Mef2c in Tregs switches off the expression of Il10 and Icos and leads to enhanced antitumor immunity in syngeneic models of lung cancer. Mechanistically, Mef2c does not directly bind the regulatory elements of Icos and Il10, but its loss-of-function in Tregs induces the expression of the transcriptional repressor, Hdac9. As a consequence, Mef2d, the more abundant member of the Mef2 family, is converted by Hdac9 into a transcriptional repressor on these loci. This leads to the impairment of Treg suppressive properties in vivo and to enhanced anti-cancer immunity. These data further highlight the central role played by the Mef2/Hdac9 axis in the regulation of CD4+Foxp3+ Treg function and adds a new level of complexity to the analysis and study of Treg biology.


INTRODUCTION
The role played by Treg cells in restraining anti-tumor immunity (1) and in limiting transplant rejection (2) is well established and has made the study of Treg differentiation, stability, subpopulations and homeostasis a major focus of immunology (3). Foxp3 is the master transcription factor (TF) involved in sustaining Treg suppressive identity and mutations leading to Foxp3 loss-of-function are associated clinically with the severe autoimmunity that presents as IPEX syndrome (Immune dysregulation, Polyendocrinopathy, Enteropathy, X-linked) (4)(5)(6). However, Foxp3 has been demonstrated to be necessary, but not sufficient, for triggering and maintaining the Treg phenotype (7). The exceptional interest in Treg biology has made Foxp3 one of the most studied TFs so far (8). Foxp3 transcriptional activities are finely modulated through four main mechanisms: i) the transcriptional and post-translational regulation of its expression (9); ii) the heterogeneity of Foxp3 partners involved in sustaining (9), but also positively or negatively modulating its transcriptional functions (10); iii) the large number of epigenetic regulators that act as pioneering factors (11,12) or plastically modulate Foxp3 activities (13,14); and iv) the protein-protein complexes and transcription factors that antagonistically or agonistically integrate Foxp3 responses (15)(16)(17).
We recently identified Mef2d as a TF that supports and integrates the transcriptional responses of Foxp3, allowing Tregs to acquire the phenotype of effector Tregs (eTregs) (18). Similarly to Foxp3, the four Mef2 paralogues (Mef2a, b, c, d) assemble into multi-protein complexes which modulate the resulting transcriptional responses to control various and complex differentiative and adaptive programs (19,20). The dynamic nature of Mef2 protein complexes has been little investigated in T effector cells and is completely unexplored in Treg cell. Here, by exploring the roles of Mef2c in CD4+ Foxp3+ Treg cells, we discovered the existence of a feed-back circuit involving Mef2c, Mef2d and Hdac9. Interference in this circuit decreases the immunosuppressive properties of Foxp3+ Tregs in vivo. Our results stress the need to deepen the study of this signaling pathway so as to more fully understand the transcriptional dynamics of Treg cells.

Flow Cytometry
Single-cell suspensions from secondary lymphoid tissues or tumors were prepared as previously described (15)

RNA-Seq and Real-Time qPCR
RNeasy kits (QIAGEN) were used to isolate mRNA. mRNA with a RIN>7 were used to prepare libraries and perform RNA-sequencing by Novogene (Sacramento, CA) on the Illumina Platform PE150. The edgeR package was used to identify the differentially expressed genes (p-value <0.05 and fold change >1.3). GSEA (26) was performed to interrogate datasets with defined genesets, as described (27). The expression levels of individual genes were verified by qPCR. For this purpose, RNA was reverse transcribed to cDNA (Applied Biosystems) and Taqman primers and probe sets were used to perform RT-qPCR. Data were normalized to endogenous 18s and relative expression was determined by the formula 2 -DCT .

Statistical analysis
Data were analyzed using GraphPad Prism 8.0 and Excel. Data are presented as mean ± standard error. Statistical comparisons between two groups were done with a 2-tailed Student's t test.
Comparisons between multiple samples were performed by using a 1-way ANOVA test with corresponding Tukey's multiple comparison test. Graft survival was evaluated with Kaplan-Meier followed by log-rank test. We marked with *p<0.05, **p<0.01, ***p<0.005.

Study Approval
Animal studies were approved by the Institutional Animal Care and Use Committee of the Children's Hospital of Philadelphia (protocols 17-001047 and 19-000561).

Primers
The following primers were used for ChIP-qPCR analysis: Itk FW: GTGCGACTGAAGGAGAGGAG, Itk RV: CATCAGAG G A G G G A G C T C A G , H d a c 9 F W : C T C C A G A G G G TGTCCTCCTA, Hdac9 RV: GGCTTTGGTGGGGTATTTTT, Icos 1 FW: CCTCAGTCAGAAGGGTCGTC, Icos 1 RV: CAGA AATTCCTGGTCATGTTTT, Icos 2 FW: AGTCTGCCATAG GGTTGGTG, Icos 2 RV: TCAGTCATTTTCTCCCCCTTT, Il10

Mef2c Deletion in Tregs Only Partially Represses Their Suppressive Properties In Vitro
As noted in the literature, the various paralogues of MEF2 can play redundant (29,30) or differentiative and adaptive roles (31,32). Despite its low expression in murine Tregs ( Figure 1A), we deleted Mef2c expression in Foxp3+ Treg cells by crossing Mef2c fl/fl mice and Foxp3 YFP/Cre mice so as to investigate the existence of alternative, supplementary or redundant transcriptional programs with respect to those controlled by Mef2d, its more abundant paralogue. Mef2c fl/fl Foxp3 YFP/Cre (hereafter Mef2c-/-) mice were born at expected Mendelian ratios and were characterized by normal numbers of Tregs in secondary lymphoid tissues, though CD4+Foxp3+ Tregs were increased in the thymus of Mef2c-/-mice ( Figures 1B, C). Whether this was due to increased generation, stability or re-circulation to the thymus is presently unknown, but was previously also observed in Mef2d fl/ fl Foxp3 YFP/Cre mice (18). The in vitro suppressive functions of Mef2c-/-Tregs were increased, as shown by cell-sorting ( Figure 1D) and testing of the ability of purified Tregs to inhibit the proliferation of conventional T cells. Area-undercurve data are shown in Figure 1E and a triplicate assay, representative of 2 such assays, is shown in Figure 1F.  (Figure 2A). Gene set enrichment analysis (GSEA) showed that in Tregs, Mef2c controls general transcriptional responses previously associated with Mef2 TFs in different cell types, like E2F and peroxisome proliferator-activated receptor-g agonists/retinoic acid signaling ( Figure 2B). Alongside the perturbation of these signaling pathways, we found that Mef2c controls the expression of more specific signatures related to the activation of Tregs and the execution of their immunosuppressive properties ( Figure 2B). This analysis suggests that Mef2c is required to sustain the activation of Tregs and the transcriptional program activated by Blimp1 ( Figure 2B). We validated the RNA-seq data by means of qRT-PCR on a selected pool of transcripts found to be dysregulated after Mef2c depletion; in each case there was agreement with the high-throughput data ( Figure 2C).

Mef2c Depletion Switches on Hdac9 Transcription and Triggers a Feedback Response That Limits the Transcriptional Activity of Mef2d
GSEA analysis showed that many aspects of the transcriptional perturbation obtained after the depletion of Mef2c in Tregs are superimposable to that obtained by depleting Mef2d ( Figure 3A). We hypothesized that this could be due to the reorganization of Mef2d transcriptional complexes achieved as a consequence of Mef2c KO. Though suggested many times in the literature (23,27,31,(34)(35)(36)(37), the stoichiometry as well as the dynamic composition of MEF2 heterodimeric complexes in living cells have never been investigated in detail. We approached this issue by looking at the perturbations of mRNA abundance of the privileged partners of MEF2 TFs observed after Mef2c or Mef2d depletion ( Figure 3B). Among the partners analyzed, only Hdac9 displayed differences between the two KOs, being induced after Mef2c deletion ( Figure 3B). We confirmed this upregulation in Mef2c-/-Tregs both at the RNA ( Figure 3C) and protein ( Figure 3D) levels, and that upregulation was likely to be due to increased transcription as a consequence of the establishment of a proficient chromatin environment at the promoter level ( Figure 3E). Increased levels of Hdac9 can flip the balance of protein complexes assembled on Mef2d, thus altering its transcriptional effects (23,38,39). Indeed, most of the transcripts repressed in Mef2c-/-Tregs ( Figure 3F) were upregulated in Hdac9-/-Tregs (33,40). Levels of Mef2d engaged in protein complexes with Hdac9 in Mef2c-/-Tregs were almost double that of WT Tregs ( Figure 3G). In Mef2c-/-Tregs, this resulted in the establishment of a repressive chromatin at the level of the Mef2d-bound (18) regulative elements of Icos and IL-10 ( Figure 3H). The cooccurrence of Hdac9 binding suggests, but does not directly prove, its involvement in decreasing the acetylation of H3K27 on these loci ( Figure 3H). These data suggest that the release of Hdac9 transcription achieved after Mef2c KO could be involved in the establishment of repressive complexes on Mef2d, thus leading to the perturbation of eTreg properties, similarly to what we reported for Mef2d-/-Tregs (18).

Mef2c Deletion Dampens Treg Function In Vivo
We used two animal models to assess the effects of Mef2c deletion on Treg suppressive functions in vivo.
First, we performed cardiac allografting using BALB/c donors and WT or Mef2c-/-C57BL/6 recipients in the presence of costimulation blockade with CD154 mAb/DST ( Figure 4A) (24). While co-stimulation blockade induced long-term allograft survival (>100 d) in WT recipients, acute rejection was observed in Mef2c-/-recipients (p<0.01) ( Figure 4B). This was unexpected, given that Mef2c deletion had led to increased Treg suppressive function in vitro ( Figure 1). Nevertheless, histologic comparison of grafts harvested from the 2 groups at 21 days post-transplant showed only focal interstitial infiltrates in WT recipients but dense mononuclear infiltrates with multifocal myocyte necrosis and vascular injury in conditionally deleted Mef2c-/-recipients ( Figure 4C); such Mef2c-associated pathology is very similar to that seen previously following Mef2d deletion in Foxp3+ Treg cells (18). Using real-time qPCR, we compared intragraft gene expression at the 21 days post-transplant time-point in WT vs. Mef2c-/-recipients. The levels of CD4, CD8, CD19 and Foxp3 were comparable between groups, as was IL-2 expression, but Mef2c-/-recipients had increased intragraft expression of IFN-g, granzyme-B and Ctla4 ( Figure 4D). Hence our in vivo allograft studies are consistent with conditional deletion of Mef2c leading to impaired Treg function.
Second, we studied the effects of Mef2c deletion on antitumor immunity, by using syngeneic models of lung cancer (TC1 tumor cells injected subcutaneously). After the successful engraftment and an initial proliferative phase, the growth of  Figures 5B, C), while Mef2c-/-Tregs showed impaired production of IL-10 ( Figures 5B, C). The fast kinetics of tumor rejection did not allow us to quantify the tumor infiltration of CD4+ and CD8+ cells in Mef2c-/-mice, but we observed higher activation (CD44 high / CD62L low ) and greater IFN-g production by CD4 and CD8 ( Figures 5D, E) T cells in the tumor-draining lymph nodes and spleens of Mef2c-/-mice versus WT tumor-bearing mice. As with the allograft data (Figure 4), these tumor studies indicated impaired Treg function in vivo after conditional Mef2c deletion. Together, these data are indicative of a marked impairment of Mef2c-/-Tregs in vivo with characteristics and a magnitude of responses very similar to those described for the conditional deletion of Mef2d (18).

DISCUSSION
While our in vitro analysis suggested Mef2c-/-Tregs have enhanced suppressive function, our molecular, biochemical and in vivo data all clearly point to disruption of normal Treg function upon deletion of Mef2c. Such a disconnect has been reported by us (15) and others (41)(42)(43) and is thought to reflect the limitations of the standard in vitro Treg suppression assay. Indeed, the originator of the in vitro Treg suppression assay has reviewed this topic (44), concluding that there are dozens of examples of genes whose deletion in Tregs has differing effects in vitro and in vivo, and that there is major need for improvements to the in vitro assay so as to better predict likely in vivo effects. We conclude that Mef2c is another example of this disconnect and going forward have chosen to focus on the molecular, biochemical and in vivo effects of Mef2c deletion on Treg biology.
The complex and dynamic nature of the protein complexes that dictate Treg identity obey biological, biochemical and epigenetic features that similarly control many other differentiative and adaptive processes (11). Similarly to Foxp3, Mef2 TFs assemble and disassemble dynamically with coactivators and co-repressors to pursue the plastic control of the transcriptional program (19). In Treg cells, Mef2d is required to sustain and integrate the Foxp3 transcriptional repertoire (18); Mef2d deletion in Foxp3+ cells impairs the acquisition of an eTreg phenotype and leads to an increase in anti-cancer immunity and transplant rejection (18). Although Mef2c is expressed at very low levels in Tregs, Mef2c-/-mice rapidly reject heart allografts and restrain cancer growth in syngeneic models, similarly to Mef2d-/-mice. The de-repression of Hdac9 achieved after Mef2c KO plays a central key role in blocking Mef2d responses and Treg immune suppressive functions. Hdac9 is transcribed by Mef2d as part of a physiological feedback aimed at switching-off Mef2d transcriptional program (23,45). This feedback mechanism is active in Tregs (18) and Hdac9-/-Tregs have increased suppressive functions (33,46). Interestingly, the strength of this feedback mechanism is altered in some pathological conditions, such as cardiac hypertrophy (47) and cancer (23,48). However, the genetic and epigenetic factors that regulate this feedback mechanism are not yet known.  Figure 5. In the case of IL-10 production, isolated cells were stimulated for 4 h with PMA/ionomycin before the staining. (n = 4, t-test between the two groups, **p < 0.01, ***p < 0.005). (D, E) Analysis of the activation status and IFN-g production in conventional CD4+ T cells and CD8+ T cells in single-cell suspensions obtained from the draining lymph node and spleen harvested from four representative mice injected as in (A). (n = 4, t-test between the two groups, **p < 0.01, ***p < 0.005).
Here we have added a new level of complexity to the wellknown molecular circuit involving Mef2 TFs and their strongest repressors, the class IIa HDACs. We report the existence of an alternative mechanism through which two different paralogues of Mef2 family, Mef2c and Mef2d, respectively repress and promote Hdac9 transcription. Although we have not yet clarified the detailed mechanism by which Mef2c maintains the repression of Hdac9, a suggestive hypothesis that is corroborated by the first ChIP data is that two antagonist complexes are assembled on Hdac9 promoter: an activator complex bound to Mef2d and a repressive one complexed to Mef2c. The presence of such delicate balances underlines the central role played by Mef2 TFs in supporting Treg identity and encourages further studies to clarify the molecular details of these interactions.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi. nlm.nih.gov/, GSE139480.

ETHICS STATEMENT
Animal studies were approved by the Institutional Animal Care and Use Committee of the Children's Hospital of Philadelphia (protocols 17-001047 and 19-000561).

AUTHOR CONTRIBUTIONS
EG designed and performed experiments, analyzed data and drafted the manuscript. LW performed cardiac allografts, performed experiments, and analyzed data. YX performed experiments and analyzed data. LC performed experiments and analyzed data. TA performed experiments and analyzed data. RH provided technical assistance. AB performed experiments. MT provided technical support. CB analyzed data. UB provided assistance with RNA-seq studies. WH analyzed data and oversaw the experimental design and writing of the manuscript. All authors contributed to the article and approved the submitted version.

FUNDING
This work was supported by NIH grants R01 AI12324 and R01 CA177852 (to WH). EG received a "Fondazione Umberto Veronesi Fellowship".

SUPPLEMENTARY MATERIAL
The  Figure 1F (total T cell responders). These experiments were repeated two times by using three mice per group.
Supplementary Figure 2 | Mef2c deletion promotes anti-cancer immunity. Graphs representing the detailed tumor growth data for each of the 10 WT and 10 Mef2c-/-mice during a 19-day-long observation period after the subcutaneous injection of 1.2x10 6 TC1 cells. The experiment was repeated twice with comparable results.